66-svctl

Controls an already supervised service at live defined in a tree.

Interface

    66-svctl [ -h ] [ -z ] [ -v verbosity ] [ -l live ] [ -t tree ] [ -T timeout ] [ -n death ] [ -u | d | r | K | X ] service(s)

This tool expects to find an already supervised service at live defined in the given tree and an already running scandir.

(!) This tool only deals with classic services—see 66‑dbctl for other types of services.

Multiple services can be handled by separating their names with a space. 66‑svctl gathers each of these services in a list called selection. The command is sent to the selection asynchronously.

Exit codes

• 0 success
• 100 wrong usage
• 111 system call failed

Options

• -h : prints this help.

• -z : use color.

• -v verbosity : increases/decreases the verbosity of the command.

  • 1 : only print error messages. This is the default.
  • 2 : also print warning messages.
  • 3 : also print tracing messages.
  • 4 : also print debugging messages.

• -l live : changes the supervision directory of service to live. By default this will be /run/66. The default can also be changed at compile time by passing the --livedir=live option to ./configure. An existing absolute path is expected and should be within a writable and executable filesystem - likely a RAM filesystem—see 66-scandir.

• -t : handles the selection of the given tree. This option is mandatory except if a tree was marked as 'current'—see 66-tree.

• -T timeout : specify a timeout in milliseconds after which 66-svctl will stop trying to reach the desired state of the service; defaults to 3000. timeout is applied to each service of the selection. Separate timeouts for different services can be set in their respective frontend files using the @timeout‑up or @timeout‑down key. The frontend key is prioritized over this option.

• -n death : specifiy the maximum number of service death events that the supervisor will keep track of. If the service dies more than this number of times 66‑svctl will stop trying to reach the desired state of the service. The frontend key @maxdeath is prioritized over this option—see frontend. If this option is not specified and no max-death-tally file exists then the maximum allowed deaths will be 5 by default.

• -u : send an up signal to service. It searches for a notification-fd file—see service startup notifications and frontend. If this file exists 66‑svctl will automatically change to uppercase U.

• -d : send a down signal to service. It searches for a notification‑fd file—see service startup notifications and frontend. If this file exists 66‑svctl will automatically change to uppercase D.

• -r : send a reload signal to service. By default the signal is a SIGTERM; this can be configured with the @down-signal field in the frontend service file. It searches for a notification-fd file—see service startup notifications and frontend. If this file exists 66‑svctl will automatically change to uppercase R.

• -X : exit; the supervisor of the service will exit too. This command should normally never be used on a working system.

• -K : send a SIGKILL to service and keep it down.

s6-svc corresponding signal

The 66‑svctl signals correspond to s6‑svc in the following manner:

• -u = -u
• -U = -uwU
• -d = -d
• -D = -uwD
• -r = -r
• -X = -xd
• -K = -kd

Notes

You can also send a signal in a similar fashion to a classic or longrun service with the s6‑svc tool which has the philosophy "launch and forget" which unlike 66‑svctl does not care for the exit status of the service when a signal is sent.

As an extra convenience the -n option allows to deal directly with max-death-tally from the command line even if it was not specified explicitely in the services frontend file and no corresponding file exists.