Introduce stats groups that will isolate accounting for logically
different transferring operations. That way multiple accounting
operations can be done in parallel without interfering with each other
stats.
Using groups is optional. There is dedicated global stats that will be
used by default if no group is specified. This is operating mode for CLI
usage which is just fire and forget operation.
For running rclone as rc http server each request will create it's own
group. Also there is an option to specify your own group.
This is done to make clear ownership over accounting object and prepare
for removing global stats object.
Stats elapsed time calculation has been altered to account for actual
transfer time instead of stats creation time.
This was started by Fionera, finished off by Laura with fixes and more
docs from Nick.
Co-authored-by: Fionera <fionera@fionera.de>
Co-authored-by: Nick Craig-Wood <nick@craig-wood.com>
We now have a backend (fichier) which doesn't support 0 length
files. Therefore all 0 length files in the tests have been replaced
with length 1.
In a future commit we will implement a test for 0 length files.
- 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.
Before this change when using "rclone config create" it wasn't
possible to add passwords in one go, it was necessary to call "rclone
config password" to add the passwords afterwards as "rclone config
create" didn't obscure passwords.
After this change "rclone config create" and "rclone config update"
will obscure passwords as necessary as will the corresponding API
calls config/create and config/update.
This makes "rclone config password" and its API config/password
obsolete, however they will be left for backwards compatibility.
Before this change serving bucket based objects
`[remote:bucket]/path/to/object` would fail with 404 not found.
This was because the leading `/` in `/path/to/object` was being passed
to NewObject.
When sorting fs.DirEntries we sort by DirEntry type and
when synchronizing files let the directories be before objects,
so when the destintation fs doesn't support duplicate names,
we will only lose duplicated object instead of whole directory.
The enables synchronisation to work with a file and a directory of the same name
which is reasonably common on bucket based remotes.
Before this change, when using rclone copy or move with --backup-dir
and the source was a single file, rclone would fail to use the backup
directory.
This change looks up the backup directory in the Fs cache and uses it
as appropriate.
This affects any commands which call operations.MoveFile or
operations.CopyFile which includes rclone move/moveto/copy/copyto
where the source is a single file.
Fixes#3219
Before this change we calculated the checksum which is potentially
time consuming and then ignored the result. After the change we don't
calculate the checksum if we are about to ignore it.
Tests have been randomly failing with messages like
listen tcp 127.0.0.1:51778: bind: address already in use
Rework all the test servers so they choose a random free port on
startup and use that for the tests to avoid.
In c5ac96e9e7 we made --files-from only read the objects specified and
don't scan directories.
This caused problems with Google drive (very very slow) and B2
(excessive API consumption) so it was decided to make the old
behaviour (traversing the directories) the default with --files-from
and use the existing --no-traverse flag (which has exactly the right
semantics) to enable the new non scanning behaviour.
See: https://forum.rclone.org/t/using-files-from-with-drive-hammers-the-api/8726Fixes#3102Fixes#3095
Before the fix we were only de-duping the ListR batches.
Afterwards we dedupe everything.
This will have the consequence that rclone uses more memory as it will
build a map of all the directory names, not just the names in a given
directory.
This dramatically increases the speed (7x in my tests) of the de-dupe
as google drive supports ListR directly and dedupe did not work with
`--fast-list`.
Fixes#2902
This will increase speed for backends which support ListR and will not
have the memory overhead of using --fast-list.
It also means that errors are queued until the end so as much of the
remote will be listed as possible before returning an error.
Commands affected are:
- lsf
- ls
- lsl
- lsjson
- lsd
- md5sum/sha1sum/hashsum
- size
- delete
- cat
- settier
It otherwise has the nearly the same interface as walk.Walk which it
will fall back to if it can't use ListR.
Using walk.ListR will speed up file system operations by default and
use much less memory and start immediately compared to if --fast-list
had been supplied.
Make the pacer package more flexible by extracting the pace calculation
functions into a separate interface. This also allows to move features
that require the fs package like logging and custom errors into the fs
package.
Also add a RetryAfterError sentinel error that can be used to signal a
desired retry time to the Calculator.
This brings it up to par with lsjson.
This commit also reworks the framework to use ListJSON internally
which removes duplicated code and makes testing easier.
This puts a shim on the reader opened by Copy so that if an error is
returned, the reader is re-opened at the correct seek point.
This should make downloading very large files more reliable.
This replaces the `sync.Pool` allocator with lib/pool. This
implements a pool of buffers of up to 64MB which can be re-used but is
flushed every 5 seconds.
If `--use-mmap` is set then rclone will use mmap for memory
allocations which is much better at returning memory to the OS.
* drive: don't run teamdrive config if auto confirm set
* onedrive: don't run extra config if auto confirm set
* make Confirm results customisable by config
Fixes#1010
This fixes several things wrong with the layout of the stats.
Transfers which haven't started are printed in the same format as
those which have so the stats with `--progress` don't show horrible
artifacts.
Checkers and transfers now get a ": checkers" and ": transfers" label
on the end of the stats line. Transfers will have the transfer stats
when the transfer has started instead of this.
There was a bug in the routine which shortened the file names (it
always produces strings 1 too long). This is now fixed with a test.
The formatting string was wrong with a fixed width of 45 - this is now
replaces with the value of `--stats-file-name-length`.
This also meant that there were unecessary leading spaces in the file
names. So the default `--stats-file-name-length` was raised to 45
from 40.
Cookies are handled by cookiejar in memory with fshttp module through
the entire session.
One useful scenario is, with HTTP storage system where index server
adds authentication cookie while redirecting to CDN for actual files.
Also, it can be helpful to reuse fshttp in other storage systems
requiring cookie.
This means that rclone will pick up tokens from concurrently running
rclones. This helps for Box which only allows each refresh token to
be used once.
Without this fix, rclone caches the refresh token at the start of the
run, then when the token expires the refresh token may have been used
already by a concurrently running rclone.
This also will retry the oauth up to 5 times at 1 second intervals.
See: https://forum.rclone.org/t/box-token-refresh-timing/8175
Before this change rclone would read the list of files from the
files-from parameter and check they existed one at a time. This could
take a very long time for lots of files.
After this change, rclone will check up to --checkers in parallel.
The --no-traverse flag was not implemented when the new sync routines
(using the march package) was implemented.
This re-implements --no-traverse in march by trying to find a match
for each object with NewObject rather than from a directory listing.
Before this change TestPurge would remove a container and subsequent
tests would fail because the container was still being deleted so
couldn't be created.
This was fixed by introducing an fstest.NewRunIndividual() test runner
for TestPurge which causes the test to be run on a new container.
If `--rc-user` or `--rc-pass` is set then the URL that is opened with
`--rc-files` will have the authorization in the URL in the
`http://user:pass@localhost/` style.
Before this change, Purge on the fallback path would try to delete
directories starting from the root rather than the dir passed in.
Rmdirs would also attempt to delete the root.
Before this change using --files-from would scan all the directories
that the files could possibly be in causing rclone to do more work
that was necessary.
After this change, rclone constructs an in memory tree using the
--fast-list mechanism but from all of the files in the --files-from
list and without scanning any directories.
Any objects that are not found in the --files-from list are ignored
silently.
This mechanism is used for sync/copy/move (march) and all of the
listing commands ls/lsf/md5sum/etc (walk).
Use the same function to join the root paths for the wrapping remotes
alias, cache and crypt.
The new function fspath.JoinRootPath is equivalent to path.Join, but if
the first non empty element starts with "//", this is preserved to allow
Windows network path to be used in these remotes.
Before this change remotes without server side Move (eg swift, s3,
gcs) would not be able to rename files.
After it means nearly all remotes will be able to rename files on
rclone mount with the notable exceptions of b2 and yandex.
This changes checks to see if the remote can do Move or Copy then
calls `operations.Move` to do the actual move. This will do a server
side Move or Copy but won't download and re-upload the file.
It also checks to see if the destination exists first which avoids
conflicts or duplicates.
Fixes#1965Fixes#2569
Before this change the moving average for the individual file stats
would start at 0 and only converge to the correct value over 15-30
seconds.
This change starts the weighting period as 1 and moves it up once per
sample which gets the average to a better value instantly.
Previous to this change package used for this
github.com/VividCortex/ewma took a 0 average to mean reset the
statistics. This happens quite often when transferring files though a
buffer.
Replace that implementation with a simple home grown one (with about
the same constant), without that feature.
This change allows remotes to be created on the fly without a config
file by using the remote type prefixed with a : as the remote name, Eg
:s3: to make an s3 remote.
This assumes the user is supplying the backend config via command line
flags or environment variables.
The race detector currently detects a race with len(chan) against
close(chan).
See: https://github.com/golang/go/issues/27070
Skip the tests which trip this bug under the race detector.
--max-backlog controls the queue length.
Add statistics for the check/upload/rename queues.
This means that checking can complete before the uploads which will
give rclone the ability to show exactly what is outstanding.
