Posted	Nick	Remark
#openstack-nova - 2017-07-31
17:57:36	jaypipes	cdent: ok, so do you have all the answers and ideas you need to work on that bug?
17:57:51	cdent	jaypipes: apparently not
17:57:58	cdent	as we keep dismissing solutions
17:58:52	dansmith	is it time for our daily hangout?
17:58:55	jaypipes	cdent: well, all of these patches I pretty much consider just "hey, here's one solution to this problem, can you all check it out".
17:58:58	cdent	or should I just go ahead and do the the VCPU introspection
17:59:04	jaypipes	dansmith: I'm game
17:59:15	cdent	i've got a different hangout now :(
17:59:21	jaypipes	cdent: no, I think we're recommending trying the "just sum it" approach.
17:59:30	jaypipes	from Mr. dansmith
17:59:35	cdent	but we said we can't clean up the sum it approach?
18:00:23	dansmith	if we're the only RP in the allocations,
18:00:25	jaypipes	cdent: for shared providers, we will heal that on the *destination host* but after the move operation is entirely ended..
18:00:36	dansmith	jaypipes: we will?
18:00:39	dansmith	I don't think we will
18:01:08	jaypipes	dansmith: yeah, because _update_usage_from_instance() will overwrite the allocations to match a single amount of the flavor.
18:01:23	dansmith	it can't
18:01:26	dansmith	until after confirm
18:01:37	jaypipes	right, which is what I said above, no?
18:01:38	dansmith	and the destination doesn't know about confirm
18:01:57	jaypipes	dansmith: "but after the move operation is entirely ended.."
18:02:22	dansmith	oh
18:02:28	dansmith	yeah, and you can't
18:02:36	dansmith	the destination host does not know when the move has ended
18:02:57	dansmith	jaypipes: https://hangouts.google.com/call/2phok3vj6nhipcx6gp62zvaly4u
18:23:59	cdent	dansmith, jaypipes still hanging out?
18:24:05	dansmith	cdent: yes
18:24:14	dansmith	cdent: just getting to the "wtf now?" phase
18:30:53	dansmith	mriedem: are you aware of a patch up that adds uuid for migration objects?
18:47:28	openstackgerrit	Merged openstack/python-novaclient master: Updated from global requirements https://review.openstack.org/488283
18:50:29	chohoor	I have a question, driver.deallocate_networks_on_reschedule(instance) will return True or False when nova do reschedulter(/nova/compute/manager.py:1822), why only ironic driver return True but other drivers return False?
19:31:06	mriedem	dansmith: nope
19:34:23	mriedem	sorry, been catching up on internal things that i put off all last week
19:37:02	cdent	dansmith, jaypipes: did you fix everything and make the world safe for moves?
19:37:19	dansmith	cdent: no, but we identified that suicide is the easiest path out
19:37:33	cdent	but it brings on many changes
19:42:32	cdent	dansmith, jaypipes: I have a horrible idea: proxy_id = consumer_uuid.replace(‘-‘, ‘+’)
19:42:50	cdent	because it’s the doubling id
19:47:18	jaypipes	cdent: that doesn't solve the problem we're currently stuck on though, w.r.t. old Ocata computes migrating instances to newer Pike computes
19:49:31	openstackgerrit	Sean Dague proposed openstack/nova master: always show urls in list_cells https://review.openstack.org/487860
19:49:37	sdague	mriedem: I added the cacert test case
19:49:49	sdague	but the rest of the suggested url adds are all actually invalid urls
19:50:06	sdague	I was looking at https://tools.ietf.org/html/rfc3986 to figure out why they were parsing weird
19:54:32	openstackgerrit	Eric Fried proposed openstack/nova master: nova.utils.get_endpoint_data() https://review.openstack.org/488137
19:55:47	cdent	jaypipes: that’s a less negative reaction than I was expecting
19:56:35	openstackgerrit	Sean Dague proposed openstack/nova master: always show urls in list_cells https://review.openstack.org/487860
19:56:38	sdague	efried: also, the typo
19:59:12	sdague	efried: and, any chance you can get powervm to run this patch to figure out what the hostname mismatch problem is - https://review.openstack.org/#/c/488381/ ?
20:02:32	efried	esberglu ^^
20:03:01	efried	sdague - esberglu ran our CI against that patch (with our new CI changes) and it's all good.
20:03:32	esberglu	efried: sdague: Yep. Everything looked fine, it was an issue on our end that's resolved now
20:04:22	efried	sdague Possibly TMI, but the deal is: We have to do some hacking to set up our internal API to run remotely (i.e. not localhost). Before, we were waiting until after stack.sh finished, doing that hacking, and restarting the compute service. We were just accepting that it would fail to start up during stack itself.
20:04:43	efried	sdague Now we've figured out how to hack it before stack.sh, so the compute service should come alive during stack.
20:09:27	sdague	efried: ok, good stuff
20:09:39	sdague	so, I think that just leaves the ironic ordering problem
20:09:49	sdague	which I can special case
20:12:01	efried	sdague Is there a reason you removed the ? from the regex?
20:12:36	efried	sdague This will now do the wrong thing if there's an @ further along in the URL (e.g. in the querystring (which may be the only place it's legal without being escaped))
20:12:49	sdague	efried: no, it's matching on nethost
20:13:08	sdague	the query string is out of bounds for that replacement
20:13:22	efried	ahhhh, cool.
20:20:43	dansmith	cdent: jaypipes: so the bottom patch in that series,
20:20:55	dansmith	to fix the allocations thing by removing the where clause..
20:20:59	dansmith	doesn't seem to work for me
20:21:15	dansmith	gibi's test on top of that still ends up with allocations for both computes after the confirm
20:22:18	cdent	the way gibi_ changed it originally was less instrusive: it simply changed the existing and clause to one condition (the consumer uuid)
20:23:10	cdent	which ought to be the same thing
20:23:33	dansmith	isn't that way jay did?
20:24:19	cdent	i’m looking up the discussion, one sec
20:24:30	openstackgerrit	Jay Pipes proposed openstack/nova master: remove source provider allocs in confirm_resize() https://review.openstack.org/488510
20:24:40	jaypipes	dansmith: have a gander ^
20:24:54	jaypipes	dansmith: still needs tests but I want to get your early feedback.
20:25:25	dansmith	jaypipes: okay but see my question above?
20:25:25	cdent	http://p.anticdent.org/4oEf
20:25:47	jaypipes	dansmith: about the bottom patch in the series?
20:26:03	dansmith	yes
20:26:13	jaypipes	dansmith: haven't looked into that yet. unrelated...
20:26:18	jaypipes	will do so now
20:26:31	cdent	dansmith: that ^ is gibi reporting on what he did, which may not be that all the tests pass but the particular situation was resolved
20:26:55	dansmith	cdent: okay but the specific part of gibi's test that should be fixed by thebottom patch isn't
20:27:17	dansmith	namely, after confirm, and after PUTing a singular allocation, we still pull the doubled allocation out of placement
20:28:14	cdent	dansmith: yeah, and what gibi changed when he ran his local confirmation, was not exactly the same as what jaypipes did in the “bottom change"
20:28:58	mriedem	i'm going to poke around and see if there is anything wrong with the test
20:29:18	dansmith	cdent: his description seems identical to me, what did gibi do differently?
20:30:27	cdent	i’m not certain
20:30:42	cdent	i’m trying to get myself spun up
20:30:50	dansmith	okay
20:31:09	dansmith	I feel like this is one of those times where getting all of us in a room with a (big ass) whiteboard for a week would really help
20:31:40	mriedem	sssshhhh
20:32:11	dansmith	yeah, re-reading that whole discussion, it sure seems like jay's patch is what gibi did
20:33:27	dansmith	oh, hmm
20:33:35	dansmith	I think maybe he didn't update his allocations after confirming
20:34:24	dansmith	ah hah, yep
20:34:52	dansmith	mriedem: if you haven't already, I can fix this and rebase on jay's latest for everyone to see
20:34:57	mriedem	go nuts
20:35:01	jaypipes	go for it.
20:35:11	cdent	dansmith: ?
20:36:02	mriedem	yeah i see it
20:36:12	mriedem	needs to do: allocations = self._get_allocations_by_server_uuid(server['id'])
20:36:14	mriedem	after confirming the resize

Earlier   Later