| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-09-10 | |||
| 14:48:48 | smcginnis | sean-k-mooney: Yep, non-client library freeze was last week. So those are already done for victoria. | |
| 14:49:04 | smcginnis | Sneaks up quick. | |
| 14:53:50 | sean-k-mooney | oh ya i know i am just used to waiting till the main project reach rc1 for the master branch to reopen and stable brances to be created | |
| 14:54:14 | sean-k-mooney | smcginnis: but creating the branch early also works | |
| 14:54:45 | sean-k-mooney | smcginnis: also its nice to not have to manually do this in gerrit anymore although i guess i have not done that in a few releases at this point | |
| 14:55:24 | smcginnis | Unofficial projects would need to, but anything official should all be going through the releases repo. | |
| 14:55:42 | sean-k-mooney | yep | |
| 15:24:38 | bauzas | lyarwood: gibi: we don't have better docs for glance image cloning but a spec ? https://specs.openstack.org/openstack/nova-specs/specs/victoria/approved/nova-image-download-via-rbd.html | |
| 15:26:39 | gibi | bauzas: there is an open doc patch https://review.opendev.org/#/c/731577/ | |
| 15:27:20 | bauzas | uh ok | |
| 15:28:08 | bauzas | gibi: so we have 2 different things related to RBD | |
| 15:28:14 | bauzas | gibi: one is the glance multistore | |
| 15:28:16 | gibi | yes | |
| 15:28:24 | bauzas | the other being the precaching | |
| 15:28:29 | gibi | yes | |
| 15:28:36 | bauzas | honestly, I'd then write two different bullets | |
| 15:28:56 | bauzas | because most ops probably don't give a clue about what a glance multistore is | |
| 15:29:06 | bauzas | and both are independent | |
| 15:29:27 | bauzas | ie. you can precache with or without multistore, and the other way too | |
| 15:29:35 | bauzas | dansmith: amirite ? | |
| 15:30:02 | dansmith | precache in nova has nothing to do with glance multistore | |
| 15:30:32 | bauzas | that's what I understood | |
| 15:30:44 | bauzas | the rbd dowload, that's the precache, right? | |
| 15:30:55 | dansmith | no | |
| 15:31:01 | bauzas | holy shit then | |
| 15:31:04 | bauzas | I'm confused | |
| 15:31:22 | dansmith | the rbd download thing is just letting nova download the image direct from rbd instead of streaming it through glance, if it has access to the rbd that backs glance | |
| 15:31:49 | bauzas | ok, that's what I understood, but then what's precaching ? | |
| 15:32:04 | bauzas | oh | |
| 15:32:11 | bauzas | pulling it to libvirt cache? | |
| 15:32:30 | bauzas | whether you pull it thru Glance or from the RBD directly | |
| 15:32:34 | bauzas | correct ? | |
| 15:32:52 | dansmith | yes, just triggering the image cache to download the image, separately from an actual instance boot | |
| 15:32:58 | bauzas | ok | |
| 15:33:06 | bauzas | so, what was merged in V ? | |
| 15:33:14 | bauzas | from the notes, I see precaching | |
| 15:33:20 | bauzas | ah na | |
| 15:33:22 | bauzas | nvm | |
| 15:33:25 | dansmith | all of that is separate from the glance multistore support in nova :) | |
| 15:33:30 | bauzas | it's RBD download | |
| 15:33:37 | bauzas | precaching was a U thing | |
| 15:33:45 | bauzas | correct ? | |
| 15:33:57 | bauzas | U or T, I don't remind exactly | |
| 15:34:32 | dansmith | I think precaching was T yeah | |
| 15:34:33 | dansmith | sorry, U | |
| 15:35:14 | sean-k-mooney | so the two new things in V are the rbd direct download and seperate the multi store support | |
| 15:35:41 | dansmith | yes, but I'm not sure the rbd download thing is worth a special callout in the release notes, if that's what we're discussing here | |
| 15:35:57 | sean-k-mooney | in the prelude ya | |
| 15:36:14 | bauzas | dansmith: kk, now I understand, thanks | |
| 15:36:22 | bauzas | yeah, for the prelude, we'll see | |
| 15:36:34 | bauzas | we don't have a shit ton of new stuff, even a single microversion to provide | |
| 15:36:51 | bauzas | but I agree, this sounds just a small improvement | |
| 15:37:01 | bauzas | not worth a prelude | |
| 15:37:12 | bauzas | (even if that went somehow to the cycle highlights :p ) | |
| 15:40:39 | openstackgerrit | Stephen Finucane proposed openstack/nova master: tox: Enable parallel docs build https://review.opendev.org/751034 | |
| 15:42:03 | sean-k-mooney | stephenfin: cool i basically only did the tox.ini part of that locally and didnt really know what to do for the extentions | |
| 15:42:30 | stephenfin | sean-k-mooney: it won't work though. oslo.config is borked | |
| 15:42:56 | stephenfin | I've pushed a patch for it but it could be a while before it lands. It's a bug but also not massively high priority | |
| 15:42:57 | sean-k-mooney | oh so they actully are not parralel safe | |
| 15:43:07 | stephenfin | afraid not | |
| 15:43:17 | sean-k-mooney | ah ok | |
| 15:43:34 | sean-k-mooney | well ya not super high priority the release notes job does seam to work properly | |
| 15:43:55 | sean-k-mooney | so maybe we coudl start with that and oter that work and do the rest later | |
| 15:44:05 | sean-k-mooney | but ya this is just a nice to have | |
| 15:48:20 | gibi | nova meeting will start in 12 minutes on openstack-meeting-3 | |
| 15:48:29 | lyarwood | bauzas / elod ; would you mind hitting this for artom please https://review.opendev.org/#/q/topic:bug/1879787+branch:stable/ussuri+status:open | |
| 15:49:12 | bauzas | lyarwood: I can, | |
| 15:49:29 | bauzas | but rushing up to finish the prelude before the meeting | |
| 15:54:42 | artom | I'm gonna have to skip the meeting, need lunch before running a time-sensitive errand | |
| 15:56:26 | gibi | artom: ack | |
| 15:56:32 | gibi | artom: enjoy | |
| 15:56:54 | elod | lyarwood: will do after the meeting, if I'm still needed there :) | |
| 15:58:28 | sean-k-mooney | what the env argument to get extra loging in func test | |
| 15:58:36 | sean-k-mooney | is it OS_DEBUG? | |
| 15:59:04 | artom | sean-k-mooney, yarp | |
| 15:59:19 | artom | But remember that if it passes, nothing is printed | |
| 15:59:35 | sean-k-mooney | its failing | |
| 15:59:43 | sean-k-mooney | also holy crap that a lot of logs | |
| 16:00:12 | artom | One might say it's a forrest :D | |
| 16:00:41 | bauzas | sean-k-mooney: OS_DEBUG=1 | |
| 16:00:50 | bauzas | and asserFalse | |
| 16:01:00 | bauzas | assertFalse(True) if you wanna get'em | |
| 16:01:05 | bauzas | or pdb | |
| 16:01:21 | sean-k-mooney | yep OS_DEBUG=True tox -e functional -- 1815153 works | |
| 16:01:29 | sean-k-mooney | but it pring all the db migration too | |
| 16:01:33 | bauzas | right | |
| 16:01:46 | bauzas | I generally pipe the output in a file | |
| 16:02:00 | bauzas | 2>&1 | tee myfile | |
| 16:02:03 | sean-k-mooney | ya i think ill send it to lnav or something | |
| 16:02:35 | bauzas | but most of the time, it's just for debugging purpose and in this case, I pdb directly | |
| 16:02:48 | bauzas | way faster | |
| 16:03:04 | sean-k-mooney | ya so the issue not really simple to pdb debug | |
| 16:03:38 | sean-k-mooney | i mean i can but if im going that far ill proaly just run it in a ide | |
| 16:04:22 | sean-k-mooney | for some reason changing form using the az filter to using placement for az is chaning the behviaor fo some tests | |
| 16:04:36 | sean-k-mooney | not in all cases but in like 8 tests | |
| 16:05:01 | sean-k-mooney | well also the compute status filter | |
| 16:05:43 | sean-k-mooney | this particalar case is not az related | |
| 16:13:28 | openstackgerrit | Stephen Finucane proposed openstack/nova master: WIP: Support old & new versions of svm and vmx traits https://review.opendev.org/660515 | |
| 16:17:37 | bauzas | stephenfin: can't see xenapi removal notes in https://docs.openstack.org/releasenotes/nova/unreleased.html | |
| 16:17:46 | bauzas | stephenfin: could you please point to me the patch itself ? | |