summary refs log tree commit homepage
AgeCommit message (Collapse)AuthorFilesLines
2011-04-30unicorn 1.1.7 - major fixes to minor components v1.1.7 1.1.x-stableEric Wong2-3/+3
No changes to the core code, so this release only affects users of the Unicorn::OobGC and Unicorn::ExecCGI modules. Unicorn::OobGC was totally broken by the fix in the v1.1.6 release and is now reimplemented. Unicorn::ExecCGI (which hardly anybody uses) now returns proper HTTP status codes.
2011-04-30oob_gc: reimplement to fix breakage and add testsEric Wong6-83/+227
This was broken since v3.3.1[1] and v1.1.6[2] since nginx relies on a closed socket (and not Content-Length/Transfer-Encoding) to detect a response completion. We have to close the client socket before invoking GC to ensure the client sees the response in a timely manner. [1] - commit b72a86f66c722d56a6d77ed1d2779ace6ad103ed [2] - commit b7a0074284d33352bb9e732c660b29162f34bf0e (cherry picked from commit faeb3223636c39ea8df4017dc9a9d39ac649b26d) Conflicts: examples/big_app_gc.rb lib/unicorn/oob_gc.rb
2011-01-21exec_cgi: handle Status header in CGI responseEric Wong1-1/+2
We no longer blindly return 200 if the CGI returned another error code. We also don't want two Status headers in our output since we no longer filter it out. (cherry picked from commit 6cca8e61c66c1c2a8ebe260813fa83e44530a768)
2011-01-06unicorn 1.1.6 - one minor, esoteric bugfix v1.1.6Eric Wong2-3/+3
We now close the client socket after closing the response body. This does not affect most applications that run under Unicorn, in fact, it may not affect any.
2011-01-05close client socket after closing response bodyEric Wong5-12/+45
Response bodies may capture the block passed to each and save it for body.close, so don't close the socket before we have a chance to call body.close (cherry picked from commit b72a86f66c722d56a6d77ed1d2779ace6ad103ed) Conflicts: lib/unicorn/http_server.rb test/unit/test_response.rb
2010-10-27unicorn 1.1.5 v1.1.5Eric Wong2-3/+3
This maintenance release fixes several long-standing but recently-noticed bugs. SIGHUP reloading now correctly restores default values if they're erased or commented-out in the Unicorn configuration file. Delays/slowdowns in signal handling since 0.990 are fixed, too.
2010-10-27doc: stop using deprecated rdoc CLI optionsEric Wong2-2/+2
-N and -a switches no longer exist in rdoc 2.5 (cherry picked from commit 054c7df93db61839648925cfd881ae880709a210)
2010-10-27gemspec: depend on Isolate 3.0.0 for devEric Wong1-1/+1
No reason to not use the latest and greatest! (cherry picked from commit 570a57c07fd8c3d24b7337637e0dd30136b3a11a) Conflicts: unicorn.gemspec
2010-10-27configurator: reloading with unset values restores defaultEric Wong2-5/+93
If a configuration directive is set at startup and later unset, it correctly restores the original default value as if it had never been set in the first place. This applies to the majority of the configuration values with a few exceptions: * This only applies to stderr_path and stdout_path when daemonized (the usual case, they'll be redirected to "/dev/null"). When NOT daemonized, we cannot easily redirect back to the original stdout/stderr destinations. * Unsetting working_directory does not restore the original working directory where Unicorn was started. As far as we can tell unsetting this after setting it is rarely desirable and greatly increases the probability of user error. (cherry picked from commit 51b2b90284000aee8d79b37a5406173c45ae212d)
2010-10-27configurator: use "__send__" instead of "send"Eric Wong1-1/+1
It's less ambiguous since this is a network server after all. (cherry picked from commit f62c5850d7d17d7b5e301a494f8bdf5be3674411)
2010-10-27Rakefile: capture prerelease tagsEric Wong1-1/+1
Since we do those, now. (cherry picked from commit 1d1a2b1bd5bdd89f774f19bf8ad24c2f5f8a2d4c)
2010-10-27Rakefile: don't post freshmeat on empty changelogsEric Wong1-2/+6
We don't want to flood or monopolize freshmeat.
2010-10-27fix delays in signal handlingEric Wong1-5/+2
There is no need to loop in the master_sleep method at all, as the rest of the code is designed to function even on interrupted sleeps. This change is included as part of a larger cleanup in master. (commit bdc79712e5ac53d39c51e80dfe50aff950e5053f). From maint: (cherry picked from commit 10037f2aabb3fab4296fc90c615e7caa9f4a9b53) Conflicts: lib/unicorn.rb
2010-10-04unicorn 1.1.4 - small bug fix and doc updates v1.1.4Eric Wong2-3/+3
We no longer unlinking actively listening sockets upon startup (but continue to unlink dead ones). This bug could trigger downtime and nginx failures if a user makes an error and attempts to start Unicorn while it is already running. Thanks to Jordan Ritter for the detailed bug report leading to this fix. ref: http://mid.gmane.org/8D95A44B-A098-43BE-B532-7D74BD957F31@darkridge.com There are also minor documentation and test updates pulled in from master. This is hopefully the last bugfix release of the 1.1.x series.
2010-10-04update Rails 3 tests to use Rails 3 finalEric Wong3-3/+3
Rails 3 is out, and requires no code changes on our end to work (as far as our tests show :) (cherry picked from commit da272fc48ffaa808456fe94dd7a3e01bc9799832)
2010-10-04avoid unlinking actively listening socketsEric Wong3-3/+95
While we've always unlinked dead sockets from nuked/leftover processes, blindly unlinking them can cause unnecessary failures when an active process is already listening on them. We now make a simple connect(2) check to ensure the socket is not in use before unlinking it. Thanks to Jordan Ritter for the detailed bug report leading to this fix. ref: http://mid.gmane.org/8D95A44B-A098-43BE-B532-7D74BD957F31@darkridge.com (cherry picked from commit 1a2363b17b1d06be6b35d347ebcaed6a0c940200)
2010-10-04doc: update HACKING for documentation contributionsEric Wong1-3/+10
We switched to RDoc 2.5.x long ago and this should clarify some documentation preferences I have. (cherry picked from commit 505a9e72d320fe3ae521ceb0f381c1c0f5ae4389)
2010-10-04doc: update Sandbox document for BundlerEric Wong1-0/+9
Thanks to Lawrence Pit, Jamie Wilkinson, and Eirik Dentz Sinclair. ref: mid.gmane.org/4C8986DA.7090603@gmail.com ref: mid.gmane.org/5F1A02DB-CBDA-4302-9E26-8050C2D72433@efficiency20.com (cherry picked from commit 1a75966a5d1a1f6307ed3386e2f91a28bbb72ca0)
2010-08-30TUNING: more on socket buffer sizesEric Wong1-1/+7
Large buffers can hurt as well as help. And the difference in real apps that do a lot of things other than I/O often makes it not worth it. (cherry picked from commit f9a7a19a361fd674bab4e2df7e0897015528bba7)
2010-08-28unicorn 1.1.3 - small bug fixes v1.1.3Eric Wong2-3/+3
This release fixes race conditions during SIGUSR1 log cycling. This bug mainly affects Rainbows! users serving static files, but some Rack apps use threads internally even under Unicorn. Other small fixes: * SIGTTIN works as documented after SIGWINCH * --help output from `unicorn` and `unicorn_rails` is more consistent
2010-08-28make log reopens even more robust in threaded appsEric Wong1-10/+10
A follow-up to 4b23693b9082a84433a9e6c1f358b58420176b27 If multithreaded programming can be compared to juggling chainsaws, then multithreaded programming with signal handlers in play is akin to juggling chainsaws on a tightrope over shark-infested waters. (cherry picked from commit feab35fe531843066db3418598874cf9f9419614)
2010-08-28update Rails3 tests to use 3.0.0rc2Eric Wong3-3/+3
No code changes needed, thankfully. (cherry picked from commit 18968f6aff2fa5ba5a7e3e3d47c9cc05cd6c260d)
2010-08-28make log reopens more robust in multithreaded appsEric Wong1-2/+16
IOError may occur due to race conditions as another thread may close the file immediately after we call File#closed? to check. Errno::EBADF may occur in some applications that close a file descriptor without notifying Ruby (or if two IO objects refer to the same descriptor, possibly one of them using IO#for_fd). (cherry picked from commit 4b23693b9082a84433a9e6c1f358b58420176b27)
2010-08-28bin/*: more consistent --help outputEric Wong2-5/+6
This fixes a long-standing bug in the output of "unicorn_rails" where the program name was missing. (cherry picked from commit 096afc1a8e958cc09b4ce8b3bfe76ce056c7ed69)
2010-07-16SIGTTIN works after SIGWINCHEric Wong2-0/+61
In addition to SIGHUP, it should be possible to gradually bring workers back up (to avoid overloading the machine) when rolling back upgrades after SIGWINCH. Noticed-by: Lawrence Pit ref: http://mid.gmane.org/4C3F8C9F.2090903@gmail.com (cherry picked from commit f1d33c80dd6c5650f960f7087f4e08f809754d34)
2010-07-13unicorn 1.1.2 - fixing upgrade rollbacks v1.1.2Eric Wong2-3/+3
This release is fixes a long-standing bug where the original PID file is not restored when rolling back from a USR2 upgrade. Presumably most upgrades aren't rolled back, so it took over a year to notice this issue. Thanks to Lawrence Pit for discovering and reporting this issue.
2010-07-13Merge commit 'v1.0.1' into 1.1.x-stableEric Wong0-0/+0
* commit 'v1.0.1': unicorn 1.0.1 - bugfixes only SIGHUP deals w/ dual master pid path scenario launcher: do not re-daemonize when USR2 upgrading tee_input: safer record separator ($/) handling
2010-07-13unicorn 1.0.1 - bugfixes only v1.0.1Eric Wong2-3/+3
The first maintenance release of 1.0.x, this release is primarily to fix a long-standing bug where the original PID file is not restored when rolling back from a USR2 upgrade. Presumably most upgrades aren't rolled back, so it took over a year to notice this issue. Thanks to Lawrence Pit for discovering and reporting this issue. There is also a pedantic TeeInput bugfix which shouldn't affect real apps from the 1.1.x series and a test case fix for OSX, too.
2010-07-13SIGHUP deals w/ dual master pid path scenarioEric Wong3-0/+118
As described in our SIGNALS documentation, sending SIGHUP to the old master (to respawn SIGWINCH-ed children) while the new master (spawned from SIGUSR2) is active is useful for backing out of an upgrade before sending SIGQUIT to the new master. Unfortunately, the SIGHUP signal to the old master will cause the ".oldbin" pid file to be reset to the non-".oldbin" version and thus attempt to clobber the pid file in use by the to-be-terminated new master process. Thanks to the previous commit to prevent redaemonization in the new master, the old master can reliably detect if the new master is active while it is reloading the config file. Thanks to Lawrence Pit for discovering this bug. ref: http://mid.gmane.org/4C3BEACF.7040301@gmail.com (cherry picked from commit c13bec3449396b21795966101367838161612d61)
2010-07-13launcher: do not re-daemonize when USR2 upgradingEric Wong1-5/+1
This was accidentally enabled when ready_pipe was developed. While re-daemonizing appears harmless in most cases this makes detecting backed-out upgrades from the original master process impossible. (cherry picked from commit 3f0f9d6d72cf17b34c130b86eb933bbc513b24b3)
2010-07-13SIGHUP deals w/ dual master pid path scenarioEric Wong3-0/+118
As described in our SIGNALS documentation, sending SIGHUP to the old master (to respawn SIGWINCH-ed children) while the new master (spawned from SIGUSR2) is active is useful for backing out of an upgrade before sending SIGQUIT to the new master. Unfortunately, the SIGHUP signal to the old master will cause the ".oldbin" pid file to be reset to the non-".oldbin" version and thus attempt to clobber the pid file in use by the to-be-terminated new master process. Thanks to the previous commit to prevent redaemonization in the new master, the old master can reliably detect if the new master is active while it is reloading the config file. Thanks to Lawrence Pit for discovering this bug. ref: http://mid.gmane.org/4C3BEACF.7040301@gmail.com (cherry picked from commit c13bec3449396b21795966101367838161612d61)
2010-07-13launcher: do not re-daemonize when USR2 upgradingEric Wong1-5/+1
This was accidentally enabled when ready_pipe was developed. While re-daemonizing appears harmless in most cases this makes detecting backed-out upgrades from the original master process impossible. (cherry picked from commit 3f0f9d6d72cf17b34c130b86eb933bbc513b24b3)
2010-07-11unicorn 1.1.1 - fixing cleanups gone bad :x v1.1.1Eric Wong3-4/+4
Unicorn::TeeInput constant resolution for Unicorn::ClientError got broken simplifying code for RDoc. This affects users of Rainbows! and Zbatery.
2010-07-11tee_input: fix constant resolution for client EOFEric Wong1-2/+2
Noticed while hacking on a Zbatery-using application (cherry picked from commit ac15513bb81a345cd12c67702a81a585b8b0514e)
2010-07-08unicorn 1.1.0 - small changes and cleanups v1.1.0Eric Wong1-2/+2
This is a small, incremental feature release with some internal changes to better support upcoming versions of the Rainbows! and Zbatery web servers. There is no need to upgrade if you're happy with 1.0.0, but also little danger in upgrading. There is one pedantic bugfix which shouldn't affect anyone and small documentation updates as well.
2010-07-08cleanup "stringio" requireEric Wong4-3/+1
"stringio" is part of the Ruby distro and we use it in multiple places, so avoid re-requiring it. (cherry picked from commit 0fea004ab093ec4f59d919915a505a136326bd8a)
2010-07-08tee_input: safer record separator ($/) handlingEric Wong1-3/+4
Different threads may change $/ during execution, so cache it at function entry to a local variable for safety. $/ may also be of a non-binary encoding, so rely on Rack::Utils.bytesize to portably capture the correct size. Our string slicing is always safe from 1.9 encoding: both our socket and backing temporary file are opened in binary mode, so we'll always be dealing with binary strings in this class (in accordance to the Rack spec). (cherry picked from commit 1cd698f8c7938b1f19e9ba091708cb4515187939)
2010-07-08prefer "[]" to "first"/"last" where possibleEric Wong2-7/+7
"[]" is slightly faster under Ruby 1.9 (but slightly slower under 1.8). (cherry picked from commit 5ece8c1c33f10e6496dfe5ae1d0d368293278d2d)
2010-07-08tee_input: safer record separator ($/) handlingEric Wong1-3/+4
Different threads may change $/ during execution, so cache it at function entry to a local variable for safety. $/ may also be of a non-binary encoding, so rely on Rack::Utils.bytesize to portably capture the correct size. Our string slicing is always safe from 1.9 encoding: both our socket and backing temporary file are opened in binary mode, so we'll always be dealing with binary strings in this class (in accordance to the Rack spec). (cherry picked from commit 1cd698f8c7938b1f19e9ba091708cb4515187939)
2010-07-06socket_helper: disable documentationEric Wong1-1/+1
(cherry picked from commit 98c51edf8b6f031a655a93b52808c9f9b78fb6fa)
2010-07-06socket_helper: cleanup FreeBSD accf_* detectionEric Wong1-21/+12
Instead of detecting at startup if filters may be used, just try anyways and log the error. It is better to ask for forgiveness than permission :) (cherry picked from commit 2b4b15cf513f66dc7a5aabaae4491c17895c288c)
2010-07-06socket_helper: no reason to check for logger methodEric Wong1-8/+3
We only use this module in HttpServer and our unit test mocks it properly. (cherry picked from commit e0ea1e1548a807d152c0ffc175915e98addfe1f2)
2010-07-06configurator: cleanup RDoc, un-indentEric Wong1-472/+466
No point in redeclaring the Unicorn module in here. (cherry picked from commit e4d2c7c302e96ee504d82376885ac6b1897c666a)
2010-07-06configurator: documentation for new accept optionsEric Wong1-0/+35
The defaults should be reasonable, but there may be folks who want to experiment. (cherry picked from commit 686281a90a9b47bac4dfd32a72a97e6e8d26afa1)
2010-07-06socket_helper: move defaults to the DEFAULTS constantEric Wong1-11/+21
This is to allow Rainbows! to override the defaults. (cherry picked from commit ef8f888ba1bacc759156f7336d39ba9b947e3f9d)
2010-07-06doc: recommend absolute paths for -c/--config-fileEric Wong2-2/+8
Suggested-by: Jeremy Evans ref: http://mid.gmane.org/AANLkTintT4vHGEdueuG45_RwJqFCToHi5pm2-WKDSUMz@mail.gmail.com (cherry picked from commit d7695c25c5e3b1c90e63bf15a5c5fdf68bfd0c34)
2010-07-06socket_helper: tunables for tcp_defer_accept/accept_filterEric Wong2-7/+53
Under Linux, this allows users to tune the time (in seconds) to defer connections before allowing them to be accepted. The behavior of TCP_DEFER_ACCEPT changed with Linux 2.6.32 and idle connections may still be accept()-ed after the specified value in seconds. A small value of '1' remains the default for Unicorn as Unicorn does not worry about slow clients. Higher values provide better DoS protection for Rainbows! but also increases kernel memory usage. Allowing "dataready" for FreeBSD accept filters will allow SSL sockets to be used in the future for HTTPS, too. (cherry picked from commit 646cc762cc9297510102fc094f3af8a5a9e296c7)
2010-07-06http_response: this should be a module, not a classEric Wong1-60/+55
This affects Rainbows!, but Rainbows! is still using the Unicorn 1.x branch. While we're at it, avoid redeclaring the "Unicorn" module, it makes documentation noisier. (cherry picked from commit 5769f313793ca84100f089b1911f2e22d0a31e9d)
2010-07-06http: avoid (re-)declaring the Unicorn moduleEric Wong1-1/+1
It makes for messy documentation. (cherry picked from commit b8b979d75519be1c84818f32b83d85f8ec5f6072)
2010-07-06tee_input: undent, avoid (re)-declaring "module Unicorn"Eric Wong1-204/+203
It makes RDoc look better and cleaner, since we don't do anything in the Unicorn namespace. (cherry picked from commit 6f720afd95d8131a2657c643b97cb18c750ed9f8)