| *** lpiwowar|PTO is now known as lpiwowar | 14:33 | |
| frickler | dansmith: just saw another oom on the integrated storage job, failing in gate this time. seems concurrency==6 is rather ambitious, there were 11 qemus running when lvcreate(sic) oomed https://zuul.opendev.org/t/openstack/build/976a98609e92465986d11a5d507261e9 | 18:25 |
|---|---|---|
| dansmith | frickler: yeah, the number of VMs we can start with concurrency=6 is my primary concern | 18:26 |
| dansmith | however, lvcreate OOMing is weird | 18:26 |
| dansmith | oh, | 18:26 |
| dansmith | we were crossing words I think | 18:26 |
| dansmith | lvcreate invoked the killer but mysql is what was killed | 18:27 |
| dansmith | so that fits with what I'd expect and what we see everywhre | 18:27 |
| dansmith | when you said you saw a python3 OOM, I thought you meant it was chosen by the killer which would have been very strange | 18:27 |
| dansmith | anything that allocates memory can invoke the killer, even if it's using almost no memory, so it's not usually the culprit | 18:28 |
| dansmith | and the killer kills so that process can continue as expected, so it usually isn't even the thing that fails | 18:28 |
| frickler | ah, right, I didn't check that closely enough, sorry | 18:30 |
| opendevreview | Jakub Skunda proposed openstack/tempest master: Remove test duplication between tempest and n-t-p QuotasTest https://review.opendev.org/c/openstack/tempest/+/892677 | 21:39 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!