Thursday, 2026-09-17

gibi_ashigupt: I will try to get to it but at the moment it is somewhere in the middle of my todo lit07:03
gibi_lis07:03
gibi_list07:03
*** bauzas7 is now known as bauzas07:54
bauzashey folks, back08:09
bauzasgibi_: I mentioned https://review.opendev.org/c/openstack/nova/+/1005727 but this requires a bit of cleanup on my side per the last comments, fwiw08:16
gibi_ack08:25
gibi_I push what I have and then start looking at yours08:26
opendevreviewBalazs Gibizer proposed openstack/nova master: Fix PCI allocation leak on failed cold migration  https://review.opendev.org/c/openstack/nova/+/100600309:00
gibi_bauzas: that is mine ^^ it reuses an existing RPC call originally used for live migration during cold migration to clean up the leaked MoveClaim09:01
bauzasack09:01
gibi_bauzas: left questions in the fix09:13
bauzasty, still trying to untangle all the potential issues09:13
gibi_ack09:14
*** gibi_ is now known as gibi09:14
bauzasas a reminder, the best solution is something persisted09:14
bauzasbut still trying to find a short-term fix for being backportable shortly09:14
gibiahh and I realized that I asked the same question as Dan 09:15
gibiabout how the two parallel request can overlap09:15
bauzasnp09:15
gibiOK moved my comment to the proper place09:26
opendevreviewBalazs Gibizer proposed openstack/nova master: Reproduce bug 2166786  https://review.opendev.org/c/openstack/nova/+/100465809:27
opendevreviewBalazs Gibizer proposed openstack/nova master: Fix PCI allocation leak on failed cold migration  https://review.opendev.org/c/openstack/nova/+/100600309:27
opendevreviewSylvain Bauza proposed openstack/nova master: Guard against deleting a concurrent request's BDM on timeout  https://review.opendev.org/c/openstack/nova/+/100572710:38
opendevreviewSylvain Bauza proposed openstack/nova master: Reject stale reserve_block_device_name RPC on compute  https://review.opendev.org/c/openstack/nova/+/100601110:38
opendevreviewMerged openstack/placement master: pre-commit: Add codespell  https://review.opendev.org/c/openstack/placement/+/93180611:36
gibigmaan: does this clear_events_for_instance needs to run for failed resizes as well https://github.com/openstack/nova/blob/94de0576ca02e0668233bedfbe7e18b87316c708/nova/compute/manager.py#L6418 ? E.g. if driver.migrate_disk_and_power_off raises do we leak a pending resize task?13:11
bauzasgibi: gmaan: I barely paid attention to the gate issues with threading, but I just saw that one https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_c5e/openstack/c5e5e75ba3ba4ea1bbc056c4f9299476/testr_results.html13:52
bauzascontext : https://review.opendev.org/c/openstack/nova/+/100572713:53
gibihttps://bugs.launchpad.net/nova/+bug/216514913:55
gibihttps://review.opendev.org/c/openstack/nova/+/100358613:56
opendevreviewribaudr proposed openstack/nova master: Fix swap disk creation skipped on NFS during cold migration  https://review.opendev.org/c/openstack/nova/+/100603613:56
gibibauzas: that is also one of the todos I have13:56
bauzasack ok13:56
bauzasI didn't wanted to blindly recheck13:56
gibiwe have open bug and a local fix that does not work on the gate :)13:56
bauzasgibi: great questions by the way on my second patch for preveting BDM creation after RPC timeout14:00
bauzasI wasn't having that level of expertise, you're pushing me into looking at all the oslo.service details :)14:00
gibiyeah my feeling is that we are getting deeper and deeper into a rabbithole14:00
gibibut I haven't had the brain power to offer a real backportable alternative14:01
bauzasfwiw, I care about the first patch in the series, I just offered help for the other patch14:02
bauzasif we don't find a short-term solution for the BDMs being created without calling Cinder, I'm all good14:03
bauzasI just want nova to not delete the wrong BDM (the other case) :-)14:03
gibiOK. we can refocus on the first14:03
gibiI understand your priorities ;)14:03
gibibtw mine is also more complicated than what I pushed. we might want to fixes on that prevent the problem to reapper(that I pushed) and a fix in the periodic to clear the old leaks14:04
gibi*two fixes one that14:04
gibiso I'm poking at the periodic14:05
gibiwhile fighiting a race condition in the original reproduce as we revert stuff in an untestable order14:05
gibiaaand I will drop for a wile in 25mins so running out of time too14:06
bauzasyeah I need some brain cycles now to look at your patch14:06
gibiif you look just focus on the RPC implications, the rest might change14:09
bauzasI need to load a bit of context, that's it14:09

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