Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2026-02-26
14:55:42 Uggla dansmith yes, just re-enforcing the msg. :)
14:59:54 Uggla sean-k-mooney, gmaan super cool for openapi. Thanks for the effort. 👍
15:06:17 cardoe Uggla: I'm hoping we can get https://review.opendev.org/c/openstack/nova/+/972601 into the next release.
15:06:40 cardoe I'm trying to follow up with sean-k-mooney cause his -1 reason is literally what this patch is correcting.
15:07:33 cardoe For context, KVM/libvirt driver does not use the prepare_networks_before_block_device_mapping() while the ironic driver does. Which is what is the root cause of this.
15:07:51 bauzas cardoe: maybe you don't know but today is FeatureFreeze, we prioritize for the moment reviews for blueprints
15:08:07 bauzas cardoe: during the next two weeks, we can then look at bugfixes like this one
15:08:14 Uggla cardoe, that's a bug right. So it is not impacted by FF.
15:08:46 cardoe bauzas: okay I didn't know.
15:09:05 cardoe I was just trying to provide context.
15:09:16 bauzas cardoe: no worries, you can usually find the release cadence there : https://releases.openstack.org/gazpacho/schedule.html
15:09:51 Uggla cardoe, so as bauzas explains we could merge it in the following weeks. Assuming all comments will be addressed
15:09:53 bauzas cardoe: again, no worries, my only concern is about context switching, like for the moment I'm trying to make sure we could accept a SEV-ES usecase
15:10:16 bauzas cardoe: but please ping us next week around Tuesday, we could look at the bug
15:10:21 cardoe Not asking you folks to context switch right now. Just putting info in here so that you can follow up async.
15:10:27 cardoe But I can ping next Tuesday.
15:10:38 bauzas thanks
15:14:39 Uggla cardoe also good to put it in our dev plan: https://etherpad.opendev.org/p/nova-2026.1-status so we should not missed it. (note: I did it for you)
15:15:04 sean-k-mooney cardoe: i was confused by the wroding you had in relese note
15:15:12 sean-k-mooney ill take anouther look
15:15:41 cardoe We've just had 4 bugs opened this cycle alone with users hitting that which was surprising.
15:18:04 sean-k-mooney cardoe: part of the concern i had with that patch was i did not see you addinga unit test that clearly showed what the meesage would be
15:18:46 sean-k-mooney cardoe: i.e. it was not clear at a glance what you were actully fixign when i looked at it
15:19:19 cardoe I can if you'd like. All the other tests to decide what exceptions come up to the user just validate the exception that's raised and not the text of the exception
15:19:28 cardoe So the test would be different than all the others.
15:20:08 cardoe To me it looked like you guys white list certain exceptions that can make it to the user otherwise all other exceptions get wrapped such that the error message to the user is a generic "this step failed"
15:20:22 cardoe But I can change it to validate the string of the exception.
15:21:40 gibi Uggla: I spotted a potential bug in the sev refactor series in https://review.opendev.org/c/openstack/nova/+/967972
15:28:35 tkajinam Returning 400 to user: Invalid input for field/attribute addresses. Value: {'': [{'version': 4, 'addr': '10.1.1.162', 'OS-EXT-IPS:type': 'fixed', 'OS-EXT-IPS-MAC:mac_addr': 'fa:16:3e:cb:ae:2e'}]}. '' does not match any of the regexes: '^.+$'
15:28:47 tkajinam it seems some of the GET requests are broken by schema validation
15:37:23 gibi tkajinam: is it a fresh one? did we landed some new schema validation thst break it?
15:37:58 tkajinam these were caught in https://zuul.opendev.org/t/openstack/build/d293798e270240c18bbfbb8d02d1ed4b which completed a few hours ago
15:38:05 tkajinam https://bugs.launchpad.net/nova/+bug/2142767
15:38:11 tkajinam https://bugs.launchpad.net/nova/+bug/2142765
15:38:24 tkajinam I have to dig into the latter because it is blocking instance creation by heat, it seems
15:40:07 tkajinam hmm I guess these two are connected because the failure might be detected during returning a response
15:41:33 bauzas I reviewed the first patch and I'm looking into the second, but I still have a bit of concerns, so I guess we gonna have to make a draw
15:49:26 opendevreview Takashi Kajinami proposed openstack/nova master: Accept an empty key for addresses https://review.opendev.org/c/openstack/nova/+/978089
15:49:32 gibi both seems to complaim that addresses are empty
15:49:42 Uggla gibi, bauzas ok
15:50:25 tkajinam gibi, yeah. I think the problem is the top key under the addresses are empty and IIUC these keys are network names in neutron
15:50:49 tkajinam and ... name in neutron networks is optional.
15:51:44 tkajinam testing the above patch by https://review.opendev.org/c/openstack/ceilometermiddleware/+/978090 ...
15:54:13 opendevreview Merged openstack/nova master: db: Move regex helpers to utils https://review.opendev.org/c/openstack/nova/+/958745
15:54:32 opendevreview Merged openstack/nova master: tests: Clean up flavors tests https://review.opendev.org/c/openstack/nova/+/958746
15:56:29 sean-k-mooney cardoe: we only allow exepciton that have no hostspecif info to get to the user
15:56:55 sean-k-mooney or ratehr we often only show the name of the exctpion to non admins
15:57:27 cardoe yeah this makes the behavior identical to KVM/libvirt driver for this error
15:58:08 cardoe because the ironic driver has prepare_networks_before_block_device_mapping, it sets up the networks earlier than the KVM/libvirt does.
15:58:34 sean-k-mooney on the nova side we treat any exposure of the hypervior/compute node name as a security hardenign opertunity
15:59:14 sean-k-mooney not nessiarly a cve but we try to treat that as privldage so we are a bit carfeul with the raw tracebacks as a result
16:00:08 sean-k-mooney cardoe: ya idely the behvior shoudl be the same give me a sec to finsih up something now im off a call and ill take a look again
16:01:58 gibi tkajinam: make sense +2
16:20:56 opendevreview Merged openstack/nova master: api: Simplify API version check for flavor description https://review.opendev.org/c/openstack/nova/+/958747
16:59:33 opendevreview Merged openstack/nova master: Deprecate unlimited compute actions https://review.opendev.org/c/openstack/nova/+/975833
16:59:50 opendevreview Merged openstack/nova master: [compute]Use single long task executor https://review.opendev.org/c/openstack/nova/+/977251
17:02:54 sean-k-mooney gibi: ^ :)
17:06:58 gibi \○/
17:07:10 gibi one more to go
17:07:25 sean-k-mooney im watching nova-next finish its tempest run
17:07:28 gibi then we can root for the graceful series on top
17:08:42 sean-k-mooney yep also https://review.opendev.org/c/openstack/nova/+/958748/6 if we can land that
17:14:37 gmaan finally :)
17:20:31 gmaan I think there are 3 changes competing for microversion 2.101
17:20:49 dansmith I just approved the async volume attach one that wants 101
17:20:51 gmaan this one is in gate, which should get it https://review.opendev.org/c/openstack/nova/+/971068
17:20:56 gmaan dansmith: yeah
17:21:44 gmaan I am reviewing the remove-os-volumes_boot which is almost ready but remember the async volume one
17:22:11 gmaan stephenfin: do you want to update remove-os-volumes_boot on top of async vol one?
17:31:34 gmaan stephenfin: I am holding the +w and one comment to update the spec whihc mentioned to return 410 instead of 404
17:32:01 gmaan sean-k-mooney: also, you tagged this cleanup as part of this BP but that is not related, it will be good to keep that separate https://review.opendev.org/c/openstack/nova/+/958863
17:32:30 gmaan stephenfin: ^^
17:33:01 gmaan sean-k-mooney: please ignore, i need to start my auto tab after typing two letter :)
17:41:17 sean-k-mooney ya i mented that seperatly to stephenfin, that it woudl be better to flip the order of that and teh one on top
17:42:07 sean-k-mooney gmaan: i was debeaitn if i shoudl rasie the 404 vs 410 point
17:42:36 sean-k-mooney i decied not to becasue i think the pattern we have is we only return 410 if we remove it form all micoverions
17:43:07 sean-k-mooney but use a 404 if its only remove in a specific microverison
17:43:59 gmaan yes, because we have same patter for other APIs also for example proxy one. and the one which are gone for all microversion they return 410
17:44:46 opendevreview Merged openstack/nova master: Run nova-compute in native threading mode https://review.opendev.org/c/openstack/nova/+/965467
17:45:02 opendevreview Merged openstack/nova master: Use 2nd RPC server in compute operations https://review.opendev.org/c/openstack/nova/+/975588
17:45:15 gibi \○/
17:45:16 gmaan \o/ gibi grab a beer
17:45:30 gibi I'm done for today :)
17:45:46 opendevreview Merged openstack/nova master: Prepare resize/cold migration for graceful shutdown https://review.opendev.org/c/openstack/nova/+/977182
17:45:48 gmaan o/
17:45:52 sean-k-mooney gmaan: so this doesnt remvoe it outright right
17:45:56 gibi o/
17:46:28 opendevreview Merged openstack/nova master: Add more test for graceful shutdown https://review.opendev.org/c/openstack/nova/+/977405
17:46:45 gibi when things start moving...
17:46:47 opendevreview Merged openstack/nova master: Add operator document for graceful shutdown https://review.opendev.org/c/openstack/nova/+/977590
17:46:48 gmaan sean-k-mooney: did not get, you mean for spec update?
17:47:05 sean-k-mooney i ment https://review.opendev.org/c/openstack/nova/+/958864
17:47:35 gibi gmaan: you reached early beer state today :)
17:47:46 sean-k-mooney so ya im jsut clarifyign are you asking for a 410 to be returend or the spec to be updated to 404
17:47:52 gmaan yeah, gate check our patient level and gives blessing
17:48:02 gmaan gibi: heh yeah
17:48:40 gmaan sean-k-mooney: ohk, I am saying code is good i am +2 there. i am asking spec to update once we land the code
17:48:46 gmaan just paper work nothing else
17:48:59 sean-k-mooney ack stephenfin ^ you might want to read back
17:49:45 gmaan which can be done later but I am holding +w due to microversion conflict only

Earlier   Later