You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Could probably make the wrapper_timeout variable rather than default to 1.1 * Runtime as it is now... I suspect it's just that with the stress test it takea a bit longer than 6 seconds to settle everything properly at the close.
At full lenght it adds 720 seconds to the run time to account for teardown. I'd suggest running the stress-ng command itself and timing how long it takes to tear down at 60 seconds and get an average... e.g. run this 10 times:
and average the runtime... on my server with 128 cores it completes like this:
ubuntu@barbos:~$ for x in seq 1 10; do /usr/bin/time -f %e stress-ng --aggressive --verify --timeout 60 -q --tz --times --af-alg 0 --bsearch 0 --context 0 --cpu 0 --crypt 0 --hsearch 0 --longjmp 0 --lsearch 0 --matrix 0 --qsort 0 --str 0 --stream 0 --tsearch 0 --vecmath 0 --wcs 0; done
68.97
64.94
70.64
64.05
66.45
63.23
64.87
62.80
64.33
63.08
So there are certainly several cases where that extra 6 seconds is not enough...
Perhaps the fix then is to eiither make the wrapper timeout user-modifyable, or to just increase it to 20 seconds OR 10% of runtime, whichever is greater
Bug Description
stress_cpu function add 10% to runtime which not seems enough for machine have more cores
To Reproduce
Environment
Relevant log output
Additional context
No response
The text was updated successfully, but these errors were encountered: