| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2018-06-14 | |||
| 14:43:34 | sahid | artom: look at conductor/tasks/live_migrate.py at this point the scheduler (or user) selected a destination | |
| 14:44:55 | mriedem | cdent: i always thought that song was about weed, but i'm not so sure | |
| 14:44:56 | sahid | there is a rpcapi check_can_live_migrate_destination. so it's here we you we have to claim the resources | |
| 14:45:17 | cdent | it's a metaphor of at least two things, including weed | |
| 14:45:21 | jangutter | Re: AIX, 15 years ago at Uni, I did a bit of tech support on one of our top researcher's AIX workstation. He did everything in TeX. He also had no subdirectories. He knew every filename by heart. | |
| 14:45:29 | jangutter | that's my memory of AIX. | |
| 14:45:59 | cdent | jangutter: that sounds like every memory of AIX I've ever heard | |
| 14:46:13 | mriedem | i think my dad knew every filename on his desktop by heart because that's where every file went | |
| 14:46:18 | mriedem | miles and miles of desktop | |
| 14:46:35 | dansmith | my dad makes all the important files named in ALL CAPS | |
| 14:46:43 | mriedem | at least he has a system | |
| 14:46:46 | mriedem | *SYSTEM | |
| 14:46:52 | dansmith | which means he has a directory full of files with ALL CAPS names | |
| 14:46:59 | dansmith | because they're all important when you create them | |
| 14:46:59 | mriedem | i had a girlfriend in college that tapped in all caps, | |
| 14:47:05 | mriedem | sounded like she was yelling at me constantly | |
| 14:47:05 | dansmith | um | |
| 14:47:12 | mriedem | *typed | |
| 14:47:14 | dansmith | oh | |
| 14:47:34 | mriedem | it was a short and painful relationship | |
| 14:47:42 | dansmith | you mean PAINFUL | |
| 14:47:49 | cdent | mriedem: the song is also about being on the dole | |
| 14:47:59 | dansmith | mriedem: I slapped that ironic host manager fix into the pike backport | |
| 14:48:21 | dansmith | will experimental run an ironic job on that? | |
| 14:48:28 | mriedem | we have a check queue ironic job | |
| 14:48:30 | mriedem | non-voting | |
| 14:48:39 | dansmith | I didn't see it fail | |
| 14:49:03 | mriedem | yeah... | |
| 14:49:07 | mriedem | http://logs.openstack.org/48/573248/4/check/ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa/c7bd653/ | |
| 14:49:10 | mriedem | what gives | |
| 14:49:27 | dansmith | was it on resource classes already? | |
| 14:49:54 | mriedem | we're not using the ironic host manager apparently in that job http://logs.openstack.org/48/573248/4/check/ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa/c7bd653/logs/etc/nova/nova.conf.txt.gz | |
| 14:50:00 | mriedem | but i bet tripleo or someone else | |
| 14:50:00 | dansmith | yeah | |
| 14:50:01 | mriedem | *is | |
| 14:50:12 | mriedem | which is how tonyb was beaten over the head | |
| 14:50:18 | dansmith | experimental runs tripleo right? | |
| 14:50:46 | mriedem | hold dear caller | |
| 14:51:31 | mriedem | don't think so, unless that's this http://logs.openstack.org/48/573248/4/check/ironic-tempest-dsvm-ipa-wholedisk-bios-agent_ipmitool-tinyipa/c7bd653/logs/etc/nova/nova.conf.txt.gz | |
| 14:51:34 | mriedem | but it doesn't say tripleo | |
| 14:51:39 | edleafe | cdent: uh, no, it isn't. http://www.songfacts.com/detail.php?id=397 | |
| 14:52:08 | mriedem | gdi wrong link | |
| 14:52:20 | mriedem | dansmith: answer is i don't know | |
| 14:52:23 | dansmith | mriedem: hmm, yeah, not running a tripleo job | |
| 14:52:27 | dansmith | I just started experimental | |
| 14:53:45 | cdent | edleafe: on the weed or the dole? because here in the uk "everyone" acknowledges it is both, because of the kutchie to dutchie doubling | |
| 14:56:49 | sahid | mriedem, mdbooth_: if you can have a look at this when you have a moment https://review.openstack.org/#/c/570528/ | |
| 14:56:59 | edleafe | cdent: How does a cooking pot relate to being on the dole? | |
| 14:57:01 | mriedem | jmlowe: so what was the https issue with nova client? | |
| 14:57:09 | cdent | edleafe: sharing | |
| 14:57:16 | sahid | this is half fixing an issue, the real fix that would handle all the case is a bit more complicated | |
| 14:57:30 | sahid | wanted to get you idea | |
| 14:57:38 | jmlowe | It wouldn't parse the endpoint correctly to pick up on https unless there was a trailing slash | |
| 14:58:22 | jmlowe | /v2.1 doesn't work /v2.1/ does | |
| 14:59:09 | cdent | edleafe: important to remember that any pop song up until the maybe late 80s that can't otherwise be explained is probably about the dole: https://www.youtube.com/watch?v=RZ2oXzrnti4 (of course that one is pretty easy to explain) | |
| 14:59:34 | cdent | s/pop song/pop song in the uk/ | |
| 15:02:22 | mriedem | sahid: i don't like that patch ,see comments inline | |
| 15:03:02 | jmlowe | mriedem: oddly osc does parse correctly nova cli doesn't | |
| 15:06:34 | mriedem | jmlowe: ok i thought that was resolved in the patch i linked from andreykurilin yesterday, but apparently not | |
| 15:06:38 | mriedem | b/c i'm pretty sure you had that fix | |
| 15:06:40 | kashyap | mriedem: Yeah, I agree with your comment here: https://review.openstack.org/#/c/570528/5//COMMIT_MSG@17 | |
| 15:07:38 | jmlowe | mriedem: I did, however I'll probably need the workaround in place forever, once a broken client gets out it's nearly impossible to fix | |
| 15:07:41 | mriedem | kashyap: the only problem is, in check_can_live_migrate_destination, i don't think we actually *know* the source instance image backend | |
| 15:08:22 | kashyap | mriedem: Nod; yeah, I was going to look into that code | |
| 15:08:26 | openstackgerrit | Merged openstack/nova master: XenAPI: define a new image handler to use vdi streaming https://review.openstack.org/486475 | |
| 15:08:35 | openstackgerrit | Merged openstack/nova master: XenAPI: update the document related to vdi streaming https://review.openstack.org/568444 | |
| 15:08:38 | kashyap | On a call, will look further once off of it | |
| 15:08:56 | sahid | mriedem: yes it means we have to add new field in migrate_data | |
| 15:09:12 | sahid | your comment makes sense i don't know what is the use case behind | |
| 15:09:25 | sahid | perhaps they try to use live-migration to do conversions | |
| 15:10:38 | sahid | if we add that new field for migrate_data, i guess we could so make this possible instead of raising an exception but.. ofcourse if there is a real use-case | |
| 15:11:27 | mriedem | sahid: yeah i think we can put the image type in the migrate data object during check_can_live_migrate_destination and then check that in check_can_live_migrate_source and fail if they are incompatible | |
| 15:12:13 | mriedem | that won't be backportable, but it's ovh and i assume they'll run with a patch for whatever they need anyway | |
| 15:12:55 | mriedem | dansmith: this makes me think, we should likely also use your long rpc timeout thing for check_can_live_migrate_destination | |
| 15:13:10 | mriedem | since that's an rpc call from conductor to the dest, which then calls back to the source which then returns back to conductor | |
| 15:13:25 | dansmith | mriedem: cool | |
| 15:13:29 | sahid | aarents: ^ | |
| 15:15:19 | mriedem | dansmith: i'm guessing you want to do that in a followup | |
| 15:15:28 | mriedem | or i / someone can | |
| 15:15:37 | mriedem | i left a comment for reminders | |
| 15:18:42 | mriedem | jmlowe: so you're hitting this i guess? https://github.com/openstack/python-novaclient/blob/master/novaclient/v2/versions.py#L91 | |
| 15:19:25 | dansmith | mriedem: I can spark a follow up yeah | |
| 15:19:49 | jmlowe | yeah, best guess is that endswith v2.1 doesn't actually match but "/v2.1/" in url.path does | |
| 15:20:06 | mriedem | jmlowe: ok did you report a novaclient bug? | |
| 15:20:14 | mriedem | did/can | |
| 15:20:37 | mriedem | i assume keystoneauth1 has some magic that we can use to just replace this | |
| 15:20:41 | mriedem | or osc-lib | |
| 15:21:04 | jmlowe | osc works just fine | |
| 15:21:15 | mriedem | right, but i'd like to fix nova client too :) | |
| 15:21:37 | jmlowe | does help when trying to use heat | |
| 15:21:52 | mriedem | yeah, if you can report a novaclient bug that would be awesome | |
| 15:26:13 | aarents | mriedem: sahid yes our needs is to convert host and instance on it from qcow2 to raw in "live". we can run live local storage migration (blockcopy) but during sometime for adminitrative purpose we need to have live-migration working even if they are not yet converted | |
| 15:27:05 | mriedem | aarents: if live migrating from qcow2 to raw is safe, then we could probably build that into a pre-live migration check for allowable conversions | |
| 15:27:11 | mriedem | so there would be 2 rules: | |
| 15:27:16 | mriedem | 1. exact image backend type match - ok | |
| 15:27:24 | mriedem | 2. qcow2 -> raw: ok? | |
| 15:27:30 | aarents | but I will dig into migrate_data / check_can_live_migrate_destination as suggested to avoid incompatible migration | |
| 15:27:32 | mriedem | everything else is not a match and we need to find another dest host | |
| 15:27:40 | aarents | yes | |
| 15:27:44 | mriedem | i honestly don't know what dragons lurk on #2 | |