The s6-svc program

s6-svc sends commands to a running s6-supervise process. In other words, it's used to control a supervised process; among other benefits, it allows an administrator to send signals to daemons without knowing their PIDs, and without using horrible hacks such as .pid files.

Interface

     s6-svc [ -wu | -wU | -wd | -wD | -wr | -wR ] [ -T timeout ] [ -s signal | -abqhkti12pcyrPCK ] [ -oduDUxO ] servicedir

s6-svc sends the given series of commands to the s6-supervise process monitoring the servicedir directory, then exits 0. It exits 111 if it cannot send a command, or 100 if no s6-supervise process is running on servicedir.

Options

-a : send a SIGALRM to the supervised process
-b : send a SIGABRT to the supervised process
-q : send a SIGQUIT to the supervised process
-h : send a SIGHUP to the supervised process
-k : send a SIGKILL to the supervised process
-t : send a SIGTERM to the supervised process
-i : send a SIGINT to the supervised process
-1 : send a SIGUSR1 to the supervised process
-2 : send a SIGUSR2 to the supervised process
-p : send a SIGSTOP to the supervised process
-c : send a SIGCONT to the supervised process
-y : send a SIGWINCH to the supervised process
-s signal : send signal to the supervised process. signal can be given as its name (case- insensitive) or its number, but only the signals listed above are accepted - you cannot, for instance, manually send a SIGSEGV to the supervised process.

-o : once. Equivalent to "-uO".
-d : down. If the supervised process is up, send it a SIGTERM (by default) then a SIGCONT (to make sure even stopped processes receive the signal aimed to kill them) and do not restart it. The SIGTERM default can be changed by editing the ./down-signal file in the service directory.
-D : down, and create a ./down file so the service does not restart automatically if the supervisor dies. This option is mostly used by automated systems working on top of s6; as a human user, you probably don't need it.
-u : up. If the supervised process is down, start it. Automatically restart it when it dies.
-U : up, and remove any ./down file that may exist, in order to make sure the service is automatically restarted even if the supervisor dies. This option is mostly used by automated systems working on top of s6; as a human user, you probably don't need it.
-x : exit. When the service is asked to be down and the supervised process dies, s6-supervise will exit too. This command should normally never be used on a working system. Note that if this command is sent and a ./finish script exists for the service, the last ./finish invocation before s6-supervise exits will run with its stdin and stdout redirected to /dev/null.
-O : mark the service to run once at most. iow: do not restart the supervised process when it dies. If it is down when the command is received, do not even start it.
-Q : once at most, and create a ./down file. Like -D, but do not terminate the service if it is currently running.
-r : If the service is up, restart it, by sending it a signal to kill it and letting s6-supervise start it again. By default, the signal is a SIGTERM; this can be configured via the ./down-signal file in the service directory.
-T timeout : if the -wstate option has been given, -T specifies a timeout (in milliseconds) after which s6-svc will exit 1 with an error message if the service still hasn't reached the desired state. By default, the timeout is 0, which means that s6-svc will block indefinitely.
-wd : s6-svc will not exit until the service is down, i.e. until the run process has died.
-wD : s6-svc will not exit until the service is down and ready to be brought up, i.e. a possible finish script has exited.
-wu : s6-svc will not exit until the service is up, i.e. there is a process running the run executable.
-wU : s6-svc will not exit until the service is up and ready as notified by the daemon itself. If the service directory does not contain a notification-fd file to tell s6-supervise to accept readiness notification, s6-svc will print a warning and act as if the -wu option had been given instead.
-wr : s6-svc will not exit until the service has been started or restarted.
-wR : s6-svc will not exit until the service has been started or restarted and has notified readiness.

-P : send a SIGSTOP to the process group of the supervised process
-C : send a SIGCONT to the process group of the supervised process
-K : send a SIGKILL to the process group of the supervised process

Usage examples

 s6-svc -h /service/httpd

Send a SIGHUP to the process represented by the /service/httpd service directory. Traditionally, this makes web servers reload their configuration file.

 s6-svc -r /service/sshd

Kill (and automatically restart, if the wanted state of the service is up) the process represented by the /service/sshd service directory - typically the sshd server.

 s6-svc -wD -d /service/ftpd

Take down the ftpd server and block until the process is down and the finish script has completed.

 s6-svc -wU -T 5000 -u /service/ftpd

Bring up the ftpd server and block until it has sent notification that it is ready. Exit 1 if it is still not ready after 5 seconds.

 s6-svc -wR -t /service/ftpd

Send a SIGTERM to the ftpd server; wait for s6-supervise to restart it, and block until it has notified that it is ready to serve again. See the NOTES section below for a caveat.

 s6-svc -a /service/httpd/log

Send a SIGALRM to the logger process for the httpd server. If this logger process is s6-log, this triggers a log rotation.

Internals

s6-svc writes control commands into the servicedir/supervise/control FIFO. An s6-supervise process running on servicedir will be listening to this FIFO, and will read and interpret those commands.
When invoked with one of the -w options, s6-svc executes into s6-svlisten1, which will listen to service state changes and spawn another s6-svc instance (without the -w option) that will send the commands to the service. Any error message written during the waiting period will mention it is being written by s6-svlisten1; this is normal.

Notes

The -t and -r options make s6-supervise send a signal to the service if it is up; but if the service is currently down, they do nothing, and in particular they do not instruct s6-supervise to bring the service up. Consequently, s6-svc -rwr servicedir may wait forever for the service to be up, if it is currently wanted down. To avoid that, make sure your service is wanted up by using s6-svc -ruwr servicedir instead.
The U and D letters, which convey the same idea as u and d (up and down) but with added emphasis, do not have the same meaning in the -U/-D and -wU/-wD options. In the -U/-D case, they mean "change the external service configuration to match what the supervisor has been instructed that the starting state of the service should be". In the -wU/-wD case, they mean "wait until the service has reached the wanted state and also is ready" (or "ready to be started again" for -wD). The thing to remember is "it's up/down, with something more", but the "something" isn't the same.