| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-05 | |||
| 18:33:55 | mriedem | that will proxy through to cinder | |
| 18:36:13 | ildikov | with volume-attachments it works | |
| 18:36:25 | mriedem | volume-attachments doesn't call cinder | |
| 18:36:30 | mriedem | it uses the block_device_mappings table in the nova db | |
| 18:36:53 | mriedem | i've got a newish devstack, i'll see if i can reproduce | |
| 18:38:58 | mriedem | this may take awhile, the cafeteria was playing don henley and i can't get it out of my head | |
| 18:39:06 | mriedem | in a bad way | |
| 18:39:59 | ildikov | pfff, ok, I will pick another one then as volume-list wasn't in the list | |
| 18:40:07 | mriedem | yeah we deleted that... | |
| 18:40:08 | mriedem | forgot | |
| 18:40:48 | mriedem | i'm going to install an older novaclient in a venv | |
| 18:41:35 | scottda | ildikov: I haven't tested that in a while. Let me set it up and see.... | |
| 18:42:22 | ildikov | scottda: if that works I will build that in to my patch, I'm currently only trying to make catalog_info work with Cinder v3 | |
| 18:42:34 | ildikov | scottda: using the latest attach code is hardcoded atm | |
| 18:42:50 | ildikov | mriedem: I was trying volume-attach | |
| 18:43:49 | scottda | ildikov: Well, just adding 'catalog_info = volumev3:cinderv3:publicURL' should just work, without any patches, for existing nova-cinder API | |
| 18:44:10 | scottda | If it doesn't, there's a problem. | |
| 18:44:25 | ildikov | mriedem: that's from the API log: GET http://127.0.0.1:8776/v3/0f810b0913db4fc0a77135d2e40dfd23/volumes/3277e3a3-3c4a-43dc-b6b2-001536c6de89 | |
| 18:45:02 | ildikov | mriedem: and that's the compute log: EndpointNotFound: publicURL endpoint for volumev3 service named cinderv3 in RegionOne region not found | |
| 18:45:23 | ildikov | in the same attach flow on the same Devstack single node | |
| 18:45:36 | ildikov | scottda: I'm experiencing the above ^^ | |
| 18:47:58 | scottda | hmmmm....I get a real IP from "openstack endpoint list", not 127.0.0.1 | |
| 18:48:23 | scottda | ie. 192.168.122.197 | |
| 18:48:35 | scottda | ildikov: What do you get? | |
| 18:48:37 | mriedem | scottda: ildikov: works for me | |
| 18:49:10 | ildikov | scottda: I think I set it this way, it's 127.0.0.1 everywhere | |
| 18:49:39 | ildikov | scottda: but that shouldn't affect whether nova can read the catalog or not I guess | |
| 18:49:43 | scottda | ok | |
| 18:50:11 | ildikov | mriedem: hmm, then I might tear down the whole thing and try again | |
| 18:50:17 | mriedem | cinder.catalog_info = volumev3:cinderv3:publicURL | |
| 18:50:17 | mriedem | this is what i have | |
| 18:50:39 | mriedem | this is that part of the catalog http://paste.openstack.org/show/594037/ | |
| 18:50:40 | ildikov | mine is the same | |
| 18:51:45 | ildikov | mriedem: have you ever experienced that the API and the compute in a single node devstack install behaves differently regarding finding or not the endpoints? | |
| 18:51:52 | mriedem | nope | |
| 18:52:39 | ildikov | I always knew I can do magic, I only wish it would not mess up things all the time but the other way around... :S :) | |
| 18:53:17 | scottda | BTW mriedem , what are your thoughts on making v3 endpoint the default for Nova to use? | |
| 18:53:34 | mriedem | scottda: umm, | |
| 18:53:47 | ildikov | scottda: mriedem: I would be pretty happy :) | |
| 18:53:50 | scottda | I'm not sure if Cinder folks are ready for the switch. Just bringing it up. | |
| 18:53:50 | mriedem | we don't ci that way at all anywhere in nova's gate | |
| 18:54:08 | mriedem | scottda: i think we start by switching one of the nova jobs over to using cinder v3 | |
| 18:54:10 | mriedem | probably the placement job | |
| 18:54:17 | mriedem | which is our test bed for new things | |
| 18:54:28 | scottda | mriedem: OK, just bringing it up since we're talking about it. | |
| 18:54:43 | mriedem | is there a devstack configure to set the cinder catalog info in nova.conf? | |
| 18:54:48 | mriedem | *config | |
| 18:55:09 | scottda | I don't think so, but not sure. | |
| 18:55:14 | mriedem | CINDER_NOVA_CATALOG_INFO | |
| 18:55:28 | mriedem | oh that's to configure cinder | |
| 18:56:07 | mriedem | if there were a devstack variable we could piggyback on from the cinder side that would be good, | |
| 18:56:20 | mriedem | because we can't yet pass localrc through from the job defintiion through to devstack | |
| 18:56:33 | mriedem | and the devstack cores aren't happy about adding more one-off config variables like that | |
| 18:57:09 | mriedem | i don't see anything | |
| 18:57:34 | mriedem | you could talk to sdague about the plumbing bits...but my guess is he'll say he wants it done with localrc | |
| 18:58:26 | dansmith | mriedem: well, progress. now all that fails is "all the listing tests" which is what that patch touches | |
| 18:58:55 | mriedem | from tempest? | |
| 18:59:17 | mriedem | well, baby steps :) | |
| 18:59:29 | mriedem | that's probably an easier fix than what we started thinking was broken this morning | |
| 18:59:54 | dansmith | yeah | |
| 18:59:56 | dansmith | that's my point | |
| 19:00:03 | mriedem | unless the cells v1 api isn't getting any instances from the cell mapping | |
| 19:00:07 | mriedem | that might suck | |
| 19:00:19 | dansmith | well, it's likely easier to resolve still | |
| 19:00:22 | mriedem | does map_cell_and_hosts also call map_instances or whatever? | |
| 19:00:30 | mriedem | my guess is the instances aren't mapped? | |
| 19:01:09 | dansmith | there are no instances at that point, and all created ones should be mapped | |
| 19:01:10 | mriedem | https://github.com/openstack/nova/blob/master/nova/cmd/manage.py#L1182 | |
| 19:01:14 | mriedem | oh right | |
| 19:04:10 | openstackgerrit | Matt Riedemann proposed openstack/nova: WIP Get Cinder v3 microversion https://review.openstack.org/385682 | |
| 19:04:11 | mriedem | scottda: ildikov: ^ | |
| 19:05:06 | mriedem | that's not going to work right no | |
| 19:05:07 | mriedem | *now | |
| 19:05:15 | ildikov | mriedem: sweet, thank you! | |
| 19:06:06 | scottda | ildikov: Just tested using cinder v3 in nova.conf and it works for me. | |
| 19:07:01 | openstackgerrit | Matt Riedemann proposed openstack/nova: WIP Get Cinder v3 microversion https://review.openstack.org/385682 | |
| 19:07:54 | mriedem | ok well ^ should see if things work in a tempest run | |
| 19:08:51 | openstackgerrit | Dan Smith proposed openstack/nova: Multicell support for instance listing https://review.openstack.org/396775 | |
| 19:08:52 | openstackgerrit | Dan Smith proposed openstack/nova: Make the SingleCellSimple fixture a little more comprehensive https://review.openstack.org/410472 | |
| 19:08:52 | openstackgerrit | Dan Smith proposed openstack/nova: Fix non-parameterized service id in hypervisors sample tests https://review.openstack.org/410471 | |
| 19:12:49 | mriedem | speaking of, | |
| 19:12:56 | mriedem | i always thought they said 'fucked up music' in https://www.youtube.com/watch?v=HV3zWSawJiw | |
| 19:12:58 | mriedem | is that just me? | |
| 19:13:06 | mriedem | i'm pretty sure it's 'funked up' | |
| 19:14:02 | openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/415978 | |
| 19:14:16 | ildikov | scottda: thanks, I will reinstall my Devstack then and see | |
| 19:17:03 | ildikov | mriedem: does not really sound 'funked up' to me, but maybe I'm not native enough to hear it especially after having a hint of what I should hear instead :) | |
| 19:17:44 | mriedem | heh | |
| 19:17:48 | mriedem | well, | |
| 19:17:50 | mriedem | i wouldn't go out tonight, | |
| 19:17:55 | mriedem | there is a bathroom on the right | |
| 19:18:17 | mriedem | https://www.youtube.com/watch?v=ExViB7ArQNA | |
| 19:19:49 | ildikov | LOL :) | |
| 19:22:08 | openstackgerrit | Ildiko Vancsa proposed openstack/nova: Implement new attach/detach Cinder flow https://review.openstack.org/330285 | |
| 19:22:09 | openstackgerrit | Ildiko Vancsa proposed openstack/nova: Remove check_attach https://review.openstack.org/335358 | |
| 19:25:03 | ildikov | mriedem: BTW, as you don't seem to go out tonight, if you would happen to have time to check the Remove check_attach patch ^^ that would be great :) | |
| 19:25:52 | openstackgerrit | Sujitha proposed openstack/nova: Add post hook script to run tempest with old n-api https://review.openstack.org/401395 | |
| 19:26:00 | mriedem | my dance card is full all of a sudden | |
| 19:31:53 | ildikov | dansmith: enjoy lunch :) | |
| 19:33:30 | openstackgerrit | Matt Riedemann proposed openstack/nova: Return uuid attribute for aggregates https://review.openstack.org/415031 | |