Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-01-30
09:45:59 openstackgerrit Feodor Tersin proposed openstack/nova: libvirt: Call snapshot operations for not file based backends only https://review.openstack.org/414710
09:45:59 openstackgerrit Feodor Tersin proposed openstack/nova: libvirt: Remove backward compatibility checks for disk snapshots https://review.openstack.org/382031
09:46:00 openstackgerrit Feodor Tersin proposed openstack/nova: libvirt: Correct ploop imagebackend attribute https://review.openstack.org/414709
09:59:01 openstackgerrit Rikimaru Honjo proposed openstack/nova: Add checks for preventing unexpected detaching volume https://review.openstack.org/354617
10:07:26 sfinucan lyarwood: Morning. Interesting question on https://review.openstack.org/#/c/425087
10:07:49 sfinucan It appears the test I've added has highlighted a latent Python 3 issue, which has since been fixed on master
10:08:40 sfinucan Do I (a) modify the change to fix said Python issue, (b) backport the Python 3 issue fix first, or (c) remove the test from the patch?
10:10:17 lyarwood sfinucan: I'd go with a and document it as a conflict in the change
10:10:41 lyarwood sfinucan: assuming the fix on master also fixes other py3 issues
10:10:41 sfinucan lyarwood: and that is what I'll do. Cheers :)
10:10:54 sfinucan Yeah, a lot of them
10:11:17 lyarwood sfinucan: yeah cool, then just fix this test up and document, that should be enough IMHO
10:23:53 sfinucan lyarwood: Think that's done now
10:24:23 lyarwood sfinucan: thanks, I'll take a look once CI comes backs.
10:24:37 lyarwood sfinucan: would you mind taking a look at https://review.openstack.org/#/c/399735/ this morning if you have time
10:24:51 sfinucan sure
10:42:27 mdbooth Morning! Now we're out of feature crunch, looking for another +2 on this bugfix which has been sitting a while: https://review.openstack.org/#/c/383859/13
10:50:06 cdent johnthetubaguy, bauzas : I'm a little lost on the state of things. Any salient updates to share?
10:50:36 bauzas cdent: waiting for +2 on the grenade fix
10:50:45 johnthetubaguy cdent: I think bauzas's change is waiting on grenade
10:50:46 johnthetubaguy yeah, that
10:51:05 johnthetubaguy the ironic one isn't read yet: https://review.openstack.org/#/c/404472
10:51:13 johnthetubaguy I was thinking about tidying that up right now actually
10:51:19 johnthetubaguy but I suspect its a bit late to get that in now
10:51:55 cdent johnthetubaguy: cool, that was the one I was wondering about most.
10:52:18 bauzas johnthetubaguy: cdent: well, that one seems at risk
10:52:43 cdent I guess jay will be back today and will have some input. It all remains a bit fuzzy about what the actual intent was.
10:53:11 bauzas cdent: question, how can we know it works with the ironic driver ?
10:53:18 bauzas I can only see one ironic job
10:53:28 bauzas which tells me it's green http://logs.openstack.org/72/404472/25/check/gate-tempest-dsvm-ironic-ipa-wholedisk-agent_ipmitool-tinyipa-ubuntu-xenial-nv/0b27bd3/
10:53:42 bauzas but I'm not sure we're covering the patch with that job
10:54:04 cdent bauzas: That's pretty much the source of my question. Without knowing the overall arc of intent, it is hard to discuss coverage.
10:54:11 cdent Too many people saying "I'm not sure".
10:54:44 bauzas cdent: I'm just trying to explain that I discovered most of the corner cases when I investigated the grenade issue
10:54:50 bauzas wrt the scheduler change
10:55:27 bauzas I'd be far less concerned if I was somehow seeing that some job is running an ironic cloud that does reporting to nova
10:55:27 johnthetubaguy bauzas: the follow up patch was trying to get coverage
10:55:42 bauzas I know the tripleo folks kinda use ironic for their undercloud
10:55:53 bauzas so we could perhaps see if their jobs are happy
10:56:26 cdent Personally, in a perfect universe, I don't think of grenade or tempest as coverage at all. They are smoke. If there aren't functional tests in tree, we're faking it and causing us the problems we encountered at the end of last week. So the _next_ time we come round to this, we should start from tests that do something (not unit tests).
10:56:55 cdent But, given that we are not in a perfect universe, and now is not the next time, it's good you did that digging, and it's what we've got...
10:57:45 cdent I do know that jroll created a dnm devstack change to exercise the ironic-related change. I'm not sure of the current state of that.
10:58:34 johnthetubaguy https://review.openstack.org/#/c/426296
10:58:36 jroll cdent: https://review.openstack.org/#/c/426296/
10:58:44 cdent thanks
10:58:55 jroll er, rebase
10:58:59 bauzas jroll: go to bed :p
10:59:13 jroll I did that once already
10:59:18 bauzas ah, nah, it's just early for you :)
10:59:27 jroll indeed :P
11:00:49 jroll ok so my DNM patch looks happy on the latest code
11:01:01 jroll er no, I can't read
11:01:13 cdent jroll: because you been readin' the wrong stuff
11:01:34 jroll http://logs.openstack.org/96/426296/3/check/gate-tempest-dsvm-ironic-ipa-wholedisk-agent_ipmitool-tinyipa-ubuntu-xenial-nv/fd3d15e/logs/screen-n-cpu.txt.gz?level=WARNING
11:01:44 bauzas jroll: what does this job ?
11:02:03 jroll bauzas: nova+ironic+boot a thing, mostly
11:02:05 cdent (this is a story from a friend's teacher who got upset in some depths of kentucky school: "You know why Johnny can't read? Because he's been reading the wrong stuff!"
11:02:45 johnthetubaguy yeah, I am fixing up that bug right now
11:03:07 cdent Unless there's something I should be doing? I'm going to go back to evaluating the number of requests the placement api is receiving.
11:03:09 jroll "Created resource class record via placement API for resource class CUSTOM_LOL." <- so we're halfway there :P
11:03:30 bauzas cool
11:03:42 bauzas jroll: good to know, so it's non-voting on check ?
11:04:05 jroll bauzas: yes, we've had a job like this non-voting on check for years :)
11:04:14 bauzas jroll: that I know
11:04:23 bauzas jroll: :)
11:04:30 johnthetubaguy FWIW, this is the failure: http://logs.openstack.org/96/426296/3/check/gate-tempest-dsvm-ironic-ipa-wholedisk-agent_ipmitool-tinyipa-ubuntu-xenial-nv/fd3d15e/logs/screen-n-cpu.txt.gz#_2017-01-28_00_25_20_187
11:04:30 jroll :P
11:11:47 jroll bauzas: cdent: btw, we've an experimental job on nova/ironic that uses resource classes (triggering this code), I just have a bit of devstack plugin code left to make that work, should finish that early this week
11:12:05 bauzas cool
11:55:00 artom_ johnthetubaguy, can I ask you to take a look at https://review.openstack.org/#/c/426030/7 when you have a moment?
11:55:30 artom_ It's the device tag 'tag' attribute disappearing issue - you mentioned it needs a microversion in https://review.openstack.org/#/c/423952/, so a spec has been whipped up :)
11:59:46 johnthetubaguy artom_: ah, its that bug fix that needs a microversion, gotcha
11:59:56 openstackgerrit John Garbutt proposed openstack/nova: placement: RT now adds proper Ironic inventory https://review.openstack.org/404472
11:59:59 johnthetubaguy artom_: I will try, honestly it feels like that should be in pike now
12:00:08 johnthetubaguy as RC1 us being cut on thursday
12:00:25 johnthetubaguy but not 100% sure
12:00:57 artom_ johnthetubaguy, I know we're short for time :( But if we can avoid a whole release cycle without device tagging...
12:01:57 artom_ (Yeah, I know the client can be version-pinned)
12:02:07 artom_ johnthetubaguy, anyways, much thanks in advance :)
12:04:25 artom johnthetubaguy, and for what it's worth, the patch is already up and Alex and gmann have given it one look over each: https://review.openstack.org/#/c/424759/
12:07:38 tdurakov hi folks
12:07:49 tdurakov do we have any issues with devstack and cells?
12:08:01 tdurakov fresh multinode env: {"message": "Host 'node2' is not mapped to any cell", "code": 400, "created": "2017-01-30T12:04:26Z"}
12:12:58 sfinucan johnthetubaguy: Could I get some eyes on https://review.openstack.org/#/c/392643/ - we're seeing the referenced bug pop up in the real world
12:13:15 sfinucan artom: Would appreciate it if you could take a look at ^ too, please :)
12:23:46 amrith andreykurilin, thank you
12:32:03 artom sfinucan, sure, I'll get to it this afternoon :)
12:32:18 sfinucan artom: Thanking you kindly, sir
13:09:48 openstackgerrit John Garbutt proposed openstack/nova: DNM: hack ironic with resource providers https://review.openstack.org/426296
13:32:10 edleafe johnthetubaguy: I see that bug and can fix it, but was waiting until after the scheduler meeting discussion for overall direction
13:41:21 johnthetubaguy edleafe: which bug?
13:42:31 edleafe johnthetubaguy: sorry, was reading scrollback
13:42:34 edleafe johnthetubaguy: AttributeError: 'SchedulerReportClient' object has no attribute '_compute_node_to_inventory_dict'
13:43:02 johnthetubaguy edleafe: ah, gotcha, pushed a fix for that, to see if the ironic tests will pass, just rebased that patch with those tests on
13:43:21 edleafe johnthetubaguy: ok, thanks. Still catching up
13:44:12 johnthetubaguy np
13:45:05 johnthetubaguy edleafe: I think what you have should work, although not 100% sure it gets us enough to push it into ocata
13:45:23 edleafe johnthetubaguy: me neither
13:45:47 johnthetubaguy edleafe: if we claimed the new resources we are reporting, then I think that would be good to have
13:45:52 edleafe johnthetubaguy: was really hoping to discuss with jaypipes before we move forward

Earlier   Later