3 Occasionally odd {issues}[link:ISSUES.html] arise without a transparent or
4 acceptable solution. Those issues are documented here.
6 * Some libraries/applications may install signal handlers which conflict
7 with signal handlers unicorn uses. Leaving "preload_app false"
8 (the default) will allow unicorn to always override existing signal
11 * Issues with FreeBSD jails can be worked around as documented by Tatsuya Ono:
12 http://mid.gmane.org/CAHBuKRj09FdxAgzsefJWotexw-7JYZGJMtgUp_dhjPz9VbKD6Q@mail.gmail.com
14 * PRNGs (pseudo-random number generators) loaded before forking
15 (e.g. "preload_app true") may need to have their internal state
16 reset in the after_fork hook. Starting with \Unicorn 3.6.1, we
17 have builtin workarounds for Kernel#rand and OpenSSL::Random users,
18 but applications may use other PRNGs.
20 * Under some versions of Ruby 1.8, it is necessary to call +srand+ in an
21 after_fork hook to get correct random number generation. We have a builtin
22 workaround for this starting with \Unicorn 3.6.1
24 See http://redmine.ruby-lang.org/issues/show/4338
26 * On Ruby 1.8 prior to Ruby 1.8.7-p248, *BSD platforms have a broken
27 stdio that causes failure for file uploads larger than 112K. Upgrade
28 your version of Ruby or continue using Unicorn 1.x/3.4.x.
30 * For notes on sandboxing tools such as Bundler or Isolate,
31 see the {Sandbox}[link:Sandbox.html] page.
33 * nginx with "sendfile on" under FreeBSD 8 is broken when
34 uploads are buffered to disk. Disabling sendfile is required to
35 work around this bug which should be fixed in newer versions of FreeBSD.
37 * When using "preload_app true", with apps using background threads
38 need to restart them in the after_fork hook because threads are never
39 shared with child processes. Additionally, any synchronization
40 primitives (Mutexes, Monitors, ConditionVariables) should be
41 reinitialized in case they are held during fork time to avoid
42 deadlocks. The core Ruby Logger class needlessly uses a MonitorMutex
43 which can be disabled with a {monkey patch}[link:examples/logger_mp_safe.rb]
47 * Under Ruby 1.9.1, methods like Array#shuffle and Array#sample will
48 segfault if called after forking. Upgrade to Ruby 1.9.2 or call
49 "Kernel.rand" in your after_fork hook to reinitialize the random
52 See http://redmine.ruby-lang.org/issues/show/2962 for more details
54 * Rails 2.3.2 bundles its own version of Rack. This may cause subtle
55 bugs when simultaneously loaded with the system-wide Rack Rubygem
56 which Unicorn depends on. Upgrading to Rails 2.3.4 (or later) is
57 strongly recommended for all Rails 2.3.x users for this (and security
58 reasons). Rails 2.2.x series (or before) did not bundle Rack and are
59 should be unnaffected. If there is any reason which forces your
60 application to use Rails 2.3.2 and you have no other choice, then
61 you may edit your Unicorn gemspec and remove the Rack dependency.
63 ref: http://mid.gmane.org/20091014221552.GA30624@dcvr.yhbt.net
64 Note: the workaround described in the article above only made
65 the issue more subtle and we didn't notice them immediately.
67 * WONTFIX: code reloading and restarts with Sinatra 0.3.x (and likely older
68 versions) apps is broken. The workaround is to force production
69 mode to disable code reloading as well as disabling "run" in your
73 Since this is no longer an issue with Sinatra 0.9.x apps, this will not be
74 fixed on our end. Since Unicorn is itself the application launcher, the
75 at_exit handler used in old Sinatra always caused Mongrel to be launched
76 whenever a Unicorn worker was about to exit.
78 Also remember we're capable of replacing the running binary without dropping
79 any connections regardless of framework :)