Thursday, 2023-01-12

opendevreviewMerged openstack/nova master: Test reschedule with PCI in placement  https://review.opendev.org/c/openstack/nova/+/85462605:06
opendevreviewMerged openstack/nova master: Test multi create with PCI in placement  https://review.opendev.org/c/openstack/nova/+/85466305:12
opendevreviewTobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990007:26
tobias-urdingibi: dansmith ^ maybe that07:26
auniyalHi sean-k-mooney, gibi 08:42
auniyalhow can I validate this bug - https://bugs.launchpad.net/nova/+bug/199673208:42
auniyallike after my changes 08:42
gibistephenfin: if you have time could you get back to this https://review.opendev.org/c/openstack/nova/+/854924 this is now the bottom of the PCI series and I fixed your comments08:46
gibitobias-urdin: added to my queue08:46
gibiauniyal: I think you can trigger late affinity check failures in a func test by injecting a fault there08:48
gibiand observing the effect of the BuildFailureWeigher can be done by crafting the settings of the existing weighers so that thd BuildFailureWeigher is overrepresented in the weighing result08:50
gibitobias-urdin: I left feedback there08:54
gibithanks for proposing a fix for this08:54
tobias-urdingibi: ack, thx!08:56
gibibauzas: I've approved the privsep refactor spec09:10
bauzasthanks09:10
bauzasI was looking into it09:10
bauzasbut I need to go to Uggla's spec 09:10
gibiI saw your +2 from before and I had a -1 that was answered so with sean-k-mooney +2 I felt we have corum09:11
opendevreviewMerged openstack/nova-specs master: Review usage of oslo-privsep library on Nova  https://review.opendev.org/c/openstack/nova-specs/+/86543209:18
opendevreviewLukas Piwowarski proposed openstack/nova stable/yoga: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991209:42
opendevreviewLukas Piwowarski proposed openstack/nova stable/xena: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991309:47
opendevreviewLukas Piwowarski proposed openstack/nova stable/wallaby: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991409:51
opendevreviewLukas Piwowarski proposed openstack/nova stable/victoria: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991509:52
opendevreviewMerged openstack/nova-specs master: Use extend volume completion action  https://review.opendev.org/c/openstack/nova-specs/+/85549009:53
opendevreviewLukas Piwowarski proposed openstack/nova stable/ussuri: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991609:53
opendevreviewLukas Piwowarski proposed openstack/nova stable/train: DNM: Test change in run-tempest role  https://review.opendev.org/c/openstack/nova/+/86991709:55
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Replace usage of compareCPU() with compareHypervisorCPU()  https://review.opendev.org/c/openstack/nova/+/86995010:38
kashyapgibi: When you're about --^10:39
auniyalgibi, right now I have single node devstack setup in that, during this test sql-connection is getting failed at here - https://github.com/openstack/nova/blob/2eb358cdcec36fcfe5388ce6982d2961ca949d0a/nova/compute/manager.py#L175311:13
auniyalI think its a env issue, so will be deploying new devstack11:13
auniyalshould I go with node have 2 compute node ? or single node would be enough ?11:14
auniyalexact error I am getting is this - "oslo_messaging.rpc.client.RemoteError: Remote error: CantStartEngineError No sql_connection parameter is established"11:14
*** labedz_ is now known as labedz11:29
sean-k-mooneygibi: regardign the hacking check for privsep12:16
sean-k-mooneywe dont have to stickly enforce "never import privladged fucntions" it might be nice to but as long as we are deliberate in keeping the contract narrow some limited shareing is proably ok12:17
sean-k-mooneywe just dont want to repate teh current issues we have with the shared ones.12:17
sean-k-mooneyi dont really know how easy it woudl be to do what you suggested in hacking but its not something im against trying12:18
gibisean-k-mooney: ack12:31
gibiI agree that this is not a super imporant thing12:31
gibikashyap: I will check it 12:35
gibiauniyal: "No sql_connection parameter is established" means the configuration is incomplete. Yeah try to rebuild devstack12:36
kashyapgibi: 🙏12:37
gibiauniyal: if you want to reproduce the problem in devstack first then you need two nodes as the BuildFailureWeigher won't fail a build but deproritize the affected compute12:37
gibiauniyal: but I'm not sure how easy it is to fail the late affinity check in devstack without modifying the nova code itself12:37
sean-k-mooneyin genreally if you are wokring on schudler or move op related code its good to do at least some testign in a multi node devstack env but i agree with gibi that repoducing in a functional test is goign to be easier then repoducign in a devstack env12:40
gibikashyap: I'm OK with the direction of https://review.opendev.org/c/openstack/nova/+/86995012:43
gibisean-k-mooney: this ^^ is now a patch that do a drop in replacement of the old compare cpu with a new compar hypervisor cpu, much cleaner than the past trials12:43
sean-k-mooneyack that is what i was hoping for ill take a look12:45
kashyapsean-k-mooney: gibi: Thanks!  Yeah, it looks clean.  I have also asked a Red Hat person to get it tested in a real env12:46
sean-k-mooneythis does not seam to be taking account fo the cpu_extra flags12:47
kashyapgibi: Heh, about the wrapping, I did it locally and didn't just commit it 12:47
gibisean-k-mooney: did the old code took account of it?12:50
gibiif so the I think this will too12:50
sean-k-mooneygibi: no which was a bug in the current implemenation12:50
gibiahh12:50
sean-k-mooneyi also dont think compareHypervisorCPU is correct12:57
opendevreviewMerged openstack/nova-specs master: Allow local scaphandre directory to be mapped to an instance using virtiofs  https://review.opendev.org/c/openstack/nova-specs/+/86188112:59
*** blarnath is now known as d34dh0r5313:05
*** dasm|off is now known as dasm13:10
kashyapsean-k-mooney: What do you mean by not correct?13:16
kashyapgibi: sean-k-mooney: Again, the extra flags are handled elsewhere and will be considered "together"13:16
kashyapExtra flags are handled here: _get_guest_cpu_model_config()13:17
kashyap(gibi's reading is correc here.)13:17
sean-k-mooneyin that case that does not coralate with the know bug13:18
kashyapsean-k-mooney: I gave the fix for the buggy environment to a Red Hat colleague.  I'm waiting for that result too13:18
sean-k-mooneykashyap: myu questions are basically related to the fixutre https://review.opendev.org/c/openstack/nova/+/869950/1/nova/tests/fixtures/libvirt.py13:21
sean-k-mooneyyou seam to have change very little and its surpising that that is sufficent to emulate the new behavior13:21
sean-k-mooneyof the new api13:21
kashyapsean-k-mooney: Exactly, I didn't change much of the fixture, as it is indeed a "drop-in" replacement and I don't think we should implement all the internal mechanics of libvirt13:22
kashyapsean-k-mooney: Oh, we can also keep the vendor check there too.  I don't think that'll affect it either13:23
sean-k-mooneywell then its identical to the ohter one :)13:23
sean-k-mooneyso we can just have one function call the other or alias thme13:23
kashyapYes :)13:23
kashyapHow do I alias? 13:23
sean-k-mooneyhave you any concenrs withthe delta in behavior of the fixture/real code 13:24
sean-k-mooneycompareHypervisorCPU = compareCPU13:24
kashyapI don't have any concerns.  I'm just waiting for a real-life test to come back 13:24
kashyapThanks for looking!13:25
sean-k-mooneyactually the alais wont work in this case13:29
sean-k-mooneykashyap: the fucntion signiture is differnt13:29
kashyapsean-k-mooney: Err, oh, right.  Indeed13:30
sean-k-mooneybut you can jsut havc compareHypervisorCPU call compareCPU in the fixutre 13:30
sean-k-mooneyassuming the vendor bit does not matter13:30
kashyapYep, it does not matter.  I'll do the call and run the tests13:30
kashyapsean-k-mooney: Good you raised the fixture point, I caught a few more buggy unit tests :)13:49
kashyapHmm, I fixed all but 2 of 'em seem a bit tricky; /me digs13:51
opendevreviewTobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990014:09
tobias-urdinsean-k-mooney: weird question, if one implements a scheduler host filter the RequestSpec.image.properties is a ImageProps class, does that get populated for BFV instances? i.e would it read cached properties of image props for that volume and populate ImageProps or read existing image props for image volume spawned from an populate ImageProps?14:20
tobias-urdinI'm guessing that is cached since it exists in RequestSpec from when it was spawned14:20
sean-k-mooneytobias-urdin: yes its gets populated for BFV instances if the volume was created from a glance image it will have the image properties stored in the volume metadata and we also copy them into the nova databasein two locaionts14:29
sean-k-mooneythe request spec and teh instance_system_metadata in teh cell db14:29
sean-k-mooneytobias-urdin: the request spec copy is used for schdulign to avoid needing to get the info out of the cell db14:31
sean-k-mooneyhttps://github.com/openstack/nova/blob/master/nova/objects/request_spec.py#L6414:31
sean-k-mooneythe image filed is an nova imagemeta object which has the image properies embded in it https://github.com/openstack/nova/blob/master/nova/objects/image_meta.py#L7514:32
tobias-urdinsean-k-mooney: thanks that good then it was like I expected :) as always, thanks for the detailed answer!14:35
opendevreviewSofia Enriquez proposed openstack/nova master: WIP: Implement encryption on backingStore  https://review.opendev.org/c/openstack/nova/+/87001214:45
opendevreviewSofia Enriquez proposed openstack/nova master: WIP: Implement encryption on backingStore  https://review.opendev.org/c/openstack/nova/+/87001214:50
opendevreviewTobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990015:11
opendevreviewTobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990015:12
bauzasshall I propose chatgpt for nova-core ? https://twitter.com/sylvainbauza/status/161355664394567270415:24
kashyapbauzas: :D15:24
kashyapbauzas: Please ask it to also fix my unit tests!15:24
kashyap(Or even better, just I submit a blueprint to it, write the rest of the code + docs + tests :P)15:24
bauzaskashyap: its trained data dates from 202115:24
bauzaskashyap: but you can ask it why your test is failing, I'm sure it will find you the bug15:25
sean-k-mooneyhow do you know im not chatgpt in disguise15:26
sean-k-mooneyoh chatgpt can spell15:27
sean-k-mooneythat a problem if i want to use it to automate my job in secret15:27
sean-k-mooneyi should ask it if can respond in seanspeak15:27
dansmithhah15:27
kashyapsean-k-mooney: LOL15:41
kashyapsean-k-mooney: I'd totally believe if you're SeanGPT515:41
kashyap!15:41
tobias-urdinhehe15:45
bauzassean-k-mooney: hahaha15:47
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Replace usage of compareCPU() with compareHypervisorCPU()  https://review.opendev.org/c/openstack/nova/+/86995015:49
bauzassean-k-mooney: well, chatgpt sometimes says some wrong answers, but eventually it finds the right one15:49
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Replace usage of compareCPU() with compareHypervisorCPU()  https://review.opendev.org/c/openstack/nova/+/86995015:51
noonedeadpunkhey folks! I was trying to find a way to get `failed_builds` counter through API but I'm not sure I udnerstand what provides ComputeNode object there. As it's not hypervisor, hosts are deperecated...15:57
noonedeadpunkit looks a bit like internal scheduler thing to me at this point...15:58
noonedeadpunkI mean specifically this thing https://opendev.org/openstack/nova/src/branch/master/nova/scheduler/host_manager.py#L261-L26215:59
noonedeadpunkWhat I'm trying to do is monitor this counter to see when node falls out from scheduling and fix the issue/restart nova-compute to return it back16:01
noonedeadpunkI do see it's in cell database but still no idea how to fetch with api :(16:11
noonedeadpunk(in compute_nodes table)16:12
opendevreviewKashyap Chamarthy proposed openstack/nova master: libvirt: Replace usage of compareCPU() with compareHypervisorCPU()  https://review.opendev.org/c/openstack/nova/+/86995017:07
kashyapgibi: --^ Fixed your nits.  Thanks for your continued reviews.17:10
gmanndansmith: did you get chance to review RBAC flag switch change https://review.opendev.org/c/openstack/nova/+/86621817:52
dansmithgmann: I haven't since your update yet sorry17:52
gmanndansmith: ok, please check its depends-on also when you have time https://review.opendev.org/c/openstack/placement/+/869525  https://review.opendev.org/c/openstack/devstack/+/86978117:53
melwittsean-k-mooney: I'm +2 on the ironic vnc spec after my questions answered, if you want to send it https://review.opendev.org/c/openstack/nova-specs/+/86377318:10
sean-k-mooneymelwitt: i can i was more or less holding off for you to get your questions answered without interferring18:13
sean-k-mooneyi know the responce form the previous iteration but i was away i was filling in context that was not in the spec18:13
melwittsean-k-mooney: I know. wanted to let you know I'm good so you can go ahead18:13
sean-k-mooneyso wanted ot let you get happy with the spec content without that bias18:13
melwittI appreciate that :)18:13
melwittI knew you were +2 already but I didn't feel comfortable +2ing until I understood it better18:14
sean-k-mooneyhehe well i was almost -2 intilaly when it was storign password. with the pivot to addign the ironic vnc proxy that removed most of my concners and this just became a ironic dirver change to add partiy for an exisitn api18:17
sean-k-mooneythe double proxying is a little odd but i see use in the ironic vnc proxy for standalone ironic too18:18
melwittI was also pretty much -2 with the initial proposal. so I saw that was removed which was good but didn't quite understand where the second proxy fit into the flow18:19
melwitt*saw that password storage was removed18:20
sean-k-mooneyya so the current vnc proxy really just wraps the vnc tcp session provided by qemu in a web socket18:20
sean-k-mooneyin the ironic case its more or less the same it will take the tcp session form the ironic proxy and convert it to a websocket for horizon18:21
melwittyeah, I think it makes sense now but at first I was confused18:22
opendevreviewMerged openstack/nova-specs master: new spec: support of vnc console for ironic  https://review.opendev.org/c/openstack/nova-specs/+/86377318:26
gmannsean-k-mooney: gibi: I know you might have this in your list but just a review reminder for placement RBAC change  https://review.opendev.org/c/openstack/placement/+/86561818:40
sean-k-mooneyit is on it but i can try and give it more priority :)18:41
sean-k-mooneyis this a depency for the nova patch to merge or was that something else18:41
opendevreviewMerged openstack/nova master: Remove deleted projects from flavor access list  https://review.opendev.org/c/openstack/nova/+/84913118:50
gmannsean-k-mooney: that was something else. there is no deps.18:52
sean-k-mooneyok that was changing the jobs to enable srbac right18:52
sean-k-mooneywe needed to do that for placment then nova18:52
gmannsean-k-mooney: yes18:52
sean-k-mooneyok and this is adding the service role18:53
sean-k-mooneyand the other changes for the srbac role18:53
sean-k-mooneyok ill take a look shortly18:53
gmannsean-k-mooney: for placement it was a change in fixture which is needed for nova to enable new defaults, this one https://review.opendev.org/c/openstack/placement/+/86952518:53
gmannsean-k-mooney: thanks 18:53
sean-k-mooneyill look at that after so18:56
opendevreviewTobias Urdin proposed openstack/nova master: Use get_rpc_client helper from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990019:24
opendevreviewTobias Urdin proposed openstack/nova master: Use new get_rpc_client API from oslo.messaging  https://review.opendev.org/c/openstack/nova/+/86990019:25
opendevreviewDan Smith proposed openstack/nova master: Persist existing node uuids locally  https://review.opendev.org/c/openstack/nova/+/86391821:20
opendevreviewDan Smith proposed openstack/nova master: Make resource tracker use UUIDs instead of names  https://review.opendev.org/c/openstack/nova/+/86391921:20
opendevreviewDan Smith proposed openstack/nova master: WIP: Detect host renames and abort startup  https://review.opendev.org/c/openstack/nova/+/86392021:21
*** dasm is now known as dasm|off22:31
opendevreviewMerged openstack/nova master: Allow enabling PCI scheduling in Placement  https://review.opendev.org/c/openstack/nova/+/85492423:08

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!