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