Framework may have a problem in waiting for the SIGQUIT exit

[thiagomacieira]

Seen in an extracted log:

"Sep 19, 2022 @ 18:05:48.381","# Executing test testname 'xxx'"
"Sep 19, 2022 @ 18:10:48.573","# Could not start 'ps': No such file or directory"
"Sep 19, 2022 @ 18:11:09.459","not ok 1842 testname       # timed out"

The 5 minutes between the first line and the second are the standard timeout

opendcdiag/framework/sandstone.cpp

Lines 690 to 692 in 34a2a7a

	Duration result = regular_duration * 5 + 30s;
	if (result < 300s)
	result = 300s;

The interval between the second and third lines is 21 seconds, which is suspiciously close to the timeout for the SIGQUIT:

opendcdiag/framework/sandstone.cpp

Lines 1483 to 1486 in 34a2a7a

	/* timed out, force the child to exit */
	debug_hung_child(child);
	terminate_child();
	if (!do_wait(20s)) {

The "could not start 'ps'" message comes from the debug_hung_child call.

Is it possible this do_wait is failing to realise that the process has exited/crashed and always times out, though gets the result right?