github-mirror/rclone
2
0
Fork 0
mirror of https://github.com/rclone/rclone.git synced 2024-11-26 02:09:55 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
109b695621
rclone/backend/s3/s3.go

2999 lines
91 KiB
Go
Raw Normal View History

Fix golint warnings
2015-09-23 01:47:16 +08:00
// Package s3 provides an interface to Amazon S3 oject storage
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
package s3
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
import (
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
"bytes"
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
"context"
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
"crypto/md5"
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
"encoding/base64"
"encoding/hex"
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
"encoding/xml"
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
"fmt"
"io"
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
"net/http"
s3: fix encoding for control characters - Fixes #3345
2019-07-23 19:24:10 +08:00
"net/url"
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
"path"
"regexp"
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
"sort"
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
"strconv"
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
"strings"
s3: Reduce memory usage streaming files by reducing max stream upload size Before this change rclone would allow the user to stream (eg with rclone mount, rclone rcat or uploading google photos or docs) 5TB files. This meant that rclone allocated 4 * 525 MB buffers per transfer which is way too much memory by default. This change makes rclone use the configured chunk size for streamed uploads. This is 5MB by default which means that rclone can stream upload files up to 48GB by default staying below the 10,000 chunks limit. This can be increased with --s3-chunk-size if necessary. If rclone detects that a file is being streamed to s3 it will make a single NOTICE level log stating the limitation. This fixes the enormous memory usage. Fixes #3568 See: https://forum.rclone.org/t/how-much-memory-does-rclone-need/12743
2019-11-06 18:41:03 +08:00
"sync"
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
"time"
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
"github.com/aws/aws-sdk-go/aws"
"github.com/aws/aws-sdk-go/aws/awserr"
s3: Fix after upstream API changes in aws-sdk-go/aws - fixes #114
2015-08-28 15:47:41 +08:00
"github.com/aws/aws-sdk-go/aws/corehandlers"
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
"github.com/aws/aws-sdk-go/aws/credentials"
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
"github.com/aws/aws-sdk-go/aws/credentials/ec2rolecreds"
s3: use aws web identity role provider
2020-01-06 03:49:31 +08:00
"github.com/aws/aws-sdk-go/aws/credentials/stscreds"
s3: add support for ECS task IAM roles ECS container IAM metadata is in a different place than EC2 IAM metadata. Use defaults' RemoteCredProvider function to query the standard locations for the credentials. Give the ECS role precedence over the role available from the underlying EC2 instance.
2017-11-23 05:21:36 +08:00
"github.com/aws/aws-sdk-go/aws/defaults"
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
"github.com/aws/aws-sdk-go/aws/ec2metadata"
s3: Use regional s3 us-east-1 endpoint
2020-06-26 21:09:29 +08:00
"github.com/aws/aws-sdk-go/aws/endpoints"
s3: Fix after upstream API changes in aws-sdk-go/aws - fixes #114
2015-08-28 15:47:41 +08:00
"github.com/aws/aws-sdk-go/aws/request"
Update to AWS SDK 0.10.0 Tested with S3 and Dreamhost Here is a link to the release notes: http://aws.amazon.com/releasenotes/5476699172355228
2015-10-30 19:50:45 +08:00
"github.com/aws/aws-sdk-go/aws/session"
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
"github.com/aws/aws-sdk-go/service/s3"
Remove filesystem flags and put in config file with editor
2014-03-16 00:06:11 +08:00
"github.com/ncw/swift"
Convert to using github.com/pkg/errors everywhere
2016-06-12 22:06:02 +08:00
"github.com/pkg/errors"
build: fix up package paths after repo move git grep -l github.com/ncw/rclone | xargs -d'\n' perl -i~ -lpe 's|github.com/ncw/rclone|github.com/rclone/rclone|g' goimports -w `find . -name \*.go`
2019-07-29 01:47:38 +08:00
"github.com/rclone/rclone/fs"
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
"github.com/rclone/rclone/fs/config"
build: fix up package paths after repo move git grep -l github.com/ncw/rclone | xargs -d'\n' perl -i~ -lpe 's|github.com/ncw/rclone|github.com/rclone/rclone|g' goimports -w `find . -name \*.go`
2019-07-29 01:47:38 +08:00
"github.com/rclone/rclone/fs/config/configmap"
"github.com/rclone/rclone/fs/config/configstruct"
"github.com/rclone/rclone/fs/fserrors"
"github.com/rclone/rclone/fs/fshttp"
"github.com/rclone/rclone/fs/hash"
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
"github.com/rclone/rclone/fs/operations"
build: fix up package paths after repo move git grep -l github.com/ncw/rclone | xargs -d'\n' perl -i~ -lpe 's|github.com/ncw/rclone|github.com/rclone/rclone|g' goimports -w `find . -name \*.go`
2019-07-29 01:47:38 +08:00
"github.com/rclone/rclone/fs/walk"
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
"github.com/rclone/rclone/lib/atexit"
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
"github.com/rclone/rclone/lib/bucket"
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
"github.com/rclone/rclone/lib/encoder"
build: fix up package paths after repo move git grep -l github.com/ncw/rclone | xargs -d'\n' perl -i~ -lpe 's|github.com/ncw/rclone|github.com/rclone/rclone|g' goimports -w `find . -name \*.go`
2019-07-29 01:47:38 +08:00
"github.com/rclone/rclone/lib/pacer"
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
"github.com/rclone/rclone/lib/pool"
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
"github.com/rclone/rclone/lib/readers"
build: fix up package paths after repo move git grep -l github.com/ncw/rclone | xargs -d'\n' perl -i~ -lpe 's|github.com/ncw/rclone|github.com/rclone/rclone|g' goimports -w `find . -name \*.go`
2019-07-29 01:47:38 +08:00
"github.com/rclone/rclone/lib/rest"
s3: fix --header-upload - Fixes #4303 Before this change we were setting the headers on the PUT request for normal and multipart uploads. For normal uploads this caused the error 403 Forbidden: There were headers present in the request which were not signed After this fix we set the headers in the object upload request itself as the s3 SDK expects. This means that we only support a limited range of headers - Cache-Control - Content-Disposition - Content-Encoding - Content-Language - Content-Type - X-Amz-Tagging - X-Amz-Meta- Note for the last of those are for setting custom metadata in the form "X-Amz-Meta-Key: value". This now works for multipart uploads and single part uploads See also #59
2020-06-05 18:45:54 +08:00
"github.com/rclone/rclone/lib/structs"
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
"golang.org/x/sync/errgroup"
Remove filesystem flags and put in config file with editor
2014-03-16 00:06:11 +08:00
)
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
// Register with Fs
func init() {
Create separate interface for object information. Take out read-only information about a Fs in a separate struct to limit access. See discussion at #282.
2016-02-18 19:35:25 +08:00
fs.Register(&fs.RegInfo{
Add help to remote chooser in rclone config - fixes #43
2016-02-16 02:11:53 +08:00
Name: "s3",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
Description: "Amazon S3 Compliant Storage Provider (AWS, Alibaba, Ceph, Digital Ocean, Dreamhost, IBM COS, Minio, etc)",
Add help to remote chooser in rclone config - fixes #43
2016-02-16 02:11:53 +08:00
NewFs: NewFs,
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
CommandHelp: commandHelp,
Remove filesystem flags and put in config file with editor
2014-03-16 00:06:11 +08:00
Options: []fs.Option{{
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Name: fs.ConfigProvider,
Help: "Choose your S3 provider.",
Examples: []fs.OptionExample{{
Value: "AWS",
Help: "Amazon Web Services (AWS) S3",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
}, {
Value: "Alibaba",
Help: "Alibaba Cloud Object Storage System (OSS) formerly Aliyun",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "Ceph",
Help: "Ceph Object Storage",
}, {
Value: "DigitalOcean",
Help: "Digital Ocean Spaces",
}, {
Value: "Dreamhost",
Help: "Dreamhost DreamObjects",
}, {
Value: "IBMCOS",
Help: "IBM COS S3",
}, {
Value: "Minio",
Help: "Minio Object Storage",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
}, {
Value: "Netease",
Help: "Netease Object Storage (NOS)",
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
}, {
Value: "Scaleway",
Help: "Scaleway Object Storage",
s3: Add StackPath Object Storage Support
2020-01-31 07:21:24 +08:00
}, {
Value: "StackPath",
Help: "StackPath Object Storage",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "Wasabi",
Help: "Wasabi Object Storage",
}, {
Value: "Other",
Help: "Any other S3 compatible provider",
}},
}, {
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Name: "env_auth",
Help: "Get AWS credentials from runtime (environment variables or EC2/ECS meta data if no env vars).\nOnly applies if access_key_id and secret_access_key is blank.",
Default: false,
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Examples: []fs.OptionExample{{
Value: "false",
Help: "Enter AWS credentials in the next step",
}, {
Value: "true",
Help: "Get AWS credentials from the environment (env vars or IAM)",
}},
}, {
Name: "access_key_id",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "AWS Access Key ID.\nLeave blank for anonymous access or runtime credentials.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "secret_access_key",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "AWS Secret Access Key (password)\nLeave blank for anonymous access or runtime credentials.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "region",
Help: "Region to connect to.",
Provider: "AWS",
Examples: []fs.OptionExample{{
Value: "us-east-1",
Help: "The default endpoint - a good choice if you are unsure.\nUS Region, Northern Virginia or Pacific Northwest.\nLeave location constraint empty.",
}, {
Value: "us-east-2",
Help: "US East (Ohio) Region\nNeeds location constraint us-east-2.",
}, {
Value: "us-west-2",
Help: "US West (Oregon) Region\nNeeds location constraint us-west-2.",
}, {
Value: "us-west-1",
Help: "US West (Northern California) Region\nNeeds location constraint us-west-1.",
}, {
Value: "ca-central-1",
Help: "Canada (Central) Region\nNeeds location constraint ca-central-1.",
}, {
Value: "eu-west-1",
Help: "EU (Ireland) Region\nNeeds location constraint EU or eu-west-1.",
}, {
Value: "eu-west-2",
Help: "EU (London) Region\nNeeds location constraint eu-west-2.",
s3: add aws endpoint eu-north-1
2019-02-03 03:44:26 +08:00
}, {
Value: "eu-north-1",
Help: "EU (Stockholm) Region\nNeeds location constraint eu-north-1.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "eu-central-1",
Help: "EU (Frankfurt) Region\nNeeds location constraint eu-central-1.",
}, {
Value: "ap-southeast-1",
Help: "Asia Pacific (Singapore) Region\nNeeds location constraint ap-southeast-1.",
}, {
Value: "ap-southeast-2",
Help: "Asia Pacific (Sydney) Region\nNeeds location constraint ap-southeast-2.",
}, {
Value: "ap-northeast-1",
Help: "Asia Pacific (Tokyo) Region\nNeeds location constraint ap-northeast-1.",
}, {
Value: "ap-northeast-2",
Help: "Asia Pacific (Seoul)\nNeeds location constraint ap-northeast-2.",
}, {
Value: "ap-south-1",
Help: "Asia Pacific (Mumbai)\nNeeds location constraint ap-south-1.",
s3: Add new region Asia Patific (Hong Kong)
2020-01-02 19:10:48 +08:00
}, {
Value: "ap-east-1",
Help: "Asia Patific (Hong Kong) Region\nNeeds location constraint ap-east-1.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "sa-east-1",
Help: "South America (Sao Paulo) Region\nNeeds location constraint sa-east-1.",
}},
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
}, {
Name: "region",
Help: "Region to connect to.",
Provider: "Scaleway",
Examples: []fs.OptionExample{{
Value: "nl-ams",
Help: "Amsterdam, The Netherlands",
}, {
Value: "fr-par",
Help: "Paris, France",
}},
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "region",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "Region to connect to.\nLeave blank if you are using an S3 clone and you don't have a region.",
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
Provider: "!AWS,Alibaba,Scaleway",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Examples: []fs.OptionExample{{
Value: "",
Help: "Use this if unsure. Will use v4 signatures and an empty region.",
}, {
Value: "other-v2-signature",
Help: "Use this only if v4 signatures don't work, eg pre Jewel/v10 CEPH.",
}},
}, {
Name: "endpoint",
Help: "Endpoint for S3 API.\nLeave blank if using AWS to use the default endpoint for the region.",
Provider: "AWS",
}, {
Name: "endpoint",
Help: "Endpoint for IBM COS S3 API.\nSpecify if using an IBM COS On Premise.",
Provider: "IBMCOS",
Examples: []fs.OptionExample{{
Value: "s3-api.us-geo.objectstorage.softlayer.net",
Help: "US Cross Region Endpoint",
}, {
Value: "s3-api.dal.us-geo.objectstorage.softlayer.net",
Help: "US Cross Region Dallas Endpoint",
}, {
Value: "s3-api.wdc-us-geo.objectstorage.softlayer.net",
Help: "US Cross Region Washington DC Endpoint",
}, {
Value: "s3-api.sjc-us-geo.objectstorage.softlayer.net",
Help: "US Cross Region San Jose Endpoint",
}, {
Value: "s3-api.us-geo.objectstorage.service.networklayer.com",
Help: "US Cross Region Private Endpoint",
}, {
Value: "s3-api.dal-us-geo.objectstorage.service.networklayer.com",
Help: "US Cross Region Dallas Private Endpoint",
}, {
Value: "s3-api.wdc-us-geo.objectstorage.service.networklayer.com",
Help: "US Cross Region Washington DC Private Endpoint",
}, {
Value: "s3-api.sjc-us-geo.objectstorage.service.networklayer.com",
Help: "US Cross Region San Jose Private Endpoint",
}, {
Value: "s3.us-east.objectstorage.softlayer.net",
Help: "US Region East Endpoint",
}, {
Value: "s3.us-east.objectstorage.service.networklayer.com",
Help: "US Region East Private Endpoint",
}, {
Value: "s3.us-south.objectstorage.softlayer.net",
Help: "US Region South Endpoint",
}, {
Value: "s3.us-south.objectstorage.service.networklayer.com",
Help: "US Region South Private Endpoint",
}, {
Value: "s3.eu-geo.objectstorage.softlayer.net",
Help: "EU Cross Region Endpoint",
}, {
Value: "s3.fra-eu-geo.objectstorage.softlayer.net",
Help: "EU Cross Region Frankfurt Endpoint",
}, {
Value: "s3.mil-eu-geo.objectstorage.softlayer.net",
Help: "EU Cross Region Milan Endpoint",
}, {
Value: "s3.ams-eu-geo.objectstorage.softlayer.net",
Help: "EU Cross Region Amsterdam Endpoint",
}, {
Value: "s3.eu-geo.objectstorage.service.networklayer.com",
Help: "EU Cross Region Private Endpoint",
}, {
Value: "s3.fra-eu-geo.objectstorage.service.networklayer.com",
Help: "EU Cross Region Frankfurt Private Endpoint",
}, {
Value: "s3.mil-eu-geo.objectstorage.service.networklayer.com",
Help: "EU Cross Region Milan Private Endpoint",
}, {
Value: "s3.ams-eu-geo.objectstorage.service.networklayer.com",
Help: "EU Cross Region Amsterdam Private Endpoint",
}, {
Value: "s3.eu-gb.objectstorage.softlayer.net",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Endpoint",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "s3.eu-gb.objectstorage.service.networklayer.com",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Private Endpoint",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "s3.ap-geo.objectstorage.softlayer.net",
Help: "APAC Cross Regional Endpoint",
}, {
Value: "s3.tok-ap-geo.objectstorage.softlayer.net",
Help: "APAC Cross Regional Tokyo Endpoint",
}, {
Value: "s3.hkg-ap-geo.objectstorage.softlayer.net",
Help: "APAC Cross Regional HongKong Endpoint",
}, {
Value: "s3.seo-ap-geo.objectstorage.softlayer.net",
Help: "APAC Cross Regional Seoul Endpoint",
}, {
Value: "s3.ap-geo.objectstorage.service.networklayer.com",
Help: "APAC Cross Regional Private Endpoint",
}, {
Value: "s3.tok-ap-geo.objectstorage.service.networklayer.com",
Help: "APAC Cross Regional Tokyo Private Endpoint",
}, {
Value: "s3.hkg-ap-geo.objectstorage.service.networklayer.com",
Help: "APAC Cross Regional HongKong Private Endpoint",
}, {
Value: "s3.seo-ap-geo.objectstorage.service.networklayer.com",
Help: "APAC Cross Regional Seoul Private Endpoint",
}, {
Value: "s3.mel01.objectstorage.softlayer.net",
Help: "Melbourne Single Site Endpoint",
}, {
Value: "s3.mel01.objectstorage.service.networklayer.com",
Help: "Melbourne Single Site Private Endpoint",
}, {
Value: "s3.tor01.objectstorage.softlayer.net",
Help: "Toronto Single Site Endpoint",
}, {
Value: "s3.tor01.objectstorage.service.networklayer.com",
Help: "Toronto Single Site Private Endpoint",
}},
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
}, {
// oss endpoints: https://help.aliyun.com/document_detail/31837.html
Name: "endpoint",
Help: "Endpoint for OSS API.",
Provider: "Alibaba",
Examples: []fs.OptionExample{{
Value: "oss-cn-hangzhou.aliyuncs.com",
Help: "East China 1 (Hangzhou)",
}, {
Value: "oss-cn-shanghai.aliyuncs.com",
Help: "East China 2 (Shanghai)",
}, {
Value: "oss-cn-qingdao.aliyuncs.com",
Help: "North China 1 (Qingdao)",
}, {
Value: "oss-cn-beijing.aliyuncs.com",
Help: "North China 2 (Beijing)",
}, {
Value: "oss-cn-zhangjiakou.aliyuncs.com",
Help: "North China 3 (Zhangjiakou)",
}, {
Value: "oss-cn-huhehaote.aliyuncs.com",
Help: "North China 5 (Huhehaote)",
}, {
Value: "oss-cn-shenzhen.aliyuncs.com",
Help: "South China 1 (Shenzhen)",
}, {
Value: "oss-cn-hongkong.aliyuncs.com",
Help: "Hong Kong (Hong Kong)",
}, {
Value: "oss-us-west-1.aliyuncs.com",
Help: "US West 1 (Silicon Valley)",
}, {
Value: "oss-us-east-1.aliyuncs.com",
Help: "US East 1 (Virginia)",
}, {
Value: "oss-ap-southeast-1.aliyuncs.com",
Help: "Southeast Asia Southeast 1 (Singapore)",
}, {
Value: "oss-ap-southeast-2.aliyuncs.com",
Help: "Asia Pacific Southeast 2 (Sydney)",
}, {
Value: "oss-ap-southeast-3.aliyuncs.com",
Help: "Southeast Asia Southeast 3 (Kuala Lumpur)",
}, {
Value: "oss-ap-southeast-5.aliyuncs.com",
Help: "Asia Pacific Southeast 5 (Jakarta)",
}, {
Value: "oss-ap-northeast-1.aliyuncs.com",
Help: "Asia Pacific Northeast 1 (Japan)",
}, {
Value: "oss-ap-south-1.aliyuncs.com",
Help: "Asia Pacific South 1 (Mumbai)",
}, {
Value: "oss-eu-central-1.aliyuncs.com",
Help: "Central Europe 1 (Frankfurt)",
}, {
Value: "oss-eu-west-1.aliyuncs.com",
Help: "West Europe (London)",
}, {
Value: "oss-me-east-1.aliyuncs.com",
Help: "Middle East 1 (Dubai)",
}},
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
}, {
Name: "endpoint",
Help: "Endpoint for Scaleway Object Storage.",
Provider: "Scaleway",
Examples: []fs.OptionExample{{
Value: "s3.nl-ams.scw.cloud",
Help: "Amsterdam Endpoint",
}, {
Value: "s3.fr-par.scw.cloud",
Help: "Paris Endpoint",
}},
s3: Add StackPath Object Storage Support
2020-01-31 07:21:24 +08:00
}, {
Name: "endpoint",
Help: "Endpoint for StackPath Object Storage.",
Provider: "StackPath",
Examples: []fs.OptionExample{{
Value: "s3.us-east-2.stackpathstorage.com",
Help: "US East Endpoint",
}, {
Value: "s3.us-west-1.stackpathstorage.com",
Help: "US West Endpoint",
}, {
Value: "s3.eu-central-1.stackpathstorage.com",
Help: "EU Endpoint",
}},
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "endpoint",
Help: "Endpoint for S3 API.\nRequired when using an S3 clone.",
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
Provider: "!AWS,IBMCOS,Alibaba,Scaleway,StackPath",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Examples: []fs.OptionExample{{
s3: Update Dreamhost endpoint - fixes #2974
2019-02-14 05:10:43 +08:00
Value: "objects-us-east-1.dream.io",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Help: "Dream Objects endpoint",
Provider: "Dreamhost",
}, {
Value: "nyc3.digitaloceanspaces.com",
Help: "Digital Ocean Spaces New York 3",
Provider: "DigitalOcean",
}, {
Value: "ams3.digitaloceanspaces.com",
Help: "Digital Ocean Spaces Amsterdam 3",
Provider: "DigitalOcean",
}, {
Value: "sgp1.digitaloceanspaces.com",
Help: "Digital Ocean Spaces Singapore 1",
Provider: "DigitalOcean",
}, {
Value: "s3.wasabisys.com",
s3: add config info for Wasabi's US-West endpoint Wasabi has two location, US East and US West, with different endpoint URLs. When configuring S3 to use Wasabi, provide the endpoint information for both locations.
2018-11-18 00:24:00 +08:00
Help: "Wasabi US East endpoint",
Provider: "Wasabi",
}, {
Value: "s3.us-west-1.wasabisys.com",
Help: "Wasabi US West endpoint",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Provider: "Wasabi",
s3: add config info for Wasabi's EU Central endpoint Wasabi has a EU Central endpoint for a couple months now, so add it to the list. Signed-off-by: Robert Marko <robimarko@gmail.com>
2019-05-15 16:22:06 +08:00
}, {
Value: "s3.eu-central-1.wasabisys.com",
Help: "Wasabi EU Central endpoint",
Provider: "Wasabi",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}},
}, {
Name: "location_constraint",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "Location constraint - must be set to match the Region.\nUsed when creating buckets only.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Provider: "AWS",
Examples: []fs.OptionExample{{
Value: "",
Help: "Empty for US Region, Northern Virginia or Pacific Northwest.",
}, {
Value: "us-east-2",
Help: "US East (Ohio) Region.",
}, {
Value: "us-west-2",
Help: "US West (Oregon) Region.",
}, {
Value: "us-west-1",
Help: "US West (Northern California) Region.",
}, {
Value: "ca-central-1",
Help: "Canada (Central) Region.",
}, {
Value: "eu-west-1",
Help: "EU (Ireland) Region.",
}, {
Value: "eu-west-2",
Help: "EU (London) Region.",
s3: add aws endpoint eu-north-1
2019-02-03 03:44:26 +08:00
}, {
Value: "eu-north-1",
Help: "EU (Stockholm) Region.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "EU",
Help: "EU Region.",
}, {
Value: "ap-southeast-1",
Help: "Asia Pacific (Singapore) Region.",
}, {
Value: "ap-southeast-2",
Help: "Asia Pacific (Sydney) Region.",
}, {
Value: "ap-northeast-1",
Help: "Asia Pacific (Tokyo) Region.",
}, {
Value: "ap-northeast-2",
Help: "Asia Pacific (Seoul)",
}, {
Value: "ap-south-1",
Help: "Asia Pacific (Mumbai)",
s3: Add new region Asia Patific (Hong Kong)
2020-01-02 19:10:48 +08:00
}, {
Value: "ap-east-1",
Help: "Asia Pacific (Hong Kong)",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "sa-east-1",
Help: "South America (Sao Paulo) Region.",
}},
}, {
Name: "location_constraint",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "Location constraint - must match endpoint when using IBM Cloud Public.\nFor on-prem COS, do not make a selection from this list, hit enter",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Provider: "IBMCOS",
Examples: []fs.OptionExample{{
Value: "us-standard",
Help: "US Cross Region Standard",
}, {
Value: "us-vault",
Help: "US Cross Region Vault",
}, {
Value: "us-cold",
Help: "US Cross Region Cold",
}, {
Value: "us-flex",
Help: "US Cross Region Flex",
}, {
Value: "us-east-standard",
Help: "US East Region Standard",
}, {
Value: "us-east-vault",
Help: "US East Region Vault",
}, {
Value: "us-east-cold",
Help: "US East Region Cold",
}, {
Value: "us-east-flex",
Help: "US East Region Flex",
}, {
Value: "us-south-standard",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "US South Region Standard",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "us-south-vault",
Help: "US South Region Vault",
}, {
Value: "us-south-cold",
Help: "US South Region Cold",
}, {
Value: "us-south-flex",
Help: "US South Region Flex",
}, {
Value: "eu-standard",
Help: "EU Cross Region Standard",
}, {
Value: "eu-vault",
Help: "EU Cross Region Vault",
}, {
Value: "eu-cold",
Help: "EU Cross Region Cold",
}, {
Value: "eu-flex",
Help: "EU Cross Region Flex",
}, {
Value: "eu-gb-standard",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Standard",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "eu-gb-vault",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Vault",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "eu-gb-cold",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Cold",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "eu-gb-flex",
backend: fix misspellings
2019-02-08 01:41:17 +08:00
Help: "Great Britain Flex",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "ap-standard",
Help: "APAC Standard",
}, {
Value: "ap-vault",
Help: "APAC Vault",
}, {
Value: "ap-cold",
Help: "APAC Cold",
}, {
Value: "ap-flex",
Help: "APAC Flex",
}, {
Value: "mel01-standard",
Help: "Melbourne Standard",
}, {
Value: "mel01-vault",
Help: "Melbourne Vault",
}, {
Value: "mel01-cold",
Help: "Melbourne Cold",
}, {
Value: "mel01-flex",
Help: "Melbourne Flex",
}, {
Value: "tor01-standard",
Help: "Toronto Standard",
}, {
Value: "tor01-vault",
Help: "Toronto Vault",
}, {
Value: "tor01-cold",
Help: "Toronto Cold",
}, {
Value: "tor01-flex",
Help: "Toronto Flex",
}},
}, {
Name: "location_constraint",
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Help: "Location constraint - must be set to match the Region.\nLeave blank if not sure. Used when creating buckets only.",
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
Provider: "!AWS,IBMCOS,Alibaba,Scaleway,StackPath",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "acl",
s3: set ACL for server side copies to that provided by the user - fixes #2691 Before this change the ACL for objects which were server side copied was left at the default "private" settings. S3 doesn't copy the ACL from the source when you copy an object, you have to set it afresh which is what this does.
2018-10-26 05:19:36 +08:00
Help: `Canned ACL used when creating buckets and storing or copying objects.
s3: add --s3-bucket-acl to control bucket ACL - fixes #2918 Before this change buckets were created with the same ACL as objects. After this change, the user can set just --s3-acl to set the ACL of buckets and objects, or use --s3-bucket-acl as well to have a different ACL used for bucket creation. This also logs at INFO level the creation and deletion of buckets.
2019-01-17 01:23:37 +08:00
This ACL is used for creating objects and if bucket_acl isn't set, for creating buckets too.
s3: set ACL for server side copies to that provided by the user - fixes #2691 Before this change the ACL for objects which were server side copied was left at the default "private" settings. S3 doesn't copy the ACL from the source when you copy an object, you have to set it afresh which is what this does.
2018-10-26 05:19:36 +08:00
For more info visit https://docs.aws.amazon.com/AmazonS3/latest/dev/acl-overview.html#canned-acl
Note that this ACL is applied when server side copying objects as S3
doesn't copy the ACL from the source but rather writes a fresh one.`,
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Examples: []fs.OptionExample{{
Value: "private",
Help: "Owner gets FULL_CONTROL. No one else has access rights (default).",
Provider: "!IBMCOS",
}, {
Value: "public-read",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ access.",
Provider: "!IBMCOS",
}, {
Value: "public-read-write",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ and WRITE access.\nGranting this on a bucket is generally not recommended.",
Provider: "!IBMCOS",
}, {
Value: "authenticated-read",
Help: "Owner gets FULL_CONTROL. The AuthenticatedUsers group gets READ access.",
Provider: "!IBMCOS",
}, {
Value: "bucket-owner-read",
Help: "Object owner gets FULL_CONTROL. Bucket owner gets READ access.\nIf you specify this canned ACL when creating a bucket, Amazon S3 ignores it.",
Provider: "!IBMCOS",
}, {
Value: "bucket-owner-full-control",
Help: "Both the object owner and the bucket owner get FULL_CONTROL over the object.\nIf you specify this canned ACL when creating a bucket, Amazon S3 ignores it.",
Provider: "!IBMCOS",
}, {
Value: "private",
Help: "Owner gets FULL_CONTROL. No one else has access rights (default). This acl is available on IBM Cloud (Infra), IBM Cloud (Storage), On-Premise COS",
config,s3: hierarchical configuration support #2140 This introduces a method of making provider specific configuration within a remote. This is useful particularly in s3. This commit does the basic configuration in S3 for IBM COS.
2018-04-13 00:05:53 +08:00
Provider: "IBMCOS",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "public-read",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ access. This acl is available on IBM Cloud (Infra), IBM Cloud (Storage), On-Premise IBM COS",
Provider: "IBMCOS",
}, {
Value: "public-read-write",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ and WRITE access. This acl is available on IBM Cloud (Infra), On-Premise IBM COS",
config,s3: hierarchical configuration support #2140 This introduces a method of making provider specific configuration within a remote. This is useful particularly in s3. This commit does the basic configuration in S3 for IBM COS.
2018-04-13 00:05:53 +08:00
Provider: "IBMCOS",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Value: "authenticated-read",
Help: "Owner gets FULL_CONTROL. The AuthenticatedUsers group gets READ access. Not supported on Buckets. This acl is available on IBM Cloud (Infra) and On-Premise IBM COS",
Provider: "IBMCOS",
}},
s3: add --s3-bucket-acl to control bucket ACL - fixes #2918 Before this change buckets were created with the same ACL as objects. After this change, the user can set just --s3-acl to set the ACL of buckets and objects, or use --s3-bucket-acl as well to have a different ACL used for bucket creation. This also logs at INFO level the creation and deletion of buckets.
2019-01-17 01:23:37 +08:00
}, {
Name: "bucket_acl",
Help: `Canned ACL used when creating buckets.
For more info visit https://docs.aws.amazon.com/AmazonS3/latest/dev/acl-overview.html#canned-acl
Note that this ACL is applied when only when creating buckets. If it
isn't set then "acl" is used instead.`,
Advanced: true,
Examples: []fs.OptionExample{{
Value: "private",
Help: "Owner gets FULL_CONTROL. No one else has access rights (default).",
}, {
Value: "public-read",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ access.",
}, {
Value: "public-read-write",
Help: "Owner gets FULL_CONTROL. The AllUsers group gets READ and WRITE access.\nGranting this on a bucket is generally not recommended.",
}, {
Value: "authenticated-read",
Help: "Owner gets FULL_CONTROL. The AuthenticatedUsers group gets READ access.",
}},
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "server_side_encryption",
Help: "The server-side encryption algorithm used when storing this object in S3.",
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
Provider: "AWS,Ceph,Minio",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Examples: []fs.OptionExample{{
Value: "",
Help: "None",
}, {
Value: "AES256",
Help: "AES256",
s3: add support for KMS Key ID - fixes #2217 This code supports aws:kms and the kms key id for the s3 backend.
2018-08-31 00:08:27 +08:00
}, {
Value: "aws:kms",
Help: "aws:kms",
}},
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
}, {
Name: "sse_customer_algorithm",
Help: "If using SSE-C, the server-side encryption algorithm used when storing this object in S3.",
Provider: "AWS,Ceph,Minio",
Advanced: true,
Examples: []fs.OptionExample{{
Value: "",
Help: "None",
}, {
Value: "AES256",
Help: "AES256",
}},
s3: add support for KMS Key ID - fixes #2217 This code supports aws:kms and the kms key id for the s3 backend.
2018-08-31 00:08:27 +08:00
}, {
Name: "sse_kms_key_id",
Help: "If using KMS ID you must provide the ARN of Key.",
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
Provider: "AWS,Ceph,Minio",
s3: add support for KMS Key ID - fixes #2217 This code supports aws:kms and the kms key id for the s3 backend.
2018-08-31 00:08:27 +08:00
Examples: []fs.OptionExample{{
Value: "",
Help: "None",
}, {
Value: "arn:aws:kms:us-east-1:*",
Help: "arn:aws:kms:*",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}},
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
}, {
Name: "sse_customer_key",
doc: fix typos throughout docs and code
2020-05-20 18:39:20 +08:00
Help: "If using SSE-C you must provide the secret encryption key used to encrypt/decrypt your data.",
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
Provider: "AWS,Ceph,Minio",
Advanced: true,
Examples: []fs.OptionExample{{
Value: "",
Help: "None",
}},
}, {
Name: "sse_customer_key_md5",
Help: "If using SSE-C you must provide the secret encryption key MD5 checksum.",
Provider: "AWS,Ceph,Minio",
Advanced: true,
Examples: []fs.OptionExample{{
Value: "",
Help: "None",
}},
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}, {
Name: "storage_class",
docs: move documentation for options from docs/content into backends In the following commit, the documentation will be autogenerated.
2018-10-02 01:36:15 +08:00
Help: "The storage class to use when storing new objects in S3.",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
Provider: "AWS",
Examples: []fs.OptionExample{{
Value: "",
Help: "Default",
}, {
Value: "STANDARD",
Help: "Standard storage class",
}, {
Value: "REDUCED_REDUNDANCY",
Help: "Reduced redundancy storage class",
}, {
Value: "STANDARD_IA",
Help: "Standard Infrequent Access storage class",
}, {
Value: "ONEZONE_IA",
Help: "One Zone Infrequent Access storage class",
s3: add GLACIER storage class Fixes #923
2018-12-07 05:53:05 +08:00
}, {
Value: "GLACIER",
Help: "Glacier storage class",
s3: add support for "Glacier Deep Archive" storage class - fixes #3088
2019-04-05 11:14:05 +08:00
}, {
Value: "DEEP_ARCHIVE",
Help: "Glacier Deep Archive storage class",
s3: Add INTELLIGENT_TIERING storage class For Intelligent-Tiering: https://aws.amazon.com/s3/storage-classes/#Unknown_or_changing_access
2019-04-05 01:04:28 +08:00
}, {
Value: "INTELLIGENT_TIERING",
Help: "Intelligent-Tiering storage class",
s3: add in config for all the supported S3 providers #2140 These are AWS, Ceph, Dreamhost, IBM COS S3, Minio, Wasabi and Other. This configures endpoints where known and makes sure config doesn't appear where it isn't valid where possible.
2018-04-13 23:08:00 +08:00
}},
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
}, {
s3: Add Alibaba OSS to integration tests and fix storage classes
2019-01-13 04:41:47 +08:00
// Mapping from here: https://www.alibabacloud.com/help/doc-detail/64919.htm
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
Name: "storage_class",
Help: "The storage class to use when storing new objects in OSS.",
Provider: "Alibaba",
Examples: []fs.OptionExample{{
s3: Add Alibaba OSS to integration tests and fix storage classes
2019-01-13 04:41:47 +08:00
Value: "",
Help: "Default",
}, {
Value: "STANDARD",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
Help: "Standard storage class",
}, {
s3: Add Alibaba OSS to integration tests and fix storage classes
2019-01-13 04:41:47 +08:00
Value: "GLACIER",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
Help: "Archive storage mode.",
}, {
s3: Add Alibaba OSS to integration tests and fix storage classes
2019-01-13 04:41:47 +08:00
Value: "STANDARD_IA",
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
Help: "Infrequent access storage mode.",
}},
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
}, {
// Mapping from here: https://www.scaleway.com/en/docs/object-storage-glacier/#-Scaleway-Storage-Classes
Name: "storage_class",
Help: "The storage class to use when storing new objects in S3.",
Provider: "Scaleway",
Examples: []fs.OptionExample{{
Value: "",
Help: "Default",
}, {
Value: "STANDARD",
Help: "The Standard class for any upload; suitable for on-demand content like streaming or CDN.",
}, {
Value: "GLACIER",
Help: "Archived storage; prices are lower, but it needs to be restored first to be accessed.",
}},
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
}, {
Name: "upload_cutoff",
Help: `Cutoff for switching to chunked upload
Any files larger than this will be uploaded in chunks of chunk_size.
The minimum is 0 and the maximum is 5GB.`,
Default: defaultUploadCutoff,
Advanced: true,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
}, {
docs: move documentation for options from docs/content into backends In the following commit, the documentation will be autogenerated.
2018-10-02 01:36:15 +08:00
Name: "chunk_size",
Help: `Chunk size to use for uploading.
s3: Reduce memory usage streaming files by reducing max stream upload size Before this change rclone would allow the user to stream (eg with rclone mount, rclone rcat or uploading google photos or docs) 5TB files. This meant that rclone allocated 4 * 525 MB buffers per transfer which is way too much memory by default. This change makes rclone use the configured chunk size for streamed uploads. This is 5MB by default which means that rclone can stream upload files up to 48GB by default staying below the 10,000 chunks limit. This can be increased with --s3-chunk-size if necessary. If rclone detects that a file is being streamed to s3 it will make a single NOTICE level log stating the limitation. This fixes the enormous memory usage. Fixes #3568 See: https://forum.rclone.org/t/how-much-memory-does-rclone-need/12743
2019-11-06 18:41:03 +08:00
When uploading files larger than upload_cutoff or files with unknown
size (eg from "rclone rcat" or uploaded with "rclone mount" or google
photos or google docs) they will be uploaded as multipart uploads
using this chunk size.
docs: move documentation for options from docs/content into backends In the following commit, the documentation will be autogenerated.
2018-10-02 01:36:15 +08:00
Note that "--s3-upload-concurrency" chunks of this size are buffered
in memory per transfer.
If you are transferring large files over high speed links and you have
s3: Reduce memory usage streaming files by reducing max stream upload size Before this change rclone would allow the user to stream (eg with rclone mount, rclone rcat or uploading google photos or docs) 5TB files. This meant that rclone allocated 4 * 525 MB buffers per transfer which is way too much memory by default. This change makes rclone use the configured chunk size for streamed uploads. This is 5MB by default which means that rclone can stream upload files up to 48GB by default staying below the 10,000 chunks limit. This can be increased with --s3-chunk-size if necessary. If rclone detects that a file is being streamed to s3 it will make a single NOTICE level log stating the limitation. This fixes the enormous memory usage. Fixes #3568 See: https://forum.rclone.org/t/how-much-memory-does-rclone-need/12743
2019-11-06 18:41:03 +08:00
enough memory, then increasing this will speed up the transfers.
Rclone will automatically increase the chunk size when uploading a
large file of known size to stay below the 10,000 chunks limit.
Files of unknown size are uploaded with the configured
chunk_size. Since the default chunk size is 5MB and there can be at
most 10,000 chunks, this means that by default the maximum size of
file you can stream upload is 48GB. If you wish to stream upload
larger files then you will need to increase chunk_size.`,
backend: add fstests.ChunkedUploadConfig - azureblob - b2 - drive - dropbox - onedrive - s3 - swift
2018-09-07 19:02:27 +08:00
Default: minChunkSize,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Advanced: true,
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
}, {
Name: "max_upload_parts",
Help: `Maximum number of parts in a multipart upload.
This option defines the maximum number of multipart chunks to use
when doing a multipart upload.
This can be useful if a service does not support the AWS S3
specification of 10,000 chunks.
Rclone will automatically increase the chunk size when uploading a
large file of a known size to stay below this number of chunks limit.
`,
Default: maxUploadParts,
Advanced: true,
s3: add --s3-copy-cutoff for size to switch to multipart copy Before this change we used the same (relatively low limits) for server side copy as we did for multipart uploads. It doesn't make sense to use the same limits since no data is being downloaded or uploaded for a server side copy. This change introduces a new parameter --s3-copy-cutoff to control when the switch from single to multipart server size copy happens and defaults it to the maximum 5GB. This makes server side copies much more efficient. It also fixes the erroneous error when trying to set the modification time of a file bigger than 5GB. See #3778
2019-12-03 01:14:57 +08:00
}, {
Name: "copy_cutoff",
Help: `Cutoff for switching to multipart copy
Any files larger than this that need to be server side copied will be
copied in chunks of this size.
The minimum is 0 and the maximum is 5GB.`,
Default: fs.SizeSuffix(maxSizeForCopy),
Advanced: true,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
}, {
s3: improve docs for --s3-disable-checksum
2020-04-24 02:47:48 +08:00
Name: "disable_checksum",
Help: `Don't store MD5 checksum with object metadata
Normally rclone will calculate the MD5 checksum of the input before
uploading it so it can add it to metadata on the object. This is great
for data integrity checking but can cause long delays for large files
to start uploading.`,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Default: false,
Advanced: true,
s3: implement `profile` and `shared_credentials_file` options It is impossible to use two different profiles at the same time - these config vars enable that. See: https://forum.rclone.org/t/s3-source-destination-named-profile/17417
2020-06-26 18:14:40 +08:00
}, {
Name: "shared_credentials_file",
Help: `Path to the shared credentials file
If env_auth = true then rclone can use a shared credentials file.
If this variable is empty rclone will look for the
"AWS_SHARED_CREDENTIALS_FILE" env variable. If the env value is empty
it will default to the current user's home directory.
Linux/OSX: "$HOME/.aws/credentials"
Windows: "%USERPROFILE%\.aws\credentials"
`,
Advanced: true,
}, {
Name: "profile",
Help: `Profile to use in the shared credentials file
If env_auth = true then rclone can use a shared credentials file. This
variable controls which profile is used in that file.
If empty it will default to the environment variable "AWS_PROFILE" or
"default" if that environment variable is also not set.
`,
Advanced: true,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
}, {
Name: "session_token",
Help: "An AWS session token",
Advanced: true,
}, {
docs: move documentation for options from docs/content into backends In the following commit, the documentation will be autogenerated.
2018-10-02 01:36:15 +08:00
Name: "upload_concurrency",
Help: `Concurrency for multipart uploads.
This is the number of chunks of the same file that are uploaded
concurrently.
If you are uploading small numbers of large file over high speed link
and these uploads do not fully utilize your bandwidth, then increasing
this may help to speed up the transfers.`,
s3: change --s3-upload-concurrency default to 4 to increase perfomance #2772 Increasing the --s3-upload-concurrency to 4 (from 2) gives an additional 45% throughput at the cost of 10MB extra memory per transfer. After testing the upload perfoc
2018-12-03 01:51:14 +08:00
Default: 4,
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
Advanced: true,
s3: Add --s3-force-path-style - fixes #2401
2018-07-18 23:40:59 +08:00
}, {
docs: move documentation for options from docs/content into backends In the following commit, the documentation will be autogenerated.
2018-10-02 01:36:15 +08:00
Name: "force_path_style",
Help: `If true use path style access if false use virtual hosted style.
If this is true (the default) then rclone will use path style access,
if false then rclone will use virtual path style. See [the AWS S3
docs](https://docs.aws.amazon.com/AmazonS3/latest/dev/UsingBucket.html#access-bucket-intro)
for more info.
s3: force path style bucket access to off for AWS deprecation AWS are deprecating path style bucket access so rclone should stop using it by default for this provider. This change shouldn't break any workflows as all AWS endpoints support virtual hosted style lookups of buckets. It may even improve performance. See: https://aws.amazon.com/blogs/aws/amazon-s3-path-deprecation-plan-the-rest-of-the-story/
2020-01-06 01:53:45 +08:00
Some providers (eg AWS, Aliyun OSS or Netease COS) require this set to
false - rclone will do this automatically based on the provider
setting.`,
s3: Add --s3-force-path-style - fixes #2401
2018-07-18 23:40:59 +08:00
Default: true,
Advanced: true,
s3: make --s3-v2-auth flag This is an alternative to setting the region to "other-v2-signature" which is inconvenient for multi-region providers.
2018-10-09 20:03:37 +08:00
}, {
Name: "v2_auth",
Help: `If true use v2 authentication.
If this is false (the default) then rclone will use v4 authentication.
If it is set then rclone will use v2 authentication.
Use this only if v4 signatures don't work, eg pre Jewel/v10 CEPH.`,
Default: false,
Advanced: true,
s3: suppport S3 Accelerated endpoints with --s3-use-accelerate-endpoint Fixes #3123
2019-04-26 17:19:00 +08:00
}, {
Name: "use_accelerate_endpoint",
Provider: "AWS",
Help: `If true use the AWS S3 accelerated endpoint.
See: [AWS S3 Transfer acceleration](https://docs.aws.amazon.com/AmazonS3/latest/dev/transfer-acceleration-examples.html)`,
Default: false,
Advanced: true,
s3: add option for multipart failiure behaviour This is needed for resuming uploads across different sessions.
2019-07-24 17:03:38 +08:00
}, {
Name: "leave_parts_on_error",
Provider: "AWS",
Help: `If true avoid calling abort upload on a failure, leaving all successfully uploaded parts on S3 for manual recovery.
It should be set to true for resuming uploads across different sessions.
WARNING: Storing parts of an incomplete multipart upload counts towards space usage on S3 and will add additional costs if not cleaned up.
`,
Default: false,
Advanced: true,
s3: introduce list_chunk option for bucket listing The S3 ListObject API returns paginated bucket listings, with "MaxKeys" items for each GET call. The default value is 1000 entries, but for buckets with millions of objects it might make sense to request more elements per request, if the backend supports it. This commit adds a "list_chunk" option for the user to specify a lower or higher value. This commit does not add safe guards around this value - if a user decides to request a too large list, it might result in connection timeouts (on the server or client). In AWS S3, there is a fixed limit of 1000, some other services might have one too. In Ceph, this can be configured in RadosGW.
2019-12-26 19:05:00 +08:00
}, {
Name: "list_chunk",
Help: `Size of listing chunk (response list for each ListObject S3 request).
This option is also known as "MaxKeys", "max-items", or "page-size" from the AWS S3 specification.
Most services truncate the response list to 1000 objects even if requested more than that.
In AWS S3 this is a global maximum and cannot be changed, see [AWS S3](https://docs.aws.amazon.com/cli/latest/reference/s3/ls.html).
In Ceph, this can be increased with the "rgw list buckets max chunk" option.
`,
Default: 1000,
Advanced: true,
s3: add --s3-no-check-bucket for minimising rclone transactions and perms Fixes #4449
2020-07-22 19:02:17 +08:00
}, {
Name: "no_check_bucket",
Help: `If set don't attempt to check the bucket exists or create it
This can be useful when trying to minimise the number of transactions
rclone does if you know the bucket exists already.
`,
Default: false,
Advanced: true,
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
}, {
Name: config.ConfigEncoding,
Help: config.ConfigEncodingHelp,
Advanced: true,
backends: move encoding definitions from fs/encodings
2020-01-15 05:51:49 +08:00
// Any UTF-8 character is valid in a key, however it can't handle
// invalid UTF-8 and / have a special meaning.
//
// The SDK can't seem to handle uploading files called '.'
//
// FIXME would be nice to add
// - initial / encoding
// - doubled / encoding
// - trailing / encoding
// so that AWS keys are always valid file names
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
Default: encoder.EncodeInvalidUtf8 |
backends: move encoding definitions from fs/encodings
2020-01-15 05:51:49 +08:00
encoder.EncodeSlash |
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
encoder.EncodeDot,
}, {
Name: "memory_pool_flush_time",
Default: memoryPoolFlushTime,
Advanced: true,
Help: `How often internal memory buffer pools will be flushed.
Uploads which requires additional buffers (f.e multipart) will use memory pool for allocations.
This option controls how often unused buffers will be removed from the pool.`,
}, {
Name: "memory_pool_use_mmap",
Default: memoryPoolUseMmap,
Advanced: true,
Help: `Whether to use mmap buffers in internal memory pool.`,
},
}})
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Constants
const (
s3: Reduce memory usage streaming files by reducing max stream upload size Before this change rclone would allow the user to stream (eg with rclone mount, rclone rcat or uploading google photos or docs) 5TB files. This meant that rclone allocated 4 * 525 MB buffers per transfer which is way too much memory by default. This change makes rclone use the configured chunk size for streamed uploads. This is 5MB by default which means that rclone can stream upload files up to 48GB by default staying below the 10,000 chunks limit. This can be increased with --s3-chunk-size if necessary. If rclone detects that a file is being streamed to s3 it will make a single NOTICE level log stating the limitation. This fixes the enormous memory usage. Fixes #3568 See: https://forum.rclone.org/t/how-much-memory-does-rclone-need/12743
2019-11-06 18:41:03 +08:00
metaMtime = "Mtime" // the meta key to store mtime in - eg X-Amz-Meta-Mtime
metaMD5Hash = "Md5chksum" // the meta key to store md5hash in
maxSizeForCopy = 5 * 1024 * 1024 * 1024 // The maximum size of object we can COPY
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
maxUploadParts = 10000 // maximum allowed number of parts in a multi-part upload
minChunkSize = fs.SizeSuffix(1024 * 1024 * 5)
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
defaultUploadCutoff = fs.SizeSuffix(200 * 1024 * 1024)
maxUploadCutoff = fs.SizeSuffix(5 * 1024 * 1024 * 1024)
minSleep = 10 * time.Millisecond // In case of error, start at 10ms sleep.
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
memoryPoolFlushTime = fs.Duration(time.Minute) // flush the cached buffers after this long
memoryPoolUseMmap = false
s3: Cap expiry duration to 1 Week and return error when sharing dir
2020-06-19 00:50:50 +08:00
maxExpireDuration = fs.Duration(7 * 24 * time.Hour) // max expiry is 1 week
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
)
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
// Options defines the configuration for this backend
type Options struct {
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
Provider string `config:"provider"`
EnvAuth bool `config:"env_auth"`
AccessKeyID string `config:"access_key_id"`
SecretAccessKey string `config:"secret_access_key"`
Region string `config:"region"`
Endpoint string `config:"endpoint"`
LocationConstraint string `config:"location_constraint"`
ACL string `config:"acl"`
BucketACL string `config:"bucket_acl"`
ServerSideEncryption string `config:"server_side_encryption"`
SSEKMSKeyID string `config:"sse_kms_key_id"`
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
SSECustomerAlgorithm string `config:"sse_customer_algorithm"`
SSECustomerKey string `config:"sse_customer_key"`
SSECustomerKeyMD5 string `config:"sse_customer_key_md5"`
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
StorageClass string `config:"storage_class"`
UploadCutoff fs.SizeSuffix `config:"upload_cutoff"`
CopyCutoff fs.SizeSuffix `config:"copy_cutoff"`
ChunkSize fs.SizeSuffix `config:"chunk_size"`
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
MaxUploadParts int64 `config:"max_upload_parts"`
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
DisableChecksum bool `config:"disable_checksum"`
s3: implement `profile` and `shared_credentials_file` options It is impossible to use two different profiles at the same time - these config vars enable that. See: https://forum.rclone.org/t/s3-source-destination-named-profile/17417
2020-06-26 18:14:40 +08:00
SharedCredentialsFile string `config:"shared_credentials_file"`
Profile string `config:"profile"`
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
SessionToken string `config:"session_token"`
UploadConcurrency int `config:"upload_concurrency"`
ForcePathStyle bool `config:"force_path_style"`
V2Auth bool `config:"v2_auth"`
UseAccelerateEndpoint bool `config:"use_accelerate_endpoint"`
LeavePartsOnError bool `config:"leave_parts_on_error"`
ListChunk int64 `config:"list_chunk"`
s3: add --s3-no-check-bucket for minimising rclone transactions and perms Fixes #4449
2020-07-22 19:02:17 +08:00
NoCheckBucket bool `config:"no_check_bucket"`
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
Enc encoder.MultiEncoder `config:"encoding"`
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
MemoryPoolFlushTime fs.Duration `config:"memory_pool_flush_time"`
MemoryPoolUseMmap bool `config:"memory_pool_use_mmap"`
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
}
s3: add support for setting storage class in config and command line
2016-09-02 05:27:50 +08:00
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
// Fs represents a remote s3 server
type Fs struct {
s3: use single memory pool Previously we had a map of pools for different chunk sizes. In practice the mapping is not very useful and requires a lock. Pools of size other that ChunkSize can only happen when we have a huge file (over 10k * ChunkSize). We need to have a bunch of identically sized huge files. In such case most likely ChunkSize should be increased. The mapping and its lock is replaced with a single initialised pool for ChunkSize, in other cases pool is allocated and freed on per file basis.
2020-04-09 18:18:58 +08:00
name string // the name of the remote
root string // root of the bucket - ignore all objects above this
opt Options // parsed options
features *fs.Features // optional features
c *s3.S3 // the connection to the s3 server
ses *session.Session // the s3 session
rootBucket string // bucket part of root (if any)
rootDirectory string // directory part of root (if any)
cache *bucket.Cache // cache for bucket creation status
pacer *fs.Pacer // To pace the API calls
srv *http.Client // a plain http client
pool *pool.Pool // memory pool
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
// Object describes a s3 object
type Object struct {
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Will definitely have everything but meta which may be nil
//
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
// List will read everything but meta & mimeType - to fill
// that in you need to call readMetaData
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
fs *Fs // what this object is part of
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
remote string // The remote path
etag string // md5sum of the object
bytes int64 // size of the object
lastModified time.Time // Last modified
meta map[string]*string // The object metadata if known - may be nil
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
mimeType string // MimeType of object - may be ""
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
storageClass string // eg GLACIER
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
// ------------------------------------------------------------
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Name of the remote (as passed into NewFs)
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (f *Fs) Name() string {
Add Name() to Fs interface to return name as passed to NewFs
2015-08-22 23:53:11 +08:00
return f.name
}
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Root of the remote (as passed into NewFs)
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (f *Fs) Root() string {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return f.root
Add Root() to Fs interface
2015-09-02 03:45:27 +08:00
}
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
// String converts this Fs to a string
func (f *Fs) String() string {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if f.rootBucket == "" {
return fmt.Sprintf("S3 root")
}
if f.rootDirectory == "" {
return fmt.Sprintf("S3 bucket %s", f.rootBucket)
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return fmt.Sprintf("S3 bucket %s path %s", f.rootBucket, f.rootDirectory)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Define a new Features() method for Fs Optional interfaces are becoming more important in rclone, --track-renames and --backup-dir both rely on them. Up to this point rclone has used interface upgrades to define optional behaviour on Fs objects. However when one Fs object wraps another it is very difficult for this scheme to work accurately. rclone has relied on specific error messages being returned when the interface isn't supported - this is unsatisfactory because it means you have to call the interface to see whether it is supported. This change enables accurate detection of optional interfaces by use of a Features struct as returned by an obligatory Fs.Features() method. The Features struct contains flags and function pointers which can be tested against nil to see whether they can be used. As a result crypt and hubic can accurately reflect the capabilities of the underlying Fs they are wrapping.
2017-01-14 01:21:47 +08:00
// Features returns the optional features of this Fs
func (f *Fs) Features() *fs.Features {
return f.features
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
// retryErrorCodes is a slice of error codes that we will retry
// See: https://docs.aws.amazon.com/AmazonS3/latest/API/ErrorResponses.html
var retryErrorCodes = []int{
backend/s3: Added 500 as retryErrorCode The error code 500 Internal Error indicates that Amazon S3 is unable to handle the request at that time. The error code 503 Slow Down typically indicates that the requests to the S3 bucket are very high, exceeding the request rates described in Request Rate and Performance Guidelines. Because Amazon S3 is a distributed service, a very small percentage of 5xx errors are expected during normal use of the service. All requests that return 5xx errors from Amazon S3 can and should be retried, so we recommend that applications making requests to Amazon S3 have a fault-tolerance mechanism to recover from these errors. https://aws.amazon.com/premiumsupport/knowledge-center/http-5xx-errors-s3/
2020-02-11 23:30:32 +08:00
500, // Internal Server Error - "We encountered an internal error. Please try again."
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
503, // Service Unavailable/Slow Down - "Reduce your request rate"
}
//S3 is pretty resilient, and the built in retry handling is probably sufficient
// as it should notice closed connections and timeouts which are the most likely
// sort of failure modes
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
func (f *Fs) shouldRetry(err error) (bool, error) {
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
// If this is an awserr object, try and extract more useful information to determine if we should retry
if awsError, ok := err.(awserr.Error); ok {
backend: fix misspellings
2019-02-08 01:41:17 +08:00
// Simple case, check the original embedded error in case it's generically retryable
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
if fserrors.ShouldRetry(awsError.OrigErr()) {
return true, err
}
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
// Failing that, if it's a RequestFailure it's probably got an http status code we can check
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
if reqErr, ok := err.(awserr.RequestFailure); ok {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// 301 if wrong region for bucket - can only update if running from a bucket
if f.rootBucket != "" {
if reqErr.StatusCode() == http.StatusMovedPermanently {
urfbErr := f.updateRegionForBucket(f.rootBucket)
if urfbErr != nil {
fs.Errorf(f, "Failed to update region for bucket: %v", urfbErr)
return false, err
}
return true, err
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
}
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
for _, e := range retryErrorCodes {
if reqErr.StatusCode() == e {
return true, err
}
}
}
}
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
// Ok, not an awserr, check for generic failure conditions
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
return fserrors.ShouldRetry(err), err
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// parsePath parses a remote 'url'
func parsePath(path string) (root string) {
root = strings.Trim(path, "/")
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// split returns bucket and bucketPath from the rootRelativePath
// relative to f.root
func (f *Fs) split(rootRelativePath string) (bucketName, bucketPath string) {
s3: use lib/encoder Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2018-11-02 20:15:30 +08:00
bucketName, bucketPath = bucket.Split(path.Join(f.root, rootRelativePath))
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
return f.opt.Enc.FromStandardName(bucketName), f.opt.Enc.FromStandardPath(bucketPath)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
}
// split returns bucket and bucketPath from the object
func (o *Object) split() (bucket, bucketPath string) {
return o.fs.split(o.remote)
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// s3Connection makes a connection to s3
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
func s3Connection(opt *Options) (*s3.S3, *session.Session, error) {
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Make the auth
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
v := credentials.Value{
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
AccessKeyID: opt.AccessKeyID,
SecretAccessKey: opt.SecretAccessKey,
SessionToken: opt.SessionToken,
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
}
s3: add support for ECS task IAM roles ECS container IAM metadata is in a different place than EC2 IAM metadata. Use defaults' RemoteCredProvider function to query the standard locations for the credentials. Give the ECS role precedence over the role available from the underlying EC2 instance.
2017-11-23 05:21:36 +08:00
lowTimeoutClient := &http.Client{Timeout: 1 * time.Second} // low timeout to ec2 metadata service
def := defaults.Get()
def.Config.HTTPClient = lowTimeoutClient
backend/s3: replace deprecated session.New() with session.NewSession()
2020-03-04 08:19:59 +08:00
// start a new AWS session
awsSession, err := session.NewSession()
if err != nil {
return nil, nil, errors.Wrap(err, "NewSession")
}
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
// first provider to supply a credential set "wins"
providers := []credentials.Provider{
// use static credentials if they're present (checked by provider)
&credentials.StaticProvider{Value: v},
// * Access Key ID: AWS_ACCESS_KEY_ID or AWS_ACCESS_KEY
// * Secret Access Key: AWS_SECRET_ACCESS_KEY or AWS_SECRET_KEY
&credentials.EnvProvider{},
s3: Look in S3 named profile files for credentials - fixes #2243
2018-04-16 19:14:35 +08:00
// A SharedCredentialsProvider retrieves credentials
// from the current user's home directory. It checks
// AWS_SHARED_CREDENTIALS_FILE and AWS_PROFILE too.
s3: implement `profile` and `shared_credentials_file` options It is impossible to use two different profiles at the same time - these config vars enable that. See: https://forum.rclone.org/t/s3-source-destination-named-profile/17417
2020-06-26 18:14:40 +08:00
&credentials.SharedCredentialsProvider{
Filename: opt.SharedCredentialsFile, // If empty will look for "AWS_SHARED_CREDENTIALS_FILE" env variable.
Profile: opt.Profile, // If empty will look gor "AWS_PROFILE" env var or "default" if not set.
},
s3: Look in S3 named profile files for credentials - fixes #2243
2018-04-16 19:14:35 +08:00
s3: add support for ECS task IAM roles ECS container IAM metadata is in a different place than EC2 IAM metadata. Use defaults' RemoteCredProvider function to query the standard locations for the credentials. Give the ECS role precedence over the role available from the underlying EC2 instance.
2017-11-23 05:21:36 +08:00
// Pick up IAM role if we're in an ECS task
defaults.RemoteCredProvider(*def.Config, def.Handlers),
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
// Pick up IAM role in case we're on EC2
&ec2rolecreds.EC2RoleProvider{
backend/s3: replace deprecated session.New() with session.NewSession()
2020-03-04 08:19:59 +08:00
Client: ec2metadata.New(awsSession, &aws.Config{
s3: add support for ECS task IAM roles ECS container IAM metadata is in a different place than EC2 IAM metadata. Use defaults' RemoteCredProvider function to query the standard locations for the credentials. Give the ECS role precedence over the role available from the underlying EC2 instance.
2017-11-23 05:21:36 +08:00
HTTPClient: lowTimeoutClient,
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
}),
backend/s3: fix ExpiryWindow value ExpiryWindow accepts duration but it was set to value 3. This changes it to 3 * time.Minute since default is 5 min.
2019-10-31 23:42:16 +08:00
ExpiryWindow: 3 * time.Minute,
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
},
s3: use aws web identity role provider
2020-01-06 03:49:31 +08:00
// Pick up IAM role if we are in EKS
&stscreds.WebIdentityRoleProvider{
ExpiryWindow: 3 * time.Minute,
},
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
}
cred := credentials.NewChainCredentials(providers)
s3: allow anonymous access to public repositories - fixes #154 When setting up the remote, leave both the access key and secret key blank.
2015-09-29 16:58:03 +08:00
switch {
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
case opt.EnvAuth:
s3: Fix empty checks in auth
2016-02-10 01:19:33 +08:00
// No need for empty checks if "env_auth" is true
case v.AccessKeyID == "" && v.SecretAccessKey == "":
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
// if no access key/secret and iam is explicitly disabled then fall back to anon interaction
cred = credentials.AnonymousCredentials
case v.AccessKeyID == "":
Update to AWS SDK 0.10.0 Tested with S3 and Dreamhost Here is a link to the release notes: http://aws.amazon.com/releasenotes/5476699172355228
2015-10-30 19:50:45 +08:00
return nil, nil, errors.New("access_key_id not found")
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
case v.SecretAccessKey == "":
Update to AWS SDK 0.10.0 Tested with S3 and Dreamhost Here is a link to the release notes: http://aws.amazon.com/releasenotes/5476699172355228
2015-10-30 19:50:45 +08:00
return nil, nil, errors.New("secret_access_key not found")
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
if opt.Region == "" {
opt.Region = "us-east-1"
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
if opt.Provider == "AWS" || opt.Provider == "Alibaba" || opt.Provider == "Netease" || opt.Provider == "Scaleway" || opt.UseAccelerateEndpoint {
s3: Support Alibaba Cloud (Aliyun) OSS The existing s3 backend passed all integration tests with OSS provided `force_path_style = false`. This makes sure that is so and adds documentation and configuration for OSS. Thanks to @luolibin for their work on the OSS backend which we ended up not needing. Fixes #1641 Fixes #1237
2019-01-13 00:46:45 +08:00
opt.ForcePathStyle = false
}
s3: add Scaleway provider - fixes #4338
2020-06-12 23:04:16 +08:00
if opt.Provider == "Scaleway" && opt.MaxUploadParts > 1000 {
opt.MaxUploadParts = 1000
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
awsConfig := aws.NewConfig().
s3: use rclone's low level retries instead of AWS SDK to fix listing retries In 5470d34740d03e15 "backend/s3: use low-level-retries as the number of SDK retries" we switched over to using the AWS SDK low level retries instead of rclone's low level retry logic. This had the unfortunate attempt that retrying listings to correct XML Syntax errors failed on non S3 backends such as CEPH. The AWS SDK was also retrying the XML Syntax error request which doesn't make sense. This change turns off the AWS SDK retries in favour of just using rclone's retry logic.
2020-03-15 01:28:29 +08:00
WithMaxRetries(0). // Rely on rclone's retry logic
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
WithCredentials(cred).
Break the fs package up into smaller parts. The purpose of this is to make it easier to maintain and eventually to allow the rclone backends to be re-used in other projects without having to use the rclone configuration system. The new code layout is documented in CONTRIBUTING.
2018-01-13 00:30:54 +08:00
WithHTTPClient(fshttp.NewClient(fs.Config)).
s3: suppport S3 Accelerated endpoints with --s3-use-accelerate-endpoint Fixes #3123
2019-04-26 17:19:00 +08:00
WithS3ForcePathStyle(opt.ForcePathStyle).
s3: Use regional s3 us-east-1 endpoint
2020-06-26 21:09:29 +08:00
WithS3UseAccelerate(opt.UseAccelerateEndpoint).
WithS3UsEast1RegionalEndpoint(endpoints.RegionalS3UsEast1Endpoint)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
if opt.Region != "" {
awsConfig.WithRegion(opt.Region)
}
if opt.Endpoint != "" {
awsConfig.WithEndpoint(opt.Endpoint)
}
s3: make v2 signatures work for ceph
2015-08-10 18:02:34 +08:00
// awsConfig.WithLogLevel(aws.LogDebugWithSigning)
s3: fix role_arn, credential_source, ... When the env_auth option is enabled, the AWS SDK's session constructor now loads configuration from ~/.aws/config and environment variables, and credentials per the selected (or default) AWS_PROFILE's settings. This is accomplished by **NOT** including any Credential provider in the aws.Config passed to the session constructor: If the Config.Credentials is non-nil, that will always be used and the user's configuration re role_arn, credential_source, source_profile, etc... from the shared config will be completely ignored. (The conditional creation and configuration of the stscreds Credential provider is complicated enough that it is not worth re-creating that logic.)
2018-11-07 10:50:28 +08:00
awsSessionOpts := session.Options{
Config: *awsConfig,
}
if opt.EnvAuth && opt.AccessKeyID == "" && opt.SecretAccessKey == "" {
// Enable loading config options from ~/.aws/config (selected by AWS_PROFILE env)
awsSessionOpts.SharedConfigState = session.SharedConfigEnable
// The session constructor (aws/session/mergeConfigSrcs) will only use the user's preferred credential source
// (from the shared config file) if the passed-in Options.Config.Credentials is nil.
awsSessionOpts.Config.Credentials = nil
}
ses, err := session.NewSessionWithOptions(awsSessionOpts)
if err != nil {
return nil, nil, err
}
c := s3.New(ses)
s3: make --s3-v2-auth flag This is an alternative to setting the region to "other-v2-signature" which is inconvenient for multi-region providers.
2018-10-09 20:03:37 +08:00
if opt.V2Auth || opt.Region == "other-v2-signature" {
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
fs.Debugf(nil, "Using v2 auth")
s3: Fix after upstream API changes in aws-sdk-go/aws - fixes #114
2015-08-28 15:47:41 +08:00
signer := func(req *request.Request) {
s3: make v2 signatures work for ceph
2015-08-10 18:02:34 +08:00
// Ignore AnonymousCredentials object
Update to AWS SDK 0.10.0 Tested with S3 and Dreamhost Here is a link to the release notes: http://aws.amazon.com/releasenotes/5476699172355228
2015-10-30 19:50:45 +08:00
if req.Config.Credentials == credentials.AnonymousCredentials {
s3: make v2 signatures work for ceph
2015-08-10 18:02:34 +08:00
return
}
Add IAM role and Env credentials This will make the s3 provider authentaction logic - Configured credentials if both key and secret available - Anonymous if key and secret missing and env_auth not set - if env_auth is set to truthy (https://golang.org/pkg/strconv/#ParseBool) - AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY environment variables - IAM role credentials as fallback
2016-02-01 21:11:27 +08:00
sign(v.AccessKeyID, v.SecretAccessKey, req.HTTPRequest)
s3: make v2 signatures work for ceph
2015-08-10 18:02:34 +08:00
}
c.Handlers.Sign.Clear()
s3: Fix after upstream API changes in aws-sdk-go/aws - fixes #114
2015-08-28 15:47:41 +08:00
c.Handlers.Sign.PushBackNamed(corehandlers.BuildContentLengthHandler)
s3: make v2 signatures work for ceph
2015-08-10 18:02:34 +08:00
c.Handlers.Sign.PushBack(signer)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Update to AWS SDK 0.10.0 Tested with S3 and Dreamhost Here is a link to the release notes: http://aws.amazon.com/releasenotes/5476699172355228
2015-10-30 19:50:45 +08:00
return c, ses, nil
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
backend: add fstests.ChunkedUploadConfig - azureblob - b2 - drive - dropbox - onedrive - s3 - swift
2018-09-07 19:02:27 +08:00
func checkUploadChunkSize(cs fs.SizeSuffix) error {
if cs < minChunkSize {
return errors.Errorf("%s is less than %s", cs, minChunkSize)
}
return nil
}
func (f *Fs) setUploadChunkSize(cs fs.SizeSuffix) (old fs.SizeSuffix, err error) {
err = checkUploadChunkSize(cs)
if err == nil {
old, f.opt.ChunkSize = f.opt.ChunkSize, cs
}
return
}
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
func checkUploadCutoff(cs fs.SizeSuffix) error {
if cs > maxUploadCutoff {
return errors.Errorf("%s is greater than %s", cs, maxUploadCutoff)
}
return nil
}
func (f *Fs) setUploadCutoff(cs fs.SizeSuffix) (old fs.SizeSuffix, err error) {
err = checkUploadCutoff(cs)
if err == nil {
old, f.opt.UploadCutoff = f.opt.UploadCutoff, cs
}
return
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// setRoot changes the root of the Fs
func (f *Fs) setRoot(root string) {
f.root = parsePath(root)
f.rootBucket, f.rootDirectory = bucket.Split(f.root)
}
s3: add support for setting storage class in config and command line
2016-09-02 05:27:50 +08:00
// NewFs constructs an Fs from the path, bucket:path
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
func NewFs(name, root string, m configmap.Mapper) (fs.Fs, error) {
// Parse config into Options struct
opt := new(Options)
err := configstruct.Set(m, opt)
if err != nil {
return nil, err
}
backend: add fstests.ChunkedUploadConfig - azureblob - b2 - drive - dropbox - onedrive - s3 - swift
2018-09-07 19:02:27 +08:00
err = checkUploadChunkSize(opt.ChunkSize)
if err != nil {
return nil, errors.Wrap(err, "s3: chunk size")
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
}
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
err = checkUploadCutoff(opt.UploadCutoff)
if err != nil {
return nil, errors.Wrap(err, "s3: upload cutoff")
}
s3: add --s3-bucket-acl to control bucket ACL - fixes #2918 Before this change buckets were created with the same ACL as objects. After this change, the user can set just --s3-acl to set the ACL of buckets and objects, or use --s3-bucket-acl as well to have a different ACL used for bucket creation. This also logs at INFO level the creation and deletion of buckets.
2019-01-17 01:23:37 +08:00
if opt.ACL == "" {
opt.ACL = "private"
}
if opt.BucketACL == "" {
opt.BucketACL = opt.ACL
}
Implement new backend config system This unifies the 3 methods of reading config * command line * environment variable * config file And allows them all to be configured in all places. This is done by making the []fs.Option in the backend registration be the master source of what the backend options are. The backend changes are: * Use the new configmap.Mapper parameter * Use configstruct to parse it into an Options struct * Add all config to []fs.Option including defaults and help * Remove all uses of pflag * Remove all uses of config.FileGet
2018-05-15 01:06:57 +08:00
c, ses, err := s3Connection(opt)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
if err != nil {
return nil, err
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
f := &Fs{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
name: name,
opt: *opt,
c: c,
ses: ses,
s3: use rclone's low level retries instead of AWS SDK to fix listing retries In 5470d34740d03e15 "backend/s3: use low-level-retries as the number of SDK retries" we switched over to using the AWS SDK low level retries instead of rclone's low level retry logic. This had the unfortunate attempt that retrying listings to correct XML Syntax errors failed on non S3 backends such as CEPH. The AWS SDK was also retrying the XML Syntax error request which doesn't make sense. This change turns off the AWS SDK retries in favour of just using rclone's retry logic.
2020-03-15 01:28:29 +08:00
pacer: fs.NewPacer(pacer.NewS3(pacer.MinSleep(minSleep))),
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
cache: bucket.NewCache(),
srv: fshttp.NewClient(fs.Config),
s3: use single memory pool Previously we had a map of pools for different chunk sizes. In practice the mapping is not very useful and requires a lock. Pools of size other that ChunkSize can only happen when we have a huge file (over 10k * ChunkSize). We need to have a bunch of identically sized huge files. In such case most likely ChunkSize should be increased. The mapping and its lock is replaced with a single initialised pool for ChunkSize, in other cases pool is allocated and freed on per file basis.
2020-04-09 18:18:58 +08:00
pool: pool.New(
time.Duration(opt.MemoryPoolFlushTime),
int(opt.ChunkSize),
opt.UploadConcurrency*fs.Config.Transfers,
opt.MemoryPoolUseMmap,
),
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
f.setRoot(root)
fs: Add CanHaveEmptyDirectories and BucketBased feature flags to all remotes
2017-08-09 22:27:43 +08:00
f.features = (&fs.Features{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
ReadMimeType: true,
WriteMimeType: true,
BucketBased: true,
BucketBasedRootOK: true,
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
SetTier: true,
GetTier: true,
fs: define SlowModTime and SlowHash features in the relevant backends
2020-06-19 17:28:34 +08:00
SlowModTime: true,
fs: Add CanHaveEmptyDirectories and BucketBased feature flags to all remotes
2017-08-09 22:27:43 +08:00
}).Fill(f)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if f.rootBucket != "" && f.rootDirectory != "" {
Implement single file operations for all file systems
2014-05-06 02:52:52 +08:00
// Check to see if the object exists
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
encodedDirectory := f.opt.Enc.FromStandardPath(f.rootDirectory)
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
req := s3.HeadObjectInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &f.rootBucket,
s3: use lib/encoder Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2018-11-02 20:15:30 +08:00
Key: &encodedDirectory,
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
err = f.pacer.Call(func() (bool, error) {
_, err = f.c.HeadObject(&req)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return f.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
Implement single file operations for all file systems
2014-05-06 02:52:52 +08:00
if err == nil {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
newRoot := path.Dir(f.root)
if newRoot == "." {
newRoot = ""
Implement single file operations for all file systems
2014-05-06 02:52:52 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
f.setRoot(newRoot)
Get rid of LimitedFs - FIXME needs docs on copying single files If remote:path points to a file make NewFs return a sentinel error fs.ErrorIsFile and an Fs which points to the parent. Use this to remove the LimitedFs and just add this file to the --files-from list. This means that server side operations can be used also. Fixes #518 Fixes #545
2016-06-22 01:01:53 +08:00
// return an error with an fs which points to the parent
return f, fs.ErrorIsFile
Implement single file operations for all file systems
2014-05-06 02:52:52 +08:00
}
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
// f.listMultipartUploads()
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return f, nil
}
Rename NewFsObject to NewObject
2016-06-26 04:58:34 +08:00
// Return an Object from a path
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
//
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
//If it can't be found it returns the error ErrorObjectNotFound.
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) newObjectWithInfo(ctx context.Context, remote string, info *s3.Object) (fs.Object, error) {
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
o := &Object{
fs: f,
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
remote: remote,
}
if info != nil {
// Set info but not meta
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
if info.LastModified == nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Failed to read last modified")
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
o.lastModified = time.Now()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
} else {
o.lastModified = *info.LastModified
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
o.etag = aws.StringValue(info.ETag)
o.bytes = aws.Int64Value(info.Size)
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
o.storageClass = aws.StringValue(info.StorageClass)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
} else {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
err := o.readMetaData(ctx) // reads info and meta, returning an error
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
if err != nil {
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
return nil, err
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
}
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
return o, nil
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
// NewObject finds the Object at remote. If it can't be found
// it returns the error fs.ErrorObjectNotFound.
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) NewObject(ctx context.Context, remote string) (fs.Object, error) {
return f.newObjectWithInfo(ctx, remote, nil)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
// Gets the bucket location
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
func (f *Fs) getBucketLocation(bucket string) (string, error) {
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
req := s3.GetBucketLocationInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &bucket,
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
}
var resp *s3.GetBucketLocationOutput
var err error
err = f.pacer.Call(func() (bool, error) {
resp, err = f.c.GetBucketLocation(&req)
return f.shouldRetry(err)
})
if err != nil {
return "", err
}
return s3.NormalizeBucketLocation(aws.StringValue(resp.LocationConstraint)), nil
}
// Updates the region for the bucket by reading the region from the
// bucket then updating the session.
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
func (f *Fs) updateRegionForBucket(bucket string) error {
region, err := f.getBucketLocation(bucket)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
if err != nil {
return errors.Wrap(err, "reading bucket location failed")
}
if aws.StringValue(f.c.Config.Endpoint) != "" {
return errors.Errorf("can't set region to %q as endpoint is set", region)
}
if aws.StringValue(f.c.Config.Region) == region {
return errors.Errorf("region is already %q - not updating", region)
}
// Make a new session with the new region
oldRegion := f.opt.Region
f.opt.Region = region
c, ses, err := s3Connection(&f.opt)
if err != nil {
return errors.Wrap(err, "creating new session failed")
}
f.c = c
f.ses = ses
fs.Logf(f, "Switched region to %q from %q", region, oldRegion)
return nil
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
// listFn is called from list to handle an object.
type listFn func(remote string, object *s3.Object, isDirectory bool) error
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// list lists the objects into the function supplied from
// the bucket and directory supplied. The remote has prefix
// removed from it and if addBucket is set then it adds the
// bucket to the start.
Add a directory parameter to Fs.List()
2016-04-24 04:46:52 +08:00
//
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
// Set recurse to read sub directories
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
func (f *Fs) list(ctx context.Context, bucket, directory, prefix string, addBucket bool, recurse bool, fn listFn) error {
if prefix != "" {
prefix += "/"
}
if directory != "" {
directory += "/"
Add a directory parameter to Fs.List()
2016-04-24 04:46:52 +08:00
}
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
delimiter := ""
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if !recurse {
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
delimiter = "/"
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
var marker *string
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
// URL encode the listings so we can use control characters in object names
// See: https://github.com/aws/aws-sdk-go/issues/1914
//
// However this doesn't work perfectly under Ceph (and hence DigitalOcean/Dreamhost) because
// it doesn't encode CommonPrefixes.
// See: https://tracker.ceph.com/issues/41870
//
// This does not work under IBM COS also: See https://github.com/rclone/rclone/issues/3345
// though maybe it does on some versions.
//
// This does work with minio but was only added relatively recently
// https://github.com/minio/minio/pull/7265
//
// So we enable only on providers we know supports it properly, all others can retry when a
// XML Syntax error is detected.
s3: Specify that Minio supports URL encoding in listings Thanks to @harshavardhana for pointing this out See #3934 for background
2020-02-09 17:45:53 +08:00
var urlEncodeListings = (f.opt.Provider == "AWS" || f.opt.Provider == "Wasabi" || f.opt.Provider == "Alibaba" || f.opt.Provider == "Minio")
s3: list an unlimited number of items - fixes #22
2015-02-11 01:58:29 +08:00
for {
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
// FIXME need to implement ALL loop
req := s3.ListObjectsInput{
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
Bucket: &bucket,
Delimiter: &delimiter,
Prefix: &directory,
s3: introduce list_chunk option for bucket listing The S3 ListObject API returns paginated bucket listings, with "MaxKeys" items for each GET call. The default value is 1000 entries, but for buckets with millions of objects it might make sense to request more elements per request, if the backend supports it. This commit adds a "list_chunk" option for the user to specify a lower or higher value. This commit does not add safe guards around this value - if a user decides to request a too large list, it might result in connection timeouts (on the server or client). In AWS S3, there is a fixed limit of 1000, some other services might have one too. In Ceph, this can be configured in RadosGW.
2019-12-26 19:05:00 +08:00
MaxKeys: &f.opt.ListChunk,
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
Marker: marker,
}
if urlEncodeListings {
req.EncodingType = aws.String(s3.EncodingTypeUrl)
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
var resp *s3.ListObjectsOutput
var err error
err = f.pacer.Call(func() (bool, error) {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
resp, err = f.c.ListObjectsWithContext(ctx, &req)
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
if err != nil && !urlEncodeListings {
if awsErr, ok := err.(awserr.RequestFailure); ok {
if origErr := awsErr.OrigErr(); origErr != nil {
if _, ok := origErr.(*xml.SyntaxError); ok {
// Retry the listing with URL encoding as there were characters that XML can't encode
urlEncodeListings = true
req.EncodingType = aws.String(s3.EncodingTypeUrl)
fs.Debugf(f, "Retrying listing because of characters which can't be XML encoded")
return true, err
}
}
}
}
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return f.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
s3: list an unlimited number of items - fixes #22
2015-02-11 01:58:29 +08:00
if err != nil {
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if awsErr, ok := err.(awserr.RequestFailure); ok {
if awsErr.StatusCode() == http.StatusNotFound {
err = fs.ErrorDirNotFound
}
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if f.rootBucket == "" {
// if listing from the root ignore wrong region requests returning
// empty directory
if reqErr, ok := err.(awserr.RequestFailure); ok {
// 301 if wrong region for bucket
if reqErr.StatusCode() == http.StatusMovedPermanently {
fs.Errorf(f, "Can't change region for bucket %q with no bucket specified", bucket)
return nil
}
}
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
return err
}
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if !recurse {
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
for _, commonPrefix := range resp.CommonPrefixes {
if commonPrefix.Prefix == nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(f, "Nil common prefix received")
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
continue
s3: list an unlimited number of items - fixes #22
2015-02-11 01:58:29 +08:00
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
remote := *commonPrefix.Prefix
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
if urlEncodeListings {
remote, err = url.QueryUnescape(remote)
if err != nil {
fs.Logf(f, "failed to URL decode %q in listing common prefix: %v", *commonPrefix.Prefix, err)
continue
}
s3: fix encoding for control characters - Fixes #3345
2019-07-23 19:24:10 +08:00
}
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
remote = f.opt.Enc.ToStandardPath(remote)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if !strings.HasPrefix(remote, prefix) {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(f, "Odd name received %q", remote)
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
continue
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
remote = remote[len(prefix):]
if addBucket {
remote = path.Join(bucket, remote)
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
if strings.HasSuffix(remote, "/") {
remote = remote[:len(remote)-1]
}
err = fn(remote, &s3.Object{Key: &remote}, true)
if err != nil {
return err
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
for _, object := range resp.Contents {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
remote := aws.StringValue(object.Key)
s3: only ask for URL encoded directory listings if we need them on Ceph This works around a bug in Ceph which doesn't encode CommonPrefixes when using URL encoded directory listings. See: https://tracker.ceph.com/issues/41870
2019-09-17 03:25:55 +08:00
if urlEncodeListings {
remote, err = url.QueryUnescape(remote)
if err != nil {
fs.Logf(f, "failed to URL decode %q in listing: %v", aws.StringValue(object.Key), err)
continue
}
s3: fix encoding for control characters - Fixes #3345
2019-07-23 19:24:10 +08:00
}
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
remote = f.opt.Enc.ToStandardPath(remote)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if !strings.HasPrefix(remote, prefix) {
fs.Logf(f, "Odd name received %q", remote)
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
continue
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
remote = remote[len(prefix):]
s3: ignore directory markers at the root also See: https://forum.rclone.org/t/issue-with-lsf-r-files-only-first-line-is-blank/15229/
2020-03-31 18:43:51 +08:00
isDirectory := remote == "" || strings.HasSuffix(remote, "/")
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if addBucket {
remote = path.Join(bucket, remote)
}
s3: ignore zero length directory markers - fixes #1621
2018-03-20 01:41:46 +08:00
// is this a directory marker?
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if isDirectory && object.Size != nil && *object.Size == 0 {
s3: ignore zero length directory markers - fixes #1621
2018-03-20 01:41:46 +08:00
continue // skip directory marker
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
err = fn(remote, object, false)
if err != nil {
return err
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
s3: list an unlimited number of items - fixes #22
2015-02-11 01:58:29 +08:00
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
if !aws.BoolValue(resp.IsTruncated) {
break
}
// Use NextMarker if set, otherwise use last Key
if resp.NextMarker == nil || *resp.NextMarker == "" {
s3: fix crash if a bad listing is received - fixes #1927 Caringo Swarm is returning a listing with IsTruncated set but no NextMarker and no Keys. Rclone doesn't know how to continue the listing at this point, so it returns an error rather than truncating the listing or risking a loop.
2017-12-21 00:40:41 +08:00
if len(resp.Contents) == 0 {
return errors.New("s3 protocol error: received listing with IsTruncated set, no NextMarker and no Contents")
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
marker = resp.Contents[len(resp.Contents)-1].Key
} else {
marker = resp.NextMarker
}
s3: fix url decoding of NextMarker - fixes #3799 Before this patch we were failing to URL decode the NextMarker when url encoding was used for the listing. The result of this was duplicated listings entries for directories with >1000 entries where the NextMarker was a file containing a space.
2019-12-12 01:23:52 +08:00
if urlEncodeListings {
*marker, err = url.QueryUnescape(*marker)
if err != nil {
return errors.Wrapf(err, "failed to URL decode NextMarker %q", *marker)
}
}
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
}
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
return nil
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
}
fs: rename BasicInfo to DirEntry
2017-06-30 17:54:14 +08:00
// Convert a list item into a DirEntry
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) itemToDirEntry(ctx context.Context, remote string, object *s3.Object, isDirectory bool) (fs.DirEntry, error) {
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if isDirectory {
size := int64(0)
if object.Size != nil {
size = *object.Size
}
Create fs.Directory interface and use it everywhere
2017-06-30 20:37:29 +08:00
d := fs.NewDir(remote, time.Time{}).SetSize(size)
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
return d, nil
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
o, err := f.newObjectWithInfo(ctx, remote, object)
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if err != nil {
return nil, err
}
return o, nil
}
// listDir lists files and directories to out
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
func (f *Fs) listDir(ctx context.Context, bucket, directory, prefix string, addBucket bool) (entries fs.DirEntries, err error) {
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
// List the objects and directories
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
err = f.list(ctx, bucket, directory, prefix, addBucket, false, func(remote string, object *s3.Object, isDirectory bool) error {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
entry, err := f.itemToDirEntry(ctx, remote, object, isDirectory)
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if err != nil {
return err
}
if entry != nil {
entries = append(entries, entry)
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
return nil
})
if err != nil {
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
return nil, err
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
azureblob,b2,gcs,qingstor,s3,swift: Don't check for bucket/container presense if listing was OK In a typical rclone copy to a bucket/container based remote, before this change we were doing a list, followed by a HEAD of the bucket to check it existed before doing the copy. The fact the list succeeded means the bucket exists so mark it OK at that point. Issue #1421
2018-03-01 20:11:34 +08:00
// bucket must be present if listing succeeded
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
f.cache.MarkOK(bucket)
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
return entries, nil
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
// listBuckets lists the buckets to out
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
func (f *Fs) listBuckets(ctx context.Context) (entries fs.DirEntries, err error) {
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
req := s3.ListBucketsInput{}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
var resp *s3.ListBucketsOutput
err = f.pacer.Call(func() (bool, error) {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
resp, err = f.c.ListBucketsWithContext(ctx, &req)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return f.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
if err != nil {
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
return nil, err
Refactor the List and ListDir interface Gives more accurate error propagation, control of depth of recursion and short circuit recursion where possible. Most of the the heavy lifting is done in the "fs" package, making file system implementations a bit simpler. This commit contains some code originally by Klaus Post. Fixes #316
2016-04-22 03:06:21 +08:00
}
for _, bucket := range resp.Buckets {
backend: adjust backends to have encoding parameter Fixes #3761 Fixes #3836 Fixes #3841
2020-01-15 01:33:35 +08:00
bucketName := f.opt.Enc.ToStandardName(aws.StringValue(bucket.Name))
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
f.cache.MarkOK(bucketName)
d := fs.NewDir(bucketName, aws.TimeValue(bucket.CreationDate))
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
entries = append(entries, d)
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
}
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
return entries, nil
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
// List the objects and directories in dir into entries. The
// entries can be returned in any order but should be for a
// complete directory.
//
// dir should be "" to list the root, and should not have
// trailing slashes.
//
// This should return ErrDirNotFound if the directory isn't
// found.
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) List(ctx context.Context, dir string) (entries fs.DirEntries, err error) {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, directory := f.split(dir)
if bucket == "" {
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
if directory != "" {
return nil, fs.ErrorListBucketRequired
}
return f.listBuckets(ctx)
s3: support sub-bucket paths
2014-05-06 01:25:32 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return f.listDir(ctx, bucket, directory, f.rootDirectory, f.rootBucket == "")
Implement fs.ListDir / lsd command
2013-01-24 06:43:20 +08:00
}
Add placeholder support for ListR interface. The ListR interface will be implemented by remotes that can do a recursive directory listing more efficiently than just recursing through the directories. These include the bucket based remotes.
2017-06-05 23:14:24 +08:00
// ListR lists the objects and directories of the Fs starting
// from dir recursively into out.
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
//
// dir should be "" to start from the root, and should not
// have trailing slashes.
//
// This should return ErrDirNotFound if the directory isn't
// found.
//
// It should call callback for each tranche of entries read.
// These need not be returned in any particular order. If
// callback returns an error then the listing will stop
// immediately.
//
// Don't implement this unless you have a more efficient way
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
// of listing recursively than doing a directory traversal.
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) ListR(ctx context.Context, dir string, callback fs.ListRCallback) (err error) {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, directory := f.split(dir)
Break the fs package up into smaller parts. The purpose of this is to make it easier to maintain and eventually to allow the rclone backends to be re-used in other projects without having to use the rclone configuration system. The new code layout is documented in CONTRIBUTING.
2018-01-13 00:30:54 +08:00
list := walk.NewListRHelper(callback)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
listR := func(bucket, directory, prefix string, addBucket bool) error {
return f.list(ctx, bucket, directory, prefix, addBucket, true, func(remote string, object *s3.Object, isDirectory bool) error {
entry, err := f.itemToDirEntry(ctx, remote, object, isDirectory)
if err != nil {
return err
}
return list.Add(entry)
})
}
if bucket == "" {
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
entries, err := f.listBuckets(ctx)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if err != nil {
return err
}
for _, entry := range entries {
err = list.Add(entry)
if err != nil {
return err
}
bucket := entry.Remote()
err = listR(bucket, "", f.rootDirectory, true)
if err != nil {
return err
}
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
// bucket must be present if listing succeeded
f.cache.MarkOK(bucket)
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
}
} else {
err = listR(bucket, directory, f.rootDirectory, f.rootBucket == "")
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
if err != nil {
return err
}
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
// bucket must be present if listing succeeded
f.cache.MarkOK(bucket)
Change List interface and add ListR optional interface This simplifies the implementation of remotes. The only required interface is now `List` which is a simple one level directory list. Optionally remotes may implement `ListR` if they have an efficient way of doing a recursive list.
2017-06-12 05:43:31 +08:00
}
return list.Flush()
Add placeholder support for ListR interface. The ListR interface will be implemented by remotes that can do a recursive directory listing more efficiently than just recursing through the directories. These include the bucket based remotes.
2017-06-05 23:14:24 +08:00
}
Rename NewFsObject to NewObject
2016-06-26 04:58:34 +08:00
// Put the Object into the bucket
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) Put(ctx context.Context, in io.Reader, src fs.ObjectInfo, options ...fs.OpenOption) (fs.Object, error) {
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
// Temporary Object under construction
fs := &Object{
fs: f,
Create separate interface for object information. Take out read-only information about a Fs in a separate struct to limit access. See discussion at #282.
2016-02-18 19:35:25 +08:00
remote: src.Remote(),
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
}
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
return fs, fs.Update(ctx, in, src, options...)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: allow streaming upload of files with unknown file size (see #1614)
2017-09-16 02:20:32 +08:00
// PutStream uploads to the remote path with the modTime given of indeterminate size
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) PutStream(ctx context.Context, in io.Reader, src fs.ObjectInfo, options ...fs.OpenOption) (fs.Object, error) {
return f.Put(ctx, in, src, options...)
s3: allow streaming upload of files with unknown file size (see #1614)
2017-09-16 02:20:32 +08:00
}
s3: Check if directory exists during Mkdir If you dont have privs to create a bucket in S3 but it exists, don't fail with an auth error, but detect that the mkdir was not needed and return successfully.
2016-02-24 09:58:55 +08:00
// Check if the bucket exists
s3: work around eventual consistency in bucket creation Deleting a bucket then testing its existence can give the wrong result. Work around by keeping a flag as to whether we have deleted the bucket.
2017-06-29 19:26:14 +08:00
//
// NB this can return incorrect results if called immediately after bucket deletion
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
func (f *Fs) bucketExists(ctx context.Context, bucket string) (bool, error) {
s3: Check if directory exists during Mkdir If you dont have privs to create a bucket in S3 but it exists, don't fail with an auth error, but detect that the mkdir was not needed and return successfully.
2016-02-24 09:58:55 +08:00
req := s3.HeadBucketInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &bucket,
s3: Check if directory exists during Mkdir If you dont have privs to create a bucket in S3 but it exists, don't fail with an auth error, but detect that the mkdir was not needed and return successfully.
2016-02-24 09:58:55 +08:00
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
err := f.pacer.Call(func() (bool, error) {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
_, err := f.c.HeadBucketWithContext(ctx, &req)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return f.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
s3: Check if directory exists during Mkdir If you dont have privs to create a bucket in S3 but it exists, don't fail with an auth error, but detect that the mkdir was not needed and return successfully.
2016-02-24 09:58:55 +08:00
if err == nil {
return true, nil
}
if err, ok := err.(awserr.RequestFailure); ok {
if err.StatusCode() == http.StatusNotFound {
return false, nil
}
}
return false, err
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Mkdir creates the bucket if it doesn't exist
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) Mkdir(ctx context.Context, dir string) error {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, _ := f.split(dir)
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
return f.makeBucket(ctx, bucket)
}
// makeBucket creates the bucket if it doesn't exist
func (f *Fs) makeBucket(ctx context.Context, bucket string) error {
s3: add --s3-no-check-bucket for minimising rclone transactions and perms Fixes #4449
2020-07-22 19:02:17 +08:00
if f.opt.NoCheckBucket {
return nil
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return f.cache.Create(bucket, func() error {
req := s3.CreateBucketInput{
Bucket: &bucket,
ACL: &f.opt.BucketACL,
s3: work around eventual consistency in bucket creation Deleting a bucket then testing its existence can give the wrong result. Work around by keeping a flag as to whether we have deleted the bucket.
2017-06-29 19:26:14 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
if f.opt.LocationConstraint != "" {
req.CreateBucketConfiguration = &s3.CreateBucketConfiguration{
LocationConstraint: &f.opt.LocationConstraint,
}
s3: work around eventual consistency in bucket creation Deleting a bucket then testing its existence can give the wrong result. Work around by keeping a flag as to whether we have deleted the bucket.
2017-06-29 19:26:14 +08:00
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
err := f.pacer.Call(func() (bool, error) {
_, err := f.c.CreateBucketWithContext(ctx, &req)
return f.shouldRetry(err)
})
if err == nil {
fs.Infof(f, "Bucket %q created with ACL %q", bucket, f.opt.BucketACL)
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
s3: fix detection of BucketAlreadyOwnedByYou and BucketAlreadyExists error This was being silently ignored until this commit e2bf91452a263ff8 s3: report errors on bucket creation (mkdir) correctly
2020-04-23 01:01:59 +08:00
if awsErr, ok := err.(awserr.Error); ok {
if code := awsErr.Code(); code == "BucketAlreadyOwnedByYou" || code == "BucketAlreadyExists" {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
err = nil
}
S3: Fix mkdir not to error if bucket exists and implement SetModTime
2013-01-09 06:31:16 +08:00
}
s3: report errors on bucket creation (mkdir) correctly Before this fix errors on bucket creation were being silently swallowed. See: https://forum.rclone.org/t/rclone-with-brand-new-aws-account-for-s3/15590
2020-04-15 20:13:13 +08:00
return err
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
}, func() (bool, error) {
return f.bucketExists(ctx, bucket)
})
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
swift, s3, googlecloudstorage: Don't delete the container/bucket if fs wasn't at root - fixes #172
2015-11-07 23:31:04 +08:00
// Rmdir deletes the bucket if the fs is at the root
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
//
// Returns an error if it isn't empty
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) Rmdir(ctx context.Context, dir string) error {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, directory := f.split(dir)
if bucket == "" || directory != "" {
swift, s3, googlecloudstorage: Don't delete the container/bucket if fs wasn't at root - fixes #172
2015-11-07 23:31:04 +08:00
return nil
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return f.cache.Remove(bucket, func() error {
req := s3.DeleteBucketInput{
Bucket: &bucket,
}
err := f.pacer.Call(func() (bool, error) {
_, err := f.c.DeleteBucketWithContext(ctx, &req)
return f.shouldRetry(err)
})
if err == nil {
fs.Infof(f, "Bucket %q deleted", bucket)
}
return err
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Precision of the remote
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (f *Fs) Precision() time.Duration {
Implement Precision interface in all filesystems * drive: add mime type on upload * drive: correct info on upload to fix crash * local: measure time precision by making a local file and using Chtimes * swift: move snet parameter here * core: add -modify-window and compute the optimum one * core: use modify window when checking times
2013-01-19 07:21:02 +08:00
return time.Nanosecond
}
s3: fix server side copy and set modtime on files with + in - fixes #2001 This was broken in 64ea94c1a4e9d7a104b393a17d2e883593e20809 when putting a work-around for Digital Ocean. PathEscape has now been adjusted so it works with both providers.
2018-01-23 18:50:50 +08:00
// pathEscape escapes s as for a URL path. It uses rest.URLPathEscape
// but also escapes '+' for S3 and Digital Ocean spaces compatibility
func pathEscape(s string) string {
return strings.Replace(rest.URLPathEscape(s), "+", "%2B", -1)
}
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
// copy does a server side copy
//
// It adds the boiler plate to the req passed in and calls the s3
// method
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
func (f *Fs) copy(ctx context.Context, req *s3.CopyObjectInput, dstBucket, dstPath, srcBucket, srcPath string, srcSize int64) error {
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
req.Bucket = &dstBucket
req.ACL = &f.opt.ACL
req.Key = &dstPath
source := pathEscape(path.Join(srcBucket, srcPath))
req.CopySource = &source
if f.opt.ServerSideEncryption != "" {
req.ServerSideEncryption = &f.opt.ServerSideEncryption
}
if f.opt.SSEKMSKeyID != "" {
req.SSEKMSKeyId = &f.opt.SSEKMSKeyID
}
if req.StorageClass == nil && f.opt.StorageClass != "" {
req.StorageClass = &f.opt.StorageClass
}
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
s3: add --s3-copy-cutoff for size to switch to multipart copy Before this change we used the same (relatively low limits) for server side copy as we did for multipart uploads. It doesn't make sense to use the same limits since no data is being downloaded or uploaded for a server side copy. This change introduces a new parameter --s3-copy-cutoff to control when the switch from single to multipart server size copy happens and defaults it to the maximum 5GB. This makes server side copies much more efficient. It also fixes the erroneous error when trying to set the modification time of a file bigger than 5GB. See #3778
2019-12-03 01:14:57 +08:00
if srcSize >= int64(f.opt.CopyCutoff) {
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
return f.copyMultipart(ctx, req, dstBucket, dstPath, srcBucket, srcPath, srcSize)
}
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
return f.pacer.Call(func() (bool, error) {
_, err := f.c.CopyObjectWithContext(ctx, req)
return f.shouldRetry(err)
})
}
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
func calculateRange(partSize, partIndex, numParts, totalSize int64) string {
start := partIndex * partSize
var ends string
if partIndex == numParts-1 {
s3: fix multipart copy - fixes #3778 Before this change multipart copies were giving the error Range specified is not valid for source object of size This was due to an off by one error in the range source introduced in 7b1274e29ac51231 "s3: support for multipart copy"
2019-12-03 01:00:54 +08:00
if totalSize >= 1 {
ends = strconv.FormatInt(totalSize-1, 10)
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
}
} else {
ends = strconv.FormatInt(start+partSize-1, 10)
}
return fmt.Sprintf("bytes=%v-%v", start, ends)
}
func (f *Fs) copyMultipart(ctx context.Context, req *s3.CopyObjectInput, dstBucket, dstPath, srcBucket, srcPath string, srcSize int64) (err error) {
var cout *s3.CreateMultipartUploadOutput
if err := f.pacer.Call(func() (bool, error) {
var err error
cout, err = f.c.CreateMultipartUploadWithContext(ctx, &s3.CreateMultipartUploadInput{
Bucket: &dstBucket,
Key: &dstPath,
})
return f.shouldRetry(err)
}); err != nil {
return err
}
uid := cout.UploadId
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
defer atexit.OnError(&err, func() {
// Try to abort the upload, but ignore the error.
fs.Debugf(nil, "Cancelling multipart copy")
_ = f.pacer.Call(func() (bool, error) {
_, err := f.c.AbortMultipartUploadWithContext(context.Background(), &s3.AbortMultipartUploadInput{
Bucket: &dstBucket,
Key: &dstPath,
UploadId: uid,
RequestPayer: req.RequestPayer,
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
})
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
return f.shouldRetry(err)
})
})()
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
s3: add --s3-copy-cutoff for size to switch to multipart copy Before this change we used the same (relatively low limits) for server side copy as we did for multipart uploads. It doesn't make sense to use the same limits since no data is being downloaded or uploaded for a server side copy. This change introduces a new parameter --s3-copy-cutoff to control when the switch from single to multipart server size copy happens and defaults it to the maximum 5GB. This makes server side copies much more efficient. It also fixes the erroneous error when trying to set the modification time of a file bigger than 5GB. See #3778
2019-12-03 01:14:57 +08:00
partSize := int64(f.opt.CopyCutoff)
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
numParts := (srcSize-1)/partSize + 1
var parts []*s3.CompletedPart
for partNum := int64(1); partNum <= numParts; partNum++ {
if err := f.pacer.Call(func() (bool, error) {
partNum := partNum
uploadPartReq := &s3.UploadPartCopyInput{
Bucket: &dstBucket,
Key: &dstPath,
PartNumber: &partNum,
UploadId: uid,
CopySourceRange: aws.String(calculateRange(partSize, partNum-1, numParts, srcSize)),
// Args copy from req
CopySource: req.CopySource,
CopySourceIfMatch: req.CopySourceIfMatch,
CopySourceIfModifiedSince: req.CopySourceIfModifiedSince,
CopySourceIfNoneMatch: req.CopySourceIfNoneMatch,
CopySourceIfUnmodifiedSince: req.CopySourceIfUnmodifiedSince,
CopySourceSSECustomerAlgorithm: req.CopySourceSSECustomerAlgorithm,
CopySourceSSECustomerKey: req.CopySourceSSECustomerKey,
CopySourceSSECustomerKeyMD5: req.CopySourceSSECustomerKeyMD5,
RequestPayer: req.RequestPayer,
SSECustomerAlgorithm: req.SSECustomerAlgorithm,
SSECustomerKey: req.SSECustomerKey,
SSECustomerKeyMD5: req.SSECustomerKeyMD5,
}
uout, err := f.c.UploadPartCopyWithContext(ctx, uploadPartReq)
if err != nil {
return f.shouldRetry(err)
}
parts = append(parts, &s3.CompletedPart{
PartNumber: &partNum,
ETag: uout.CopyPartResult.ETag,
})
return false, nil
}); err != nil {
return err
}
}
return f.pacer.Call(func() (bool, error) {
_, err := f.c.CompleteMultipartUploadWithContext(ctx, &s3.CompleteMultipartUploadInput{
Bucket: &dstBucket,
Key: &dstPath,
MultipartUpload: &s3.CompletedMultipartUpload{
Parts: parts,
},
RequestPayer: req.RequestPayer,
UploadId: uid,
})
return f.shouldRetry(err)
})
}
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
// Copy src to this remote using server side copy operations.
//
// This is stored with the remote path given
//
// It returns the destination Object and a possible error
//
// Will only be called if src.Fs().Name() == f.Name()
//
// If it isn't possible then return fs.ErrorCantCopy
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (f *Fs) Copy(ctx context.Context, src fs.Object, remote string) (fs.Object, error) {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
dstBucket, dstPath := f.split(remote)
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
err := f.makeBucket(ctx, dstBucket)
swift, s3, gcs: create container if necessary on server side copy
2017-06-29 04:14:53 +08:00
if err != nil {
return nil, err
}
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
srcObj, ok := src.(*Object)
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
if !ok {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Debugf(src, "Can't copy - not same remote type")
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
return nil, fs.ErrorCantCopy
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
srcBucket, srcPath := srcObj.split()
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
req := s3.CopyObjectInput{
MetadataDirective: aws.String(s3.MetadataDirectiveCopy),
}
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
err = f.copy(ctx, &req, dstBucket, dstPath, srcBucket, srcPath, srcObj.Size())
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
if err != nil {
return nil, err
}
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
return f.NewObject(ctx, remote)
Implement server side copies if possible - fixes #99 Add optional fs.Copier interface Implemented for * swift * s3 * drive * dropbox * google cloud storage
2015-02-15 02:48:08 +08:00
}
Add support for multiple hash types. Add support for multiple hash types with negotiation of common hash types for comparison. Manually rebased version of #277 (see discussion there)
2016-01-11 20:39:33 +08:00
// Hashes returns the supported hash sets.
Break the fs package up into smaller parts. The purpose of this is to make it easier to maintain and eventually to allow the rclone backends to be re-used in other projects without having to use the rclone configuration system. The new code layout is documented in CONTRIBUTING.
2018-01-13 00:30:54 +08:00
func (f *Fs) Hashes() hash.Set {
config/hash: rename more symbols after factoring into own package
2018-01-19 04:27:52 +08:00
return hash.Set(hash.MD5)
Add support for multiple hash types. Add support for multiple hash types with negotiation of common hash types for comparison. Manually rebased version of #277 (see discussion there)
2016-01-11 20:39:33 +08:00
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
func (f *Fs) getMemoryPool(size int64) *pool.Pool {
s3: use single memory pool Previously we had a map of pools for different chunk sizes. In practice the mapping is not very useful and requires a lock. Pools of size other that ChunkSize can only happen when we have a huge file (over 10k * ChunkSize). We need to have a bunch of identically sized huge files. In such case most likely ChunkSize should be increased. The mapping and its lock is replaced with a single initialised pool for ChunkSize, in other cases pool is allocated and freed on per file basis.
2020-04-09 18:18:58 +08:00
if size == int64(f.opt.ChunkSize) {
return f.pool
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
s3: use single memory pool Previously we had a map of pools for different chunk sizes. In practice the mapping is not very useful and requires a lock. Pools of size other that ChunkSize can only happen when we have a huge file (over 10k * ChunkSize). We need to have a bunch of identically sized huge files. In such case most likely ChunkSize should be increased. The mapping and its lock is replaced with a single initialised pool for ChunkSize, in other cases pool is allocated and freed on per file basis.
2020-04-09 18:18:58 +08:00
return pool.New(
time.Duration(f.opt.MemoryPoolFlushTime),
int(size),
f.opt.UploadConcurrency*fs.Config.Transfers,
f.opt.MemoryPoolUseMmap,
)
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
}
s3: implement link sharing with PublicLink
2020-06-05 05:03:12 +08:00
// PublicLink generates a public link to the remote path (usually readable by anyone)
func (f *Fs) PublicLink(ctx context.Context, remote string, expire fs.Duration, unlink bool) (link string, err error) {
s3: Cap expiry duration to 1 Week and return error when sharing dir
2020-06-19 00:50:50 +08:00
if strings.HasSuffix(remote, "/") {
return "", fs.ErrorCantShareDirectories
}
s3: implement link sharing with PublicLink
2020-06-05 05:03:12 +08:00
if _, err := f.NewObject(ctx, remote); err != nil {
return "", err
}
s3: Cap expiry duration to 1 Week and return error when sharing dir
2020-06-19 00:50:50 +08:00
if expire > maxExpireDuration {
fs.Logf(f, "Public Link: Reducing expiry to %v as %v is greater than the max time allowed", maxExpireDuration, expire)
expire = maxExpireDuration
}
s3: implement link sharing with PublicLink
2020-06-05 05:03:12 +08:00
bucket, bucketPath := f.split(remote)
httpReq, _ := f.c.GetObjectRequest(&s3.GetObjectInput{
Bucket: &bucket,
Key: &bucketPath,
})
return httpReq.Presign(time.Duration(expire))
}
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
var commandHelp = []fs.CommandHelp{{
Name: "restore",
Short: "Restore objects from GLACIER to normal storage",
Long: `This command can be used to restore one or more objects from GLACIER
to normal storage.
Usage Examples:
rclone backend restore s3:bucket/path/to/object [-o priority=PRIORITY] [-o lifetime=DAYS]
rclone backend restore s3:bucket/path/to/directory [-o priority=PRIORITY] [-o lifetime=DAYS]
rclone backend restore s3:bucket [-o priority=PRIORITY] [-o lifetime=DAYS]
This flag also obeys the filters. Test first with -i/--interactive or --dry-run flags
rclone -i backend restore --include "*.txt" s3:bucket/path -o priority=Standard
All the objects shown will be marked for restore, then
rclone backend restore --include "*.txt" s3:bucket/path -o priority=Standard
It returns a list of status dictionaries with Remote and Status
keys. The Status will be OK if it was successfull or an error message
if not.
[
{
"Status": "OK",
"Path": "test.txt"
},
{
"Status": "OK",
"Path": "test/file4.txt"
}
]
`,
Opts: map[string]string{
"priority": "Priority of restore: Standard|Expedited|Bulk",
"lifetime": "Lifetime of the active copy in days",
"description": "The optional description for the job.",
},
s3: implement cleanup and backend command to list & remove multipart uploads This implements `rclone cleanup` to remove multipart uploads over 24 hours old. It also implements the backend command `list-multipart-uploads` to see which ones are available and `cleanup` to delete them with a configurable expiry interval. See #4302
2020-06-25 23:11:05 +08:00
}, {
Name: "list-multipart-uploads",
Short: "List the unfinished multipart uploads",
Long: `This command lists the unfinished multipart uploads in JSON format.
rclone backend list-multipart s3:bucket/path/to/object
It returns a dictionary of buckets with values as lists of unfinished
multipart uploads.
You can call it with no bucket in which case it lists all bucket, with
a bucket or with a bucket and path.
{
"rclone": [
{
"Initiated": "2020-06-26T14:20:36Z",
"Initiator": {
"DisplayName": "XXX",
"ID": "arn:aws:iam::XXX:user/XXX"
},
"Key": "KEY",
"Owner": {
"DisplayName": null,
"ID": "XXX"
},
"StorageClass": "STANDARD",
"UploadId": "XXX"
}
],
"rclone-1000files": [],
"rclone-dst": []
}
`,
}, {
Name: "cleanup",
Short: "Remove unfinished multipart uploads.",
Long: `This command removes unfinished multipart uploads of age greater than
max-age which defaults to 24 hours.
Note that you can use -i/--dry-run with this command to see what it
would do.
rclone backend cleanup s3:bucket/path/to/object
rclone backend cleanup -o max-age=7w s3:bucket/path/to/object
Durations are parsed as per the rest of rclone, 2h, 7d, 7w etc.
`,
Opts: map[string]string{
"max-age": "Max age of upload to delete",
},
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
}}
// Command the backend to run a named command
//
// The command run is name
// args may be used to read arguments from
// opts may be used to read optional arguments from
//
// The result should be capable of being JSON encoded
// If it is a string or a []string it will be shown to the user
// otherwise it will be JSON encoded and shown to the user like that
func (f *Fs) Command(ctx context.Context, name string, arg []string, opt map[string]string) (out interface{}, err error) {
switch name {
case "restore":
req := s3.RestoreObjectInput{
//Bucket: &f.rootBucket,
//Key: &encodedDirectory,
RestoreRequest: &s3.RestoreRequest{},
}
if lifetime := opt["lifetime"]; lifetime != "" {
ilifetime, err := strconv.ParseInt(lifetime, 10, 64)
if err != nil {
return nil, errors.Wrap(err, "bad lifetime")
}
req.RestoreRequest.Days = &ilifetime
}
if priority := opt["priority"]; priority != "" {
req.RestoreRequest.GlacierJobParameters = &s3.GlacierJobParameters{
Tier: &priority,
}
}
if description := opt["description"]; description != "" {
req.RestoreRequest.Description = &description
}
type status struct {
Status string
Remote string
}
var (
outMu sync.Mutex
out = []status{}
)
err = operations.ListFn(ctx, f, func(obj fs.Object) {
// Remember this is run --checkers times concurrently
o, ok := obj.(*Object)
st := status{Status: "OK", Remote: obj.Remote()}
defer func() {
outMu.Lock()
out = append(out, st)
outMu.Unlock()
}()
if operations.SkipDestructive(ctx, obj, "restore") {
return
}
if !ok {
st.Status = "Not an S3 object"
return
}
bucket, bucketPath := o.split()
reqCopy := req
reqCopy.Bucket = &bucket
reqCopy.Key = &bucketPath
err = f.pacer.Call(func() (bool, error) {
_, err = f.c.RestoreObject(&reqCopy)
return f.shouldRetry(err)
})
if err != nil {
st.Status = err.Error()
}
})
if err != nil {
return out, err
}
return out, nil
s3: implement cleanup and backend command to list & remove multipart uploads This implements `rclone cleanup` to remove multipart uploads over 24 hours old. It also implements the backend command `list-multipart-uploads` to see which ones are available and `cleanup` to delete them with a configurable expiry interval. See #4302
2020-06-25 23:11:05 +08:00
case "list-multipart-uploads":
return f.listMultipartUploadsAll(ctx)
case "cleanup":
maxAge := 24 * time.Hour
if opt["max-age"] != "" {
maxAge, err = fs.ParseDuration(opt["max-age"])
if err != nil {
return nil, errors.Wrap(err, "bad max-age")
}
}
return nil, f.cleanUp(ctx, maxAge)
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
default:
return nil, fs.ErrorCommandNotFound
}
}
s3: implement cleanup and backend command to list & remove multipart uploads This implements `rclone cleanup` to remove multipart uploads over 24 hours old. It also implements the backend command `list-multipart-uploads` to see which ones are available and `cleanup` to delete them with a configurable expiry interval. See #4302
2020-06-25 23:11:05 +08:00
// listMultipartUploads lists all outstanding multipart uploads for (bucket, key)
//
// Note that rather lazily we treat key as a prefix so it matches
// directories and objects. This could suprise the user if they ask
// for "dir" and it returns "dirKey"
func (f *Fs) listMultipartUploads(ctx context.Context, bucket, key string) (uploads []*s3.MultipartUpload, err error) {
var (
keyMarker *string
uploadIDMarker *string
)
uploads = []*s3.MultipartUpload{}
for {
req := s3.ListMultipartUploadsInput{
Bucket: &bucket,
MaxUploads: &f.opt.ListChunk,
KeyMarker: keyMarker,
UploadIdMarker: uploadIDMarker,
Prefix: &key,
}
var resp *s3.ListMultipartUploadsOutput
err = f.pacer.Call(func() (bool, error) {
resp, err = f.c.ListMultipartUploads(&req)
return f.shouldRetry(err)
})
if err != nil {
return nil, errors.Wrapf(err, "list multipart uploads bucket %q key %q", bucket, key)
}
uploads = append(uploads, resp.Uploads...)
if !aws.BoolValue(resp.IsTruncated) {
break
}
keyMarker = resp.NextKeyMarker
uploadIDMarker = resp.NextUploadIdMarker
}
return uploads, nil
}
func (f *Fs) listMultipartUploadsAll(ctx context.Context) (uploadsMap map[string][]*s3.MultipartUpload, err error) {
uploadsMap = make(map[string][]*s3.MultipartUpload)
bucket, directory := f.split("")
if bucket != "" {
uploads, err := f.listMultipartUploads(ctx, bucket, directory)
if err != nil {
return uploadsMap, err
}
uploadsMap[bucket] = uploads
return uploadsMap, nil
}
entries, err := f.listBuckets(ctx)
if err != nil {
return uploadsMap, err
}
for _, entry := range entries {
bucket := entry.Remote()
uploads, listErr := f.listMultipartUploads(ctx, bucket, "")
if listErr != nil {
err = listErr
fs.Errorf(f, "%v", err)
}
uploadsMap[bucket] = uploads
}
return uploadsMap, err
}
// cleanUpBucket removes all pending multipart uploads for a given bucket over the age of maxAge
func (f *Fs) cleanUpBucket(ctx context.Context, bucket string, maxAge time.Duration, uploads []*s3.MultipartUpload) (err error) {
fs.Infof(f, "cleaning bucket %q of pending multipart uploads older than %v", bucket, maxAge)
for _, upload := range uploads {
if upload.Initiated != nil && upload.Key != nil && upload.UploadId != nil {
age := time.Since(*upload.Initiated)
what := fmt.Sprintf("pending multipart upload for bucket %q key %q dated %v (%v ago)", bucket, *upload.Key, upload.Initiated, age)
if age > maxAge {
fs.Infof(f, "removing %s", what)
if operations.SkipDestructive(ctx, what, "remove pending upload") {
continue
}
req := s3.AbortMultipartUploadInput{
Bucket: &bucket,
UploadId: upload.UploadId,
Key: upload.Key,
}
_, abortErr := f.c.AbortMultipartUpload(&req)
if abortErr != nil {
err = errors.Wrapf(abortErr, "failed to remove %s", what)
fs.Errorf(f, "%v", err)
}
} else {
fs.Debugf(f, "ignoring %s", what)
}
}
}
return err
}
// CleanUp removes all pending multipart uploads
func (f *Fs) cleanUp(ctx context.Context, maxAge time.Duration) (err error) {
uploadsMap, err := f.listMultipartUploadsAll(ctx)
if err != nil {
return err
}
for bucket, uploads := range uploadsMap {
cleanErr := f.cleanUpBucket(ctx, bucket, maxAge, uploads)
if err != nil {
fs.Errorf(f, "Failed to cleanup bucket %q: %v", bucket, cleanErr)
err = cleanErr
}
}
return err
}
// CleanUp removes all pending multipart uploads older than 24 hours
func (f *Fs) CleanUp(ctx context.Context) (err error) {
return f.cleanUp(ctx, 24*time.Hour)
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// ------------------------------------------------------------
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Fs returns the parent Fs
Create separate interface for object information. Take out read-only information about a Fs in a separate struct to limit access. See discussion at #282.
2016-02-18 19:35:25 +08:00
func (o *Object) Fs() fs.Info {
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
return o.fs
Factor the generic code into fs and add some more intefaces
2014-03-29 01:56:04 +08:00
}
// Return a string version
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (o *Object) String() string {
Factor the generic code into fs and add some more intefaces
2014-03-29 01:56:04 +08:00
if o == nil {
return "<nil>"
}
return o.remote
}
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Remote returns the remote path
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (o *Object) Remote() string {
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
return o.remote
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: ignore etags from multipart uploads which aren't md5sums - fixes #56
2015-05-09 17:37:43 +08:00
var matchMd5 = regexp.MustCompile(`^[0-9a-f]{32}$`)
Add support for multiple hash types. Add support for multiple hash types with negotiation of common hash types for comparison. Manually rebased version of #277 (see discussion there)
2016-01-11 20:39:33 +08:00
// Hash returns the Md5sum of an object returning a lowercase hex string
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) Hash(ctx context.Context, t hash.Type) (string, error) {
config/hash: rename more symbols after factoring into own package
2018-01-19 04:27:52 +08:00
if t != hash.MD5 {
return "", hash.ErrUnsupported
Add support for multiple hash types. Add support for multiple hash types with negotiation of common hash types for comparison. Manually rebased version of #277 (see discussion there)
2016-01-11 20:39:33 +08:00
}
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
hash := strings.Trim(strings.ToLower(o.etag), `"`)
s3: ignore etags from multipart uploads which aren't md5sums - fixes #56
2015-05-09 17:37:43 +08:00
// Check the etag is a valid md5sum
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
if !matchMd5.MatchString(hash) {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
err := o.readMetaData(ctx)
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
if err != nil {
return "", err
}
if md5sum, ok := o.meta[metaMD5Hash]; ok {
md5sumBytes, err := base64.StdEncoding.DecodeString(*md5sum)
if err != nil {
return "", err
}
hash = hex.EncodeToString(md5sumBytes)
} else {
hash = ""
}
s3: ignore etags from multipart uploads which aren't md5sums - fixes #56
2015-05-09 17:37:43 +08:00
}
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
return hash, nil
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
// Size returns the size of an object in bytes
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (o *Object) Size() int64 {
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
return o.bytes
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
// readMetaData gets the metadata if it hasn't already been fetched
//
// it also sets the info
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) readMetaData(ctx context.Context) (err error) {
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
if o.meta != nil {
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return nil
}
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, bucketPath := o.split()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
req := s3.HeadObjectInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &bucket,
Key: &bucketPath,
s3: make reading metadata more reliable to work around eventual consistency problems
2014-07-29 05:32:15 +08:00
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
var resp *s3.HeadObjectOutput
err = o.fs.pacer.Call(func() (bool, error) {
var err error
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
resp, err = o.fs.c.HeadObjectWithContext(ctx, &req)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return o.fs.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
if err != nil {
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
if awsErr, ok := err.(awserr.RequestFailure); ok {
if awsErr.StatusCode() == http.StatusNotFound {
s3: fix upload of single files into buckets without create permission Before this change, attempting to upload a single file into an s3 bucket which did not have create permission gave AccessDenied: Access Denied error when it tried to create the bucket. This was masked until e2bf91452a263ff8babb3e5e153a2d2c7b4b3719 was fixed. This fix marks the bucket as OK if a fetch on an object indicates it is OK. This stops rclone thinking it has to create the bucket in the first place. Fixes #4297
2020-06-02 21:29:42 +08:00
// NotFound indicates bucket was OK
// NoSuchBucket would be returned if bucket was bad
if awsErr.Code() == "NotFound" {
o.fs.cache.MarkOK(bucket)
}
Make NewObject return an error * make it return an error * make a canonical error fs.ErrorNotFound * make a test for it * remove logs/debugs of error
2016-06-26 04:23:20 +08:00
return fs.ErrorObjectNotFound
}
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return err
}
s3: fix upload of single files into buckets without create permission Before this change, attempting to upload a single file into an s3 bucket which did not have create permission gave AccessDenied: Access Denied error when it tried to create the bucket. This was masked until e2bf91452a263ff8babb3e5e153a2d2c7b4b3719 was fixed. This fix marks the bucket as OK if a fetch on an object indicates it is OK. This stops rclone thinking it has to create the bucket in the first place. Fixes #4297
2020-06-02 21:29:42 +08:00
o.fs.cache.MarkOK(bucket)
s3: Treat missing Content-Length as 0 for some ceph installations
2014-05-16 23:27:53 +08:00
var size int64
// Ignore missing Content-Length assuming it is 0
// Some versions of ceph do this due their apache proxies
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
if resp.ContentLength != nil {
size = *resp.ContentLength
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
o.etag = aws.StringValue(resp.ETag)
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
o.bytes = size
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
o.meta = resp.Metadata
s3: fix nil pointer reference if no metadata returned for object Fixes #3651 Fixes #3652
2019-10-23 15:16:22 +08:00
if o.meta == nil {
o.meta = map[string]*string{}
}
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
o.storageClass = aws.StringValue(resp.StorageClass)
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
if resp.LastModified == nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Failed to read last modified from HEAD: %v", err)
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
o.lastModified = time.Now()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
} else {
o.lastModified = *resp.LastModified
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
o.mimeType = aws.StringValue(resp.ContentType)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return nil
}
// ModTime returns the modification time of the object
//
// It attempts to read the objects mtime and if that isn't present the
// LastModified returned in the http headers
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) ModTime(ctx context.Context) time.Time {
s3,swift: Add --use-server-modtime `--use-server-modtime` stops s3 and swift retrieving the modtime from metadata which enables a fast sync mode with the `--update` flag.
2018-04-13 20:32:17 +08:00
if fs.Config.UseServerModTime {
return o.lastModified
}
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
err := o.readMetaData(ctx)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
if err != nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Failed to read metadata: %v", err)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return time.Now()
}
// read mtime out of metadata if available
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
d, ok := o.meta[metaMtime]
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
if !ok || d == nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
// fs.Debugf(o, "No metadata")
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
return o.lastModified
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
modTime, err := swift.FloatStringToTime(*d)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
if err != nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Failed to read mtime from object: %v", err)
Refactor into sub filesystems
2013-06-28 03:13:07 +08:00
return o.lastModified
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
return modTime
}
Fix golint warnings
2015-09-23 01:47:16 +08:00
// SetModTime sets the modification time of the local fs object
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) SetModTime(ctx context.Context, modTime time.Time) error {
err := o.readMetaData(ctx)
S3: Fix mkdir not to error if bucket exists and implement SetModTime
2013-01-09 06:31:16 +08:00
if err != nil {
Change the interface of SetModTime to return an error - #348
2016-03-22 23:07:10 +08:00
return err
S3: Fix mkdir not to error if bucket exists and implement SetModTime
2013-01-09 06:31:16 +08:00
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
o.meta[metaMtime] = aws.String(swift.TimeToFloatString(modTime))
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
// Can't update metadata here, so return this error to force a recopy
if o.storageClass == "GLACIER" || o.storageClass == "DEEP_ARCHIVE" {
return fs.ErrorCantSetModTime
}
s3: Fix corrupting Content-Type on mod time update This fixes an issue where updating the modification time resets the content-type to the S3 default of binary/octet-stream which breaks static websites that expect an html file to have a content-type of text/html.
2016-01-02 16:58:48 +08:00
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
// Copy the object to itself to update the metadata
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, bucketPath := o.split()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
req := s3.CopyObjectInput{
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
ContentType: aws.String(fs.MimeType(ctx, o)), // Guess the content type
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
Metadata: o.meta,
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
MetadataDirective: aws.String(s3.MetadataDirectiveReplace), // replace metadata with that passed in
s3: make SetModTime work for GLACIER while syncing - Fixes #3224 Before this change rclone would fail with Failed to set modification time: InvalidObjectState: Operation is not valid for the source object's storage class when attempting to set the modification time of an object in GLACIER. After this change rclone will re-upload the object as part of a sync if it needs to change the modification time. See: https://forum.rclone.org/t/suspected-bug-in-s3-or-compatible-sync-logic-to-glacier/10187
2019-06-03 22:28:19 +08:00
}
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
return o.fs.copy(ctx, &req, bucket, bucketPath, bucket, bucketPath, o.bytes)
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Fix golint warnings
2015-09-23 01:47:16 +08:00
// Storable raturns a boolean indicating if this object is storable
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
func (o *Object) Storable() bool {
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
return true
}
// Open an object for read
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) Open(ctx context.Context, options ...fs.OpenOption) (in io.ReadCloser, err error) {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, bucketPath := o.split()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
req := s3.GetObjectInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &bucket,
Key: &bucketPath,
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
if o.fs.opt.SSECustomerAlgorithm != "" {
req.SSECustomerAlgorithm = &o.fs.opt.SSECustomerAlgorithm
}
if o.fs.opt.SSECustomerKey != "" {
req.SSECustomerKey = &o.fs.opt.SSECustomerKey
}
if o.fs.opt.SSECustomerKeyMD5 != "" {
req.SSECustomerKeyMD5 = &o.fs.opt.SSECustomerKeyMD5
}
s3: add support for HTTPOption
2020-02-10 17:02:04 +08:00
httpReq, resp := o.fs.c.GetObjectRequest(&req)
azureblob/b2/dropbox/gcs/koofr/qingstor/s3: fix 0 length files In 0386d22cc946dfc5 we introduced a test for 0 length files read the way mount does. This test failed on these backends which we fix up here.
2019-08-06 22:18:08 +08:00
fs.FixRangeOption(options, o.bytes)
Add options for Open and implement Range for all remotes
2016-09-10 18:29:57 +08:00
for _, option := range options {
switch option.(type) {
case *fs.RangeOption, *fs.SeekOption:
_, value := option.Header()
req.Range = &value
s3: add support for HTTPOption
2020-02-10 17:02:04 +08:00
case *fs.HTTPOption:
key, value := option.Header()
httpReq.HTTPRequest.Header.Add(key, value)
Add options for Open and implement Range for all remotes
2016-09-10 18:29:57 +08:00
default:
if option.Mandatory() {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Unsupported mandatory option: %v", option)
Add options for Open and implement Range for all remotes
2016-09-10 18:29:57 +08:00
}
}
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
err = o.fs.pacer.Call(func() (bool, error) {
var err error
s3: add support for HTTPOption
2020-02-10 17:02:04 +08:00
httpReq.HTTPRequest = httpReq.HTTPRequest.WithContext(ctx)
err = httpReq.Send()
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return o.fs.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
s3: Error message for objects in glacier
2017-09-09 20:02:26 +08:00
if err, ok := err.(awserr.RequestFailure); ok {
if err.Code() == "InvalidObjectState" {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
return nil, errors.Errorf("Object in GLACIER, restore first: bucket=%q, key=%q", bucket, bucketPath)
s3: Error message for objects in glacier
2017-09-09 20:02:26 +08:00
}
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
if err != nil {
return nil, err
}
return resp.Body, nil
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
s3: Reduce memory usage streaming files by reducing max stream upload size Before this change rclone would allow the user to stream (eg with rclone mount, rclone rcat or uploading google photos or docs) 5TB files. This meant that rclone allocated 4 * 525 MB buffers per transfer which is way too much memory by default. This change makes rclone use the configured chunk size for streamed uploads. This is 5MB by default which means that rclone can stream upload files up to 48GB by default staying below the 10,000 chunks limit. This can be increased with --s3-chunk-size if necessary. If rclone detects that a file is being streamed to s3 it will make a single NOTICE level log stating the limitation. This fixes the enormous memory usage. Fixes #3568 See: https://forum.rclone.org/t/how-much-memory-does-rclone-need/12743
2019-11-06 18:41:03 +08:00
var warnStreamUpload sync.Once
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
func (o *Object) uploadMultipart(ctx context.Context, req *s3.PutObjectInput, size int64, in io.Reader) (err error) {
f := o.fs
// make concurrency machinery
concurrency := f.opt.UploadConcurrency
if concurrency < 1 {
concurrency = 1
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
tokens := pacer.NewTokenDispenser(concurrency)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
uploadParts := f.opt.MaxUploadParts
if uploadParts < 1 {
uploadParts = 1
} else if uploadParts > maxUploadParts {
uploadParts = maxUploadParts
}
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
// calculate size of parts
partSize := int(f.opt.ChunkSize)
// size can be -1 here meaning we don't know the size of the incoming file. We use ChunkSize
// buffers here (default 5MB). With a maximum number of parts (10,000) this will be a file of
// 48GB which seems like a not too unreasonable limit.
if size == -1 {
warnStreamUpload.Do(func() {
fs.Logf(f, "Streaming uploads using chunk size %v will have maximum file size of %v",
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
f.opt.ChunkSize, fs.SizeSuffix(int64(partSize)*uploadParts))
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
})
} else {
// Adjust partSize until the number of parts is small enough.
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
if size/int64(partSize) >= uploadParts {
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
// Calculate partition size rounded up to the nearest MB
s3: add `max_upload_parts` support - fixes #4159 * s3: add `max_upload_parts` support This allows to configure a maximum amount of chunks used to upload file: - Support Scaleway which has a limit of 1k chunks currently - Reduce a cost on S3 when each request costs some money at the expense of memory used Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
2020-06-09 01:22:34 +08:00
partSize = int((((size / uploadParts) >> 20) + 1) << 20)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
}
}
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
memPool := f.getMemoryPool(int64(partSize))
s3: fix --header-upload - Fixes #4303 Before this change we were setting the headers on the PUT request for normal and multipart uploads. For normal uploads this caused the error 403 Forbidden: There were headers present in the request which were not signed After this fix we set the headers in the object upload request itself as the s3 SDK expects. This means that we only support a limited range of headers - Cache-Control - Content-Disposition - Content-Encoding - Content-Language - Content-Type - X-Amz-Tagging - X-Amz-Meta- Note for the last of those are for setting custom metadata in the form "X-Amz-Meta-Key: value". This now works for multipart uploads and single part uploads See also #59
2020-06-05 18:45:54 +08:00
var mReq s3.CreateMultipartUploadInput
structs.SetFrom(&mReq, req)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
var cout *s3.CreateMultipartUploadOutput
err = f.pacer.Call(func() (bool, error) {
var err error
s3: fix --header-upload - Fixes #4303 Before this change we were setting the headers on the PUT request for normal and multipart uploads. For normal uploads this caused the error 403 Forbidden: There were headers present in the request which were not signed After this fix we set the headers in the object upload request itself as the s3 SDK expects. This means that we only support a limited range of headers - Cache-Control - Content-Disposition - Content-Encoding - Content-Language - Content-Type - X-Amz-Tagging - X-Amz-Meta- Note for the last of those are for setting custom metadata in the form "X-Amz-Meta-Key: value". This now works for multipart uploads and single part uploads See also #59
2020-06-05 18:45:54 +08:00
cout, err = f.c.CreateMultipartUploadWithContext(ctx, &mReq)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
return f.shouldRetry(err)
})
if err != nil {
return errors.Wrap(err, "multipart upload failed to initialise")
}
uid := cout.UploadId
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
defer atexit.OnError(&err, func() {
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
if o.fs.opt.LeavePartsOnError {
return
}
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
fs.Debugf(o, "Cancelling multipart upload")
errCancel := f.pacer.Call(func() (bool, error) {
_, err := f.c.AbortMultipartUploadWithContext(context.Background(), &s3.AbortMultipartUploadInput{
Bucket: req.Bucket,
Key: req.Key,
UploadId: uid,
RequestPayer: req.RequestPayer,
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
})
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
return f.shouldRetry(err)
})
if errCancel != nil {
fs.Debugf(o, "Failed to cancel multipart upload: %v", errCancel)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
}
s3: cancel in progress multipart uploads and copies on rclone exit #4300
2020-06-04 18:09:27 +08:00
})()
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
var (
g, gCtx = errgroup.WithContext(ctx)
finished = false
partsMu sync.Mutex // to protect parts
parts []*s3.CompletedPart
off int64
)
for partNum := int64(1); !finished; partNum++ {
backend/s3: use memory pool for buffer allocations Currently each multipart upload allocated his own buffers, which after file upload was garbaged. Next files couldn't leverage already allocated memory which resulted in inefficent memory management. This change introduces backend memory pool keeping memory chunks which can be used during object operations. Fixes #3967
2020-02-19 18:17:25 +08:00
// Get a block of memory from the pool and token which limits concurrency.
tokens.Get()
buf := memPool.Get()
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
s3: don't leak memory or tokens in edge cases for multipart upload
2020-05-14 14:48:18 +08:00
free := func() {
// return the memory and token
memPool.Put(buf)
tokens.Put()
}
backend/s3: Fail fast multipart upload When a part upload request fails error is returned and gCtx is cancelled. This does not prevent from other parts being tried. They immediately fail due to a canceled context, but are retried by rclone anyway... Example AWS debug output ``` ----------------------------------------------------- 2020/02/11 14:12:17 DEBUG: Retrying Request s3/UploadPart, attempt 4 2020/02/11 14:12:17 DEBUG: Request s3/UploadPart Details: ---[ REQUEST POST-SIGN ]----------------------------- PUT /backuptest-rclone/huge/file.db?partNumber=11&uploadId=190939b4-3c43-4b98-ac11-92303e3f11b0 HTTP/1.1 Host: 192.168.100.99:9000 User-Agent: aws-sdk-go/1.23.8 (go1.13.1; linux; amd64) Content-Length: 5242880 Authorization: AWS4-HMAC-SHA256 Credential=miniouser/20200211/us-east-1/s3/aws4_request, SignedHeaders=content-length;content-md5;expect;host;x-amz-content-sha256;x-amz-date, Signature=3fc03a01f651cec09b05290459e9ceb26db9a8aa00c4e1b16e8cf5617eb81da8 Content-Md5: XzY+DlipXwbL6bvGYsXftg== Expect: 100-Continue X-Amz-Content-Sha256: c036cbb7553a909f8b8877d4461924307f27ecb66cff928eeeafd569c3887e29 X-Amz-Date: 20200211T131217Z Accept-Encoding: gzip ----------------------------------------------------- http://192.168.100.99:9000/backuptest-rclone/huge/file.db?partNumber=11&uploadId=190939b4-3c43-4b98-ac11-92303e3f11b0 2020/02/11 14:12:17 DEBUG: Response s3/UploadPart Details: ---[ RESPONSE ]-------------------------------------- HTTP/1.1 500 InternalServerError Content-Length: 0 ----------------------------------------------------- UploadPartWithContext() error InternalError: We encountered an internal error. Please try again status code: 500, request id: , host id: 2020/02/11 14:12:18 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled 2020/02/11 14:12:20 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled 2020/02/11 14:12:22 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled ``` This adds a fail fast behaviour in case the context was cancelled.
2020-02-12 00:12:08 +08:00
// Fail fast, in case an errgroup managed function returns an error
// gCtx is cancelled. There is no point in uploading all the other parts.
if gCtx.Err() != nil {
s3: don't leak memory or tokens in edge cases for multipart upload
2020-05-14 14:48:18 +08:00
free()
backend/s3: Fail fast multipart upload When a part upload request fails error is returned and gCtx is cancelled. This does not prevent from other parts being tried. They immediately fail due to a canceled context, but are retried by rclone anyway... Example AWS debug output ``` ----------------------------------------------------- 2020/02/11 14:12:17 DEBUG: Retrying Request s3/UploadPart, attempt 4 2020/02/11 14:12:17 DEBUG: Request s3/UploadPart Details: ---[ REQUEST POST-SIGN ]----------------------------- PUT /backuptest-rclone/huge/file.db?partNumber=11&uploadId=190939b4-3c43-4b98-ac11-92303e3f11b0 HTTP/1.1 Host: 192.168.100.99:9000 User-Agent: aws-sdk-go/1.23.8 (go1.13.1; linux; amd64) Content-Length: 5242880 Authorization: AWS4-HMAC-SHA256 Credential=miniouser/20200211/us-east-1/s3/aws4_request, SignedHeaders=content-length;content-md5;expect;host;x-amz-content-sha256;x-amz-date, Signature=3fc03a01f651cec09b05290459e9ceb26db9a8aa00c4e1b16e8cf5617eb81da8 Content-Md5: XzY+DlipXwbL6bvGYsXftg== Expect: 100-Continue X-Amz-Content-Sha256: c036cbb7553a909f8b8877d4461924307f27ecb66cff928eeeafd569c3887e29 X-Amz-Date: 20200211T131217Z Accept-Encoding: gzip ----------------------------------------------------- http://192.168.100.99:9000/backuptest-rclone/huge/file.db?partNumber=11&uploadId=190939b4-3c43-4b98-ac11-92303e3f11b0 2020/02/11 14:12:17 DEBUG: Response s3/UploadPart Details: ---[ RESPONSE ]-------------------------------------- HTTP/1.1 500 InternalServerError Content-Length: 0 ----------------------------------------------------- UploadPartWithContext() error InternalError: We encountered an internal error. Please try again status code: 500, request id: , host id: 2020/02/11 14:12:18 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled 2020/02/11 14:12:20 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled 2020/02/11 14:12:22 DEBUG ERROR: Request s3/UploadPart: ---[ REQUEST DUMP ERROR ]----------------------------- context canceled ------------------------------------------------------ UploadPartWithContext() error RequestCanceled: request context canceled caused by: context canceled ``` This adds a fail fast behaviour in case the context was cancelled.
2020-02-12 00:12:08 +08:00
break
}
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
// Read the chunk
var n int
n, err = readers.ReadFill(in, buf) // this can never return 0, nil
if err == io.EOF {
s3: fix multipart upload uploading 0 length files This regression was introduced by the recent re-write of the s3 multipart upload code.
2020-01-05 19:24:15 +08:00
if n == 0 && partNum != 1 { // end if no data and if not first chunk
s3: don't leak memory or tokens in edge cases for multipart upload
2020-05-14 14:48:18 +08:00
free()
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
break
}
finished = true
} else if err != nil {
s3: don't leak memory or tokens in edge cases for multipart upload
2020-05-14 14:48:18 +08:00
free()
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
return errors.Wrap(err, "multipart upload failed to read source")
}
buf = buf[:n]
partNum := partNum
fs.Debugf(o, "multipart upload starting chunk %d size %v offset %v/%v", partNum, fs.SizeSuffix(n), fs.SizeSuffix(off), fs.SizeSuffix(size))
off += int64(n)
g.Go(func() (err error) {
s3: don't leak memory or tokens in edge cases for multipart upload
2020-05-14 14:48:18 +08:00
defer free()
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
partLength := int64(len(buf))
// create checksum of buffer for integrity checking
md5sumBinary := md5.Sum(buf)
md5sum := base64.StdEncoding.EncodeToString(md5sumBinary[:])
err = f.pacer.Call(func() (bool, error) {
uploadPartReq := &s3.UploadPartInput{
Body: bytes.NewReader(buf),
Bucket: req.Bucket,
Key: req.Key,
PartNumber: &partNum,
UploadId: uid,
ContentMD5: &md5sum,
ContentLength: &partLength,
RequestPayer: req.RequestPayer,
SSECustomerAlgorithm: req.SSECustomerAlgorithm,
SSECustomerKey: req.SSECustomerKey,
SSECustomerKeyMD5: req.SSECustomerKeyMD5,
}
uout, err := f.c.UploadPartWithContext(gCtx, uploadPartReq)
if err != nil {
if partNum <= int64(concurrency) {
return f.shouldRetry(err)
}
// retry all chunks once have done the first batch
return true, err
}
partsMu.Lock()
parts = append(parts, &s3.CompletedPart{
PartNumber: &partNum,
ETag: uout.ETag,
})
partsMu.Unlock()
return false, nil
})
if err != nil {
return errors.Wrap(err, "multipart upload failed to upload part")
}
return nil
})
}
err = g.Wait()
if err != nil {
return err
}
// sort the completed parts by part number
sort.Slice(parts, func(i, j int) bool {
return *parts[i].PartNumber < *parts[j].PartNumber
})
err = f.pacer.Call(func() (bool, error) {
_, err := f.c.CompleteMultipartUploadWithContext(ctx, &s3.CompleteMultipartUploadInput{
Bucket: req.Bucket,
Key: req.Key,
MultipartUpload: &s3.CompletedMultipartUpload{
Parts: parts,
},
RequestPayer: req.RequestPayer,
UploadId: uid,
})
return f.shouldRetry(err)
})
if err != nil {
return errors.Wrap(err, "multipart upload failed to finalise")
}
return nil
}
Factor Fs.Put into Object.Update and call Update rather than Put if possible
2014-04-19 00:04:21 +08:00
// Update the Object from in with modTime and size
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) Update(ctx context.Context, in io.Reader, src fs.ObjectInfo, options ...fs.OpenOption) error {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, bucketPath := o.split()
s3,b2,googlecloudstorage,swift,qingstor,azureblob: fixes after code review #3421 - change the interface of listBuckets() removing dir parameter and adding context - add makeBucket() and use in place of Mkdir("") - this fixes some corner cases in Copy/Update - mark all the listed buckets OK in ListR Thanks to @yparitcher for the review.
2019-08-23 04:30:55 +08:00
err := o.fs.makeBucket(ctx, bucket)
swift, b2, gcs, s3: Fix moveto and copyto We now make sure the container/bucket is created before creating any objects.
2017-06-07 21:16:50 +08:00
if err != nil {
return err
}
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
modTime := src.ModTime(ctx)
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
size := src.Size()
Create separate interface for object information. Take out read-only information about a Fs in a separate struct to limit access. See discussion at #282.
2016-02-18 19:35:25 +08:00
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
multipart := size < 0 || size >= int64(o.fs.opt.UploadCutoff)
Factor Fs.Put into Object.Update and call Update rather than Put if possible
2014-04-19 00:04:21 +08:00
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
// Set the mtime in the meta data
metadata := map[string]*string{
metaMtime: aws.String(swift.TimeToFloatString(modTime)),
}
s3: fix missing x-amz-meta-md5chksum headers for multipart uploads This reverts "s3: fix DisableChecksum condition" which introduced the problem. This reverts commit c05bb63f9679e09ecb078e7477f3efd02846c2ad. The code was correct as it stands - the comment was incorrect and this commit updates it. See: https://forum.rclone.org/t/s3-upload-md5-check-sum/13706
2020-01-08 03:23:08 +08:00
// read the md5sum if available
// - for non multpart
// - so we can add a ContentMD5
// - for multipart provided checksums aren't disabled
// - so we can add the md5sum in the metadata as metaMD5Hash
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
var md5sum string
s3: fix missing x-amz-meta-md5chksum headers for multipart uploads This reverts "s3: fix DisableChecksum condition" which introduced the problem. This reverts commit c05bb63f9679e09ecb078e7477f3efd02846c2ad. The code was correct as it stands - the comment was incorrect and this commit updates it. See: https://forum.rclone.org/t/s3-upload-md5-check-sum/13706
2020-01-08 03:23:08 +08:00
if !multipart || !o.fs.opt.DisableChecksum {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
hash, err := src.Hash(ctx, hash.MD5)
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
if err == nil && matchMd5.MatchString(hash) {
hashBytes, err := hex.DecodeString(hash)
if err == nil {
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
md5sum = base64.StdEncoding.EncodeToString(hashBytes)
if multipart {
metadata[metaMD5Hash] = &md5sum
}
s3: set/get the hash for multipart files - #523
2018-01-06 22:30:10 +08:00
}
}
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
// Guess the content type
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
mimeType := fs.MimeType(ctx, src)
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
req := s3.PutObjectInput{
Bucket: &bucket,
ACL: &o.fs.opt.ACL,
Key: &bucketPath,
ContentType: &mimeType,
Metadata: metadata,
}
if md5sum != "" {
req.ContentMD5 = &md5sum
}
if o.fs.opt.ServerSideEncryption != "" {
req.ServerSideEncryption = &o.fs.opt.ServerSideEncryption
}
backend/s3: add SSE-C support for AWS, Ceph, and MinIO
2020-03-30 18:26:52 +08:00
if o.fs.opt.SSECustomerAlgorithm != "" {
req.SSECustomerAlgorithm = &o.fs.opt.SSECustomerAlgorithm
}
if o.fs.opt.SSECustomerKey != "" {
req.SSECustomerKey = &o.fs.opt.SSECustomerKey
}
if o.fs.opt.SSECustomerKeyMD5 != "" {
req.SSECustomerKeyMD5 = &o.fs.opt.SSECustomerKeyMD5
}
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
if o.fs.opt.SSEKMSKeyID != "" {
req.SSEKMSKeyId = &o.fs.opt.SSEKMSKeyID
}
if o.fs.opt.StorageClass != "" {
req.StorageClass = &o.fs.opt.StorageClass
}
s3: fix --header-upload - Fixes #4303 Before this change we were setting the headers on the PUT request for normal and multipart uploads. For normal uploads this caused the error 403 Forbidden: There were headers present in the request which were not signed After this fix we set the headers in the object upload request itself as the s3 SDK expects. This means that we only support a limited range of headers - Cache-Control - Content-Disposition - Content-Encoding - Content-Language - Content-Type - X-Amz-Tagging - X-Amz-Meta- Note for the last of those are for setting custom metadata in the form "X-Amz-Meta-Key: value". This now works for multipart uploads and single part uploads See also #59
2020-06-05 18:45:54 +08:00
// Apply upload options
for _, option := range options {
key, value := option.Header()
lowerKey := strings.ToLower(key)
switch lowerKey {
case "":
// ignore
case "cache-control":
req.CacheControl = aws.String(value)
case "content-disposition":
req.ContentDisposition = aws.String(value)
case "content-encoding":
req.ContentEncoding = aws.String(value)
case "content-language":
req.ContentLanguage = aws.String(value)
case "content-type":
req.ContentType = aws.String(value)
case "x-amz-tagging":
req.Tagging = aws.String(value)
default:
const amzMetaPrefix = "x-amz-meta-"
if strings.HasPrefix(lowerKey, amzMetaPrefix) {
metaKey := lowerKey[len(amzMetaPrefix):]
req.Metadata[metaKey] = aws.String(value)
} else {
fs.Errorf(o, "Don't know how to set key %q on upload", key)
}
}
}
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
if multipart {
s3: re-implement multipart upload to fix memory issues There have been quite a few reports of problems with the multipart uploader using too much memory and not retrying possible errors. Before this change the multipart uploader used the s3manager abstraction in the AWS SDK. There are numerous bug reports of this using up too much memory. This change re-implements a much simplified version of the s3manager code specialized for rclone's purposes. This should use much less memory and retry chunks properly. See: https://forum.rclone.org/t/memory-usage-s3-alike-to-glacier-without-big-directories/13563 See: https://forum.rclone.org/t/copy-from-local-to-s3-has-high-memory-usage/13405 See: https://forum.rclone.org/t/big-file-upload-to-s3-fails/13575
2019-12-31 07:17:06 +08:00
err = o.uploadMultipart(ctx, &req, size, in)
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
if err != nil {
return err
}
} else {
// Create the request
putObj, _ := o.fs.c.PutObjectRequest(&req)
// Sign it so we can upload using a presigned request.
//
// Note the SDK doesn't currently support streaming to
// PutObject so we'll use this work-around.
url, headers, err := putObj.PresignRequest(15 * time.Minute)
if err != nil {
return errors.Wrap(err, "s3 upload: sign request")
}
s3: fix signature v2_auth headers When used with v2_auth = true, PresignRequest doesn't return signed headers, so remote dest authentication would be fail. This commit copying back HTTPRequest.Header to headers. Tested with RiakCS v2.1.0. Signed-off-by: Anthony Rusdi <33247310+antrusd@users.noreply.github.com>
2019-09-21 09:30:45 +08:00
if o.fs.opt.V2Auth && headers == nil {
headers = putObj.HTTPRequest.Header
}
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
// Set request to nil if empty so as not to make chunked encoding
if size == 0 {
in = nil
}
// create the vanilla http request
httpReq, err := http.NewRequest("PUT", url, in)
if err != nil {
return errors.Wrap(err, "s3 upload: new request")
}
Add context to all http.NewRequest #3257 When we drop support for go1.12 we can use http.NewRequestWithContext
2019-09-05 03:21:10 +08:00
httpReq = httpReq.WithContext(ctx) // go1.13 can use NewRequestWithContext
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
// set the headers we signed and the length
httpReq.Header = headers
httpReq.ContentLength = size
err = o.fs.pacer.CallNoRetry(func() (bool, error) {
resp, err := o.fs.srv.Do(httpReq)
if err != nil {
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return o.fs.shouldRetry(err)
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
}
body, err := rest.ReadBody(resp)
if err != nil {
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return o.fs.shouldRetry(err)
s3: implement --s3-upload-cutoff for single part uploads below this - fixes #2772 Before this change rclone would use multipart uploads for any size of file. However multipart uploads are less efficient for smaller files and don't have MD5 checksums so it is advantageous to use single part uploads if possible. This implements single part uploads for all files smaller than the upload_cutoff size. Streamed files must be uploaded as multipart files though.
2018-11-27 05:09:23 +08:00
}
if resp.StatusCode >= 200 && resp.StatusCode < 299 {
return false, nil
}
err = errors.Errorf("s3 upload: %s: %s", resp.Status, body)
return fserrors.ShouldRetryHTTP(resp, retryErrorCodes), err
})
if err != nil {
return err
}
s3: Read metadata after Update or Put
2014-07-19 19:37:11 +08:00
}
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
s3: Read metadata after Update or Put
2014-07-19 19:37:11 +08:00
// Read the metadata from the newly created object
swift, s3, dropbox: fix metadata read on Update() This was causing changed files to be marked as corrupted on upload
2014-07-20 18:23:05 +08:00
o.meta = nil // wipe old metadata
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
err = o.readMetaData(ctx)
Factor Fs.Put into Object.Update and call Update rather than Put if possible
2014-04-19 00:04:21 +08:00
return err
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Remove an object
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) Remove(ctx context.Context) error {
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
bucket, bucketPath := o.split()
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
req := s3.DeleteObjectInput{
s3: make all operations work from the root #3421
2019-08-09 18:29:36 +08:00
Bucket: &bucket,
Key: &bucketPath,
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
}
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
err := o.fs.pacer.Call(func() (bool, error) {
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
_, err := o.fs.c.DeleteObjectWithContext(ctx, &req)
s3: Auto detect region for buckets on operation failure - fixes #2915 If an incorrect region error is returned while using a bucket then the region is updated, the session is remade and the operation is retried.
2019-01-16 21:35:19 +08:00
return o.fs.shouldRetry(err)
S3: Use (custom) pacer, to retry operations when reasonable - fixes #2503
2018-09-03 12:41:04 +08:00
})
s3: use official github.com/aws/aws-sdk-go including multipart upload - fixes #101
2014-12-23 20:09:02 +08:00
return err
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
}
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
// MimeType of an Object if known, "" otherwise
Add context propagation to rclone - Change rclone/fs interfaces to accept context.Context - Update interface implementations to use context.Context - Change top level usage to propagate context to lover level functions Context propagation is needed for stopping transfers and passing other request-scoped values.
2019-06-17 16:34:30 +08:00
func (o *Object) MimeType(ctx context.Context) string {
err := o.readMetaData(ctx)
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
if err != nil {
Rename log functions and factor into own file
2017-02-09 19:01:20 +08:00
fs.Logf(o, "Failed to read metadata: %v", err)
Make ContentType be preserved for cloud -> cloud copies - fixes #733
2016-09-22 05:13:24 +08:00
return ""
}
return o.mimeType
}
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
// SetTier performs changing storage class
func (o *Object) SetTier(tier string) (err error) {
ctx := context.TODO()
tier = strings.ToUpper(tier)
bucket, bucketPath := o.split()
req := s3.CopyObjectInput{
MetadataDirective: aws.String(s3.MetadataDirectiveCopy),
StorageClass: aws.String(tier),
}
s3: support for multipart copy Fixes #2375 Fixes #3579
2019-10-04 23:49:06 +08:00
err = o.fs.copy(ctx, &req, bucket, bucketPath, bucket, bucketPath, o.bytes)
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
if err != nil {
return err
}
o.storageClass = tier
return err
}
// GetTier returns storage class as string
func (o *Object) GetTier() string {
if o.storageClass == "" {
return "STANDARD"
}
return o.storageClass
}
Add initial S3 support * Requires a patched goamz to enable metadata reading and writing * Still todo: updating metadata * EU region not working yet
2013-01-09 02:53:35 +08:00
// Check the interfaces are satisfied
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
var (
s3: allow streaming upload of files with unknown file size (see #1614)
2017-09-16 02:20:32 +08:00
_ fs.Fs = &Fs{}
_ fs.Copier = &Fs{}
_ fs.PutStreamer = &Fs{}
_ fs.ListRer = &Fs{}
s3: add backend restore command to restore objects from GLACIER See: https://forum.rclone.org/t/rclone-settier-fails-with-scaleway-entitytoolarge/17384
2020-06-24 18:02:34 +08:00
_ fs.Commander = &Fs{}
s3: implement cleanup and backend command to list & remove multipart uploads This implements `rclone cleanup` to remove multipart uploads over 24 hours old. It also implements the backend command `list-multipart-uploads` to see which ones are available and `cleanup` to delete them with a configurable expiry interval. See #4302
2020-06-25 23:11:05 +08:00
_ fs.CleanUpper = &Fs{}
s3: allow streaming upload of files with unknown file size (see #1614)
2017-09-16 02:20:32 +08:00
_ fs.Object = &Object{}
_ fs.MimeTyper = &Object{}
s3: fix SetModTime on GLACIER/ARCHIVE objects and implement set/get tier - Read the storage class for each object - Implement SetTier/GetTier - Check the storage class on the **object** before using SetModTime This updates the fix in 1a2fb52 so that SetModTime works when you are using objects which have been migrated to GLACIER but you aren't using GLACIER as a storage class. Fixes #3522
2019-09-10 03:44:50 +08:00
_ fs.GetTierer = &Object{}
_ fs.SetTierer = &Object{}
Be more constistent with naming in remotes * External objects are called Fs and Object * Object.fs always points to the Fs
2015-11-07 19:14:46 +08:00
)
Reference in New Issue Copy Permalink