Tuesday, 2026-04-28

opendevreviewmelanie witt proposed openstack/nova master: DNM test ceph ipv6  https://review.opendev.org/c/openstack/nova/+/98462300:42
opendevreviewTaketani Ryo proposed openstack/nova master: doc: clarify SEV/SEV-ES behavior with locked_memory  https://review.opendev.org/c/openstack/nova/+/97130000:53
opendevreviewGhanshyam Maan proposed openstack/nova master: Add functional test for cleanup_host cancelling queued live migrations  https://review.opendev.org/c/openstack/nova/+/98647702:18
opendevreviewGhanshyam Maan proposed openstack/nova master: DNM: Test removal of enforce_scope  https://review.opendev.org/c/openstack/nova/+/98647902:44
gibimelwitt: I will take a look04:33
gibimelwitt: the only other option I see on the code level is to change the default of wait_for_state_change04:35
gibimaybe just change it in the coverage job via an env var passed?04:35
gibiwe have OS_TEST_TIMEOUT as an example for a similar thing. We can have another var like OS_SLOW_ENV_WAIT_FACTOR that defaults to 1 and we can set it to 2 in the coverage env. Then this var can be multiplied in any sleep or retry count we see impacted by the slowness in the coverage env04:41
*** ykarel_ is now known as ykarel07:44
opendevreviewLajos Katona proposed openstack/nova master: Use SDK for Neutron subnets  https://review.opendev.org/c/openstack/nova/+/96219008:20
opendevreviewLajos Katona proposed openstack/nova master: Use SDK for Neutron extensions  https://review.opendev.org/c/openstack/nova/+/96227008:20
opendevreviewLajos Katona proposed openstack/nova master: Use SDK for Neutron floating IPs  https://review.opendev.org/c/openstack/nova/+/96260408:20
*** sambork_ is now known as sambork09:50
lajoskatonaUggla, bauzas: Hi, if you have few minutes I have a small doc update patch for mdev_class: https://review.opendev.org/c/openstack/nova/+/983089 (just a follow-up for bp/generic-mdevs  https://review.opendev.org/q/topic:%22bp/generic-mdevs%22)10:14
opendevreviewKamil Sambor proposed openstack/nova master: Change nova-alt-configurations job  https://review.opendev.org/c/openstack/nova/+/98317912:46
opendevreviewMerged openstack/nova master: Doc: add short doc section for mdev_class  https://review.opendev.org/c/openstack/nova/+/98308913:28
opendevreviewFlorian proposed openstack/nova master: Add check for PCIe devices attach limit for volume and ports  https://review.opendev.org/c/openstack/nova/+/95558413:39
opendevreviewDan Smith proposed openstack/nova-specs master: Add unpin-az spec  https://review.opendev.org/c/openstack/nova-specs/+/98653913:59
opendevreviewDan Smith proposed openstack/nova-specs master: Add unpin-az spec  https://review.opendev.org/c/openstack/nova-specs/+/98653914:27
opendevreviewBalazs Gibizer proposed openstack/nova-specs master: add pci passthrough groups spec for 2026.2  https://review.opendev.org/c/openstack/nova-specs/+/97563514:31
melwittgibi: yeah I kinda thought about a wait multiplier too15:12
opendevreviewmelanie witt proposed openstack/nova master: DNM test ceph ipv6  https://review.opendev.org/c/openstack/nova/+/98462315:20
opendevreviewmelanie witt proposed openstack/nova master: DNM test ceph ipv6  https://review.opendev.org/c/openstack/nova/+/98462316:43
melwittgmaan: another unit test stabilization for CI testenv:cover https://review.opendev.org/c/openstack/nova/+/986468 not sure if this will likely be the last "one-off" but if it's a more widespread problem, there are a couple of review comments in there with idea16:55
gmaanmelwitt: lgtm, thanks17:33
melwittthanks gmaan 17:49
melwittsean-k-mooney: was finally able to prove that ipv6 brackets in ceph mon addrs passed to libvirt without stripping the brackets, would fail https://review.opendev.org/c/openstack/nova/+/984623/1817:51
melwittI will likely need your help to figure out where and how the proper place to propose all of this ipv6 stuff haha .... for my test I crammed most of it into a pre-run script as you could see17:52
sean-k-mooneymelwitt: oh cool17:54
sean-k-mooneyi mean that works. it shoudl get added into the devstack roels17:55
sean-k-mooneybut we can figure that out now that we have a working baseline17:55
melwittyeah that was my thinking. hack it for the POC but then propose them in the proper places if it got to working17:56
sean-k-mooneyas long as we are calling it a hack im ok with it if we call it a tempory solution it will out live cochroaches17:57
melwittclaude thought some of it should go under openstack/zuul-jobs so maybe that's true, maybe it isn't haha17:57
sean-k-mooneytempory colution that work never get fixed17:57
sean-k-mooneymelwitt: yes some of it should17:57
sean-k-mooneywe do the multi node bridge configuraiton in zuul/zuul-jobs17:57
sean-k-mooneyhttps://opendev.org/zuul/zuul-jobs/src/branch/master/roles/multi-node-bridge17:58
melwittyeah. I just didn't want to make the depends-on web really big while I was trial-and-erroring on it17:58
melwittyes exactly what it was saying17:58
sean-k-mooneyso we shudl proably make that proply do ipv6 as well17:58
melwittyes exactly17:58
sean-k-mooneyhttps://opendev.org/zuul/zuul-jobs/src/branch/master/roles/multi-node-bridge/tasks/switch.yaml17:58
sean-k-mooneysame for the host file https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/multi-node-hosts-file/tasks/main.yaml17:59
melwittI cloned all of the relevant repos and claude knew where stuff should go, which was really cool considering I would have zero idea. I know you told me first but I also had claude look at all of it locally and it made all of the connections between them18:00
sean-k-mooneyso if you want to see a beautful hack that i have not updated in a long time18:00
melwittsure 18:00
sean-k-mooneyhttps://github.com/SeanMooney/ard/tree/master/molecule18:00
sean-k-mooneyso i dont know if you knwo what molecule is18:01
sean-k-mooneybtu its how you test ansible18:01
sean-k-mooneyhttps://github.com/SeanMooney/ard/blob/master/molecule/default/molecule.yml18:01
sean-k-mooneyi defiend a moluel role using vagrnat/libvtr/qemu18:01
JayFhttps://review.opendev.org/c/openstack/nova/+/983753 this is a high quality change to the Ironic virt driver which would have a significant impact on Ironic users, it's also simpler and easy to understand than most ironic virt driver changes. Your attention would be appreciated :D 18:01
sean-k-mooneythat create a contoler and comptue and then runs the devstack ansible playbooks form the ci jobs on them18:01
melwittno I didn't know what molecule is. I had heard you say it before but I never knew what you were talking about haha18:02
sean-k-mooneyso you can do molecule up adn it will use the ci roles to deploy a 2 node devstack18:02
melwittahh that's pretty cool18:02
sean-k-mooneymelwitt: i used to ocationlly use it to test jobs changes locally18:03
sean-k-mooneyJayF: ay yes that woudl be good to do and yes its pretty simple 18:03
sean-k-mooneymelwitt: i should see if it still work, but you could point it at phsyicla system too and just run it18:04
JayFyou also got a 3-in-1 ironic cores review on it (me/cli.f/ci.d in a GR-OSS team review)18:04
JayFalthough admittedly some % of that time was spent explaining resechedules in nova :D18:04
sean-k-mooneyJayF: technially they nolonger exist18:04
sean-k-mooneywe replaced it with alternitive hosts18:04
sean-k-mooneyso we can aovid going back to the schduler18:05
JayFis there a meaningful difference from an Ironic standpoint?18:05
sean-k-mooneyno18:05
sean-k-mooneythe diffent is es dont goback to the chsduelr if ther first host fails18:05
JayFNice, so you just get to avoid the double dip in the scheduler18:05
sean-k-mooneywe callulate alterntive in advnace and just select the next one18:05
sean-k-mooneyyep18:05
JayFif those alternatives are scheduled to in the meantime18:05
sean-k-mooneythe allcoation fails super early adn we go to the next one18:06
JayFack, so it is a meaningful change for Ironic users :)18:06
sean-k-mooneybasiclly the first thing we do is see if the canidate allcation is still valid and if not we go to the next18:06
JayFthere's nearly zero chance that after a timed out deployment (hours) those allocs would still be valid18:06
sean-k-mooneywell perhasp but this also changed in like pike or rocky18:07
sean-k-mooneyit was part fo the cell_v2 work18:07
JayFNice to know, albeit WAY too late, about the behavior though. Could be surprising for an Ironic user.18:07
* JayF files it away in his nova shelf18:07
JayF(I'm not saying you all had some responsibility to tell us; mainly annoyed I've had a flawed mental model for bordering-on a decade)18:08
JayFcid: clif: Worth reading the backlog if you wanna fill in the gaps around how retrying actually works now ^^^18:09
dansmithyeah I think at this point it has worked that way for longer than it worked the other way :)18:15
JayFdansmith: I still find myself un-learning things that I "learned" from the old, hacked to @%$! cloud at my first OpenStack employer :)18:16
JayFthis is added to the list, a good 12 years later :( 18:16
opendevreviewMerged openstack/nova master: Increase wait retries for resize reschedule with qos port test  https://review.opendev.org/c/openstack/nova/+/98646819:05

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!