usage: %MAIN% [options] [command ...]

Run-time control of the Strongloop process manager.

Options:
  -h,--help               Print help and exit.
  -v,--version            Print version and exit.
  -C,--control CTL        Control endpoint for process manager.

The control endpoint for the process manager is searched for if not specified,
in this order:

1. `STRONGLOOP_PM` in environment: an HTTP URL.
2. `http://localhost:8701`: a process manager running on localhost

An HTTP URL is mandatory for remote process managers, but can also be used on
localhost. If the process manager is using HTTP authentication
then valid credentials must be set in the URL directly, such as
`http://user-here:pass-here@example.com:7654`.

When using an HTTP URL, it can optionally be tunneled over ssh by changing the
protocol to `http+ssh://`. The ssh username will default to your current user
and authentication defaults to using your current ssh-agent. The username can be
overridden by setting an `SSH_USER` environment variable. The authentication can
be overridden to use an existing private key instead of an agent by setting the
`SSH_KEY` environment variable to the path of the private key to be used. The
port used for ssh defaults to the standard ssh port (22) but can be overridden
by setting the `SSH_PORT` environment variable.


Global commands: apply to the process manager itself

  info                    Information about the process manager.
  ls                      List services.
  shutdown                Shutdown the process manager.


Service commands: apply to a specific service

`SVC` is either a service ID or service Name. It can be obtained by listing
services using `ls` or `status`.

  create NAME             Create service named NAME.

  cluster-restart SVC     Restart the service SVC cluster workers.
        This is a zero-downtime restart, the workers are soft restarted
        one-by-one, so that some workers will always be available to service
        requests.

  env[-get] SVC [KEYS...]
        List specified environment variables for service SVC. If none are
        specified, list all variables.
  env-set SVC K=V...      Set one or more environment variables for service SVC.
  env-unset SVC KEYS...
        Unset one or more environment variables for service SVC. The
        environment variables are applied to the current application, and the
        application is hard restarted with the new environment after change
        (either set or unset).

  log-dump SVC [--follow]
        Empty the log buffer, dumping the contents to stdout for service SVC.
        If --follow is given the log buffer is continuously dumped to stdout.

  npmls SVC [DEPTH]       List dependencies of the service.

  remove SVC              Remove a service SVC.

  restart SVC             Hard stop and restart service SVC with new config.

  set-size SVC N          Set cluster size for service SVC to N workers.
        The default cluster size is the number of CPU cores.

  start SVC               Start service SVC.

  status [SVC]            Report status for service SVC, or all services if
        no SVC is provided. This is the default command.

  stop SVC                Hard stop service SVC.

  soft-stop SVC           Soft stop service SVC.
  soft-restart SVC        Soft stop and restart service SVC with new config.
        "Soft" stops notify workers they are being disconnected, and give them
        a grace period for any existing connections to finish. "Hard" stops
        kill the supervisor and its workers with `SIGTERM`.

  tracing-start SVC       Restart all workers with tracing on.

  tracing-stop SVC        Restart all workers with tracing off.


Worker commands: apply to a specific worker

A `WRK` specification is either `<SVC>.1.<PID>` or `<SVC>.1.<WID>`, where SVC
is the service ID, `1` is the executor ID (limited to 1 in this release), and
the final part is either the process ID, or the cluster worker ID.

The WRK specification can be copied directly from the output of the status
command.

  cpu-start WRK [TO [ST]]  Start CPU profiling on worker WRK.
        When TO is 0, starts the CPU profiler.

        Only supported on Linux:
          TO is the optional Smart Profiling timeout, in milliseconds (default
          is 0, no timeout). With a timeout, the profiler is activated when
          an event loop stalls longer than TO; i.e. when a script runs for too
          long. The profiling is suspended after the script returns to the
          event loop.

          ST is the number of stalls after which the profiler is stopped
          automatically (default is 0, never auto-stop). View an auto-stopped
          profile with `slc arc`.

  cpu-stop WRK [NAME]      Stop CPU profiling on worker WRK.
        The profile is saved as `<NAME>.cpuprofile`. CPU profiles must be
        loaded into Chrome Dev Tools. The NAME is optional, and defaults to
        `node.<WRK>`.

  heap-snapshot WRK [NAME] Save heap snapshot for worker WRK.
        The snapshot is saved as `<NAME>.heapsnapshot`.  Heap snapshots must be
        loaded into Chrome Dev Tools. The NAME is optional, and defaults to
        `node.<WRK>`.

  objects-start WRK        Start tracking objects on worker WRK.
  objects-stop WRK         Stop tracking objects on worker WRK.
        Object tracking is published as metrics, and requires configuration so
        that the `--metrics=URL` option is passed to the runner.

  patch WRK FILE           Apply patch FILE to worker WRK.

  dbg-start WRK            Enable DevTools Debugger backend in worker WRK.

  dbg-stop WRK             Disable DevTools Debugger backend in worker WRK.