| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-07-23 | |||
| 14:10:18 | jlejeune | hm ok, so a check with --help option before running the qemu-img command, just to verify that qemu(img binary is present or not, right ? | |
| 14:10:31 | sean-k-mooney | jlejeune: basically i dont want to do any of the extra setup in that function if its not installed and does not supprot the format | |
| 14:11:44 | dansmith | personally I'd just go for consistency with the one above (which the current version pretty much does) and not worry about it as much since this is all moving to oslo | |
| 14:12:11 | dansmith | I'd also say that I think a dependency on qemu-img being in path and available for tests is very reasonable | |
| 14:12:24 | dansmith | ...for anything in nova for sure | |
| 14:12:36 | dansmith | and definitely for oslo.utils, since it has a bunch of tests for qemu-img info already | |
| 14:12:42 | jlejeune | dansmith: that's was my first proposal, do the same as other tests and waiting for a global move in oslo | |
| 14:12:55 | jlejeune | :/ | |
| 14:13:13 | dansmith | jlejeune: yep, that's fine with me fwiw.. I don't think we need to probe --help, but maybe sean-k-mooney can explain more | |
| 14:14:13 | sean-k-mooney | our downsstream version dont have some format compiled in and as you are aware many disues have an unsafe qemu image so i didnt want to run it even if its just on our local system | |
| 14:14:46 | sean-k-mooney | what proaposed is fine i just wanted to aovid calllign convert/create ectra if we coudl | |
| 14:15:19 | dansmith | changing FileNotFound to Exception (as the previous one you pointed to does) will catch both cases right? | |
| 14:15:50 | sean-k-mooney | yep and i dont think we want to differenciate | |
| 14:16:02 | sean-k-mooney | ill admign because of the other things i just pushed i only quickly skimmed the patch | |
| 14:16:10 | sean-k-mooney | so i might have missed somethign | |
| 14:17:13 | dansmith | yeah, so I think just using bare exception like the case above makes the most sense, jlejeune bauzas probably just didn't realize (as mentioned by the comment above) that it covers format issues and missing file issues | |
| 14:18:04 | jlejeune | dansmith: totally agree, that was my opinion at first | |
| 14:19:44 | bauzas | yeah I missed the first catch | |
| 14:20:07 | bauzas | what I explained is that I prefer to catch specific exceptions mostly | |
| 14:20:19 | bauzas | but we have some exceptions in code where we bare-catch | |
| 14:20:39 | bauzas | here, sean-k-mooney is just pointing out an already existing pattern, so I'm cool with it | |
| 14:20:51 | bauzas | sorry if that created some misunderstanding | |
| 14:21:01 | dansmith | yep, honestly we should raise on FileNotFound and call qemu-img a requirement, IMHO, and skip on anything else, but this code shouldn't be staying in nova so let's just do the consistent thing for now | |
| 14:21:38 | jlejeune | dansmith: ok, so let's go back on my patchset1 :) | |
| 14:21:55 | bauzas | agreed with dansmith here | |
| 14:22:16 | bauzas | keep consistency for now, but I'd appreciate some further rework on that whole crazy module :) | |
| 14:22:26 | jlejeune | :) | |
| 14:27:04 | opendevreview | Julien Le Jeune proposed openstack/nova master: Fix test_vmdk_bad_descriptor_mem_limit and test_vmdk_bad_descriptor_mem_limit_stream_optimized https://review.opendev.org/c/openstack/nova/+/924148 | |
| 14:29:16 | sean-k-mooney | jlejeune: we have some thing in ci that we dont want this to conflict with so we will likely loop back to your review tomorrow or later today | |
| 14:31:20 | jlejeune | sean-k-mooney: no problem | |
| 14:31:40 | dansmith | sean-k-mooney: ++ | |
| 14:31:46 | bauzas | sean-k-mooney: about that, I missed some comment, but could you please rebase the stable patches by adding [stable-only] tag so the order wouldn't be important for merging ? | |
| 14:32:14 | dansmith | that's not necessary, IMHO | |
| 14:32:21 | dansmith | they're up, we can land them in normal order now | |
| 14:32:26 | sean-k-mooney | bauzas: i spoke to gibi about that and came to the same conlcution | |
| 14:32:30 | sean-k-mooney | we dont need to do that in this case | |
| 14:32:43 | dansmith | they just need to be posted so we have URLs for the advisory, but no need to force fast merge them or anything | |
| 14:32:59 | bauzas | well, sure | |
| 14:33:13 | sean-k-mooney | i also want a full set of check results so dont want to promot to gate directly | |
| 14:33:16 | bauzas | I'm saying that because our SLURP branches could merge before the Bobcat one | |
| 14:33:39 | sean-k-mooney | bauzas: that techncially breaking policy | |
| 14:33:40 | bauzas | but we can proceed by order, I'm fine | |
| 14:33:52 | sean-k-mooney | we could but they are equally supported while instable | |
| 14:34:23 | sean-k-mooney | we shoudl not priorities slurps over supported non slurps | |
| 14:34:47 | bauzas | it was just a pragmatic question in terms of merging | |
| 14:34:50 | bauzas | but I'm cool | |
| 14:35:17 | sean-k-mooney | with that said i have promoated zuul to its own monitor which i will be watching until thsee land :) | |
| 14:35:50 | bauzas | ++ | |
| 14:38:02 | frickler | sean-k-mooney that might be a night shift ;-/ | |
| 14:39:17 | frickler | depending on the state of the gate I would still promote them to the head of gate if needed. also ping me if re-gates are needed due to unrelated failures | |
| 14:40:12 | dansmith | we should at least wait until we see a clean run in check, but yeah, thanks | |
| 14:40:53 | frickler | yes, I won't submit them to gate before that | |
| 14:48:17 | sean-k-mooney | frickler: ya we can consider that once the iniall run is complete. at least this time its one commit per branch | |
| 14:56:15 | sean-k-mooney | awsome post failure in one of the jobs.... | |
| 14:56:21 | sean-k-mooney | and its related. | |
| 14:56:32 | sean-k-mooney | well i know what im doing for the next little while | |
| 14:56:39 | sean-k-mooney | https://4b125a6587520ffd7605-3e126156a22dab1f4ba4bdb0e9add880.ssl.cf2.rackcdn.com/924731/1/check/nova-live-migration/a83a99e/testr_results.html | |
| 14:57:19 | dansmith | oye | |
| 14:57:42 | dansmith | sean-k-mooney: thanks for doing that, I gotta run in a sec, but I'll be back later | |
| 14:58:12 | dansmith | sean-k-mooney: in glance we found some tempest jobs were uploading qcows as AMI because AMI is the first format in the disk_image_formats list, so that could be related | |
| 14:58:41 | sean-k-mooney | maybe btu theses are migration test failrues | |
| 14:58:59 | dansmith | sean-k-mooney: https://review.opendev.org/c/openstack/glance/+/923321 | |
| 14:59:03 | sean-k-mooney | but ya ill check if that job has that fixed or not | |
| 14:59:05 | dansmith | sean-k-mooney: yeah I know | |
| 14:59:09 | dansmith | just saying, something to look for | |
| 14:59:22 | sean-k-mooney | i think this is the job we use uec images in so it proably related | |
| 14:59:30 | dansmith | ahh right | |
| 14:59:38 | sean-k-mooney | if thats all it is then we can fix the ci without respining the patch | |
| 14:59:44 | sean-k-mooney | whic would be good | |
| 15:00:10 | sean-k-mooney | lets see if it passes the same test in the hybrid plug job | |
| 15:00:13 | sean-k-mooney | since that is using qcow | |
| 15:00:44 | sean-k-mooney | oh yep it did https://zuul.opendev.org/t/openstack/build/be52199567724f90a6017b7a33fd0e3b | |
| 15:00:57 | sean-k-mooney | so ya this looks liek a job config issue not a patch issue | |
| 15:06:30 | bauzas | guys, we'll have a nova meeting | |
| 15:06:39 | bauzas | but I don't want to conflict with the effort here | |
| 15:07:05 | bauzas | dansmith: sean-k-mooney: are you okay if we run the meeting for a short period of time, say 15 mins | |
| 15:07:23 | sean-k-mooney | its fine | |
| 15:07:33 | sean-k-mooney | we likely wont need to cordiate much | |
| 15:07:43 | sean-k-mooney | so run it as normal | |
| 15:07:51 | bauzas | okay, anyway I'll try to make it quick | |
| 15:08:02 | bauzas | reminder : nova meeting in ~ 1 hour here | |
| 15:14:46 | opendevreview | sean mooney proposed openstack/nova master: update nova-live-migration to use corrrect disk-formats https://review.opendev.org/c/openstack/nova/+/924746 | |
| 15:16:28 | opendevreview | Merged openstack/nova master: conf: Add '[api] response_validation' option https://review.opendev.org/c/openstack/nova/+/915740 | |
| 15:46:14 | sbauza | sorry folks, intermittent problems with my internet | |
| 15:47:06 | bauzas__ | for the moment, I'll stick with bauzas__ | |
| 15:47:27 | bauzas__ | can someone ping me please in order to doublecheck whether you can hear me ? | |
| 15:47:42 | sean-k-mooney | bauzas__: nope sorry cant do that :P | |
| 15:47:50 | bauzas__ | okay cool | |
| 15:48:10 | bauzas__ | fucking French local fiber operator | |
| 15:48:24 | gibi | sean-k-mooney: re live-migration post failure, I see that the the live-migraton job passed on 2023.2 and 2023.1 on the backports | |
| 15:48:43 | sean-k-mooney | gibi: ya i think https://review.opendev.org/c/openstack/nova/+/924746 will fix it on master | |
| 15:48:56 | sean-k-mooney | it also passed in the hybrid plug job | |
| 15:49:06 | sean-k-mooney | which uses the same storage but qcow images | |
| 15:49:08 | gibi | fingres crossed | |
| 15:49:39 | sean-k-mooney | that just completed devstack so we will know one way or another soon | |
| 15:49:57 | sean-k-mooney | if it fails ill update it with the other possibel option i need to set before the jobs completes | |
| 15:50:13 | gibi | OK | |
| 15:50:58 | sean-k-mooney | https://zuul.openstack.org/stream/9fb4f36837c644ae8ff100d21f8a0666?logfile=console.log so far so good | |
| 15:51:40 | bauzas__ | don't trust this guy, that's my znc bouncer that continues to fail ^ | |
| 15:51:42 | bauzas__ | :) | |