smooth recovery: timeout waiting for readiness notification

smooth recovery: termination while waiting for readiness notification

Test exec() failure for process service start.

Time-out during start

Test that a timeout doesn't stop a "waits for" dependent to fail to start

Test stop timeout

Unexpected termination with restart

Unexpected termination with restart, with dependent

Restart due to dependent, after unexpected termination

Restart after unexpected termination, start times out

Unexpected termination but only restarts on non-zero exit code.

Unexpected termination but only restarts on specific signals.

Failure to restart should propagate to dependent

Unexpected termination until restarts exhausted, followed by a normal start.

Unexpected termination

Stopping a restarting service, which is in the stopping phase, should prevent restart

Stopping a restarting service, which is in the stopping phase, should prevent restart (this time with a dependent)

Stopping a restarting service, which is in the stopping phase, should prevent restart, and prevent restart of a dependency which is not otherwise acti

Stopping a restarting service, which is in the stopping phase, should prevent restart, but not prevent restart of a dependency which is otherwise acti

Test interrupting start of a service marked skippable

Test stop timeout

Stopping a soft dependency doesn't cause the dependent to stop.

Stopping a milestone dependency doesn't cause the dependent to stop

A failing milestone dependency causes the dependent to fail

If a milestone dependency start is cancelled, dependent doesn't start.

Test that soft dependents of a service reattach when service starts

Test that already-started dependencies are correctly attached when starting

Test that a soft dependency restarts if set to auto-restart and dependent still running

Test that a soft dependency doesn't restart if not set to auto-restart and dependent still running

Termination via stop request

Termination via stop request, ensure reason is reset:

stop twice

Test that starting a service with a waits-for dependency on another - currently stopping - service, causes that service to re-start.

remove current head of queue, return new head:

Called when the watcher has been removed. It is guaranteed by the caller that: - the dispatch method is not currently running - the dispatch method wi

Wrapper for writev: don't fail (but retry instead) on EINTR