kaitiaki.terminal

Functions

`execute`(command[, timeout, cwd, warn])	Executes a terminal command.
`_custom_subprocess_handler`(command[, timeout, cwd])	Okay, this one deserves an explanation.

Module Contents

kaitiaki.terminal.execute(command, timeout=5 * 60, cwd=None, warn=True)

Executes a terminal command.

Has a default timeout period of 5 minutes.

Parameters:

execute. (command {string} -- The command to)

Keyword Arguments:

time (timeout {int} -- The)
(default (should execute for at most.) – {5*60})

Returns:

tuple – A 3-tuple representing (stdout, stderr, reason for termination)

kaitiaki.terminal._custom_subprocess_handler(command, timeout=5 * 60, cwd=None)

Okay, this one deserves an explanation.

I noticed that for jobs that took longer than timeout seconds to run, they wouldn’t be killed correctly. See the following SO post that I made:

https://stackoverflow.com/questions/70587181/terminate-child-process-on-subprocess-timeoutexpired/72135833

So what I ended up finding out is that Python treats subprocesses which spawn groups weirdly. STARS, I suppose, does such a thing. Essentially, the SIGKILL signal gets sent to the subprocess – but not the group. In turn, this means that the subprocesses that do spawn groups don’t get killed when a TimeoutException is thrown. So, what follows is a copy of the Python source code, with the group-killer added (see “the magic line!” below).