| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-26 | |||
| 21:24:52 | mriedem | bauzas: https://review.openstack.org/#/c/425822/ is merged, i'll update the novaclient 7.1.0 patch while you instrument things | |
| 21:27:17 | openstackgerrit | Sylvain Bauza proposed openstack/nova: Scheduler calling the Placement API https://review.openstack.org/417961 | |
| 21:28:25 | bauzas | mriedem: cool | |
| 21:28:45 | bauzas | mriedem: the logging is done, I'm just now waiting the results... | |
| 21:29:41 | mriedem | si capitan | |
| 21:30:45 | bauzas | France won and goes to final, one good thing at least | |
| 21:32:37 | dansmith | mriedem: I made dash show starred changes in blue, which makes it handy for following stuff through the gate | |
| 21:33:05 | mriedem | cool | |
| 21:38:52 | dansmith | I have been sitting in the same place since this morning, so I'm going to get up for a bit now that my change is in the gate again, and will be back in a bit | |
| 21:39:03 | dansmith | looks like bauzas' change is failing pep8 | |
| 21:39:11 | bauzas | fk | |
| 21:39:17 | bauzas | I edited inline | |
| 21:40:10 | bauzas | http://logs.openstack.org/61/417961/32/check/gate-nova-pep8-ubuntu-xenial/b7bc39a/console.html#_2017-01-26_21_33_30_781456 | |
| 21:40:15 | bauzas | o_O | |
| 21:40:35 | mriedem | bauzas: the wifi is so terrible here i can't even git review -d your novaclient release change | |
| 21:40:44 | mriedem | so i guess it's on you to update that | |
| 21:41:12 | bauzas | mriedem: roger | |
| 21:42:19 | bauzas | mriedem: FWIW, will use inline gerrit | |
| 21:45:01 | bauzas | mriedem: dansmith: do you guys think the pep8 issue would prevent us to see the logs correctly? | |
| 21:45:18 | dansmith | eh? | |
| 21:45:26 | bauzas | I don't give a shit to pep8 since I don't want to merge that like this, but I do care about whether the logs are good | |
| 21:45:39 | bauzas | dansmith: http://logs.openstack.org/61/417961/32/check/gate-nova-pep8-ubuntu-xenial/b7bc39a/console.html#_2017-01-26_21_33_30_781456 | |
| 21:45:40 | dansmith | so you want to let it run? that's fine | |
| 21:45:47 | bauzas | yeah | |
| 21:45:52 | dansmith | biab | |
| 21:46:10 | bauzas | I'll have to bail out in 1h15 so I prefer leave it running | |
| 22:07:26 | openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding vlans field to Device tagging metadata https://review.openstack.org/405951 | |
| 22:07:26 | openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: expose virtual interfaces with vlans to metadata https://review.openstack.org/405954 | |
| 22:07:44 | mriedem | vladikr: -1 on https://review.openstack.org/#/c/405954 because that's not handling that we are getting strings back from neutron for the vlan id in the port detials | |
| 22:07:46 | mriedem | *details | |
| 22:08:25 | vladikr | mriedem, just fixed it | |
| 22:08:30 | vladikr | ^^ | |
| 22:08:59 | mriedem | ah good timing :) | |
| 22:09:02 | vladikr | :) | |
| 22:10:35 | mriedem | jroll: do we have anything in CI that sets the node.resource_class field? | |
| 22:11:25 | edleafe | recheck finally worked on https://review.openstack.org/#/c/404472/ \o/ | |
| 22:13:35 | mriedem | edleafe: just wrapping up some comments in there | |
| 22:13:57 | mriedem | edleafe: what sucks is we have nothing in CI to test this new flow http://logs.openstack.org/72/404472/21/check/gate-tempest-dsvm-ironic-ipa-wholedisk-agent_ipmitool-tinyipa-ubuntu-xenial-nv/d12f128/logs/screen-n-cpu.txt.gz#_2017-01-26_18_59_36_215 | |
| 22:13:57 | edleafe | mriedem: okie dokie | |
| 22:14:00 | mriedem | we're still just doing the old thing | |
| 22:14:47 | edleafe | mriedem: yeah, the extra specs stuff for ironic flavors hasn't been done yet | |
| 22:15:36 | jroll | mriedem: I have some stuff in progress to do that thing | |
| 22:15:42 | bauzas | dansmith: around ? | |
| 22:18:22 | bauzas | dansmith: confirmed that the min service version is 16 http://logs.openstack.org/61/417961/32/check/gate-grenade-dsvm-neutron-multinode-ubuntu-xenial/f07f0e1/logs/new/screen-n-sch.txt.gz#_2017-01-26_22_02_47_735 | |
| 22:18:29 | bauzas | while it should 15 | |
| 22:18:31 | bauzas | ;( | |
| 22:18:33 | bauzas | mriedem: ^ | |
| 22:19:50 | bauzas | potentially the cache isn't working well | |
| 22:24:52 | mriedem | edleafe: comments posted | |
| 22:25:29 | mriedem | mostly simple things | |
| 22:25:31 | mriedem | and some missing tests | |
| 22:26:31 | mriedem | dansmith: you should be able to set a str to an IntegerField right? | |
| 22:27:39 | mriedem | or does coerce() just make that work | |
| 22:32:03 | openstackgerrit | Eric Brown proposed openstack/nova: Add RPC version aliases for Ocata https://review.openstack.org/425958 | |
| 22:34:11 | dansmith | bauzas: no, it shouldn't be the cache, unless we're down long enough to consider the other node down | |
| 22:34:18 | dansmith | and then start before it checks in | |
| 22:34:34 | dansmith | mriedem: I think it will coerce a string to an int yeah | |
| 22:34:46 | mriedem | blarg | |
| 22:34:49 | mriedem | ok lesson learned | |
| 22:35:08 | bauzas | dansmith: any thoughts on why then we have 16? | |
| 22:35:09 | mriedem | vladikr: +2 on your middle change too, even though a couple of the unit tests need tweaks | |
| 22:36:15 | dansmith | bauzas: how long is conductor stopped for? | |
| 22:37:35 | bauzas | dansmith: verifying the tempest test | |
| 22:37:46 | dansmith | service_down_time defaults to 60s so if it is >60s between conductor going down and scheduler coming back up, then it could be excluding the "down" service | |
| 22:39:25 | edleafe | mriedem: working on adding the missing tests | |
| 22:39:32 | bauzas | dansmith: the test doesn't assume stopping anything | |
| 22:39:42 | dansmith | bauzas: what test? | |
| 22:39:50 | bauzas | the tempest test | |
| 22:39:57 | bauzas | oh, you mean, upgrade? | |
| 22:40:02 | dansmith | during the upgrade yeah | |
| 22:40:17 | bauzas | so we try to boot on the old node | |
| 22:40:50 | bauzas | I'm asking the service version for computes | |
| 22:43:07 | dansmith | I'm not getting what you're saying | |
| 22:43:21 | bauzas | dansmith: I'm trying to understand the inner logic | |
| 22:44:01 | bauzas | I'm not getting why the scheduler could be excluding the old node if we never upgraded the latter | |
| 22:44:16 | bauzas | wrt the service down time you mentioned | |
| 22:46:34 | dansmith | because conductor was down, so the subnode couldn't refresh its record | |
| 22:49:29 | dansmith | bauzas: see my change to your grenade patch just now | |
| 22:50:52 | bauzas | dansmith: ah, I see | |
| 22:51:06 | bauzas | dansmith: we're restarting the scheduler too soon | |
| 22:51:17 | dansmith | right, that's what I'm thinking | |
| 22:51:19 | bauzas | without the compute having properly registered back | |
| 22:51:22 | dansmith | right | |
| 22:51:23 | bauzas | gotcha | |
| 22:51:35 | bauzas | okay, let's see | |
| 22:51:37 | dansmith | if it's more than 60s then the scheduler's min version calculus will exclude it | |
| 22:51:43 | bauzas | you gonna recheck ? | |
| 22:51:53 | bauzas | I mean the main change | |
| 22:52:15 | dansmith | jesus I have to do everything around here! | |
| 22:52:26 | bauzas | :p | |
| 22:52:41 | bauzas | well, I can write a gerrit comment :p | |
| 22:54:51 | bauzas | dansmith: it's pretty late now, and I have to pass my whole tomorrow on train | |
| 22:55:02 | dansmith | ...oh | |
| 22:55:02 | bauzas | dansmith: I'll have to handover :( | |
| 22:55:09 | dansmith | hmm, okay | |
| 22:55:41 | bauzas | mriedem: thoughts on me officially asking for a FFE? | |
| 22:56:15 | mriedem | granted | |
| 22:57:29 | bauzas | dansmith: FWIW, do you think I should make a note in the reno file that operators have to make sure they leave a large enough window for having all the computes registered back or nope? | |
| 22:57:59 | dansmith | bauzas: well, this is already the case for auto rpc mode, but sure | |
| 22:58:08 | bauzas | right, my point | |