3 In general, signals need only be sent to the master process. However,
4 the signals Unicorn uses internally to communicate with the worker
5 processes are documented here as well. With the exception of TTIN/TTOU,
6 signal handling matches the behavior of {nginx}[http://nginx.net/] so it
7 should be possible to easily share process management scripts between
12 * HUP - reloads config file and gracefully restart all workers.
13 If the "preload_app" directive is false (the default), then workers
14 will also pick up any application code changes when restarted. If
15 "preload_app" is true, then application code changes will have no
16 effect; USR2 + QUIT (see below) must be used to load newer code in
19 * INT/TERM - quick shutdown, kills all workers immediately
21 * QUIT - graceful shutdown, waits for workers to finish their
22 current request before finishing.
24 * USR1 - reopen all logs owned by the master and all workers
25 See Unicorn::Util.reopen_logs for what is considered a log.
27 * USR2 - reexecute the running binary. A separate QUIT
28 should be sent to the original process once the child is verified to
31 * WINCH - gracefully stops workers but keep the master running.
32 This will only work for daemonized processes.
34 * TTIN - increment the number of worker processes by one
36 * TTOU - decrement the number of worker processes by one
40 Sending signals directly to the worker processes should not normally be
41 needed. If the master process is running, any exited worker will be
42 automatically respawned.
44 * INT/TERM - Quick shutdown, immediately exit.
45 Unless WINCH has been sent to the master (or the master is killed),
46 the master process will respawn a worker to replace this one.
48 * QUIT - Gracefully exit after finishing the current request.
49 Unless WINCH has been sent to the master (or the master is killed),
50 the master process will respawn a worker to replace this one.
52 * USR1 - Reopen all logs owned by the worker process.
53 See Unicorn::Util.reopen_logs for what is considered a log.
54 Log files are not reopened until it is done processing
55 the current request, so multiple log lines for one request
56 (as done by Rails) will not be split across multiple logs.
58 It is NOT recommended to send the USR1 signal directly to workers via
59 "killall -USR1 unicorn" if you are using user/group-switching support
60 in your workers. You will encounter incorrect file permissions and
61 workers will need to be respawned. Sending USR1 to the master process
62 first will ensure logs have the correct permissions before the master
63 forwards the USR1 signal to workers.
65 === Procedure to replace a running unicorn executable
67 You may replace a running instance of unicorn with a new one without
68 losing any incoming connections. Doing so will reload all of your
69 application code, Unicorn config, Ruby executable, and all libraries.
70 The only things that will not change (due to OS limitations) are:
72 1. The path to the unicorn executable script. If you want to change to
73 a different installation of Ruby, you can modify the shebang
74 line to point to your alternative interpreter.
76 The procedure is exactly like that of nginx:
78 1. Send USR2 to the master process
80 2. Check your process manager or pid files to see if a new master spawned
81 successfully. If you're using a pid file, the old process will have
82 ".oldbin" appended to its path. You should have two master instances
83 of unicorn running now, both of which will have workers servicing
84 requests. Your process tree should look something like this:
97 3. You can now send WINCH to the old master process so only the new workers
98 serve requests. If your unicorn process is bound to an interactive
99 terminal, you can skip this step. Step 5 will be more difficult but
100 you can also skip it if your process is not daemonized.
102 4. You should now ensure that everything is running correctly with the
103 new workers as the old workers die off.
105 5. If everything seems ok, then send QUIT to the old master. You're done!
107 If something is broken, then send HUP to the old master to reload
108 the config and restart its workers. Then send QUIT to the new master