action #123646
closed
coordination #121876: [epic] Handle openQA review failures in Yam squad - SLE 15 SP5
Run statistical investigation regarding RAM/CPU for Installer regarding bsc#1207284
Added by JERiveraMoya almost 2 years ago.
Updated almost 2 years ago.
Description
Motivation¶
The idea is to check if just with the minimum 1G/1cpu we can hit the bug in similar fashion than with 2G/4cpu.
Acceptance criteria¶
AC1: Run statistical investigation (20 runs per each case) to try different configuration and follow up on #123589
Results are weird, we never hit the "rogue workqueue" message in openqa whatever the settings unless we are in the functional group, but it fails in almost every case with ""BUG: workqueue lockup - pool cpus=0 node=0 flags=0x0 nice=0 stuck for 155s!" which I assume to be the same bug.
results: https://openqa.suse.de/tests/overview?distri=sle&version=15-SP5&build=66.1-smptest&groupid=456.
it is
- sporadic with 1 cpu 1G, machine=64bit
- sporadic with 1 cpu 1G, machine=64bit-smp
- happens always with 1 cpu, 2G, machine=64bit
- happens (almost) always with 1 cpu, 2G, machine=64bit-smp (weird that occurence is more with more ram om one CPU, bad luck?)
- happens always with 2 cpu, 2G, machine=64bit-smp
- happens always with 4 cpu, 2G, machine=64bit-smp
- happens always with 4cpu, 1G, machine=64bit-smp
- Never happens with 4 cpu, 4G, machine=64bit-smp
- Never happens with 1 cpu, 4G, machine=64bit
So:
- Never happens with 4G of ram.
- Always happens with 2G of ram or less, except on single core.
- Sporadic with 1G of ram on one single core, but happens always with more ram, but not sure that's relevant
Conclusion: probably the lack of ram is what provokes the bug, but there seem to be a small difference between one or more cores on low memory.
- Status changed from In Progress to Resolved
- Parent task set to #121876
- Related to action #123691: Add installation in staging with VIDEOMODE text mode added
Also available in: Atom
PDF