3 Occasionally odd {issues}[link:ISSUES.html] arise without a transparent or
4 acceptable solution. Those issues are documented here.
6 * For notes on sandboxing tools such as Bundler or Isolate,
7 see the {Sandbox}[link:Sandbox.html] page.
9 * nginx with "sendfile on" under FreeBSD 8 is broken when
10 uploads are buffered to disk. Disabling sendfile is required to
11 work around this bug which should be fixed in newer versions of FreeBSD.
13 * When using "preload_app true", with apps using background threads
14 need to restart them in the after_fork hook because threads are never
15 shared with child processes. Additionally, any synchronization
16 primitives (Mutexes, Monitors, ConditionVariables) should be
17 reinitialized in case they are held during fork time to avoid
18 deadlocks. The core Ruby Logger class needlessly uses a MonitorMutex
19 which can be disabled with a {monkey patch}[link:examples/logger_mp_safe.rb]
23 * Under Ruby 1.9.1, methods like Array#shuffle and Array#sample will
24 segfault if called after forking. Upgrade to Ruby 1.9.2 or call
25 "Kernel.rand" in your after_fork hook to reinitialize the random
28 See http://redmine.ruby-lang.org/issues/show/2962 for more details
30 * Rails 2.3.2 bundles its own version of Rack. This may cause subtle
31 bugs when simultaneously loaded with the system-wide Rack Rubygem
32 which Unicorn depends on. Upgrading to Rails 2.3.4 (or later) is
33 strongly recommended for all Rails 2.3.x users for this (and security
34 reasons). Rails 2.2.x series (or before) did not bundle Rack and are
35 should be unnaffected. If there is any reason which forces your
36 application to use Rails 2.3.2 and you have no other choice, then
37 you may edit your Unicorn gemspec and remove the Rack dependency.
39 ref: http://mid.gmane.org/20091014221552.GA30624@dcvr.yhbt.net
40 Note: the workaround described in the article above only made
41 the issue more subtle and we didn't notice them immediately.
43 * WONTFIX: code reloading and restarts with Sinatra 0.3.x (and likely older
44 versions) apps is broken. The workaround is to force production
45 mode to disable code reloading as well as disabling "run" in your
49 Since this is no longer an issue with Sinatra 0.9.x apps, this will not be
50 fixed on our end. Since Unicorn is itself the application launcher, the
51 at_exit handler used in old Sinatra always caused Mongrel to be launched
52 whenever a Unicorn worker was about to exit.
54 Also remember we're capable of replacing the running binary without dropping
55 any connections regardless of framework :)