Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2019-11-22
07:50:51 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
10:38:20 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
11:04:47 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
11:13:50 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
11:28:24 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
11:32:34 openstackgerrit Riccardo Pittau proposed openstack/openstacksdk master: Add reset_interfaces argument to patch_node https://review.opendev.org/689414
11:39:08 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
12:10:24 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
12:57:12 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
13:05:14 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
13:07:35 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
13:19:27 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
13:31:32 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
13:38:36 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
14:19:11 openstackgerrit Daniel Bengtsson proposed openstack/python-openstackclient master: Stop testing python 2 in tox and zuul. https://review.opendev.org/695427
14:37:30 openstackgerrit Thomas Goirand proposed openstack/python-openstackclient master: Display the host field and allow filtering by host https://review.opendev.org/695586
17:40:55 efried dtroyer: Sanity check me here please: python-openstackclient is *not* considered a "client library" for purposes of "wait until XX to drop py2 support", right??
17:41:19 efried that rule is for projects that are programmatically consumed by other projects, and osc is not.
17:41:41 efried (programmatically in python; CLI scripting doesn't count)
18:56:40 dtroyer efried: OSC is considered to be a client library in the release freeze timelines because other projects have dependencies on it. For that case the Python API vs CLI distinction doesn't matter. I usually operate with that interpretation for most things without a specific reason otherwise.
18:59:10 efried dtroyer: the specific reason in this case is whether to wait to merge the patch tha removes py2 testing from tox.ini/zuul.
18:59:28 dtroyer specifcally about dropping py2, I am not pushing the envelope to do that quickly but there is a point that not doing so impacts other projects. Think of it as the inverse of lazy-loading, we'll do it when we need to unless someone comes along with the time/energy to do it sooner.
18:59:31 efried if there's nobody that consumes osc by importing it python-wise, there's no reason to wait
18:59:49 efried well, there's a patch in flight being actively worked.
19:00:07 efried https://review.opendev.org/#/c/695427/
19:05:01 dtroyer I can't say that nobody is doing that, the only ones I knew about were some plugins a while back, which is why osc-lib exists now. Hopefully they all have changed.
19:06:00 efried okay, fair enough, might be worth a procedural -2 on ^ in that case
19:06:24 efried I can't imagine how/why anybody could/would be importing python-openstackclient, but whatevs.
19:06:41 efried oh, I think I see what you're saying about plugins.
19:06:52 dtroyer I'll take a look, I'm not sure how else to find out who breaks without just merging it though, I suppose we can revert if necessary
19:06:59 efried well
19:07:15 efried the only way we actually break is if we a) merge this and then b) also merge something that doesn't work in py2
19:07:23 efried given the velocity of the project, that seems vanishingly unlikely.
19:07:56 dtroyer true, I can't remember offhand if any of that changes the jobs that other projects may consume
19:08:16 dtroyer I'll be able to look in an couple of hours
19:08:31 efried no big hurry, the patch needs at least one more PS before it would be ready to go in any case.
19:08:34 efried thanks dtroyer
#openstack-sdks - 2019-11-23
11:38:30 mordred eandersson: I think you're on the right track with that backref - but there are a couple of more backrefs - I'm trying something out with weakref to see if it helps
11:44:38 openstackgerrit Monty Taylor proposed openstack/openstacksdk master: Keep connection backrefs with weakref.proxy https://review.opendev.org/695795
11:44:59 mordred eandersson: perhaps something like that ^^ would help
21:38:41 eandersson Yep mordred first test looks good with that patch
21:39:24 eandersson This is the bug on the Senlin side that lead me down this path https://bugs.launchpad.net/senlin/+bug/1851646
21:39:24 openstack Launchpad bug 1851646 in senlin "senlin-health-manager process consume many memory" [Undecided,New]
22:01:36 eandersson Gonna run a couple of longer tests. Looks like I still need https://review.opendev.org/#/c/695628/ but at least it's a step in the right direction.
#openstack-sdks - 2019-11-25
06:51:03 openstackgerrit Yongli He proposed openstack/python-openstackclient master: Microversion 2.78 - server show topology https://review.opendev.org/680928
13:33:48 mordred gtema, Shrews: I'm not here today - but https://review.opendev.org/695795 fixes an issue eandersson is seeing in production with senlin - I'm not sure if there's any good way to add a test to that patch - but if you get a sec, a review would be good
13:34:08 gtema oki
13:47:46 Shrews I’m not here this week
15:18:35 efried eandersson: Sorry to be a pain, but does the fix work with only the first delta? I'm not sure the second one (in service_description.py) is right.
15:50:41 efried eandersson, gtema, mordred: I put up https://review.opendev.org/695934 to see if I can break it...
15:51:21 gtema efried okay
17:37:22 efried gtema, eandersson, mordred: confirmed https://zuul.opendev.org/t/openstack/build/b03dfb0bb3dd4cee868a50c101e29ea3/log/controller/logs/screen-n-cpu.txt.gz?severity=4
23:44:38 eandersson efried I'll start some tests now
23:44:55 efried eandersson: cool. I'll check back in tomorrow, about to jet for the day.
23:45:35 eandersson sounds good - will take me a few hours anyway to run the tests and verify the results
#openstack-sdks - 2019-11-26
00:42:17 openstackgerrit Yongli He proposed openstack/python-openstackclient master: Microversion 2.78 - server show topology https://review.opendev.org/680928
03:01:19 tonyb So I have a newbie question
03:01:34 tonyb I'm writing my first app using the opensatcksdk
03:01:59 tonyb I need to know how to get to the 'fualt' on a server object
03:02:44 tonyb It doesn't seem to be exposed by default, but I can see it the shell via openstack
03:02:55 tonyb Hmm I guess I could just go see what it does
04:58:29 openstackgerrit Ian Wienand proposed openstack/openstacksdk master: Create an OpenDev docker image https://review.opendev.org/696015
05:11:54 openstackgerrit Ian Wienand proposed openstack/openstacksdk master: Create an OpenDev docker image https://review.opendev.org/696015
05:40:42 openstackgerrit Ian Wienand proposed openstack/openstacksdk master: Create an OpenDev docker image https://review.opendev.org/696015
05:49:16 openstackgerrit Ian Wienand proposed openstack/openstacksdk master: Create an OpenDev docker image https://review.opendev.org/696015
15:08:07 openstackgerrit Monty Taylor proposed openstack/openstacksdk master: Keep connection backrefs with weakref.proxy https://review.opendev.org/695795
17:28:35 eandersson mordred, efried: Is it possible to cross link a bug from sdk to senlin? or do they require separate bug reports?
17:32:31 efried eandersson: sdk is storyboard, right? What's senlin?
17:33:24 eandersson launchpad
17:33:34 eandersson https://bugs.launchpad.net/senlin/+bug/1851646
17:33:34 openstack Launchpad bug 1851646 in senlin "senlin-health-manager process consume many memory" [Undecided,Confirmed]
17:33:52 efried then I think the best you can do is stuff each link in the other's bug repo
17:34:04 efried but diablo_rojo et al might have a better answer
19:08:41 umbSublime Hi, I got a bit more time to work on my "fix_allocations" script again, and just noticed this behaviour change between microversions of the nova-api. http://paste.openstack.org/show/786747/ moving forward will UUID's be used to reference hypervisors ?
19:11:16 efried umbSublime: Yes, UUIDs. Were you expecting sdk to be negotiating the latest microversion?
19:11:18 efried by default
19:11:31 efried there was a bug recently about that...
19:12:19 umbSublime yes I was assuming it would
19:12:29 umbSublime my API support microversion up to 2.60
19:12:43 umbSublime t.conn.compute.get_endpoint_data().max_microversion -> (2, 60)
19:13:05 efried umbSublime: have a look at this https://review.opendev.org/#/c/686600/2/releasenotes/notes/fix-for-microversion-70cd686b6d6e3fd0.yaml
19:13:06 umbSublime I'm alsov curious why the openstackcli doesn't also use the latest microversion
19:13:12 efried that's a different question.
19:14:24 umbSublime efried: curiously ebough though it looks like the SDK doe figure out mi max_microversion accuretely. Meaning version discovery would be working correctly right ?
19:14:36 umbSublime I can't type today sorry :/
19:14:56 efried umbSublime: It's *supposed* to be, but read that reno ^ -- it was broken prior to a fix in ksa.
19:15:15 efried so you want to make sure you have that ksa -- and once you do, you'll want the sdk fix from that patch or other things will break.
19:16:14 umbSublime but isn't ksa doing the right thing in my case: http://paste.openstack.org/show/786748/ (maybe I just don't understand the bug)
19:16:47 efried umbSublime: what ksa version do you have?
19:16:56 efried and what sdk version?
19:17:22 umbSublime Much older than the one mentioned in the bug: keystoneauth1==3.4.0
19:17:49 efried but also, ksa has about fifty different code paths for discovery, depending how it's used. Just because that particular query responds "correctly" doesn't mean whatever sdk is using for hypervisors is doing same.
19:18:19 umbSublime errr nvm, I was outisde my env when running. Im on 3.18.0
19:18:36 umbSublime efried: gotcha, thanks for explanation !
19:18:58 efried np
19:20:22 umbSublime so I guess I need to wait for sdk 0.38.0 ? or is there a 0.37.1 version ? I'd prefer not having to manage backporting a patch
19:21:08 efried oh, yeah, that fix ought to have been released, sec...
19:21:50 efried hmph, it's in stable/train, which *should* have meant it got a default release at some point...
19:22:43 efried yeah, looks like it ought to be in 0.37.0, let me confirm quick..
19:23:55 efried umbSublime: yeah, it should be in 0.37.0. Not sure why gerrit doesn't show that.

Earlier   Later