3 Occasionally odd {issues}[link:ISSUES.html] arise without a transparent or
4 acceptable solution. Those issues are documented here.
6 * Under some versions of Ruby 1.8, it is necessary to call +srand+ in an
7 after_fork hook to get correct random number generation.
9 See http://redmine.ruby-lang.org/issues/show/4338
11 * For notes on sandboxing tools such as Bundler or Isolate,
12 see the {Sandbox}[link:Sandbox.html] page.
14 * nginx with "sendfile on" under FreeBSD 8 is broken when
15 uploads are buffered to disk. Disabling sendfile is required to
16 work around this bug which should be fixed in newer versions of FreeBSD.
18 * When using "preload_app true", with apps using background threads
19 need to restart them in the after_fork hook because threads are never
20 shared with child processes. Additionally, any synchronization
21 primitives (Mutexes, Monitors, ConditionVariables) should be
22 reinitialized in case they are held during fork time to avoid
23 deadlocks. The core Ruby Logger class needlessly uses a MonitorMutex
24 which can be disabled with a {monkey patch}[link:examples/logger_mp_safe.rb]
28 * Under Ruby 1.9.1, methods like Array#shuffle and Array#sample will
29 segfault if called after forking. Upgrade to Ruby 1.9.2 or call
30 "Kernel.rand" in your after_fork hook to reinitialize the random
33 See http://redmine.ruby-lang.org/issues/show/2962 for more details
35 * Rails 2.3.2 bundles its own version of Rack. This may cause subtle
36 bugs when simultaneously loaded with the system-wide Rack Rubygem
37 which Unicorn depends on. Upgrading to Rails 2.3.4 (or later) is
38 strongly recommended for all Rails 2.3.x users for this (and security
39 reasons). Rails 2.2.x series (or before) did not bundle Rack and are
40 should be unnaffected. If there is any reason which forces your
41 application to use Rails 2.3.2 and you have no other choice, then
42 you may edit your Unicorn gemspec and remove the Rack dependency.
44 ref: http://mid.gmane.org/20091014221552.GA30624@dcvr.yhbt.net
45 Note: the workaround described in the article above only made
46 the issue more subtle and we didn't notice them immediately.
48 * WONTFIX: code reloading and restarts with Sinatra 0.3.x (and likely older
49 versions) apps is broken. The workaround is to force production
50 mode to disable code reloading as well as disabling "run" in your
54 Since this is no longer an issue with Sinatra 0.9.x apps, this will not be
55 fixed on our end. Since Unicorn is itself the application launcher, the
56 at_exit handler used in old Sinatra always caused Mongrel to be launched
57 whenever a Unicorn worker was about to exit.
59 Also remember we're capable of replacing the running binary without dropping
60 any connections regardless of framework :)