| *** akekane_ is now known as abhishekk | 06:46 | |
| ykarel | ianw, when you get chance can u revisit https://review.opendev.org/c/openstack/devstack/+/803706 | 07:12 |
|---|---|---|
| ykarel | i did more investigation on ubuntu/centos qemu 4 vs qemu 5 and added details there | 07:13 |
| ianw | ykarel: sorry had my head in react/redux stuff lately. will go back to | 07:32 |
| *** jpena|off is now known as jpena | 07:37 | |
| ykarel | Thanks ianw | 07:46 |
| *** rpittau|afk is now known as rpittau | 07:51 | |
| *** ykarel is now known as ykarel|lunch | 08:23 | |
| *** ykarel|lunch is now known as ykarel | 09:36 | |
| *** ykarel is now known as ykarel|afk | 10:53 | |
| *** dviroel|out is now known as dviroel|ruck | 11:31 | |
| *** jpena is now known as jpena|lunch | 11:33 | |
| *** jpena|lunch is now known as jpena | 12:29 | |
| opendevreview | Lee Yarwood proposed openstack/devstack stable/wallaby: nova: Enable apic removal workaround for bug #1939108 when using QEMU https://review.opendev.org/c/openstack/devstack/+/805632 | 13:02 |
| opendevreview | Lee Yarwood proposed openstack/devstack stable/victoria: nova: Enable apic removal workaround for bug #1939108 when using QEMU https://review.opendev.org/c/openstack/devstack/+/805633 | 13:04 |
| opendevreview | Lee Yarwood proposed openstack/devstack stable/ussuri: nova: Enable apic removal workaround for bug #1939108 when using QEMU https://review.opendev.org/c/openstack/devstack/+/805634 | 13:05 |
| opendevreview | Lee Yarwood proposed openstack/devstack stable/train: nova: Enable apic removal workaround for bug #1939108 when using QEMU https://review.opendev.org/c/openstack/devstack/+/805636 | 13:05 |
| opendevreview | Ade Lee proposed openstack/devstack master: Add option to set chap algoorithms for iscsid https://review.opendev.org/c/openstack/devstack/+/803800 | 13:56 |
| *** pojadhav- is now known as pojadhav | 14:14 | |
| gmann | yoctozepto: kopecmartin these devstack jobs voting things are ready https://review.opendev.org/c/openstack/devstack/+/804287/2 https://review.opendev.org/c/openstack/devstack/+/804282/2 | 14:43 |
| clarkb | ykarel: ianw: the reason swap was reduced to 1GB is that we could no longer use a fast sparse file for the swapfile. This means we have to write as many zeros as we specify the swapfile to be before the job continues. Writing 8GB of zeros is not necessarily quick | 14:56 |
| clarkb | neither is 1GB but we felt that was a reasonable compromise. | 14:56 |
| clarkb | If just really need more than 1GB of swap I personally would push back on the jobs a bit and see if they can be leaner. I've sent emails with information around this in the past and unfrotunately know that not many take action on it | 14:57 |
| ykarel | clarkb, yeap i seen past discussion regarding reasoning and seen the suggestion that instead of changing defaults update swap in jobs wherever necessary | 14:58 |
| ykarel | i see some jobs already using that | 14:58 |
| ykarel | to configure 8GB as swap | 14:59 |
| clarkb | right, when the best approach would be trimming the size of privsep which is ridiculously large compared to what utility it provides | 14:59 |
| ykarel | for devstack c8-stream i proposed to use 4 GB as qemu-5 defaults update not working with default 1GB | 14:59 |
| clarkb | I think that privsep alone is almost 1GB of memroy when you add up all of the instances of it that are running in a devstack job | 15:00 |
| clarkb | but its been a while since I did an indepth look at it (and unlikely to have time to do a current look) | 15:00 |
| ykarel | no idea about that, and haven't thought about that | 15:00 |
| ykarel | i just investigated ubuntu vs c8-stream and that resulted in to qemu-4 vs qemu-5 | 15:01 |
| clarkb | ykarel: right I'm trying to say there is a lot of excess in those jobs that should ideally be trimmed. | 15:01 |
| clarkb | While it may be qemu-4 vs qemu-5 that pushes us over the 1GB of swap limit we shouldn't be swapping at all | 15:01 |
| ykarel | clarkb, yes possibly | 15:01 |
| clarkb | and there are lots of places where we're ignoring problems | 15:01 |
| ykarel | clarkb, so yes i agree alternative can be to be a minimal job with less service and less load with running less tempest test if that's what you meant by trim | 15:03 |
| clarkb | no I mean more that we should improve the actual software. privsep is heavy and probably doesn't need to be, we can address that. I'm sure there are many more situation like that | 15:04 |
| ykarel | but if ubuntu too switches to qemu-5 in future than issue would be seen there too | 15:04 |
| ykarel | clarkb, ohkk that's may need investigation in different projects/aspects | 15:05 |
| ykarel | clarkb, can't be updated swap to 4GB be accepted until other ways is investigated and fixed? | 15:06 |
| ykarel | just for c8-stream devstack jobs | 15:06 |
| clarkb | It is certainly an option but if also the one that seems least likely to create an other improvements. I suppose if the job runtimes get very long again (also related to swapping beacuse swapping jobs are slow) that might change | 15:08 |
| ykarel | yes agree job times increases with slow jobs due to swapping | 15:13 |
| *** dviroel|ruck is now known as dviroel|ruck|lunch | 15:17 | |
| *** jpena is now known as jpena|off | 15:36 | |
| *** rpittau is now known as rpittau|afk | 15:58 | |
| *** dviroel|ruck|lunch is now known as dviroel|ruck | 16:19 | |
| *** slaweq is now known as slaweq_ | 19:50 | |
| opendevreview | Merged openstack/tempest master: Updating NON_ACTIVE_LIST https://review.opendev.org/c/openstack/tempest/+/796698 | 20:28 |
| -opendevstatus- NOTICE: The Gerrit service on review.opendev.org has been restarted for a patch version upgrade, resulting in a brief outage | 21:43 | |
| opendevreview | James Parker proposed openstack/whitebox-tempest-plugin master: Mixed Dedicated and Shared cpu instances https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/805731 | 21:47 |
| *** dviroel|ruck is now known as dviroel|out | 22:37 | |
| *** yoctozepto4 is now known as yoctozepto | 23:37 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!