Commit Graph

1079 Commits

Author SHA1 Message Date
Abiola Ibrahim
e92a911e7d Add more tests. 2016-03-11 23:44:50 +01:00
Abiola Ibrahim
84845a66ab Fix broken build. 2016-03-11 23:11:21 +01:00
Abiola Ibrahim
f3a183ecc1 Use filepath.Clean for fileserver. 2016-03-11 15:39:13 +01:00
Matthew Holt
741d7685f1 Merge branch 'master' into fastcgi-methods
# Conflicts:
#	middleware/fastcgi/fastcgi.go
2016-03-07 16:25:23 -07:00
Matthew Holt
88e3a26c99 Full changes to contributing doc
That was weird, only half of the file got committed...
2016-03-07 12:10:26 -07:00
Matthew Holt
f52b1e80f5 Update contributing doc and add issue template 2016-03-07 12:07:39 -07:00
Matt Holt
9e386fc921 Merge pull request #652 from elcore/patch-2
https: Support ECC keys
2016-03-03 08:19:20 -07:00
elcore
9099375b11 Support ECC certificates 2016-03-03 00:52:07 +00:00
Matthew Holt
36b440c04b https: Refuse start only if renewal fails on expired cert (closes #642) 2016-03-02 11:34:39 -07:00
Matthew Holt
2a46f2a14e Revert recent Content-Length-related changes and fix fastcgi return
fastcgi's ServeHTTP method originally returned the correct value (0) in
b51e8bc191. Later, I mistakenly suggested
we change that to return the status code because I forgot that status
codes aren't logged by the return value. So fastcgi broke due in
3966936bd6 due to my error.

We later had to try to make up for this with ugly Content-Length checks
like in c37ad7f677. Turns out that all we
had to do was fix the returned status here back to 0. The proxy
middleware behaves the same way, and returning 0 is correct. We should
only return a status code if the response has not been written, but with
upstream servers, we do write a response; they do not know about our
error handler.

Also clarifed this in the middleware.Handler documentation.
2016-03-02 11:33:40 -07:00
Matthew Holt
741880a38b Only obtain certificate and enable TLS if host qualifies (fixes #638) 2016-03-01 12:27:46 -07:00
Matt Holt
43c339c7e3 Merge pull request #641 from hkjn/fix-build-acme-crypt
https: Fix build after https://github.com/xenolf/lego/commit/0e26b
2016-02-27 10:21:44 -07:00
Henrik Jonsson
49c2807ba1 Fix build after https://github.com/xenolf/lego/commit/0e26b
Fix up last-second changes

Fixes #640
2016-02-27 18:06:56 +01:00
Matthew Holt
da08c94a8c Implant version information with -ldflags with help of build script
Without -ldflags, the verison information needs to be updated manually,
which is never done between releases, so development builds appear
indiscernable from stable builds using `caddy -version`.

This is part of a set of changes intended to relieve the burden of
always updating version information manually and distributing binaries
that look stable but actually may not be.

A stable build is defined as one which is produced at a git tag with
a clean working directory (no uncommitted changes). A dev build is
anything else. With this build script, `caddy -version` will now reveal
whether it is a development build and, if so, the base version, the
latest commit, the date and time of build, and the names of files with
changes as well as how many changes were made.

The output of `caddy -version` for stable builds remains the same.
2016-02-26 00:26:31 -07:00
Matthew Holt
c827a71d5d Version 0.8.2 2016-02-25 10:26:42 -07:00
Matthew Holt
2ecc837020 templates: .Truncate can truncate from end of string if length is negative 2016-02-24 20:32:26 -07:00
Matthew Holt
c37ad7f677 Only write error message/page if body not already written (fixes #567)
Based on work started in, and replaces, #614
2016-02-24 19:50:46 -07:00
Matthew Holt
737c7c4372 fastcgi: Only perform extra copy if necessary; added tests 2016-02-24 16:42:01 -07:00
Matt Holt
367397dbd6 Merge pull request #623 from xlab/f/docflags
markdown: Implement .DocFlags and tests
2016-02-24 14:45:22 -07:00
Matt Holt
a2dbfdc10e Merge pull request #636 from humboldtux/fastcgi-cl
fastcgi: Explicitly set Content-Length (fixes #626)
2016-02-24 14:39:15 -07:00
Benoit Benedetti
ef5f9c771d FastCGI: Explicitly set Content-Length #626 2016-02-24 22:11:07 +01:00
Matthew Holt
05957b4965 gzip: Implement http.Hijacker (fixes #635) 2016-02-24 12:23:50 -07:00
Matt Holt
72fcdec8d8 Merge pull request #627 from ax-nathan/master
rewrite: Allow status to be 2xx or 4xx.
2016-02-24 11:01:54 -07:00
Nathan Probst
f4bb43781c Remove unneeded Regexp from tests. 2016-02-24 10:28:06 -07:00
Matt Holt
197297b0d7 Merge pull request #628 from elcore/master
tls: Add SHA384 ciphers
2016-02-23 15:14:24 -07:00
elcore
a541eb7899 Adding new cipher suites 2016-02-23 21:23:13 +00:00
Nathan Probst
2ea6c95ac4 Allow rewrite status codes to be 2xx and 4xx. 2016-02-22 15:30:55 -07:00
Maxim Kupriianov
c7674e2060 Implement .DocFlags directive and tests. It holds all the boolean-typed front matter values. 2016-02-22 13:53:47 +03:00
Matt Holt
c12847e5ba Merge pull request #620 from humboldtux/recorder
middleware: Export ResponseRecorder and add a couple getter methods
2016-02-20 15:17:45 -07:00
Benoit Benedetti
bec130a563 Recorder: Exporting ResponseRecorder #614 2016-02-20 23:10:04 +01:00
Matthew Holt
09b7ce6c93 Try to get Go 1.6 on appveyor 2016-02-19 18:07:48 -07:00
Matt Holt
b860be01bb Merge pull request #615 from 1lann/master
errors: Set missing Content-Type for plaintext error messages
2016-02-19 14:52:43 -07:00
Matthew Holt
f7b5187bf3 server: Add "Referer" to log entry when host not found 2016-02-19 13:34:54 -07:00
Matthew Holt
09a7af8cae https: Wait as long as possible to create ACME client at startup (fixes #617) 2016-02-19 10:33:01 -07:00
Jason Chu
5f2670fdde Fix missing Content-Type for certain errors
And corrected an error in a copy and pasted comment
2016-02-20 00:42:17 +08:00
Matthew Holt
ecf913e58d Update change log 2016-02-18 20:57:38 -07:00
Matthew Holt
d05f89294e https: Minor refactoring and some new tests 2016-02-18 20:33:15 -07:00
Matthew Holt
1ef7f3c4b1 Remove path scoping for middleware slice
It was implemented for almost a year but we'll probably never use it, especially since we'll match more than the path in the future.
2016-02-17 18:11:03 -07:00
Matthew Holt
f25ae8230f Move to Go 1.6 and set CGO_ENABLED=0 in tests 2016-02-17 16:08:25 -07:00
Matthew Holt
1cfd960f3c Bug fixes and other improvements to TLS functions
Now attempt to staple OCSP even for certs that don't have an existing staple (issue #605). "tls off" short-circuits tls setup function. Now we call getEmail() when setting up an acme.Client that does renewals, rather than making a new account with empty email address. Check certificate expiry every 12 hours, and OCSP every hour.
2016-02-15 23:39:04 -07:00
Matt Holt
2dba44327a Merge pull request #600 from jacobhands/betterlogging
Rotate process log
2016-02-14 23:32:50 -07:00
Matthew Holt
cae9f7de9c gofmt -s; fix misspellings and lint; Go 1.5.3 in Travis CI 2016-02-14 00:10:57 -07:00
Matthew Holt
a11e14aca8 Fix HTTPS config for empty/no Caddyfile
This fixes a regression introduced in recent commits that enabled TLS on the default ":2015" config. This fix is possible because On-Demand TLS is no longer implicit; it must be explicitly enabled by the user by setting a maximum number of certificates to issue.
2016-02-12 13:04:24 -07:00
Jacob Hands
dc63e50172 Use rotating log files 2016-02-12 08:30:47 -06:00
Matthew Holt
04c7c442c5 https: Only create ACMEClient if it's actually going to be used
Otherwise it tries to create an account and stuff at first start, even without a Caddyfile or when serving localhost.
2016-02-11 16:20:59 -07:00
Matthew Holt
7bd2adf0dc Fix edge case related to reloaded configs and ACME challenge
If Caddy is running but not listening on port 80, reloading Caddy with a new Caddyfile that needs to obtain a TLS cert from the CA would fail, because it was just assumed that, if reloading, port 80 as already in use. That is not always the case, so we scan the servers to see if one of them is listening on port 80, and we configure the ACME client accordingly. Kind of a hack... but it works.
2016-02-11 15:37:51 -07:00
Matthew Holt
1fe39e4633 Additional mitigation for on-demand TLS
After 10 certificates are issued, no new certificate requests are allowed for 10 minutes after a successful issuance.
2016-02-11 14:27:57 -07:00
Matthew Holt
216a617249 tls: Some bug fixes, basic rate limiting, max_certs setting 2016-02-11 13:48:52 -07:00
Matthew Holt
d25a3e95e4 Merge branch 'master' into getcertificate 2016-02-11 00:06:23 -07:00
Matthew Holt
11103bd8d6 Major refactor of all HTTPS/TLS/ACME code
Biggest change is no longer using standard library's tls.Config.getCertificate function to get a certificate during TLS handshake. Implemented our own cache which can be changed dynamically at runtime, even during TLS handshakes. As such, restarts are no longer required after certificate renewals or OCSP updates.

We also allow loading multiple certificates and keys per host, even by specifying a directory (tls got a new 'load' command for that).

Renamed the letsencrypt package to https in a gradual effort to become more generic; and https is more fitting for what the package does now.

There are still some known bugs, e.g. reloading where a new certificate is required but port 80 isn't currently listening, will cause the challenge to fail. There's still plenty of cleanup to do and tests to write. It is especially confusing right now how we enable "on-demand" TLS during setup and keep track of that. But this change should basically work so far.
2016-02-11 00:06:05 -07:00