exec_helpers.Subprocess().check_call timeout parameter doesn't (seem to) kill the process while running

[rayburgemeestre]

The feature I would like to request is being able to set a timeout like this:

with exec_helpers.Subprocess() as executor:
    try:
        executor.check_call("/tmp/test.sh", shell=False, timeout=2)
    except exec_helpers.ExecHelperTimeoutError as e:
        print(e)
    except exec_helpers.CalledProcessError as e:
        print(e)

If test.sh contains: sleep 5; echo 123 it would be nice if the process throws the timeout error after 2 seconds, currently it's done after 5 seconds.

To compare with "raw" subprocess:

try:
    # this works, throws after 2 seconds
    subprocess.check_output("/tmp/test.sh", stderr=subprocess.STDOUT, timeout=2, shell=False)
except subprocess.TimeoutExpired as e:
    print(e)
except subprocess.CalledProcessError as e:
    print(e)


try:
    # this also doesn't work, throws after 5 seconds, because of the shell=True
    subprocess.check_output("/tmp/test.sh", stderr=subprocess.STDOUT, timeout=2, shell=True)
except subprocess.TimeoutExpired as e:
    print(e)
except subprocess.CalledProcessError as e:
    print(e)

[penguinolog]

Found source issue: after polling model change, lock model was not updated and after process kill we hang in output reader.
Will fix ASAP.

[penguinolog]

After investigation and debug: switch on process tree kill, it prevents zombies and helps with on-time resources unlock.

[rayburgemeestre]

Thanks, version 2.1.2. works for me!