| alaski | these are not using that fixture apparently, they're starting up compute services | 00:00 |
|---|---|---|
| *** baoli has quit IRC | 00:00 | |
| melwitt | yeah, was just looking at that. I saw only two tests use the fixture and they're about BuildRequest | 00:01 |
| *** ducttape_ has quit IRC | 00:07 | |
| *** rrecio has quit IRC | 00:08 | |
| alaski | oh, I may have it | 00:10 |
| alaski | on a scheduling failure we bail early at http://git.openstack.org/cgit/openstack/nova/tree/nova/conductor/manager.py#n493 so the BuildRequest isn't cleaned up | 00:11 |
| alaski | and instance shows then come from the buildrequest rather than the instance | 00:11 |
| melwitt | alaski: nice find. and then it wants to look at the instance fault message | 00:12 |
| melwitt | (the test does) | 00:12 |
| melwitt | and the status | 00:13 |
| *** hieulq has joined #openstack-nova | 00:14 | |
| *** raunak has quit IRC | 00:15 | |
| *** avolkov`` has joined #openstack-nova | 00:17 | |
| alaski | that fixed half the tests | 00:20 |
| *** Sukhdev has joined #openstack-nova | 00:20 | |
| *** baoli has joined #openstack-nova | 00:21 | |
| *** avolkov` has quit IRC | 00:22 | |
| *** marst has joined #openstack-nova | 00:22 | |
| melwitt | yey | 00:23 |
| *** trisq has joined #openstack-nova | 00:25 | |
| *** ducttape_ has joined #openstack-nova | 00:28 | |
| *** baoli has quit IRC | 00:30 | |
| *** ducttape_ has quit IRC | 00:31 | |
| *** browne has quit IRC | 00:32 | |
| *** baoli has joined #openstack-nova | 00:35 | |
| *** thorst has joined #openstack-nova | 00:35 | |
| alaski | ugh. So the other half failed because of a timing issue in the tests which should be fixed regardless. But a relatively simple change seems to fix everything | 00:40 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Fix service version lookups https://review.openstack.org/363888 | 00:45 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Pull from cell0 and build_requests for instance list https://review.openstack.org/356138 | 00:45 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Delete BuildRequest regardless of service_version https://review.openstack.org/363911 | 00:45 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Remove BuildRequest when scheduling fails https://review.openstack.org/364005 | 00:45 |
| alaski | I'll add tests tomorrow, but that should address it | 00:46 |
| melwitt | alaski: awesome. is the bottom change one that you want to add tests to or is that one good to go? | 00:47 |
| alaski | melwitt: yeah, that's the one that I need to test | 00:48 |
| melwitt | alaski: okay, cool | 00:48 |
| *** xinliang has quit IRC | 00:48 | |
| *** mtanino has quit IRC | 00:49 | |
| *** ducttape_ has joined #openstack-nova | 00:49 | |
| *** ociuhandu has joined #openstack-nova | 00:49 | |
| openstackgerrit | Andrew Laski proposed openstack/nova: Pull from cell0 and build_requests for instance list https://review.openstack.org/356138 | 00:50 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Delete BuildRequest regardless of service_version https://review.openstack.org/363911 | 00:50 |
| *** chrichip has quit IRC | 00:52 | |
| *** vishwanathj has quit IRC | 00:53 | |
| *** chrichip has joined #openstack-nova | 00:53 | |
| *** Sukhdev has quit IRC | 00:55 | |
| *** cheneydc has joined #openstack-nova | 00:56 | |
| openstackgerrit | Li Yingjun proposed openstack/nova: Adds view builders for hypervisors controller https://review.openstack.org/335282 | 01:00 |
| *** xinliang has joined #openstack-nova | 01:01 | |
| *** trisq has quit IRC | 01:03 | |
| *** ociuhandu has quit IRC | 01:08 | |
| mriedem | alaski: new fun failures http://logs.openstack.org/56/363956/3/check/gate-grenade-dsvm-neutron-ubuntu-trusty/09a989d/logs/grenade.sh.txt.gz#_2016-09-01_00_11_34_072 | 01:11 |
| mriedem | Instance ac03c9a7-3546-41ef-bba2-018c3542e41b is not mapped to a cell (upgrade is incomplete) | 01:11 |
| *** priteau has joined #openstack-nova | 01:11 | |
| *** Sukhdev has joined #openstack-nova | 01:12 | |
| *** Sukhdev has quit IRC | 01:13 | |
| mriedem | i think we might be getting into some kind of weird setup/upgrade loop here | 01:14 |
| *** hoangcx has joined #openstack-nova | 01:14 | |
| *** ducttape_ has quit IRC | 01:15 | |
| *** priteau has quit IRC | 01:16 | |
| openstackgerrit | Ed Leafe proposed openstack/nova: WIP create_all and delete_all for AllocationList https://review.openstack.org/363209 | 01:18 |
| edleafe | cdent: jaypipes: bauzas: ^^ update on the allocations code | 01:18 |
| *** mingyu has joined #openstack-nova | 01:18 | |
| edleafe | All the tests are passing, and I changed the _create back to _set | 01:18 |
| *** ducttape_ has joined #openstack-nova | 01:18 | |
| mriedem | alaski: oh wait a minute, the grenade change probably can't be calling cells_v2 verify_instance if we didn't run simple_cell_setup | 01:22 |
| mriedem | to do the instance mapping | 01:22 |
| mriedem | yeah that's the ticket | 01:24 |
| mriedem | i'll update that quick | 01:24 |
| *** cheneydc has quit IRC | 01:25 | |
| openstackgerrit | melanie witt proposed openstack/nova: Make req spec and RT aware of volume-backed instances https://review.openstack.org/355091 | 01:26 |
| *** wangqun has joined #openstack-nova | 01:27 | |
| *** baoli_ has joined #openstack-nova | 01:28 | |
| *** kevinz has joined #openstack-nova | 01:28 | |
| *** baoli has quit IRC | 01:28 | |
| *** baoli has joined #openstack-nova | 01:30 | |
| *** hoangcx has quit IRC | 01:31 | |
| mriedem | alaski: ok updated that grenade change, rechecked the other changes that depend on it, still watching https://review.openstack.org/#/c/363971/ to see if that's going to pass | 01:31 |
| *** trisq has joined #openstack-nova | 01:32 | |
| dansmith | need me to look at anything? | 01:33 |
| *** baoli_ has quit IRC | 01:33 | |
| *** salv-orlando has joined #openstack-nova | 01:34 | |
| mriedem | dansmith: i don't think so, unless you want to sanity check, | 01:37 |
| mriedem | but i'm just watching zuul results | 01:37 |
| dansmith | okay | 01:37 |
| dansmith | I've been drinking so I probably shouldn't be making judgment calls | 01:37 |
| dansmith | (and for the logs, I mean literally a single beer) | 01:38 |
| mriedem | all calls to nova-manage cell_v2 stuff in grenade is now conditional | 01:38 |
| *** chrichip has quit IRC | 01:38 | |
| mriedem | "i've been drinking...literally ounces of beer here" | 01:38 |
| dansmith | hah, yeah | 01:38 |
| mriedem | as for the other side of the fun, https://review.openstack.org/#/c/363209/ was updated but i'm guessing we're still waiting on jaypipes to review that | 01:39 |
| *** EinstCrazy has joined #openstack-nova | 01:39 | |
| dansmith | mriedem: I'm waiting for the client side so I can poke at it locally | 01:39 |
| *** EinstCrazy has quit IRC | 01:39 | |
| *** EinstCrazy has joined #openstack-nova | 01:39 | |
| *** chrichip has joined #openstack-nova | 01:39 | |
| dansmith | mriedem: I told him I'd pick that up tomorrow where he leaves off | 01:39 |
| mriedem | yeah i guess the client side will be the test | 01:39 |
| mriedem | if the whole stack | 01:40 |
| dansmith | yar | 01:40 |
| mriedem | *of | 01:40 |
| mriedem | looking at some of the other approved things from https://etherpad.openstack.org/p/placement-next and they are still in zuul | 01:41 |
| dansmith | yeah | 01:41 |
| *** ducttape_ has quit IRC | 01:42 | |
| *** diablo_rojo has quit IRC | 01:43 | |
| *** ayoung has joined #openstack-nova | 01:43 | |
| *** salv-orlando has quit IRC | 01:44 | |
| *** ducttape_ has joined #openstack-nova | 01:46 | |
| tonyb | mriedem: Is it your intent that novaclient 6.0.0 should be used for newton deploys (as opposed to ocata)? | 01:46 |
| melwitt | mriedem: alaski said he's going to add tests to https://review.openstack.org/#/c/363888/ tomorrow too | 01:47 |
| mriedem | melwitt: ok i missed the convo before but it sounds like someting was still failing | 01:47 |
| tonyb | mriedem: the removal of the newtorking stuff has broken horizon and therefore can't land in u-c | 01:47 |
| mriedem | tonyb: i'm not sure it matters | 01:47 |
| tonyb | mriedem: I don't mind fixing it | 01:47 |
| mriedem | tonyb: is horizon using the API or CLI? | 01:48 |
| mriedem | tonyb: the CLI should cap the networking stuff at 2.35 | 01:48 |
| tonyb | mriedem: API i'd guess | 01:48 |
| mriedem | tonyb: are they specifying 2.latest? | 01:48 |
| mriedem | for the microversion? | 01:48 |
| melwitt | mriedem: something was failing and he fixed it but said he needs to add some tests to it | 01:48 |
| tonyb | mriedem: haven't looked but I assume so | 01:48 |
| mriedem | tonyb: 2.latest is only for testing | 01:49 |
| mriedem | and crazy people | 01:49 |
| mriedem | tonyb: do you have a link to a failure? | 01:49 |
| tonyb | mriedem: http://logs.openstack.org/16/363516/1/check/gate-cross-horizon-python27-ubuntu-xenial/2644145/console.html#_2016-08-31_09_34_07_151838 | 01:49 |
| tonyb | mriedem: so should we cap @ 2.35 for simplicity in horizon for newton? | 01:50 |
| mriedem | tonyb: let me look at this test code first | 01:51 |
| mriedem | nice https://github.com/openstack/horizon/blob/master/openstack_dashboard/test/helpers.py#L442 | 01:53 |
| *** ducttape_ has quit IRC | 01:54 | |
| edleafe | dansmith: so if you're not too drunk, want to give me a sanity check on https://review.openstack.org/#/c/363209/ ? | 01:54 |
| mriedem | tonyb: so with something like this: http://logs.openstack.org/16/363516/1/check/gate-cross-horizon-python27-ubuntu-xenial/2644145/console.html#_2016-08-31_09_34_07_189395 | 01:55 |
| mriedem | the horizon unit tests are stubbing (with mox) the novaclient.Client impl | 01:55 |
| mriedem | which is kind of terrible | 01:55 |
| tonyb | :/ | 01:56 |
| mriedem | looking at this: https://github.com/openstack/horizon/blob/master/openstack_dashboard/api/nova.py#L50 | 01:57 |
| mriedem | the API code in horizon should at most be using 2.9 | 01:57 |
| mriedem | the client is created here https://github.com/openstack/horizon/blob/master/openstack_dashboard/api/nova.py#L476 | 01:58 |
| mriedem | the tests are doing this stub_novaclient thing though | 01:58 |
| mriedem | my guess would be, the stubbed client in https://github.com/openstack/horizon/blob/master/openstack_dashboard/test/helpers.py#L442 needs to set a fake api_version attr or sometihng | 01:59 |
| mriedem | and maybe default that to 2.1 | 01:59 |
| *** woodster_ has quit IRC | 01:59 | |
| mriedem | i'm not really sure, but they have to do something like that for keystoneclient https://github.com/openstack/horizon/blob/master/openstack_dashboard/test/helpers.py#L457 | 02:00 |
| mriedem | tonyb: here is a bug anyway https://bugs.launchpad.net/horizon/+bug/1619093 | 02:01 |
| openstack | Launchpad bug 1619093 in OpenStack Dashboard (Horizon) "unit tests fail with novaclient 6.0.0" [Undecided,New] | 02:01 |
| tonyb | mriedem: Thanks. | 02:01 |
| *** ijw has quit IRC | 02:01 | |
| mriedem | i can try setting a fake api_version locally and see how many tests that fixes | 02:02 |
| mriedem | wow it takes a long time to clone horizon | 02:03 |
| *** Apoorva has quit IRC | 02:03 | |
| *** EinstCra_ has joined #openstack-nova | 02:04 | |
| tonyb | mriedem: ok. I'll try to chip away at it also | 02:05 |
| *** ljxiash has joined #openstack-nova | 02:05 | |
| dansmith | edleafe: did you change anything? | 02:06 |
| dansmith | I looked at the version jay pushed | 02:06 |
| *** sdake has quit IRC | 02:06 | |
| dansmith | edleafe: ah I see | 02:07 |
| edleafe | dansmith: fixed a few loose ends and added the logic back to update | 02:07 |
| dansmith | edleafe: those are things I'd hope to find while playing with it locally, but yeah changed look sane | 02:07 |
| dansmith | *changes | 02:07 |
| *** hoangcx has joined #openstack-nova | 02:07 | |
| edleafe | dansmith: cool. Have another beer :) | 02:07 |
| *** EinstCrazy has quit IRC | 02:08 | |
| *** phuongnh has joined #openstack-nova | 02:11 | |
| mriedem | still cloning horizon...they must store a bunch of binaries in git | 02:13 |
| mriedem | tonyb: while i wait, do me a solid https://review.openstack.org/#/c/352884/ | 02:14 |
| mriedem | and https://review.openstack.org/#/c/355210/ | 02:14 |
| melwitt | I cloned nova earlier and it took forever too. thought maybe there was something afoot with github | 02:14 |
| *** hemanthm|afk is now known as hemanthm | 02:14 | |
| tonyb | mriedem: Sure np | 02:16 |
| *** unicell has quit IRC | 02:16 | |
| *** chrichip has quit IRC | 02:16 | |
| *** chrichip has joined #openstack-nova | 02:17 | |
| *** thorst has quit IRC | 02:18 | |
| *** thorst has joined #openstack-nova | 02:19 | |
| *** ducttape_ has joined #openstack-nova | 02:19 | |
| mriedem | ok running horizon unit tests | 02:22 |
| mriedem | i might be able to fix 44 of the failures | 02:22 |
| mriedem | and leave the other 10 for someone else | 02:22 |
| mriedem | wtf, do they use nosetests? | 02:25 |
| tonyb | mriedem: Adding "nova_client.Client.api_version = "2.9"" works for me with novaclient 5.1.0 and 6.0.0 | 02:25 |
| mriedem | so i tried to be cute and use the 'preferred' version from their api.nova.VERSIONS thing | 02:26 |
| mriedem | but the novaclient api version stuff doesn't like '2', it needs '2.0' | 02:26 |
| mriedem | so i'm going to just hardcode to 2.1 since that's sane | 02:26 |
| mriedem | and is the minimum you'd use | 02:26 |
| mriedem | i'll push this up in a sec | 02:27 |
| *** ducttape_ has quit IRC | 02:27 | |
| mriedem | nova_client.Client.api_version needs to be an APIVersion object | 02:27 |
| mriedem | not a string | 02:27 |
| *** thorst has quit IRC | 02:28 | |
| tonyb | mriedem: picky ;P | 02:28 |
| mriedem | well, i think using a string probably just gets lucky with the lazy compare functions we have in APIVersion | 02:28 |
| mriedem | i've thought about fixing that before | 02:28 |
| mriedem | but you know. lazy. | 02:28 |
| tonyb | mriedem: Yeah I'll do it properly this was a quick hack to see if we're onthe right track | 02:29 |
| mriedem | i'll have the patch up here in a sec, just need nose to finish | 02:29 |
| *** dimtruck is now known as zz_dimtruck | 02:30 | |
| tonyb | mriedem: ok | 02:31 |
| mriedem | tonyb: https://review.openstack.org/#/c/364029/ | 02:32 |
| mriedem | updated your reqs change with the depends-on | 02:33 |
| mriedem | dansmith: alaski: woot http://logs.openstack.org/71/363971/1/check/gate-grenade-dsvm-ubuntu-trusty/0c1458c/logs/new/localrc.txt.gz NOVA_CONFIGURE_CELLSV2=True | 02:34 |
| mriedem | and that passed | 02:34 |
| mriedem | that is from https://review.openstack.org/#/c/363971/ | 02:34 |
| mriedem | as did http://logs.openstack.org/71/363971/1/check/gate-tempest-dsvm-full-ubuntu-xenial/4d3b342/logs/localrc.txt.gz | 02:35 |
| *** ducttape_ has joined #openstack-nova | 02:35 | |
| *** EinstCra_ has quit IRC | 02:36 | |
| *** EinstCrazy has joined #openstack-nova | 02:37 | |
| openstackgerrit | zhufl proposed openstack/nova: Remove unused oslo_config https://review.openstack.org/364031 | 02:37 |
| *** TravT has joined #openstack-nova | 02:38 | |
| mriedem | ok i think i'm going to call it a day | 02:38 |
| mriedem | i expect sean diggity dague to hit a bunch of these things in the morning | 02:39 |
| *** mriedem has quit IRC | 02:39 | |
| *** jamielennox|away is now known as jamielennox | 02:41 | |
| *** ducttape_ has quit IRC | 02:43 | |
| *** salv-orlando has joined #openstack-nova | 02:43 | |
| *** armax has quit IRC | 02:45 | |
| *** zz_dimtruck is now known as dimtruck | 02:47 | |
| *** ssurana has quit IRC | 02:47 | |
| *** sudipto has joined #openstack-nova | 02:47 | |
| *** Sukhdev has joined #openstack-nova | 02:48 | |
| *** gouthamr has quit IRC | 02:49 | |
| *** salv-orlando has quit IRC | 02:50 | |
| *** amotoki has joined #openstack-nova | 02:51 | |
| *** salv-orlando has joined #openstack-nova | 02:52 | |
| *** su_zhang has joined #openstack-nova | 02:53 | |
| *** chrichip has quit IRC | 02:54 | |
| *** chrichip has joined #openstack-nova | 02:55 | |
| *** salv-orlando has quit IRC | 02:57 | |
| *** hoangcx has quit IRC | 02:58 | |
| *** dimtruck is now known as zz_dimtruck | 03:02 | |
| *** armax has joined #openstack-nova | 03:03 | |
| *** amotoki has quit IRC | 03:08 | |
| *** armax has quit IRC | 03:08 | |
| *** raunak has joined #openstack-nova | 03:17 | |
| *** yamahata has quit IRC | 03:17 | |
| *** amotoki has joined #openstack-nova | 03:21 | |
| auggy | is that a friday nick suggestion? | 03:21 |
| *** zz_dimtruck is now known as dimtruck | 03:23 | |
| *** thorst has joined #openstack-nova | 03:26 | |
| openstackgerrit | zhufl proposed openstack/nova: Remove unused oslo_config https://review.openstack.org/364031 | 03:29 |
| *** amotoki has quit IRC | 03:30 | |
| *** thorst has quit IRC | 03:33 | |
| *** amotoki has joined #openstack-nova | 03:35 | |
| *** sudipto has quit IRC | 03:37 | |
| *** ducttape_ has joined #openstack-nova | 03:43 | |
| *** ayoung has quit IRC | 03:47 | |
| *** dikonoor has joined #openstack-nova | 03:48 | |
| *** akshai_ has quit IRC | 03:48 | |
| *** raunak has quit IRC | 03:48 | |
| *** su_zhang has quit IRC | 03:48 | |
| *** chrichip has quit IRC | 03:48 | |
| *** raunak has joined #openstack-nova | 03:48 | |
| *** su_zhang has joined #openstack-nova | 03:48 | |
| *** ducttape_ has quit IRC | 03:49 | |
| *** su_zhang has quit IRC | 03:53 | |
| *** salv-orlando has joined #openstack-nova | 03:53 | |
| *** salv-orlando has quit IRC | 03:56 | |
| *** ljxiash has quit IRC | 03:56 | |
| *** ljxiash has joined #openstack-nova | 03:56 | |
| *** ljxiash has quit IRC | 04:00 | |
| *** amotoki has quit IRC | 04:01 | |
| *** ljxiash has joined #openstack-nova | 04:03 | |
| *** jamielennox is now known as jamielennox|away | 04:05 | |
| *** su_zhang has joined #openstack-nova | 04:07 | |
| *** hemanthm is now known as hemanthm|away | 04:08 | |
| *** ljxiash has quit IRC | 04:08 | |
| *** Sukhdev has quit IRC | 04:09 | |
| *** yamahata has joined #openstack-nova | 04:11 | |
| *** fragatina has quit IRC | 04:14 | |
| *** amotoki has joined #openstack-nova | 04:16 | |
| *** ekuris has joined #openstack-nova | 04:19 | |
| *** dimtruck is now known as zz_dimtruck | 04:22 | |
| *** unicell has joined #openstack-nova | 04:22 | |
| *** baoli has quit IRC | 04:24 | |
| *** su_zhang has quit IRC | 04:24 | |
| *** su_zhang has joined #openstack-nova | 04:24 | |
| *** unicell has quit IRC | 04:27 | |
| *** unicell has joined #openstack-nova | 04:27 | |
| *** su_zhang has quit IRC | 04:28 | |
| *** thorst has joined #openstack-nova | 04:30 | |
| *** prateek has quit IRC | 04:30 | |
| *** ssurana has joined #openstack-nova | 04:30 | |
| *** links has joined #openstack-nova | 04:32 | |
| *** fragatina has joined #openstack-nova | 04:34 | |
| *** krtaylor has quit IRC | 04:34 | |
| *** thorst has quit IRC | 04:37 | |
| *** NobodyCam has quit IRC | 04:38 | |
| *** scottda has quit IRC | 04:38 | |
| *** zhiyan has quit IRC | 04:38 | |
| *** nikhil has quit IRC | 04:38 | |
| *** betherly has quit IRC | 04:38 | |
| *** sballe_ has quit IRC | 04:38 | |
| *** junjie has quit IRC | 04:38 | |
| *** fmccrthy has quit IRC | 04:38 | |
| *** vdrok has quit IRC | 04:38 | |
| *** pbandark has joined #openstack-nova | 04:39 | |
| *** scottda has joined #openstack-nova | 04:40 | |
| *** Cassowary has quit IRC | 04:40 | |
| *** junjie has joined #openstack-nova | 04:40 | |
| *** nikhil has joined #openstack-nova | 04:40 | |
| *** zhiyan has joined #openstack-nova | 04:40 | |
| *** vdrok has joined #openstack-nova | 04:40 | |
| *** sballe_ has joined #openstack-nova | 04:40 | |
| *** betherly has joined #openstack-nova | 04:40 | |
| *** fmccrthy has joined #openstack-nova | 04:41 | |
| *** NobodyCam has joined #openstack-nova | 04:41 | |
| *** zz_dimtruck is now known as dimtruck | 04:42 | |
| *** yamahata has quit IRC | 04:42 | |
| *** ducttape_ has joined #openstack-nova | 04:45 | |
| *** McNinja has quit IRC | 04:46 | |
| *** ManishD has joined #openstack-nova | 04:48 | |
| *** ratailor has joined #openstack-nova | 04:49 | |
| *** ducttape_ has quit IRC | 04:50 | |
| *** McNinja has joined #openstack-nova | 04:52 | |
| *** Cassowary has joined #openstack-nova | 04:52 | |
| *** yamahata has joined #openstack-nova | 04:54 | |
| *** mamandle has joined #openstack-nova | 04:55 | |
| ratailor | Can anybody pl review https://review.openstack.org/#/c/312887/ | 04:56 |
| *** moshele has joined #openstack-nova | 04:57 | |
| *** pcaruana has quit IRC | 04:57 | |
| *** wangqun_ has joined #openstack-nova | 04:59 | |
| *** wangqun has quit IRC | 05:00 | |
| *** prateek has joined #openstack-nova | 05:00 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/363871 | 05:00 |
| *** ecelik has joined #openstack-nova | 05:02 | |
| *** salv-orlando has joined #openstack-nova | 05:03 | |
| *** carl_baldwin has quit IRC | 05:03 | |
| *** dougwig has quit IRC | 05:03 | |
| *** ediardo has quit IRC | 05:03 | |
| *** cargonza has quit IRC | 05:04 | |
| *** jamielennox|away is now known as jamielennox | 05:04 | |
| *** ediardo has joined #openstack-nova | 05:04 | |
| *** carl_baldwin has joined #openstack-nova | 05:04 | |
| *** rama_y has joined #openstack-nova | 05:05 | |
| *** cargonza has joined #openstack-nova | 05:06 | |
| *** dimtruck is now known as zz_dimtruck | 05:06 | |
| *** harlowja_at_home has joined #openstack-nova | 05:06 | |
| *** rama_y has quit IRC | 05:06 | |
| *** dougwig has joined #openstack-nova | 05:07 | |
| *** rama_y has joined #openstack-nova | 05:07 | |
| *** salv-orlando has quit IRC | 05:10 | |
| *** macsz has quit IRC | 05:13 | |
| *** fyxim has quit IRC | 05:13 | |
| *** rmcadams has quit IRC | 05:13 | |
| *** andrewbogott has quit IRC | 05:13 | |
| *** rstarmer has joined #openstack-nova | 05:13 | |
| *** salv-orlando has joined #openstack-nova | 05:14 | |
| *** rmcadams has joined #openstack-nova | 05:14 | |
| *** macsz has joined #openstack-nova | 05:14 | |
| *** fyxim has joined #openstack-nova | 05:14 | |
| *** andrewbogott has joined #openstack-nova | 05:15 | |
| *** rstarmer has quit IRC | 05:15 | |
| openstackgerrit | Slawek Kaplonski proposed openstack/nova: Fix race condition bug during live_snapshot https://review.openstack.org/363926 | 05:20 |
| *** fragatin_ has joined #openstack-nova | 05:22 | |
| *** sdake has joined #openstack-nova | 05:23 | |
| *** fragatina has quit IRC | 05:25 | |
| *** raunak has quit IRC | 05:27 | |
| *** maestropandy has joined #openstack-nova | 05:28 | |
| *** zerda2 has joined #openstack-nova | 05:31 | |
| *** nkrinner_afk is now known as nkrinner | 05:32 | |
| *** ljxiash has joined #openstack-nova | 05:32 | |
| openstackgerrit | viswesuwara nathan proposed openstack/python-novaclient: Fix incorrect output of "nova show" for long user data https://review.openstack.org/362303 | 05:34 |
| *** thorst has joined #openstack-nova | 05:35 | |
| *** prateek has quit IRC | 05:35 | |
| *** prateek has joined #openstack-nova | 05:36 | |
| *** ihrachys has joined #openstack-nova | 05:42 | |
| *** thorst has quit IRC | 05:42 | |
| *** mjura has joined #openstack-nova | 05:42 | |
| *** zz_dimtruck is now known as dimtruck | 05:42 | |
| *** Hosam_ has joined #openstack-nova | 05:43 | |
| *** wanghao_ has joined #openstack-nova | 05:43 | |
| *** ducttape_ has joined #openstack-nova | 05:46 | |
| *** ducttape_ has quit IRC | 05:50 | |
| *** harlowja_at_home has quit IRC | 05:51 | |
| *** mjura has quit IRC | 05:51 | |
| *** mjura has joined #openstack-nova | 05:51 | |
| *** TravT_ has joined #openstack-nova | 05:52 | |
| *** yamahata has quit IRC | 05:52 | |
| *** TravT has quit IRC | 05:55 | |
| *** hoangcx has joined #openstack-nova | 05:56 | |
| *** mingyu has quit IRC | 05:57 | |
| *** edmondsw has joined #openstack-nova | 05:59 | |
| *** edmondsw has quit IRC | 06:00 | |
| *** Sukhdev has joined #openstack-nova | 06:01 | |
| *** maestropandy has left #openstack-nova | 06:01 | |
| *** coolsvap has joined #openstack-nova | 06:05 | |
| *** paul-carlton2 has joined #openstack-nova | 06:06 | |
| *** BharatK_ has joined #openstack-nova | 06:06 | |
| *** kylek3h has quit IRC | 06:07 | |
| *** kylek3h has joined #openstack-nova | 06:08 | |
| *** rcernin has quit IRC | 06:08 | |
| *** dakhmetov has quit IRC | 06:08 | |
| *** salv-orl_ has joined #openstack-nova | 06:08 | |
| *** avolkov`` has quit IRC | 06:09 | |
| *** salv-orlando has quit IRC | 06:11 | |
| *** kylek3h has quit IRC | 06:12 | |
| *** dinobot_ has joined #openstack-nova | 06:13 | |
| *** mamandle has quit IRC | 06:14 | |
| *** divyakkonoor has joined #openstack-nova | 06:18 | |
| *** oanson has joined #openstack-nova | 06:18 | |
| *** Jeffrey4l__ is now known as Jeffrey4l | 06:19 | |
| *** dulek_ has joined #openstack-nova | 06:20 | |
| *** ljjjustin has quit IRC | 06:21 | |
| *** dulek has quit IRC | 06:21 | |
| *** dulek_ is now known as dulek | 06:21 | |
| *** ljjjustin has joined #openstack-nova | 06:21 | |
| *** dikonoor has quit IRC | 06:21 | |
| *** gszasz has joined #openstack-nova | 06:25 | |
| *** paul-carlton2 has quit IRC | 06:25 | |
| *** pcaruana has joined #openstack-nova | 06:30 | |
| *** ihrachys has quit IRC | 06:33 | |
| *** lpetrut has joined #openstack-nova | 06:35 | |
| *** paul-carlton2 has joined #openstack-nova | 06:36 | |
| *** sahid has joined #openstack-nova | 06:37 | |
| *** thorst has joined #openstack-nova | 06:40 | |
| *** zenoway has joined #openstack-nova | 06:40 | |
| *** ljxiash has quit IRC | 06:41 | |
| *** rcernin has joined #openstack-nova | 06:41 | |
| *** ljxiash has joined #openstack-nova | 06:41 | |
| *** dinobot_ has quit IRC | 06:41 | |
| *** sridharg has joined #openstack-nova | 06:42 | |
| *** Sukhdev has quit IRC | 06:42 | |
| *** rcernin has quit IRC | 06:45 | |
| *** rcernin has joined #openstack-nova | 06:45 | |
| *** TravT has joined #openstack-nova | 06:45 | |
| *** ljxiash has quit IRC | 06:46 | |
| *** kaisers_ has joined #openstack-nova | 06:46 | |
| *** ducttape_ has joined #openstack-nova | 06:47 | |
| *** thorst has quit IRC | 06:47 | |
| *** TravT_ has quit IRC | 06:48 | |
| *** salv-orl_ has quit IRC | 06:50 | |
| *** hoangcx has quit IRC | 06:52 | |
| *** hoangcx_ has joined #openstack-nova | 06:52 | |
| *** ducttape_ has quit IRC | 06:52 | |
| *** Hosam has joined #openstack-nova | 06:52 | |
| *** hferenc has quit IRC | 06:55 | |
| *** Hosam_ has quit IRC | 06:56 | |
| *** ljxiash has joined #openstack-nova | 06:57 | |
| *** dinobot_ has joined #openstack-nova | 06:57 | |
| *** oanson has quit IRC | 07:00 | |
| *** claudiub has joined #openstack-nova | 07:01 | |
| *** tesseract- has joined #openstack-nova | 07:03 | |
| *** mingyu has joined #openstack-nova | 07:03 | |
| *** mingyu has quit IRC | 07:04 | |
| *** andreas_s has joined #openstack-nova | 07:05 | |
| *** mingyu has joined #openstack-nova | 07:05 | |
| *** rama_y has quit IRC | 07:06 | |
| *** rama_y has joined #openstack-nova | 07:06 | |
| *** sudipto has joined #openstack-nova | 07:08 | |
| *** rama_y has quit IRC | 07:08 | |
| *** kylek3h has joined #openstack-nova | 07:08 | |
| *** rama_y has joined #openstack-nova | 07:08 | |
| *** kylek3h has quit IRC | 07:14 | |
| *** sudipto has quit IRC | 07:15 | |
| *** lpetrut has quit IRC | 07:17 | |
| *** jamielennox is now known as jamielennox|away | 07:17 | |
| *** ssurana has quit IRC | 07:18 | |
| *** BharatK_ has quit IRC | 07:18 | |
| *** sridharg has quit IRC | 07:19 | |
| *** markus_z has joined #openstack-nova | 07:20 | |
| *** dimtruck is now known as zz_dimtruck | 07:22 | |
| *** jlvillal has quit IRC | 07:22 | |
| *** rama_y has quit IRC | 07:23 | |
| *** jlvillal has joined #openstack-nova | 07:23 | |
| *** rama_y has joined #openstack-nova | 07:24 | |
| *** praneshp has quit IRC | 07:26 | |
| *** unicell has quit IRC | 07:26 | |
| *** lpetrut has joined #openstack-nova | 07:26 | |
| *** unicell has joined #openstack-nova | 07:27 | |
| *** rama_y has quit IRC | 07:29 | |
| *** bkopilov has quit IRC | 07:30 | |
| *** phuongnh has quit IRC | 07:30 | |
| *** phuongnh has joined #openstack-nova | 07:31 | |
| *** roeyc has joined #openstack-nova | 07:33 | |
| *** hoangcx_ has quit IRC | 07:35 | |
| *** ljxiash has quit IRC | 07:39 | |
| *** hoangcx has joined #openstack-nova | 07:40 | |
| *** apuimedo has joined #openstack-nova | 07:40 | |
| *** bkopilov has joined #openstack-nova | 07:40 | |
| *** ihrachys has joined #openstack-nova | 07:41 | |
| apuimedo | hieulq: ping | 07:41 |
| *** ljxiash has joined #openstack-nova | 07:42 | |
| apuimedo | I saw you worked on nova/conf/neutron.py and I wanted to ask about it | 07:43 |
| *** mgoddard has joined #openstack-nova | 07:43 | |
| apuimedo | since when I run the tox -egenconfig I do not see the neutron section get generated in the sample config file | 07:43 |
| *** sdake has quit IRC | 07:43 | |
| *** ljxiash has quit IRC | 07:43 | |
| *** sdake has joined #openstack-nova | 07:44 | |
| *** sameo has joined #openstack-nova | 07:44 | |
| *** thorst has joined #openstack-nova | 07:45 | |
| *** lpetrut has quit IRC | 07:47 | |
| *** sridharg has joined #openstack-nova | 07:47 | |
| *** ljxiash has joined #openstack-nova | 07:47 | |
| *** ducttape_ has joined #openstack-nova | 07:49 | |
| *** phuongnh has quit IRC | 07:50 | |
| openstackgerrit | Augustina Ragwitz proposed openstack/nova: Assume neutron port_binding extensions enabled https://review.openstack.org/357726 | 07:51 |
| openstackgerrit | Augustina Ragwitz proposed openstack/nova: Use admin context to create neutron ports https://review.openstack.org/357540 | 07:51 |
| * auggy passes out | 07:51 | |
| *** salv-orlando has joined #openstack-nova | 07:51 | |
| *** paul-carlton2 has quit IRC | 07:52 | |
| *** thorst has quit IRC | 07:52 | |
| *** paul-carlton2 has joined #openstack-nova | 07:52 | |
| *** ducttape_ has quit IRC | 07:53 | |
| openstackgerrit | edan david proposed openstack/nova: WIP: Assign mac address to vf interface https://review.openstack.org/364121 | 07:53 |
| openstackgerrit | JiWei proposed openstack/nova: 'unmap' is written without considering the configuration https://review.openstack.org/364122 | 07:53 |
| *** liverpooler has joined #openstack-nova | 07:54 | |
| *** ihrachys has quit IRC | 07:55 | |
| *** edand has joined #openstack-nova | 07:55 | |
| *** zzzeek has quit IRC | 08:00 | |
| *** salv-orlando has quit IRC | 08:00 | |
| *** ljjjustin has quit IRC | 08:00 | |
| *** zehua has quit IRC | 08:00 | |
| *** zzzeek has joined #openstack-nova | 08:00 | |
| *** paul-carlton2 has quit IRC | 08:00 | |
| *** jaosorior has joined #openstack-nova | 08:04 | |
| *** hoangcx has quit IRC | 08:05 | |
| *** avolkov has joined #openstack-nova | 08:05 | |
| *** mamandle has joined #openstack-nova | 08:07 | |
| *** oanson has joined #openstack-nova | 08:11 | |
| *** matrohon has joined #openstack-nova | 08:12 | |
| *** mingyu has quit IRC | 08:14 | |
| *** BharatK has joined #openstack-nova | 08:15 | |
| *** BharatK has quit IRC | 08:16 | |
| *** BharatK has joined #openstack-nova | 08:17 | |
| *** diga has joined #openstack-nova | 08:18 | |
| *** paul-carlton2 has joined #openstack-nova | 08:18 | |
| *** BharatK has quit IRC | 08:24 | |
| *** sudipto has joined #openstack-nova | 08:24 | |
| *** Hosam_ has joined #openstack-nova | 08:25 | |
| *** sudipto has quit IRC | 08:27 | |
| *** BharatK has joined #openstack-nova | 08:28 | |
| *** Hosam has quit IRC | 08:29 | |
| *** sudipto has joined #openstack-nova | 08:30 | |
| *** rkmrhj has joined #openstack-nova | 08:33 | |
| *** Hosam has joined #openstack-nova | 08:35 | |
| *** priteau has joined #openstack-nova | 08:37 | |
| *** lucas-dinner is now known as lucasagomes | 08:38 | |
| johnthetubaguy | sdague: yeah, that made me ask the question, didn't catch the debate on that... I thought it was odd that could be None. | 08:38 |
| *** Hosam_ has quit IRC | 08:39 | |
| rkmrhj | Sorry to write so late at night. I uploaded 2 bug-fixes to gerrit about 3 weeks ago. But the reviewer who reviewed my patches is only one person. | 08:44 |
| rkmrhj | Is there someone who can review my patches? My patches are https://review.openstack.org/#/c/354617/ and https://review.openstack.org/#/c/327001/ . | 08:45 |
| *** bkopilov has quit IRC | 08:45 | |
| *** houming has joined #openstack-nova | 08:46 | |
| *** sridharg has quit IRC | 08:48 | |
| *** ducttape_ has joined #openstack-nova | 08:49 | |
| *** thorst has joined #openstack-nova | 08:50 | |
| *** hwoarang_ has joined #openstack-nova | 08:54 | |
| *** derekh has joined #openstack-nova | 08:54 | |
| *** jhesketh_ has joined #openstack-nova | 08:54 | |
| *** HenryG_ has joined #openstack-nova | 08:54 | |
| *** ducttape_ has quit IRC | 08:54 | |
| *** amotoki_ has joined #openstack-nova | 08:55 | |
| *** jeblair_ has joined #openstack-nova | 08:55 | |
| *** kevinz has quit IRC | 08:56 | |
| *** jlvillal_ has joined #openstack-nova | 08:57 | |
| *** toabctl_ has joined #openstack-nova | 08:57 | |
| *** kevinz has joined #openstack-nova | 08:57 | |
| *** thorst has quit IRC | 08:57 | |
| *** sudipto has quit IRC | 08:58 | |
| *** jlvillal has quit IRC | 08:58 | |
| *** dulek has quit IRC | 08:58 | |
| *** amotoki has quit IRC | 08:58 | |
| *** cmartin has quit IRC | 08:58 | |
| *** Fdaisuke has quit IRC | 08:58 | |
| *** lennyb has quit IRC | 08:58 | |
| *** cfriesen has quit IRC | 08:58 | |
| *** kswiatek has quit IRC | 08:58 | |
| *** jhesketh has quit IRC | 08:58 | |
| *** darrenc has quit IRC | 08:58 | |
| *** tpatzig_ has quit IRC | 08:58 | |
| *** gstepanov__ has quit IRC | 08:58 | |
| *** lamt has quit IRC | 08:58 | |
| *** raj_singh has quit IRC | 08:58 | |
| *** mat128 has quit IRC | 08:58 | |
| *** bapalm has quit IRC | 08:58 | |
| *** HenryG has quit IRC | 08:58 | |
| *** anteaya has quit IRC | 08:58 | |
| *** brad[] has quit IRC | 08:58 | |
| *** aloga has quit IRC | 08:58 | |
| *** jeblair has quit IRC | 08:58 | |
| *** niedbalski has quit IRC | 08:58 | |
| *** Anticimex has quit IRC | 08:58 | |
| *** eliqiao has quit IRC | 08:58 | |
| *** ejat has quit IRC | 08:58 | |
| *** hwoarang has quit IRC | 08:58 | |
| *** toabctl has quit IRC | 08:58 | |
| *** jgrassler has quit IRC | 08:58 | |
| *** toabctl_ is now known as toabctl | 08:58 | |
| *** salv-orlando has joined #openstack-nova | 08:58 | |
| *** bapalm has joined #openstack-nova | 08:59 | |
| *** mat128 has joined #openstack-nova | 08:59 | |
| *** toabctl has quit IRC | 08:59 | |
| *** Anticimex has joined #openstack-nova | 09:00 | |
| *** aloga has joined #openstack-nova | 09:00 | |
| *** Fdaisuke has joined #openstack-nova | 09:00 | |
| *** toabctl has joined #openstack-nova | 09:00 | |
| *** jlvillal_ is now known as jlvillal | 09:00 | |
| *** sridharg has joined #openstack-nova | 09:00 | |
| *** jlvillal is now known as Guest77956 | 09:01 | |
| *** salv-orlando has quit IRC | 09:01 | |
| *** sudipto has joined #openstack-nova | 09:03 | |
| *** dulek has joined #openstack-nova | 09:03 | |
| *** lennyb has joined #openstack-nova | 09:03 | |
| *** cfriesen has joined #openstack-nova | 09:03 | |
| *** kswiatek has joined #openstack-nova | 09:03 | |
| *** darrenc has joined #openstack-nova | 09:03 | |
| *** tpatzig_ has joined #openstack-nova | 09:03 | |
| *** gstepanov__ has joined #openstack-nova | 09:03 | |
| *** lamt has joined #openstack-nova | 09:03 | |
| *** raj_singh has joined #openstack-nova | 09:03 | |
| *** anteaya has joined #openstack-nova | 09:03 | |
| *** brad[] has joined #openstack-nova | 09:03 | |
| *** niedbalski has joined #openstack-nova | 09:03 | |
| *** eliqiao has joined #openstack-nova | 09:03 | |
| *** ejat has joined #openstack-nova | 09:03 | |
| *** jgrassler has joined #openstack-nova | 09:05 | |
| *** salv-orlando has joined #openstack-nova | 09:05 | |
| *** cdent has joined #openstack-nova | 09:05 | |
| *** zenoway has quit IRC | 09:07 | |
| *** kylek3h has joined #openstack-nova | 09:09 | |
| *** salv-orlando has quit IRC | 09:10 | |
| *** zenoway has joined #openstack-nova | 09:12 | |
| *** ecelik has quit IRC | 09:12 | |
| *** mvk has joined #openstack-nova | 09:13 | |
| bauzas | p!spy | 09:14 |
| *** kylek3h has quit IRC | 09:14 | |
| *** sridharg has quit IRC | 09:14 | |
| *** sudipto has quit IRC | 09:15 | |
| *** sridharg has joined #openstack-nova | 09:15 | |
| openstackgerrit | Rong Han proposed openstack/nova-specs: iAdd a parameter of availability zone for live-migration https://review.openstack.org/364179 | 09:16 |
| *** HenryG_ is now known as HenryG | 09:17 | |
| openstackgerrit | Rong Han proposed openstack/nova-specs: Add a parameter of availability zone for live-migration https://review.openstack.org/364179 | 09:17 |
| kashyap | slaweq: Hey, oh, so that was the wrong log? Did you follow my analysis in comment#13? | 09:18 |
| *** BharatK has quit IRC | 09:20 | |
| *** zenoway has quit IRC | 09:21 | |
| kashyap | I'll let you upload another log that exhibits the issue, as you said in comment#14. | 09:21 |
| *** lpetrut has joined #openstack-nova | 09:22 | |
| *** ihrachys has joined #openstack-nova | 09:25 | |
| *** Hosam_ has joined #openstack-nova | 09:27 | |
| *** HenryG has quit IRC | 09:29 | |
| *** HenryG has joined #openstack-nova | 09:29 | |
| *** mingyu has joined #openstack-nova | 09:30 | |
| *** zenoway has joined #openstack-nova | 09:30 | |
| *** Hosam has quit IRC | 09:31 | |
| *** hwoarang_ is now known as hwoarang | 09:31 | |
| *** Hosam has joined #openstack-nova | 09:33 | |
| *** mingyu has quit IRC | 09:34 | |
| *** roeyc has quit IRC | 09:34 | |
| *** wangqun_ has quit IRC | 09:35 | |
| *** Hosam_ has quit IRC | 09:37 | |
| *** abhishekk has quit IRC | 09:38 | |
| *** mingyu has joined #openstack-nova | 09:39 | |
| *** zenoway has quit IRC | 09:39 | |
| *** TravT_ has joined #openstack-nova | 09:40 | |
| *** TravT has quit IRC | 09:43 | |
| *** salv-orlando has joined #openstack-nova | 09:45 | |
| *** daemontool has joined #openstack-nova | 09:45 | |
| *** sdague has joined #openstack-nova | 09:46 | |
| *** diga has quit IRC | 09:49 | |
| openstackgerrit | Pawel Koniszewski proposed openstack/nova: Remove block_migration argument from live migration operations https://review.openstack.org/295316 | 09:50 |
| openstackgerrit | Pawel Koniszewski proposed openstack/nova: Remove block_migration from LM post operations https://review.openstack.org/295315 | 09:50 |
| openstackgerrit | Pawel Koniszewski proposed openstack/nova: Remove block_migration from LM rollback https://review.openstack.org/292826 | 09:50 |
| *** ducttape_ has joined #openstack-nova | 09:50 | |
| cdent | sdague, dansmith, johnthetubaguy : When you guys surface and/or have a moment there are some pending questions on desired behavior on https://review.openstack.org/#/c/363209/ (writing allocations) that need some shared brain to resolve. Thanks. | 09:52 |
| *** ducttape_ has quit IRC | 09:54 | |
| *** thorst has joined #openstack-nova | 09:55 | |
| sdague | cdent: ok, just landing on my terminal, I need to finish this cup of coffee before that sql is going to even start to make sense to me | 09:57 |
| *** mingyu has quit IRC | 09:57 | |
| openstackgerrit | Pawel Koniszewski proposed openstack/nova: DO NOT MERGE - try to run LM tests with grenade https://review.openstack.org/364206 | 09:57 |
| sdague | did any client code appear yesterday? | 09:57 |
| cdent | sdague: no rush from my side, I assumed you weren't even here yet and was just leaving a message for the future | 09:58 |
| sdague | cdent: no worries. :) | 09:58 |
| *** mingyu has joined #openstack-nova | 09:58 | |
| cdent | my understanding is that jay crashed out around midnight last night after pushing the mass of sql, so I don't know about client code | 09:58 |
| sdague | ok | 09:59 |
| *** code-R has joined #openstack-nova | 09:59 | |
| cdent | for reasons unclear to me, I was up then, and did a first review on the sql stuff, edleafe fixed most of those things, and there are a few bits remaining, discovered just now, before it is soup | 10:01 |
| sdague | ok cool | 10:01 |
| *** sdake has quit IRC | 10:01 | |
| *** zenoway has joined #openstack-nova | 10:02 | |
| sdague | cdent: question... about the wsgi stack | 10:02 |
| * cdent listens | 10:02 | |
| sdague | I was looking at stuff in keystone to cross reference | 10:02 |
| *** thorst has quit IRC | 10:03 | |
| sdague | is there a reason that you didn't @webob.dec.wsgify the __call__ in handler? | 10:03 |
| * cdent tries to remember | 10:03 | |
| sdague | because that seems to be the common pattern so that you are dealing with requests earlier | 10:03 |
| openstackgerrit | Chris Dent proposed openstack/nova: Add {get_,set_}aggregates to objects.ResourceProvider https://review.openstack.org/355263 | 10:03 |
| sdague | it seems like every dispatch function is decorated that way | 10:03 |
| sdague | and it explains why I felt a little lost on some stuff yesterday | 10:04 |
| sdague | if there is a reason, that's cool. But if not, it might be better to pull that up one layer. | 10:04 |
| *** Hosam_ has joined #openstack-nova | 10:06 | |
| *** zenoway has quit IRC | 10:07 | |
| cdent | It's because of the way dispatch is called to handle the 405-maintaining of the routing and then goes on to call the dispatched-to-handlers. That preserve the wsgi interface down into the whatever gets dispatched-to, so that it is possible to have "apps" which are just regular wsgi things that we didn't right ourselves (not that that's expected, but I thought it worthwhile to preserve WSGI at the dispatching layer). | 10:07 |
| cdent | sigh | 10:07 |
| cdent | s/right/write/ | 10:07 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/363871 | 10:08 |
| *** rmart04 has joined #openstack-nova | 10:09 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Testing latest u-c https://review.openstack.org/303345 | 10:10 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Testing latest u-c https://review.openstack.org/303345 | 10:10 |
| *** Hosam has quit IRC | 10:10 | |
| cdent | the hope was that for most people doing maintenance of existing handlers or creation of new ones the only thing they need to worry about is updating the DECLARATIONS and go change/create their handler that does their thing | 10:10 |
| cdent | sdague: the mental model is that each method on each url is its own indepdendent application | 10:10 |
| cdent | (that happens to reuse stuff by way of middleware and things in the environ) | 10:11 |
| openstackgerrit | Chris Dent proposed openstack/nova: Implement get and set aggregates in the placement API https://review.openstack.org/357993 | 10:11 |
| sdague | right, so everywhere else I've seen in openstack, we wsgify early | 10:12 |
| sdague | so you are typically working with a req directly | 10:12 |
| sdague | you can always get environ back from the req | 10:12 |
| *** code-R_ has joined #openstack-nova | 10:13 | |
| *** rkmrhj has quit IRC | 10:13 | |
| sdague | I think that if you want to inject something new that's wsgi raw, that's kind of what the middleware pipeline is for | 10:13 |
| *** zenoway has joined #openstack-nova | 10:14 | |
| *** code-R has quit IRC | 10:16 | |
| cdent | that doesn't work if what you want to inject is a raw wsgi endpoint | 10:16 |
| cdent | it could certainly change, but I think it would be step in the wrong direction | 10:17 |
| cdent | it takes things away from being generic and standard-following to being implementation specific, which means that changing things later becomes more complex. A thing we want to preserve is the ability to make structural changes with low cost when we realize that something is not what we want. | 10:18 |
| cdent | Consider, for example, the ways in which a breadth and depth commitment to WSME (instead of a tool-oriented) approach has made things a lot harder for some of the projects. | 10:18 |
| sdague | sure, but converting to req early isn't wsme | 10:19 |
| sdague | and in 6 years of practice, it's never gotten in the way | 10:19 |
| *** riccardo has joined #openstack-nova | 10:19 | |
| *** jhesketh_ is now known as jhesketh | 10:20 | |
| cdent | Can you explain again what problem or issue doing so would solve? | 10:20 |
| cdent | (btw, i'm making food so I'm back and forth) | 10:22 |
| sdague | it would remove a gratuitous difference here from the rest of openstack. I get there are times to be different when you get a huge gain out of it. But I don't see the gain here. | 10:23 |
| sdague | and things like request logging would be a lot less silly if we'd already wsgified. Doing that was the first time I realized some of the implications here of pushing the wsgify down a layer | 10:23 |
| *** pilgrimstack has quit IRC | 10:24 | |
| *** maestropandy1 has joined #openstack-nova | 10:25 | |
| *** tbachman has quit IRC | 10:26 | |
| *** pilgrimstack has joined #openstack-nova | 10:27 | |
| *** scsnow has joined #openstack-nova | 10:27 | |
| riccardo | hi community, I have a Nova extension that I wanted to port from Juno to Mitaka. Now, I got informed that extensions are not anymore the right way. Therefore I would like to ask if there is a tutorial to build a new service and where i can get support. The idea (may be in a second step) is also to publish the service in oder that other OpenStack user can use and benefit from it. | 10:28 |
| *** murali__ has joined #openstack-nova | 10:28 | |
| murali__ | Someone please have a look at this and let me know ur thoughts | 10:28 |
| murali__ | https://ask.openstack.org/en/question/96376/nova-docker-setup-in-devstack/ | 10:28 |
| murali__ | I am trying for fixing this, Bt nothing of any help | 10:29 |
| cdent | sdague: is there a version of your logging bits already in gerrit somewhere? | 10:30 |
| sdague | yes, it's already approved, in a recheck | 10:31 |
| sdague | https://review.openstack.org/#/c/363130/ | 10:31 |
| *** pilgrimstack has quit IRC | 10:31 | |
| * cdent looks | 10:31 | |
| sdague | anyway, when I backed the decision on not using a framework like flask, I did not actually expect this to go so far from the norms. But the further this is away from norms, the slower the reviews will be, because people have to switch up thinking about the world | 10:33 |
| cdent | s/norms/local norms/ | 10:34 |
| sdague | cdent: sure, but that's the point | 10:34 |
| sdague | we've got a community, being different without huge gains means mostly not caring about people's review bw, because you just made them all do extra work | 10:35 |
| cdent | So on that logging, can you explain to me why we want the wsgi application to do request logging when that is usually the job of the wsgi server? I'm not being snarky, I'm genuinely trying to understand. | 10:35 |
| sdague | and at a certain point, they just don't | 10:35 |
| sdague | because the wsgi doesn't log really important things | 10:35 |
| sdague | like request-id | 10:35 |
| sdague | and project/tenant | 10:36 |
| sdague | user | 10:36 |
| cdent | let's not go down that pathway, sdague, this code has been under review for six months. If you wanted it to be different you could have said a lot more then instead of now on the day of ff. | 10:36 |
| cdent | I want it to be good and correct and make people happy. I'll try to make it so, and hopefully other people will too. | 10:37 |
| sdague | sure, which is why I'm asking for a change (post ff) to realign this | 10:38 |
| cdent | second question on the logging: why isn't done as middleware? | 10:38 |
| ratailor | can anyone review, https://review.openstack.org/#/c/312887/ , its hanging since log. | 10:38 |
| cdent | or what I mean is: is there a reason why? | 10:38 |
| sdague | cdent: maybe it could be, but it's in the main path in the rest of the projects | 10:38 |
| *** EinstCrazy has quit IRC | 10:40 | |
| *** Qiming has quit IRC | 10:40 | |
| *** ljxiash has quit IRC | 10:40 | |
| *** hehx has quit IRC | 10:40 | |
| *** chenfei has quit IRC | 10:40 | |
| sdague | which means it's a pattern we know kind of works for real in the field, and don't have to guess about when things are skipped past in the chain, or translated in a way we don't anticipate | 10:41 |
| cdent | sdague: You know, I've been trying really hard to make openstack, especially bits of nova, _better_ in ways that are commonly recognized as better and it is a constant fight against them needing to be the same as things are already done. I understand the perceived benefits of that, but it is active friction against a culture of improvement. | 10:41 |
| *** pilgrimstack has joined #openstack-nova | 10:41 | |
| *** FL1SK has quit IRC | 10:41 | |
| cdent | And I know you do too | 10:41 |
| cdent | (try really hard to make things better) | 10:41 |
| cdent | but I'm fucking burnt | 10:41 |
| *** hehx has joined #openstack-nova | 10:41 | |
| *** ljxiash has joined #openstack-nova | 10:42 | |
| cdent | There has been zero about nova experience which makes me want to recommend working on it to anyone. | 10:42 |
| cdent | But I keep soldiering on. I'm not sure why. | 10:42 |
| *** trisq has quit IRC | 10:45 | |
| *** Qiming has joined #openstack-nova | 10:45 | |
| *** chenfei has joined #openstack-nova | 10:46 | |
| *** sudipto has joined #openstack-nova | 10:47 | |
| *** Hosam has joined #openstack-nova | 10:47 | |
| *** gongysh has quit IRC | 10:49 | |
| *** Hosam_ has quit IRC | 10:50 | |
| *** maestropandy1 has quit IRC | 10:50 | |
| *** ducttape_ has joined #openstack-nova | 10:51 | |
| *** roeyc has joined #openstack-nova | 10:52 | |
| *** sridharg has quit IRC | 10:55 | |
| *** ducttape_ has quit IRC | 10:55 | |
| *** salv-orlando has quit IRC | 10:58 | |
| *** thorst has joined #openstack-nova | 11:00 | |
| cdent | sdague: I've put looking at refactoring the logging on my todo for some near future time | 11:04 |
| riccardo | hi, I have a Nova extension that I want to port from Juno to Mitaka. Now, I got informed that extensions are not anymore the right way to add new functionalities. Therefore I would like to ask if there is a tutorial to build a new service. The idea (may be in a second step) is also to publish the service in oder that other OpenStack user can use and benefit from it. Basically it is a "fairness service" that reallocates virtual resources during | 11:05 |
| riccardo | runtime to enforce fairness among cloud users. | 11:05 |
| sdague | ok | 11:05 |
| *** paul-carlton2 has quit IRC | 11:06 | |
| sdague | riccardo: there isn't as of yet. It would be a good thing to exist. | 11:06 |
| sdague | the tutorial that is | 11:06 |
| *** thorst has quit IRC | 11:08 | |
| *** kylek3h has joined #openstack-nova | 11:10 | |
| riccardo | sdague: ok, thx | 11:10 |
| riccardo | A former student did an extension to the Nova service for Juno: https://github.com/savf/novafairness (see the nova/fairness folder). Now I want to take away from Nova and have it as a separate service. Is there some framework that I can utilize and put my own code in order to build this new service? | 11:14 |
| *** gcb has joined #openstack-nova | 11:14 | |
| riccardo | the fairness service | 11:15 |
| slaweq | kashyap: hello | 11:15 |
| kashyap | slaweq: Yes? | 11:16 |
| slaweq | I just uploaded new libvirt log in https://bugs.launchpad.net/nova/+bug/1530275 - this one is from failed attempt to create live_snapshot | 11:16 |
| openstack | Launchpad bug 1530275 in OpenStack Compute (nova) "Live snapshot is corrupted (possibly race condition?)" [Undecided,In progress] - Assigned to Slawek Kaplonski (slaweq) | 11:16 |
| slaweq | kashyap: and also I propose some fix for that: https://review.openstack.org/#/c/363926/ | 11:16 |
| slaweq | maybe You could take a look on it? | 11:17 |
| kashyap | slaweq: Let me look. So, this is the log from a corrupted snapshot? | 11:17 |
| slaweq | yes | 11:17 |
| slaweq | snapshot image got different size (smaller) then image created properly (with my patch) | 11:18 |
| slaweq | and I can't boot vm from this broken snapshot (I have info that filesystem is corrupted there) | 11:18 |
| *** rtheis has joined #openstack-nova | 11:18 | |
| kashyap | slaweq: Ah-ha, indeed, in this case, as you can see the 'block-job-cancel' was issued too soon | 11:19 |
| slaweq | kashyap: I was trying to make thi log as small as possible but I have many instance on host and there is a lot of logged messages there | 11:20 |
| kashyap | slaweq: No worries, I found what I was looking for | 11:20 |
| *** paul-carlton2 has joined #openstack-nova | 11:20 | |
| sdague | riccardo: ok, sure. The reality is we should have something like an openstack-service-cookiecutter under openstack-dev, however no one has gotten there yet | 11:20 |
| kashyap | slaweq: If you see the request id "libvirt-14" from the failed log, where QEMU is queried to check the block operation status, it reports as "ready": false | 11:21 |
| *** bvanhav has joined #openstack-nova | 11:21 | |
| kashyap | slaweq: Following that, with ID "libvirt-15", libvirt aborts the snapshot operation with 'block-job-cancel' | 11:22 |
| slaweq | yes, I see | 11:22 |
| kashyap | Without waiting for "ready": true | 11:22 |
| kashyap | Hence, it seems to be corrupted | 11:22 |
| slaweq | yes, and in nova it is "visible" as status.end = 0 | 11:22 |
| slaweq | and status.cur = 0 | 11:22 |
| slaweq | so nova finish waiting for job completed and starts extracts snapshot | 11:23 |
| slaweq | which is wrong | 11:23 |
| slaweq | if nova would wait littlebit more, snapshot is created fine then | 11:23 |
| *** ljxiash has quit IRC | 11:23 | |
| kashyap | slaweq: Exactly | 11:24 |
| slaweq | great that You can confirm this issue :) | 11:24 |
| riccardo | sdague: I am not sure if I understand. Do you mean that it is just not common to build a new service, so there is no documentation? | 11:24 |
| sdague | riccardo: that there is no documentation | 11:24 |
| sdague | it it quite common, seeing how many projects exist in openstack | 11:25 |
| sdague | but, there is no easy guide for building a new one right now. That mostly just comes from reading whats out there. | 11:25 |
| tdurakov | hi folks | 11:25 |
| *** thorst has joined #openstack-nova | 11:26 | |
| sdague | tdurakov: is it always the live migration test that's falling over? | 11:26 |
| riccardo | sdague: I see. I am very new to the OpenStack community and have very little idea on how to do it. Any hint is highly appreciate. | 11:26 |
| *** zerda2 has quit IRC | 11:26 | |
| sdague | riccardo: ok, well today is feature freeze, so this is a bad week to ask those questions | 11:26 |
| tdurakov | sdague: for nfs? | 11:26 |
| tdurakov | or for no-shared storage case? | 11:26 |
| sdague | tdurakov: just subnode connecting | 11:27 |
| sdague | tdurakov: something is up with subnodes not being accessible - http://logstash.openstack.org/#/dashboard/file/logstash.json?query=message:%5C%22Killed%20timeout%20-s%209%5C%22%20AND%20build_name:*multinode* | 11:27 |
| sdague | which is causing multinode jobs to time out at 3 hours | 11:27 |
| riccardo | sdague: OK, I understand. then I will first investigate more by myself. Do you have any suggestion of a small service where i can study the structure? | 11:28 |
| tdurakov | sdague: haven't seen that before | 11:28 |
| *** roeyc has quit IRC | 11:29 | |
| tdurakov | sdague: it's for gate-tempest-dsvm-neutron-multinode-full job right? | 11:29 |
| tdurakov | are there any specific options for that? | 11:29 |
| *** maestropandy has joined #openstack-nova | 11:30 | |
| *** roeyc has joined #openstack-nova | 11:30 | |
| tdurakov | from logstash I see that valid for multinode-full and neutron multinode, and it doesn't affect live-migration job | 11:31 |
| *** maestropandy has left #openstack-nova | 11:31 | |
| *** sudipto has quit IRC | 11:32 | |
| sdague | tdurakov: ok, just starting to investigate, over in #openstack-infra to figure this out | 11:32 |
| *** lucasagomes is now known as lucas-hungry | 11:33 | |
| riccardo | Hi guys, I know you are all very busy with the today's feature freeze. So, I don't wanna bother you anymore for today. But maybe anybody has suggestion where I can find a small OpenStack example service where I can study the structure. I would like to build a new service. | 11:34 |
| mdbooth | sdague: Have you come across the issue with python-cryptography in liberty's upper constraints? | 11:34 |
| mdbooth | sdague: It's this: https://github.com/pyca/cryptography/issues/2750 | 11:35 |
| mdbooth | Essentially, it means that python-cryptography < 1.2.3 is no longer buildable against patched openssl | 11:35 |
| sdague | mdbooth: I have not | 11:35 |
| mdbooth | But we're pinning it at 1.1.2 iirc | 11:36 |
| sdague | mdbooth: well, you could look at bumping the upper constraints. By policy we don't do that automatically because it breaks things. But doing so for good reason (like this one) is probably fine | 11:37 |
| mdbooth | I've confirmed locally that upping that pin to 1.2.3 builds, but wondering what the process would be. | 11:37 |
| sdague | mdbooth: just propose the patch to upper-constraints.txt on the appropriate branch | 11:37 |
| kashyap | mdbooth: IIRC, in the past I just submitted it to stable, and let the comments roll on the review | 11:37 |
| sdague | with a detailed commit message on why | 11:37 |
| mdbooth | sdague kashyap: Will do, thanks | 11:37 |
| sdague | mdbooth: np | 11:37 |
| *** moshele has quit IRC | 11:38 | |
| *** sudipto has joined #openstack-nova | 11:38 | |
| *** baoli has joined #openstack-nova | 11:41 | |
| *** rfolco has joined #openstack-nova | 11:41 | |
| *** baoli_ has joined #openstack-nova | 11:42 | |
| *** Kevin_Zheng has quit IRC | 11:42 | |
| *** baoli has quit IRC | 11:46 | |
| *** ManishD has quit IRC | 11:47 | |
| *** gongysh has joined #openstack-nova | 11:49 | |
| *** ljxiash has joined #openstack-nova | 11:49 | |
| mdbooth | sdague: https://review.openstack.org/364249 | 11:50 |
| mdbooth | What tests will run against that, out of curiosity? | 11:50 |
| sdague | a bunch of full stack tests | 11:51 |
| sdague | so devstack + tempest in a set of different configs | 11:52 |
| *** ducttape_ has joined #openstack-nova | 11:52 | |
| mdbooth | If anybody is familiar with python ffi, I'm also really curious to know why python-cryptography embeds function signatures from an external library. I assume there's a really good reason for it, but at first glance it looks pretty evil. | 11:54 |
| mdbooth | They're passed to ffi.cdef(). I'm guessing that doesn't something magical which means a #include doesn't work. | 11:55 |
| mdbooth | s/doesn't/does/ | 11:55 |
| *** Kevin_Zheng has joined #openstack-nova | 11:56 | |
| *** ducttape_ has quit IRC | 11:56 | |
| sdague | mdbooth: dstuft probably understands all the cffi stuff better than anyone else | 11:58 |
| sdague | iirc cryptopgraphy was one of the early users, and yes, a lot of thinks are a bit funky with it | 11:59 |
| *** rfolco has quit IRC | 11:59 | |
| *** edand has quit IRC | 12:00 | |
| jroll | mdbooth: #cryptography-dev is usually pretty friendly, too, if you want to bug them | 12:00 |
| mdbooth | jroll: Nah, just scratching an itch :) | 12:00 |
| *** rfolco has joined #openstack-nova | 12:00 | |
| jroll | heh, right on | 12:00 |
| alex_xu | question: we freeze at 9/2 morning or the end of 9/2 | 12:01 |
| sdague | alex_xu: good question, we'll see what mriedem says when he gets up | 12:02 |
| alex_xu | sdague: ok, cool | 12:02 |
| alex_xu | sdague: johnthetubaguy, so just reminder, don't forget the last few API patches, https://review.openstack.org/356963 https://review.openstack.org/360943 https://review.openstack.org/322944 | 12:03 |
| sdague | right now, we're not landing any more code for the release, because of a multinode issue | 12:03 |
| alex_xu | :( | 12:03 |
| sdague | so unless that gets sorted, it doesn't really matter | 12:03 |
| *** zengine has joined #openstack-nova | 12:04 | |
| *** jaosorior has quit IRC | 12:06 | |
| *** jaosorior has joined #openstack-nova | 12:06 | |
| *** pradk has joined #openstack-nova | 12:09 | |
| *** mingyu has quit IRC | 12:09 | |
| tdurakov | has anyone seen PaulMurray today? | 12:11 |
| *** ducttape_ has joined #openstack-nova | 12:11 | |
| tdurakov | jroll: hello, could you please review these series https://review.openstack.org/#/c/347948/ ? | 12:12 |
| *** ducttape_ has quit IRC | 12:13 | |
| jroll | tdurakov: it's on my todo list for this week, it's been a busy one, sorry | 12:13 |
| tdurakov | jroll: no worries, thanks in advance | 12:13 |
| *** mingyu has joined #openstack-nova | 12:13 | |
| jroll | sdague: fwiw, matt said his plan was to submit the release by the end of today | 12:13 |
| jroll | idk if that's changed in the last 24h or so | 12:13 |
| tdurakov | pkoniszewski: hi | 12:13 |
| *** riccardo has quit IRC | 12:14 | |
| *** trisq has joined #openstack-nova | 12:14 | |
| *** BharatK has joined #openstack-nova | 12:16 | |
| *** ducttape_ has joined #openstack-nova | 12:17 | |
| *** rfolco has quit IRC | 12:17 | |
| *** rfolco has joined #openstack-nova | 12:18 | |
| *** Hosam_ has joined #openstack-nova | 12:19 | |
| *** sridharg has joined #openstack-nova | 12:20 | |
| *** oanson has quit IRC | 12:20 | |
| *** lucas-hungry is now known as lucasagomes | 12:21 | |
| *** BharatK has quit IRC | 12:21 | |
| *** Hosam has quit IRC | 12:22 | |
| *** mamandle has quit IRC | 12:22 | |
| *** moshele has joined #openstack-nova | 12:24 | |
| *** kolesovdv has joined #openstack-nova | 12:29 | |
| *** ratailor has quit IRC | 12:29 | |
| *** mvk has quit IRC | 12:30 | |
| *** mdrabe has joined #openstack-nova | 12:31 | |
| bauzas | sdague: johnthetubaguy: so I think we need to have https://review.openstack.org/#/c/362766/5 merged before our N-3 cyt | 12:35 |
| bauzas | cut | 12:35 |
| bauzas | if not, it would be sad for operatords | 12:35 |
| bauzas | and us needing to think about data migrations in Ocata | 12:35 |
| *** ducttape_ has quit IRC | 12:35 | |
| bauzas | so, I asked for a recheck (because I want to be sure our placement job is happy) but I'd ideally want to see how we could make sure we have that for Newton | 12:36 |
| *** gongysh has quit IRC | 12:36 | |
| sdague | bauzas: no, we said that was off the table yesterday | 12:37 |
| bauzas | orly? | 12:37 |
| *** mingyu has quit IRC | 12:38 | |
| bauzas | I missed that, apologies | 12:38 |
| cdent | Given the concerns raised on the same call about data correctness, I still think it is worth merging if people think it is acceptable code. Nobody _has_ to use it. | 12:38 |
| *** prateek has quit IRC | 12:38 | |
| bauzas | cdent: tbh, I want to make sure our devstack job runs it | 12:39 |
| sdague | bauzas: I don't want to stop and do the qa analysis to figure out if we're testing common case correctly | 12:39 |
| sdague | and another branch here means that's even hard | 12:39 |
| sdague | harder | 12:40 |
| bauzas | sdague: I do understand | 12:40 |
| sdague | we were really trying to scope back to minimal needed for newton | 12:40 |
| bauzas | yeah, gotcha | 12:40 |
| sdague | which still takes us past freeze, because we need the allocations reporting | 12:40 |
| openstackgerrit | Claudiu Belu proposed openstack/nova: libvirt: fixes python 3 related unit tests https://review.openstack.org/278834 | 12:40 |
| openstackgerrit | Claudiu Belu proposed openstack/nova: Fixes python 3 unit tests https://review.openstack.org/278810 | 12:40 |
| bauzas | sdague: so, if we say the above is out for Newton, then I could write a devstack patch for using it | 12:40 |
| sdague | also, don't recheck anything right now, it will fail, because of osc. A work around is in the gate | 12:41 |
| bauzas | sdague: oh my bad | 12:41 |
| sdague | which was my last 1.5 hours | 12:41 |
| *** FL1SK has joined #openstack-nova | 12:41 | |
| bauzas | oops, haven't seen that :( | 12:41 |
| cdent | bauzas: the lib/placement code already has a variable that allows one to state "I want to use the placement db" | 12:41 |
| sdague | yeh, sorry, didn't raise the issue as it was in the middle of debug | 12:42 |
| cdent | "PLACEMENT_DB_ENABLED" | 12:42 |
| bauzas | cdent: okay, I need to look again the devstack code | 12:42 |
| bauzas | nice | 12:42 |
| bauzas | so just a job modification then | 12:42 |
| openstackgerrit | Claudiu Belu proposed openstack/nova: numa: Fixes NUMA topology related unit tests https://review.openstack.org/278835 | 12:43 |
| bauzas | sdague: yeah, and I apologize, I didn't paid tribute to the IRC conversations like as usual, today's my old daughter first day at 1st grade :) | 12:43 |
| openstackgerrit | Claudiu Belu proposed openstack/nova: compute: fixes python 3 related unit tests https://review.openstack.org/300262 | 12:43 |
| sdague | bauzas: no worries | 12:43 |
| sdague | also, congrats | 12:43 |
| bauzas | well, I tried to ask our French ministry to delay that day because of the FeatureFreeze, but no luck... :) | 12:44 |
| *** Kimmo___ is now known as Kimmo_ | 12:44 | |
| bauzas | sdague: then, something else urgent to review before we cut ? | 12:46 |
| sdague | bauzas: there are some api policy patches that alex_xu's got | 12:47 |
| sdague | the db part of the placement work from cdent | 12:47 |
| sdague | I can't remember where the cell cut line currently is | 12:48 |
| *** gcb has quit IRC | 12:48 | |
| bauzas | cdent: https://review.openstack.org/#/c/362863/ is the one ? | 12:48 |
| edleafe | cdent: good point on checking for inventory on all RPs, not just that some records are found | 12:49 |
| edleafe | I can fix that, and add a test case for it | 12:49 |
| cdent | bauzas, sdague I thougt we punted on aggregates too (https://review.openstack.org/#/c/362863/) but if not, then yes, that needs review | 12:50 |
| cdent | edleafe: awesome thanks | 12:50 |
| cdent | edleafe: any ideas on the semantics riddle? | 12:50 |
| *** sudipto has quit IRC | 12:50 | |
| *** markvoelker has joined #openstack-nova | 12:50 | |
| edleafe | cdent: yes - 'set' is more correct | 12:50 |
| *** links has quit IRC | 12:51 | |
| edleafe | cdent: 'create' is simply 'create', wherease 'set' can also be 'update' | 12:51 |
| *** efried has joined #openstack-nova | 12:51 | |
| *** gouthamr has joined #openstack-nova | 12:51 | |
| cdent | edleafe: yeah, agree on that, but I meant the ordering of deleting compared to checking inventory | 12:51 |
| edleafe | cdent: oh, that | 12:52 |
| edleafe | cdent: I suppose it is better to delete first, then check for capacity, but I didn't want to refactor too much | 12:52 |
| *** vladikr has joined #openstack-nova | 12:53 | |
| *** tbachman has joined #openstack-nova | 12:53 | |
| edleafe | cdent: I can do that, too, and re-jigger the flow so it doesn't seem so hacky | 12:53 |
| cdent | is it right though? what's your justification for that ordering? | 12:53 |
| *** ducttape_ has joined #openstack-nova | 12:53 | |
| edleafe | cdent: the healing case | 12:54 |
| cdent | when we change an allocation do we need more space to accomodate both, or is it actually an atomic replace | 12:54 |
| cdent | but what about the resize case? | 12:54 |
| edleafe | resize to the same RP? | 12:54 |
| cdent | yeah | 12:54 |
| cdent | this is something that alaski had concerns about | 12:54 |
| * alaski peeks in | 12:55 | |
| edleafe | ah, since both the old and new would have the same consumer_id | 12:55 |
| sdague | cdent: yeh, I think we did punt on that | 12:55 |
| cdent | alaski your help would be appreciated on deciding something on : https://review.openstack.org/#/c/363209/ | 12:56 |
| pkoniszewski | tdurakov: hello | 12:56 |
| alaski | I'd go with blue | 12:56 |
| *** houming has quit IRC | 12:56 | |
| cdent | blue's such a conservative choice though. I was thinking either a matte black or an opalescent pearl | 12:56 |
| edleafe | cdent: wouldn't that also be a problem when deleting the old, smaller vm? Calling delete_all with that consumer_id would wipe the bigger vm's allocations out, too | 12:57 |
| alaski | hmm, change it to glossy black and we have a deal | 12:57 |
| edleafe | cdent: would we even have a way to distinguish the two? | 12:57 |
| cdent | edleafe: I don't know enough about the use case so I'm hoping when alaski gets done choosing his dress he can have a look | 12:57 |
| mdbooth | sdague: So, that cryptography bump fails in the gate: http://logs.openstack.org/49/364249/1/check/gate-tempest-dsvm-neutron-full-ubuntu-trusty/1c4b8f0/logs/devstacklog.txt.gz | 12:57 |
| tdurakov | pkoniszewski: I will be on vacation next 2 weeks till Sept, 16, so just in case PaulMurray be away too, could you please chair live-migration meeting? | 12:57 |
| cdent | because it was his comments a few weeks ago that make me even think of it | 12:57 |
| pkoniszewski | tdurakov: sure | 12:58 |
| alaski | I'm trying to find the relevant info. comments on PS 4? | 12:58 |
| cdent | let me pinpint alask | 12:58 |
| cdent | i | 12:58 |
| tdurakov | pkoniszewski: thanks a lot:) | 12:58 |
| edleafe | cdent: hmmm... well, I've only had my first pot of coffee, so I'll make some more and think about it | 12:58 |
| mdbooth | That's kinda awkward, though, because now we've got a situation where you need a different constraint depending on whether you patched openssl | 12:58 |
| sdague | mdbooth: right, agreed. Is there a cffi bump you need as well? | 12:58 |
| cdent | alaski: patchset 5, line 791 | 12:59 |
| cdent | https://review.openstack.org/#/c/363209/5/nova/objects/resource_provider.py@791 | 12:59 |
| cdent | edleafe: sound advice, me too | 12:59 |
| sdague | mdbooth: or are you saying that the fix in cryptography now makes it only work with new openssl? | 12:59 |
| tdurakov | sdague: have you figured out root cause for multinode? | 12:59 |
| sdague | tdurakov: yes | 12:59 |
| sdague | tdurakov: osc bug | 13:00 |
| *** cdelatte has joined #openstack-nova | 13:00 | |
| tdurakov | osc? | 13:00 |
| mdbooth | Actually, the error looks decidedly different. | 13:00 |
| sdague | where it hangs forever on a password prompt | 13:00 |
| mdbooth | AttributeError: 'CompiledFFI' object has no attribute 'def_extern' | 13:00 |
| * mdbooth would have expected another error about differing function signatures | 13:00 | |
| tdurakov | sdague: is it fixed already? | 13:01 |
| sdague | it's in the gate | 13:01 |
| mdbooth | Perhaps it's something else as you suggest. Will investigate more. | 13:01 |
| *** mamandle has joined #openstack-nova | 13:01 | |
| tdurakov | ok | 13:01 |
| sdague | don't recheck yet | 13:01 |
| sdague | mdbooth: yeh, I don't know, maybe stand up a liberty devstack and play with it to figure out if there is a requirements shuffle to get us past that | 13:01 |
| *** efried has quit IRC | 13:02 | |
| alaski | cdent: so in a resize scenario, which is the main trigger for an update afaik, there is a period where both allocations are needed. I'll have to check the data model but if it's a single entry then a resize will need to update to a size that accomadates old and new, then update down to just new after the resize is confirmed | 13:02 |
| openstackgerrit | JiWei proposed openstack/nova: 'unmap' is written without considering the configuration https://review.openstack.org/364122 | 13:02 |
| *** andreykurilin has left #openstack-nova | 13:02 | |
| alaski | cdent: so I think what's there is correct | 13:03 |
| *** efried has joined #openstack-nova | 13:03 | |
| *** andreykurilin has joined #openstack-nova | 13:03 | |
| *** rlrossit has joined #openstack-nova | 13:03 | |
| *** ekuris has quit IRC | 13:05 | |
| *** salv-orlando has joined #openstack-nova | 13:07 | |
| *** markvoelker has quit IRC | 13:07 | |
| *** ducttape_ has quit IRC | 13:07 | |
| bauzas | alaski: so, AFAIK, we claim differently from a resize | 13:09 |
| bauzas | for* | 13:09 |
| bauzas | alaski: that's what the scheduler gets | 13:09 |
| cdent | thanks alaski | 13:11 |
| *** johnthetubaguy_w has joined #openstack-nova | 13:11 | |
| * johnthetubaguy_w shakes fist at laptop that will not charge up any more | 13:12 | |
| * johnthetubaguy_w shakes fist at big purple windows laptop | 13:12 | |
| tdurakov | pkoniszewski: I'm not sure it's the right way to test upgrades https://review.openstack.org/#/c/364162/2/jenkins/jobs/devstack-gate.yaml | 13:13 |
| edleafe | johnthetubaguy: shake something heavier, like a bat, at it :) | 13:13 |
| *** salv-orlando has quit IRC | 13:13 | |
| *** prateek has joined #openstack-nova | 13:13 | |
| bauzas | alaski: cdent: yeah, so we double-count the resources indeed https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L277-L279 | 13:13 |
| johnthetubaguy_w | edleafe: yeah, after I try harder to fix it | 13:13 |
| edleafe | alaski: thanks, that makes things clearer. A resize will claim the sum of old and new. | 13:14 |
| tdurakov | pkoniszewski: from what I see upgrades should be verified by separate job, and covering them in dsvm is not good idea | 13:14 |
| pkoniszewski | tdurakov: at this point i only wanted to turn on grenade | 13:14 |
| pkoniszewski | tdurakov: but it did not work | 13:14 |
| bauzas | edleafe: context for the review ? | 13:14 |
| *** akshai has joined #openstack-nova | 13:14 | |
| alaski | edleafe: cdent: so digesting further, now I'm second guessing. What's in the patch doesn't double count, it just updates. there's a period where double counting is needed. | 13:14 |
| edleafe | So the steps should be: | 13:14 |
| bauzas | edleafe: see my link above | 13:14 |
| edleafe | 1) delete allocs for the rp/consumer | 13:14 |
| edleafe | 2) check space for the new allocs | 13:14 |
| edleafe | 3) write those new allocs | 13:14 |
| *** amotoki_ has quit IRC | 13:14 | |
| edleafe | 4) check generation | 13:15 |
| cdent | alaski: I was awash in a see of second guessing when i left that comment, and gave up for people who had some read context. | 13:15 |
| *** su_zhang has joined #openstack-nova | 13:15 | |
| edleafe | 5) roll it all back if anything isn't good | 13:15 |
| *** tanvirt has joined #openstack-nova | 13:16 | |
| alaski | edleafe: cdent I think the confusion here is that this only operates on one consumer. The simplification would be to have the second copy of the instance that exists during the resize be a new consumer | 13:16 |
| alaski | because really it's a second allocation, and then a deallocation | 13:16 |
| *** paul-carlton2 has quit IRC | 13:16 | |
| cdent | so maybe the issue is this: | 13:16 |
| *** gongysh has joined #openstack-nova | 13:16 | |
| edleafe | alaski: but conceptually, it's the same consumer; it's just temporarily taking up twice the space | 13:17 |
| cdent | we're trying to cover the resize and heal use cases in the same way, but they are different and the calling code needs to be more clear about what it is doing | 13:17 |
| alaski | edleafe: yeah. really depends on how you define the consumer. disk space usage, or instance using it | 13:17 |
| bauzas | cdent: edleafe: alaski: so here I think we're mixing RT concepts and Resource Providers | 13:18 |
| bauzas | cdent: edleafe: alaski: as step #1, I was thinking we would only provide to to RPs the exact semantics of what we update in DB | 13:18 |
| bauzas | without changing how the RT tracks | 13:18 |
| edleafe | alaski: the consumer is the vm | 13:19 |
| alaski | cdent: yeah. the calling code needs to handle heal and resize differently | 13:19 |
| edleafe | alaski: the usage is the allocation | 13:19 |
| bauzas | mid-term, I see a possible improvement of us getting rid of the RT and providing a better interface for allocations | 13:19 |
| cdent | bauzas: sure, but the question still remains in the current code: does it, and if so how and in what ordering, remove a pre-existing allocation that matches rp_uuid, consumer_id? | 13:19 |
| edleafe | cdent: we could have an 'add' and an 'update' call | 13:19 |
| *** yamahata has joined #openstack-nova | 13:20 | |
| cdent | the original data model was that we would only ever create and delete | 13:20 |
| cdent | but then discussion made it clear that would make it hard to heal | 13:20 |
| bauzas | cdent: the current code does a full update | 13:20 |
| alaski | edleafe: sure. I could see changing the model so the consumer is a disk that needs to be written, detached from the instance. But that's a distraction here. Just might simplify things | 13:20 |
| cdent | bauzas: yes, it does, I wrote the original version, but is it doing it correctly? | 13:21 |
| edleafe | I see the complicating issue as the eventual delete of the resized instance. | 13:21 |
| bauzas | cdent: meaning https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L687 | 13:21 |
| *** Hosam_ has quit IRC | 13:21 | |
| * cdent has too many threads, will pause | 13:21 | |
| edleafe | How do you tell which of the allocations to delete if they are the same RP and the same consumer_id? | 13:21 |
| *** paul-carlton2 has joined #openstack-nova | 13:21 | |
| alaski | edleafe: that can still be an update. but whomever is doing the update will need to update to old + new instance during the resize, then update to new instance after confirmation | 13:22 |
| * cdent draws S Q L on the ground and summons jaypipes | 13:23 | |
| *** markvoelker has joined #openstack-nova | 13:23 | |
| edleafe | alaski: as long as the resize/mig always reports the full amount that is being used, we should be goog | 13:24 |
| edleafe | good, even | 13:24 |
| alaski | edleafe: remove the allocation that matches the resources you no longer consume | 13:24 |
| alaski | if it's a separate entry | 13:24 |
| edleafe | alaski: that would be difficult and not always possible | 13:24 |
| edleafe | since it's keyed by RP and consumer_id | 13:25 |
| edleafe | and the amounts could be identical | 13:25 |
| alaski | right, but in that case does it matter which is removed? | 13:25 |
| johnthetubaguy_w | isn't there normally a source and destination resource provider, so we know which one? | 13:26 |
| johnthetubaguy_w | I guess you can resize to the same host | 13:26 |
| alaski | johnthetubaguy_w: this is in the case where source/dest are the same | 13:26 |
| alaski | yeah | 13:26 |
| johnthetubaguy_w | in which case, meh, as long as the values are all good at the end, so +1 alaski I think | 13:26 |
| edleafe | alaski: but a 'delete where rpid=x and consumer_id=y and used=N' will delete both | 13:27 |
| alaski | edleafe: limit 1? :) | 13:27 |
| edleafe | alaski: heh | 13:27 |
| alaski | it is an odd model | 13:28 |
| *** markvoelker has quit IRC | 13:28 | |
| johnthetubaguy_w | do we not have a constraint to stop that kind of duplicate? | 13:28 |
| alaski | if only we had some sort of unique id to use :) | 13:28 |
| alaski | I'm also not in the trenches like y'all are, so I may be missing something here | 13:28 |
| *** eharney has joined #openstack-nova | 13:29 | |
| edleafe | alaski: we have several. Looks like that's not enough :) | 13:29 |
| johnthetubaguy_w | the migration object could store the allocation id, if we hand that out? | 13:29 |
| tdurakov | pkoniszewski: there was an issues with multinode jobs, so I think you could recheck once it's fixed | 13:30 |
| alaski | johnthetubaguy_w: if there were a uuid, that would be a good way to handle it | 13:30 |
| alaski | edleafe: my experience with the Nova db is wherever we don't have a uuid we eventually regret it, or I just get frustrated by it | 13:30 |
| edleafe | alaski: it's an int | 13:31 |
| edleafe | alaski: because speed | 13:31 |
| pkoniszewski | tdurakov: yeah, i see, will give it some more time and it should work, thanks for information | 13:31 |
| *** amotoki has joined #openstack-nova | 13:31 | |
| edleafe | alaski: I'm with ya, brother. I proposed converting all int IDs to uuid back in Bexar | 13:32 |
| alaski | edleafe: yeah. I wouldn't want to use the id. That's burned me in the past as well, and I think jaypipes would explode | 13:32 |
| tdurakov | pkoniszewski: no worries, but still consider to test live-migration in separate job, it would be clearer. | 13:32 |
| tdurakov | p.s. good idea) | 13:32 |
| *** ducttape_ has joined #openstack-nova | 13:32 | |
| edleafe | alaski: exploding pipes is never good | 13:32 |
| *** pwnall has joined #openstack-nova | 13:33 | |
| cdent | the permissions argument on using ints is pretty bogus for a database as small as anything nova will ever have | 13:34 |
| pkoniszewski | tdurakov: yeah, you are right, i just want to learn how all this stuff work | 13:34 |
| pkoniszewski | tdurakov: as i have no idea right now :) | 13:34 |
| cdent | especially if you tune your database setting approrpirately. but that ship has sailed | 13:34 |
| cdent | johnthetubaguy_w: I do wonder if there is a missing constraint on the table | 13:35 |
| cdent | I commented elsewhere that it seemed like there should a unique constraint on rp_uuid, consumer_id | 13:35 |
| cdent | which would make duplicates (the problem that led to this weird when to delete issue) not allowed | 13:35 |
| cdent | edleafe: Were you able to extract an action out of all that? I think I got lost. | 13:36 |
| alaski | that would avoid much of this messiness. and then it's an issue of figuring out what to update to during a point when the consumer has duplicate allocations | 13:36 |
| *** mamandle has quit IRC | 13:37 | |
| edleafe | cdent: I thought it was still up in the air, depending on whether resize/migration always reports the full usage of both original and target | 13:37 |
| edleafe | cdent: of course, that's only important in the same-host situation | 13:38 |
| *** diogogmt has quit IRC | 13:39 | |
| cdent | Right now the thing that matters is avoiding a situation where reality is reflected in the artifact world in a way that will break future attempts to allocate | 13:39 |
| johnthetubaguy_w | cdent: yeah, same feeling | 13:39 |
| johnthetubaguy_w | edleafe: the better approach appears to be int id and uuid, but maybe thats what you meant | 13:39 |
| *** pwnall has quit IRC | 13:39 | |
| edleafe | johnthetubaguy_w: no, I really don't see the benefit of int except that it's smaller and faster | 13:40 |
| edleafe | johnthetubaguy_w: neither of which is critical on our scale | 13:40 |
| edleafe | johnthetubaguy_w: and both negates the 'smaller' argument | 13:40 |
| openstackgerrit | Paulo Matias proposed openstack/nova: Fill destination check data with VNC/SPICE listen addresses https://review.openstack.org/358599 | 13:42 |
| edleafe | cdent: so unless we come up with a reason it won't work in the next few minutes, I'll start changing the code to match my numbered steps from about 30 min ago | 13:43 |
| *** amotoki has quit IRC | 13:43 | |
| *** ayoung has joined #openstack-nova | 13:43 | |
| cdent | edleafe: seems sane | 13:43 |
| *** ducttape_ has quit IRC | 13:44 | |
| cdent | We need some kind of poster we can share around that has a flowchart on it that says "how to reason about <various hard stuff>" | 13:45 |
| *** diogogmt has joined #openstack-nova | 13:45 | |
| alaski | edleafe: assuming "new allocs" handles the double accounting when necessary that seems fine | 13:46 |
| johnthetubaguy_w | edleafe: both is a referece to the smaller index, apparently giving better performance than having relationishps that use long utf8 strings | 13:46 |
| *** mriedem has joined #openstack-nova | 13:46 | |
| mriedem | what's the good news? | 13:47 |
| mriedem | cellsv2 and placement stuff is all merged? | 13:47 |
| * cdent gives mriedem a comfortable chair | 13:47 | |
| dansmith | mriedem: I don't see anything from jay on the client bit | 13:47 |
| alaski | cells stuff all passed. just writing some tests for my fix from yesterday | 13:48 |
| mriedem | ok, looks like at least the cells v2 devstack and grenade changes are passing ci | 13:48 |
| mriedem | https://review.openstack.org/#/c/363971/ is clean | 13:48 |
| sdague | mriedem: a bunch of stuff bounced off the gate because of a multinode osc issue | 13:48 |
| *** cleong has joined #openstack-nova | 13:48 | |
| *** efried has quit IRC | 13:48 | |
| sdague | the fix for that is in the gate, but check queue for nova is going to be bonkers until that is in | 13:48 |
| *** openstackgerrit has quit IRC | 13:49 | |
| mriedem | ok | 13:49 |
| mriedem | sdague: dansmith: we should get this in when we can https://review.openstack.org/#/c/361436/ | 13:49 |
| mriedem | that's holding up a nova and devstack change for cellsv2 | 13:49 |
| *** openstackgerrit has joined #openstack-nova | 13:49 | |
| dansmith | mriedem: don't we not need the reordering now because alaski fixed the setup command to sync? | 13:50 |
| dansmith | we still need the config guard but not the reorder I think | 13:50 |
| *** amotoki has joined #openstack-nova | 13:50 | |
| *** diogogmt has quit IRC | 13:50 | |
| mriedem | fewer double negatives please | 13:50 |
| alaski | dansmith: right | 13:50 |
| dansmith | alaski: I will fix | 13:50 |
| dansmith | mriedem: sorry | 13:50 |
| mriedem | https://review.openstack.org/#/c/363956/ is dependent on that grenade reordering change | 13:51 |
| alaski | the reorder won't hurt anything though | 13:51 |
| mriedem | it's really dependent on the conditional | 13:51 |
| dansmith | alaski: yeah, just seems wrong | 13:51 |
| *** prateek_ has joined #openstack-nova | 13:51 | |
| sdague | mriedem: you are going to need to draw me a map for the devstack / grenade /nova cells changes, because I understood the order yesterday, and now I'm not sure that I do | 13:51 |
| dansmith | I'll leave it to sdague as to whether we undo the reorder | 13:51 |
| mriedem | i don't care if we undo the re-order, but we'd probably want to run that through check again if we do | 13:51 |
| *** efried has joined #openstack-nova | 13:51 | |
| alaski | dansmith: agreed. not doing the reordering is probably better | 13:51 |
| *** prateek has quit IRC | 13:52 | |
| mriedem | sdague: the treasure hunt starts here https://review.openstack.org/#/c/355602/ | 13:52 |
| mriedem | that's devstack | 13:52 |
| mriedem | depends on nova, which depends on grenade | 13:52 |
| mriedem | and then X marks the cellsv2 treasure | 13:52 |
| sdague | mriedem: ok, I need to go make breakfast, can you just write the map in an etherpad and I'll look when I come back | 13:52 |
| *** tblakes has joined #openstack-nova | 13:52 | |
| mriedem | yeah, thought about starting an etherpad for the cells things anyway | 13:52 |
| mriedem | https://etherpad.openstack.org/p/nova-cellsv2-newton-ff | 13:53 |
| dansmith | mriedem: I just pushed an edit to the grenade patch and rechecked 956 | 13:54 |
| *** Guest77956 is now known as jlvillal | 13:55 | |
| *** rodrigods has quit IRC | 13:55 | |
| *** rodrigods has joined #openstack-nova | 13:55 | |
| *** pwnall has joined #openstack-nova | 13:56 | |
| *** xyang1 has joined #openstack-nova | 13:56 | |
| mriedem | ok i'll run the d-g test change again | 13:56 |
| *** ducttape_ has joined #openstack-nova | 13:57 | |
| mriedem | has anyone talked to jaypipes? | 13:57 |
| *** eharney has quit IRC | 13:58 | |
| dansmith | not I | 13:58 |
| *** salv-orlando has joined #openstack-nova | 13:59 | |
| edleafe | mriedem: jay was around late last night. He said he would work on the client stuff in the (his) morning. | 13:59 |
| edleafe | mriedem: but so far nothing today | 13:59 |
| *** amotoki has quit IRC | 13:59 | |
| *** eharney has joined #openstack-nova | 14:00 | |
| *** mgoddard_ has joined #openstack-nova | 14:00 | |
| *** zz_dimtruck is now known as dimtruck | 14:00 | |
| *** kevinz has quit IRC | 14:00 | |
| *** anush__ has joined #openstack-nova | 14:00 | |
| *** su_zhang has quit IRC | 14:01 | |
| *** su_zhang has joined #openstack-nova | 14:01 | |
| *** hemanthm|away is now known as hemanthm | 14:01 | |
| *** jheroux has joined #openstack-nova | 14:02 | |
| openstackgerrit | Sergey Nikitin proposed openstack/nova-specs: Enable to share PCI devices between numa nodes https://review.openstack.org/361140 | 14:02 |
| *** parora has joined #openstack-nova | 14:02 | |
| dansmith | alaski: you're doing tests for this? https://review.openstack.org/#/c/364005/1/nova/conductor/manager.py | 14:03 |
| *** mgoddard has quit IRC | 14:03 | |
| mriedem | alaski: adding tests for this? https://review.openstack.org/#/c/364005/1 | 14:03 |
| *** oanson has joined #openstack-nova | 14:03 | |
| mriedem | ha | 14:03 |
| alaski | yep | 14:03 |
| mriedem | *all eyes are now on laski* | 14:03 |
| * alaski moves to the back of the classroom | 14:03 | |
| *** jaosorior has quit IRC | 14:04 | |
| johnthetubaguy_w | mriedem: I just added a -1 for that | 14:05 |
| *** prateek_ has quit IRC | 14:05 | |
| alaski | dansmith: I know you have concerns about how to handle object equivalence, but it would make testing easier at times to get that sorted | 14:05 |
| dansmith | alaski: we have a helper to compare two objects by primitive-izing them | 14:06 |
| *** prateek_ has joined #openstack-nova | 14:06 | |
| dansmith | does that help? | 14:06 |
| *** vishwanathj has joined #openstack-nova | 14:06 | |
| mriedem | unrelated, novaclient 6.0.0 is now in upper-constraints for newton \o/ | 14:06 |
| alaski | not in this case, it's the assert_called_with case | 14:06 |
| *** su_zhang has quit IRC | 14:06 | |
| alaski | so I just end up with MatchType() | 14:06 |
| dansmith | alaski: yeah, well, sorry | 14:06 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Fix service version lookups https://review.openstack.org/363888 | 14:06 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Pull from cell0 and build_requests for instance list https://review.openstack.org/356138 | 14:06 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Delete BuildRequest regardless of service_version https://review.openstack.org/363911 | 14:06 |
| openstackgerrit | Andrew Laski proposed openstack/nova: Remove BuildRequest when scheduling fails https://review.openstack.org/364005 | 14:06 |
| *** pilgrimstack has quit IRC | 14:07 | |
| alaski | dansmith: I mean, I don't blame you. just mentioning that it would help to figure something out | 14:07 |
| dansmith | yeah | 14:07 |
| alaski | okay, basic testing added | 14:07 |
| *** parora has quit IRC | 14:09 | |
| *** ducttape_ has quit IRC | 14:09 | |
| *** claudiub has quit IRC | 14:10 | |
| mriedem | dansmith: https://review.openstack.org/#/c/361436/11/projects/60_nova/upgrade.sh | 14:12 |
| openstackgerrit | Christopher Martin proposed openstack/nova: Fix Boot From Volume size error https://review.openstack.org/364337 | 14:12 |
| mriedem | let's remove the additional comment | 14:12 |
| dansmith | gah | 14:12 |
| mriedem | the commit message too | 14:12 |
| mriedem | this basically just becomes, make cellv2 setup conditional | 14:13 |
| sdague | mriedem: so you -1ed the bottom of that whole stack | 14:13 |
| *** parora has joined #openstack-nova | 14:13 | |
| sdague | also, the recheck will fail | 14:13 |
| mriedem | yes i know | 14:13 |
| *** efried has quit IRC | 14:13 | |
| dansmith | mriedem: done | 14:13 |
| mriedem | it's like 10 minutes old | 14:14 |
| sdague | mriedem: ok | 14:14 |
| avolkov | alaski: hi, what do you think about https://review.openstack.org/#/c/301864/? It's, possibly, last thing in https://review.openstack.org/#/q/topic:bp/cells-cell0 now. | 14:14 |
| *** efried has joined #openstack-nova | 14:14 | |
| *** amotoki has joined #openstack-nova | 14:14 | |
| sdague | well, I'm hard stop in 5.75 hours, and them I'm not back online until tuesday | 14:14 |
| sdague | to be clear what the merge window is here is for anything that needs my feedback | 14:15 |
| dansmith | sdague: you should give me temporary provisional devstack core status in case of emergency | 14:15 |
| mriedem | dansmith: alaski: on this grenade change, just so we're clear, running simple_cell_setup after db sync as it is today, just won't have the cell0 migrations run right? | 14:15 |
| *** pilgrimstack has joined #openstack-nova | 14:15 | |
| *** tongli has joined #openstack-nova | 14:15 | |
| mriedem | because alaski's change to run the migrations on cell0 in simple_cell_setup are dependent on this grenade change, so it's backward | 14:15 |
| dansmith | mriedem: no, it will because he aded it in the nova code | 14:15 |
| mriedem | dansmith: not in this order | 14:15 |
| alaski | avolkov: that would be nice to have, but given the pressure on merging other things right now may need to wait. However since it's only test changes I think we can look at that after FF | 14:15 |
| dansmith | mriedem: right, but we don't configure this to be on until after we merge the nova one | 14:15 |
| dansmith | mriedem: or.. shouldn't | 14:16 |
| mriedem | don't configure this to be on? as in flip that flag to run it? that's going into a job change | 14:16 |
| mriedem | i thought simple_cell_setup will just create the cell0 mapping, so cell0 db has to exist | 14:16 |
| mriedem | you should have run db sync before that to run the migrations, but not required, as in won't fail | 14:16 |
| *** prateek_ has quit IRC | 14:16 | |
| mriedem | won't fail simple_cell_setup i mean, but the actual code that needs to use cell0 might fail | 14:17 |
| *** prateek has joined #openstack-nova | 14:17 | |
| mriedem | until the nova change lands | 14:17 |
| dansmith | mriedem: I'm confused about what you're saying | 14:17 |
| mriedem | is it hangout time already? | 14:17 |
| alaski | yeah, I'm trying to parse all of that | 14:17 |
| dansmith | mriedem: we won't actually run cell_setup until after we land the job change to enable it | 14:17 |
| *** mingyu has joined #openstack-nova | 14:17 | |
| dansmith | we do that after we land the nova change that makes it work in this order | 14:17 |
| mriedem | true that is dependent on the devstack > nova > grenade order | 14:18 |
| *** cmartin has joined #openstack-nova | 14:18 | |
| *** parora has quit IRC | 14:19 | |
| bauzas | mriedem: dansmith: alaski: given what I've heard so far, it seems we've reached to a point where we're done with the placement bits for Newton, right? | 14:19 |
| alaski | mriedem: you're correct that without the reordering we need the nova patch to db sync | 14:19 |
| alaski | but the conditional will save us as dansmith says | 14:19 |
| dansmith | bauzas: eh? | 14:19 |
| bauzas | dansmith: mriedem: alaski: now, the last bits are about the cell0 stuff, right? | 14:19 |
| mriedem | dansmith: alaski: sdague: ok +1 on https://review.openstack.org/#/c/361436/ | 14:20 |
| dansmith | bauzas: no, we still have to land the allocation server and client bits | 14:20 |
| bauzas | dansmith: I'm a bit confused about what you folks agreed last night | 14:20 |
| alaski | bauzas: I believe people are typing as we speak | 14:20 |
| bauzas | dansmith: mmm, links ? | 14:20 |
| bauzas | dansmith: that sounds an hard task :/ | 14:20 |
| *** marst has quit IRC | 14:20 | |
| dansmith | bauzas: unfortunately we don't have a link for the client bit yet :( | 14:21 |
| bauzas | https://review.openstack.org/#/c/357993/ is the server side, right? | 14:21 |
| *** pwnall has quit IRC | 14:21 | |
| dansmith | bauzas: server side: https://review.openstack.org/#/c/339874/34 | 14:21 |
| bauzas | oops, bad copy/paste | 14:21 |
| *** coolsvap has quit IRC | 14:22 | |
| *** pwnall has joined #openstack-nova | 14:22 | |
| dansmith | bauzas: I don't understand your comment on alaski's bottom cells patch | 14:22 |
| bauzas | dansmith: you mean https://review.openstack.org/#/c/364005/2/nova/conductor/manager.py ? | 14:23 |
| dansmith | y | 14:23 |
| dansmith | eah | 14:23 |
| *** prateek_ has joined #openstack-nova | 14:23 | |
| *** rlrossit has quit IRC | 14:23 | |
| cdent | yeah, I struggled with that too | 14:23 |
| mriedem | bauzas: we're tracking things for cellsv2 in https://etherpad.openstack.org/p/nova-cellsv2-newton-ff fyi | 14:23 |
| *** liverpooler has quit IRC | 14:23 | |
| bauzas | dansmith: so, the logic is that we delete the BuildRequest once the instance found a place | 14:23 |
| dansmith | oh I think I misread the line number reference | 14:24 |
| bauzas | dansmith: the problem in that logic is that we're catching a global exception that prevents us to read that section | 14:24 |
| bauzas | because we return | 14:24 |
| *** mtanino has joined #openstack-nova | 14:24 | |
| bauzas | logically, I'd prefer to have a simplier workflow that would run that section anyway | 14:24 |
| *** praneshp has joined #openstack-nova | 14:24 | |
| openstackgerrit | Ed Leafe proposed openstack/nova: WIP create_all and delete_all for AllocationList https://review.openstack.org/363209 | 14:24 |
| alaski | right, but it has to return because the rest of that method shouldn't run | 14:24 |
| bauzas | I knoxw | 14:25 |
| edleafe | cdent: ^^ fixes the issue with only some RPs with inventory. | 14:25 |
| edleafe | cdent: working on the re-jiggering now | 14:25 |
| dansmith | yeah, I'm not sure how it could be different | 14:25 |
| cdent | awesome edleafe | 14:25 |
| alaski | bauzas: the buildrequest cleanup also can't be done after the method is run because it does a different thing for BuildRequestNotFound down below | 14:25 |
| bauzas | alaski: but I was thinking of deleting the object earlier, possibly in a finally statement of a contextmanager wrapping _schedule_instance() | 14:25 |
| alaski | notably the return before compute_rpcapi | 14:25 |
| bauzas | we 'd just reraise if NoValidHost | 14:26 |
| bauzas | but anyway, I'm nitpicking | 14:26 |
| alaski | It' can't be deleted until the instance mapping on line 516 if scheduling succeeds | 14:26 |
| *** prateek has quit IRC | 14:26 | |
| dansmith | I +2d | 14:27 |
| openstackgerrit | Sean Dague proposed openstack/nova: Report compute node inventories through placement https://review.openstack.org/363061 | 14:27 |
| dansmith | optimization aside, gotta get that in | 14:27 |
| sdague | that rebase was just to reset it in check | 14:27 |
| sdague | because it was in a recheck that was going to take another hour to timeout/fail | 14:28 |
| bauzas | alaski: well, that's because you wanted to detect that an API delete call was here | 14:28 |
| dansmith | sdague: thanks | 14:28 |
| bauzas | but okay, that becomes overcomplicated to think about an alternative and again, I don't want to nitpick | 14:28 |
| bauzas | alaski: I just held my +2 because I needed to think of any possible alternative but I struggled founding those | 14:29 |
| bauzas | alaski: so... | 14:29 |
| alaski | bauzas: okay. we can certainly look to clean it up in O if there's a better way | 14:29 |
| bauzas | alaski: it's not really about a matter of duplicating code, it's rather about taking a pause and seeing the workflow | 14:29 |
| bauzas | alaski: but I agree, that's Ocata | 14:30 |
| *** raj_singh_ has joined #openstack-nova | 14:31 | |
| sdague | johnthetubaguy_w: do you know where the demonstration patch for - https://review.openstack.org/#/c/356963/ is? | 14:32 |
| mriedem | cdent: i replied in https://review.openstack.org/#/c/364005/2//COMMIT_MSG | 14:33 |
| *** marst has joined #openstack-nova | 14:33 | |
| sdague | ok, check queue is hopefully good again | 14:33 |
| *** gongysh has quit IRC | 14:33 | |
| johnthetubaguy_w | sdague: ah... sorry, I am not sure :( he did say that was panned | 14:33 |
| *** rama_y has joined #openstack-nova | 14:33 | |
| *** dimtruck is now known as zz_dimtruck | 14:33 | |
| johnthetubaguy_w | planned | 14:33 |
| *** zz_dimtruck is now known as dimtruck | 14:33 | |
| cdent | thanks mriedem I figured it was something like that | 14:33 |
| *** jiwei has joined #openstack-nova | 14:34 | |
| *** eharney has quit IRC | 14:36 | |
| *** jheroux has quit IRC | 14:37 | |
| *** parora has joined #openstack-nova | 14:38 | |
| dansmith | bauzas: cdent: so on the placement stuff, we're waiting for jaypipes to hopefully pop up today with a working client-side allocation patch | 14:38 |
| bauzas | k | 14:38 |
| dansmith | that may not happen though, so I think maybe we should start throwing noodles at the wall | 14:38 |
| cdent | dansmith: that's my understanding, yeah | 14:38 |
| bauzas | dansmith: I appreciate the analogism | 14:38 |
| dansmith | the only reason I haven't started is that I'm not sure how and where to report these things from RT | 14:39 |
| dansmith | specifically: | 14:39 |
| dansmith | I think we can probably report from somewhere in/around the claim process that an instance is using a thing | 14:39 |
| bauzas | dansmith: well, the main problem I see with implementating that is that we do full updates every time | 14:39 |
| dansmith | however that will fail for single-node migrate, | 14:39 |
| dansmith | and will be wrong or have to be updated for resize | 14:39 |
| *** ducttape_ has joined #openstack-nova | 14:40 | |
| bauzas | dansmith: which is kinda different from the idea of putting a specific allocation against a specific resource | 14:40 |
| dansmith | bauzas: I'm not sure what you mean by full updates | 14:40 |
| mriedem | bauzas: i see what you were saying in https://review.openstack.org/#/c/364005/ i left some comments too | 14:40 |
| mriedem | i think pulling some of the chunks out of that method into smaller private methods might help with the confusion and maintaintability/readability - in ocata | 14:40 |
| alaski | dansmith: I'm +2 on https://review.openstack.org/#/c/326906 | 14:40 |
| dansmith | alaski: cool I'll look over again in a few | 14:41 |
| *** prateek_ has quit IRC | 14:41 | |
| bauzas | mriedem: yeah, I was more thinking of a contextmanager wrapping _schedule_instances() but whatever :) | 14:41 |
| *** oanson has quit IRC | 14:41 | |
| mriedem | bauzas: that too sure | 14:41 |
| mriedem | i just get lost in the returns in the middle of the function | 14:41 |
| bauzas | dansmith: sorry, I mean https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L168 | 14:42 |
| *** jheroux has joined #openstack-nova | 14:42 | |
| bauzas | dansmith: which is just us asking to update the whole ComputeNode object with all its fields | 14:42 |
| cdent | dansmith: my understanding of the resource tracker is very weak, I keep defering my "grok resource tracker" to do item, so I can't be much help. Did you see the earlier conversation between edleafe, alaski etc about dealing with single-node resize? | 14:42 |
| *** dimtruck is now known as zz_dimtruck | 14:43 | |
| bauzas | cdent: so RT is mapped to a ComputeNode | 14:43 |
| dansmith | cdent: no | 14:43 |
| dansmith | bauzas: that's where you think we need to report the per-instance stuff? | 14:43 |
| bauzas | dansmith: so, that's where the call is made to the scheduler client | 14:44 |
| bauzas | dansmith: I mean in _update() | 14:44 |
| dansmith | bauzas: for compute node stuff (i.e. inventory) | 14:44 |
| mriedem | alaski: was there additional testing needed in https://review.openstack.org/#/c/363888/ based on what melwitt and you were talking about last night? | 14:44 |
| bauzas | dansmith: that private helper method can be called either from any claim method (for a boot or a move) or periodically | 14:45 |
| *** prateek_ has joined #openstack-nova | 14:45 | |
| mriedem | compared to PS3? | 14:45 |
| cdent | dansmith: near here in the logs http://p.anticdent.org/2lOO | 14:45 |
| bauzas | dansmith: so, say we'd have self.compute containing the values, we would need to capture what actually changed | 14:45 |
| dansmith | bauzas: right, but we could put the actual per-instance reporting closer to the claim and not bundle it in the bulk update_all_the_things() method | 14:45 |
| alaski | mriedem: the additional testing was for the patch I slipped in below that one | 14:45 |
| mriedem | alaski: ah ok, cool | 14:45 |
| mriedem | sneaky sis | 14:45 |
| bauzas | dansmith: right, but that's a pure mess of shit | 14:45 |
| bauzas | dansmith: like, we have a specific helper for migration that does call it | 14:46 |
| dansmith | bauzas: it's kinda the design of the thing right? since we can | 14:46 |
| bauzas | plus, we have an update_usage() that does the calculation and call it | 14:46 |
| dansmith | bauzas: since we can't update allocations [ :~( ] we kinda have to only do it when we have a new instance or something | 14:46 |
| bauzas | dansmith: yeah, so just to explain my worries, it's just the fact that we have a couple of helpers that do calculate the new amounts and modify the object, but you only have one single helper for updating the Computenode object | 14:47 |
| alaski | dansmith: updaing allocations was what we were discussing earlier | 14:47 |
| bauzas | dansmith: so, it would mean that we would need to hook up the allocation call *before* the call to _update() | 14:47 |
| *** tonytan4ever has joined #openstack-nova | 14:47 | |
| dansmith | bauzas: but we don't have to do any calculations to write an instance allocation | 14:47 |
| alaski | I lost the link but I think there was a patch up for that | 14:47 |
| dansmith | alaski: we definitely need that long-term, I'm just not sure it's something we're going to be able to merge and I'm trying to think of the minimum possible code to make something work | 14:48 |
| bauzas | dansmith: ie. not using the existing scheduler client Facade for transforming our Computenode object into a list of allocations | 14:48 |
| bauzas | dansmith: ie. we can't use* | 14:48 |
| bauzas | not sure I'm clear :/ | 14:48 |
| *** parora has quit IRC | 14:49 | |
| bauzas | gosh I hate me | 14:49 |
| *** erhudy has joined #openstack-nova | 14:49 | |
| dansmith | I don't think we have any existing facade for doing anything like this right now so I'm not sure what your point is | 14:49 |
| mriedem | do we need a hangout? | 14:49 |
| bauzas | dansmith: that's my point | 14:49 |
| edleafe | bauzas: allocations are for the instances on the compute node. | 14:50 |
| bauzas | dansmith: the fact that we can't just modify the scheduler client code to transform a request into a list of allocations | 14:50 |
| bauzas | edleafe: I know | 14:50 |
| bauzas | mriedem: well, your call | 14:50 |
| dansmith | bauzas: there is nothing in scheduler client to transform, AFAIK | 14:50 |
| alaski | dansmith: for resize we could delete/create for now. it sucks but would keep us closer to reality than ignoring it completely | 14:51 |
| bauzas | that's my point | 14:51 |
| bauzas | dansmith: ^ | 14:51 |
| edleafe | bauzas: so transforming a compute node to a list of allocations is iterating over its instances | 14:51 |
| *** diogogmt has joined #openstack-nova | 14:51 | |
| dansmith | alaski: that's really dangerous | 14:51 |
| dansmith | alaski: we race with some new schedule and then we can't re-create our allocation | 14:51 |
| bauzas | edleafe: I'd rather prefer to not do that | 14:51 |
| bauzas | edleafe: but rather create allocations in claims | 14:51 |
| *** haplo37__ has joined #openstack-nova | 14:52 | |
| dansmith | edleafe: one option I see is to just go over all the instances, check their allocations and update/delete/create as needed | 14:52 |
| edleafe | bauzas: sure, that's for new instances | 14:52 |
| edleafe | we were considering the "healing" case | 14:52 |
| dansmith | edleafe: the other is to just create on claim and delete on unclaim (or whatever) | 14:52 |
| *** kevinz has joined #openstack-nova | 14:52 | |
| alaski | dansmith: true. so only doing it on claim means we may have more allocated than the db indicates, but compute node resource tracking will deal with that | 14:52 |
| bauzas | dansmith: edleafe: keep in mind we don't claim for live migrations | 14:52 |
| dansmith | edleafe: well, I'm considering both | 14:52 |
| edleafe | where a compute node just wants to make sure everything is correct | 14:52 |
| dansmith | alaski: but the server side will not let us double allocate | 14:52 |
| bauzas | dansmith: edleafe: so the only way for the scheduler to notice some migration change is by getting the periodic update | 14:52 |
| dansmith | bauzas: we don't care about the scheduler | 14:53 |
| dansmith | at all | 14:53 |
| edleafe | dansmith: the new stuff already worked. It was the healing case that we needed to accomodate | 14:53 |
| dansmith | all we're trying to do is get the allocations written to the db | 14:53 |
| dansmith | edleafe: eh? | 14:53 |
| bauzas | dansmith: we care about providing new allocations when a new move or boot operation is claimed, right? | 14:53 |
| edleafe | dansmith: jay's sql stuff worked for new allocations only | 14:53 |
| dansmith | yeah | 14:53 |
| dansmith | edleafe: you mean the server side | 14:53 |
| dansmith | edleafe: we have nothing for the RT/client side yet | 14:53 |
| bauzas | dansmith: so, here is the thing, if we don't claim when live-migrating, then you can't write allocations for that specific move operation | 14:54 |
| edleafe | dansmith: ah, I didn't realize you were talking client | 14:54 |
| *** Sukhdev has joined #openstack-nova | 14:54 | |
| dansmith | edleafe: I'm trying to figure out what we can get done from the client with just what we have on the server side right now | 14:54 |
| dansmith | edleafe: yep | 14:54 |
| bauzas | mriedem: I retain your idea of an hangout | 14:54 |
| dansmith | edleafe: trying to figure out what we can get done in case jay doesn't show up with the client patch | 14:54 |
| *** rrecio has joined #openstack-nova | 14:55 | |
| edleafe | dansmith: as long as we always pass the full resource usage for each consumer, it should work | 14:55 |
| mriedem | https://hangouts.google.com/call/lsmbe6rtofbrhm6cixsh7h4smye | 14:55 |
| edleafe | dansmith: things like migrations sound icky though | 14:55 |
| dansmith | migrations are the hard part | 14:55 |
| dansmith | because they will fail on single node | 14:55 |
| mriedem | i can't hear anything | 14:56 |
| mriedem | oh there we go | 14:57 |
| *** krtaylor has joined #openstack-nova | 14:58 | |
| *** kevinz has quit IRC | 14:58 | |
| *** kevinz has joined #openstack-nova | 14:58 | |
| *** TravT_ is now known as TravT | 14:58 | |
| *** mjura has quit IRC | 15:01 | |
| mriedem | at some point today i need to re-read this http://specs.openstack.org/openstack/nova-specs/specs/newton/approved/resource-providers-allocations.html | 15:01 |
| cdent | dansmith: https://review.openstack.org/363209 | 15:02 |
| *** pcaruana has quit IRC | 15:03 | |
| *** eharney has joined #openstack-nova | 15:03 | |
| openstackgerrit | Andrew Laski proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379 | 15:03 |
| *** annegentle has joined #openstack-nova | 15:04 | |
| *** hshiina has joined #openstack-nova | 15:05 | |
| *** irenab has quit IRC | 15:06 | |
| *** kolesovdv has quit IRC | 15:06 | |
| *** roeyc has quit IRC | 15:06 | |
| *** rcernin has quit IRC | 15:06 | |
| *** bvanhav has quit IRC | 15:08 | |
| *** parora has joined #openstack-nova | 15:08 | |
| *** mgoddard_ has quit IRC | 15:08 | |
| *** mgoddard has joined #openstack-nova | 15:09 | |
| *** jiwei has quit IRC | 15:09 | |
| *** zz_dimtruck is now known as dimtruck | 15:11 | |
| *** prateek_ has quit IRC | 15:11 | |
| *** nkrinner is now known as nkrinner_afk | 15:11 | |
| openstackgerrit | Andrew Laski proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379 | 15:12 |
| *** hemnafk is now known as hemna | 15:15 | |
| *** bvanhav has joined #openstack-nova | 15:17 | |
| *** ljxiash has quit IRC | 15:17 | |
| *** moshele has quit IRC | 15:17 | |
| *** rmart04 has quit IRC | 15:19 | |
| *** mvk has joined #openstack-nova | 15:19 | |
| mdbooth | sdague: It needed a cffi bump too, as you guessed: https://review.openstack.org/#/c/364249/ | 15:20 |
| *** browne has joined #openstack-nova | 15:20 | |
| sdague | mdbooth: cool, that seems to work? | 15:20 |
| mdbooth | The tests pass | 15:21 |
| sdague | great | 15:21 |
| sdague | what are the mitaka values? | 15:21 |
| sdague | is there a similar issue there? | 15:21 |
| * mdbooth checks | 15:21 | |
| sdague | so, on mitaka we have that level of cryptograpy, but cffi===1.5.2 | 15:22 |
| sdague | it's probably better if we mirror that in liberty | 15:22 |
| sdague | instead of 1.6.0 for cffi | 15:22 |
| mdbooth | Mitaka, cryptograph 1.2.3 | 15:22 |
| mdbooth | sdague: Yeah, my hg-fu wasn't strong enough to determine when def_extern was introduced in cffi | 15:22 |
| sdague | mdbooth: no worries | 15:23 |
| mdbooth | However, 1.6.0 was the first release which mentioned it, and that it had been 'fixed' | 15:23 |
| *** sdake has joined #openstack-nova | 15:23 | |
| sdague | locally what are you using with the new openssl? | 15:23 |
| sdague | just want to make sure that liberty -> mitaka doesn't dictate a downgrade | 15:23 |
| mdbooth | For sure | 15:24 |
| *** Sukhdev has quit IRC | 15:24 | |
| mdbooth | Locally I'm pulling it from pypi, so exactly as constrained | 15:24 |
| mdbooth | So, bump cffi in mitaka, or downgrade in liberty? | 15:24 |
| * mdbooth finds the cffi release note | 15:24 | |
| bhagyashri | rpodolyaka: Hi, I have addressed the review comments. could you please review https://review.openstack.org/#/c/345191/? | 15:24 |
| *** praneshp has quit IRC | 15:25 | |
| sdague | mdbooth: well, ideally, we'd change as little as possible | 15:25 |
| mdbooth | sdague: http://cffi.readthedocs.io/en/latest/whatsnew.html#v1-6 | 15:25 |
| sdague | so 1.5.2 in liberty would be that | 15:25 |
| mdbooth | sdague: Yeah, I'll bump it down | 15:26 |
| sdague | if you can get 1.5.2 to work in liberty in gate, and locally, that would be the best case scenario | 15:26 |
| sdague | mdbooth: great | 15:26 |
| mdbooth | In the absence of being able to work it out in mercurial, I literally just went from that release note about 'robustness' | 15:26 |
| sdague | and as long as that works with your patched openssl, we should be able to shephard that all through | 15:26 |
| *** armax has joined #openstack-nova | 15:26 | |
| sdague | mdbooth: no worries :) | 15:26 |
| mriedem | sdague: https://review.openstack.org/#/c/363971/ is still running jobs, but it's passed grenade and single-node dsvm jobs, so the devstack/nova/grenade series it's dependent on should be good to go | 15:30 |
| mriedem | sdague: starting here https://review.openstack.org/#/c/355602/ | 15:30 |
| *** rcernin has joined #openstack-nova | 15:31 | |
| bauzas | dansmith: so, I looked at who calls update_usage_from_instance() so that it would call out the placement API | 15:32 |
| *** annegentle has quit IRC | 15:32 | |
| *** tbachman has quit IRC | 15:32 | |
| bauzas | dansmith: that would work pretty well for new instances and for the heal operation | 15:32 |
| sdague | mriedem: ok, will go look | 15:33 |
| bauzas | dansmith: but IMHO, we would miss all the move operation updates until we reconcile with the heal | 15:33 |
| dansmith | bauzas: okay | 15:33 |
| dansmith | bauzas: let me do that place then and we can figure out after if we can do it elsewhere to cover all the bases | 15:33 |
| sdague | mriedem: ok, that patch is going to fail tests, because of the osc bug | 15:33 |
| bauzas | dansmith: I think it's reasonable enough to assess that would be sufficient for Newton | 15:34 |
| mriedem | hmm, the RT is going to need an instance of the report client | 15:34 |
| *** BharatK has joined #openstack-nova | 15:34 | |
| dansmith | mriedem: it has one | 15:34 |
| bauzas | dansmith: given the heal | 15:34 |
| *** pbandark has quit IRC | 15:34 | |
| cfriesen | bauzas: dansmith: if we have the cpu thread policy set to "isolate", and we boot an instance with 1 vcpu, essentially it also reserves N sibling CPUs. should the hypervisor "vcpus_used" count be incremented by 1 or by 1+N? | 15:34 |
| bauzas | dansmith: because we'd get a reasonable but a bit unaccurate list of allocations | 15:34 |
| *** zengine has quit IRC | 15:34 | |
| dansmith | cdent: your allocation patches aren't on top of my inventory reporting one apparently, and seem to be out of date.. right? | 15:34 |
| mriedem | dansmith: do you mean this? self.scheduler_client = scheduler_client.SchedulerClient() | 15:35 |
| mriedem | i guess the RT would have to call self.scheduler_client.reportclient.foo() | 15:35 |
| dansmith | mriedem: yeah, which merges in the reportclient | 15:35 |
| cdent | dansmith: jay and ed were last in there and I think edleafe has something pending | 15:35 |
| cdent | i can move them now if need be? | 15:35 |
| edleafe | cdent: should be ready in a few minutes. | 15:35 |
| dansmith | cdent: okay I need these all lined up in order to do this | 15:35 |
| *** andreas_s has quit IRC | 15:35 | |
| dansmith | edleafe: okay push and then ping me when done and I will rebase on top of my inventory thing and then should be good | 15:36 |
| cdent | edleafe: can you do that then. we mean this one yeah: https://review.openstack.org/#/c/363061/ | 15:36 |
| dansmith | or that | 15:36 |
| dansmith | prolly easier for edleafe to do it | 15:36 |
| cdent | after edleafe does that I will fix the api side of things so they reflect the correct methods and are happy (they may already be okay) | 15:36 |
| edleafe | cdent: I'm working on https://review.openstack.org/#/c/363209 | 15:37 |
| sdague | dansmith: ok, you want to +A - https://review.openstack.org/#/c/361436 | 15:37 |
| cdent | edleafe: right, and the link i pointed to is what to base that one on | 15:37 |
| sdague | that should flush out the nova change | 15:38 |
| edleafe | cdent: ah, I didn't get the order for the patches | 15:38 |
| sdague | and the devstack change is rebase | 15:38 |
| sdague | so that will probably pass tests this time | 15:38 |
| dansmith | sdague: yep | 15:39 |
| rpodolyaka | bhagyashri: done, lgtm! | 15:41 |
| *** jheroux has quit IRC | 15:42 | |
| *** Swami has joined #openstack-nova | 15:42 | |
| bhagyashri | rpodolyaka: Thank you. | 15:43 |
| rpodolyaka | np! | 15:43 |
| *** salv-orlando has quit IRC | 15:43 | |
| cdent | oh hey, edleafe, while you're in there, probably time to remove the "WIP"? | 15:44 |
| sdague | johnthetubaguy_w: still about? | 15:45 |
| edleafe | cdent: will do | 15:45 |
| sfinucan | snikitin: Hey - got a moment? | 15:45 |
| sdague | alaski - https://review.openstack.org/#/c/322944 is the policy tool | 15:46 |
| sdague | which is still not approved | 15:46 |
| sdague | and I think is kind of needed to help people understand | 15:46 |
| *** BharatK_ has joined #openstack-nova | 15:47 | |
| *** kaisers_ has quit IRC | 15:47 | |
| *** BharatK has quit IRC | 15:47 | |
| *** BjoernT has joined #openstack-nova | 15:49 | |
| *** raj_singh_ has quit IRC | 15:49 | |
| *** paul-carlton2 has quit IRC | 15:50 | |
| *** matrohon has quit IRC | 15:50 | |
| openstackgerrit | Alberto Planas proposed openstack/nova: Add migrate_uri for invoking the migration https://review.openstack.org/356558 | 15:51 |
| *** markvoelker has joined #openstack-nova | 15:51 | |
| openstackgerrit | Ed Leafe proposed openstack/nova: Add create_all and delete_all for AllocationList https://review.openstack.org/363209 | 15:52 |
| edleafe | cdent: dansmith: ^^ | 15:52 |
| dansmith | edleafe: did you orphan the api bit? | 15:53 |
| mriedem | sdague: also before you're gone https://review.openstack.org/#/c/363937/ | 15:53 |
| edleafe | dansmith: huh? | 15:53 |
| dansmith | edleafe: you just pushed the db patch but not the api one that was on top right? | 15:53 |
| dansmith | I need both stacked.. I can rebase, just checking | 15:53 |
| cdent | dansmith: i'll get that | 15:53 |
| dansmith | cdent: okay | 15:53 |
| cdent | because I need to test it first to reflect the current state of affairs | 15:54 |
| edleafe | dansmith: I rebased on 363061 | 15:54 |
| dansmith | edleafe: I know, nevermind, cdent is doing it | 15:54 |
| openstackgerrit | Sam Betts proposed openstack/nova: [WIP] Support Ironic interface attach/detach in nova virt https://review.openstack.org/364413 | 15:54 |
| openstackgerrit | Chris Dent proposed openstack/nova: WIP Add support for allocations in placement API https://review.openstack.org/339874 | 15:54 |
| cdent | that's _not_ it ^ | 15:54 |
| dansmith | thanks | 15:55 |
| cdent | need one test to fix an alex_xu point, and then it should be good to go | 15:55 |
| sdague | mriedem: I occured to me, we should probably rename the placement job to -next | 15:56 |
| sdague | and use it for all the forward looking changes | 15:56 |
| sdague | so standing up placement api, standing up cells v2 new things, etc etc. It's kind of the bleeding edge of our code. | 15:56 |
| mriedem | sdague: i did think about renaming it to add cellsv2 in the name | 15:57 |
| mriedem | -next would also work | 15:57 |
| mriedem | does renaming a job while changes are in flight hurt anything? | 15:57 |
| sdague | mriedem: comment left, for later consideration | 15:57 |
| sdague | +2 | 15:57 |
| mriedem | ok | 15:57 |
| sdague | no, it just picks things up later | 15:57 |
| sdague | rename won't hurt any inflight thigns | 15:58 |
| mriedem | i'll try testing this policy discovery cli change locally | 15:58 |
| *** zengine has joined #openstack-nova | 16:01 | |
| *** claudiub has joined #openstack-nova | 16:04 | |
| *** andrewbogott has quit IRC | 16:05 | |
| *** andrewbogott has joined #openstack-nova | 16:05 | |
| *** bvanhav has quit IRC | 16:05 | |
| openstackgerrit | Sam Betts proposed openstack/nova: [WIP] Support Ironic interface attach/detach in nova virt https://review.openstack.org/364413 | 16:06 |
| * bauzas goes AWOL for a couple of hours, be back around 7pm UTC | 16:06 | |
| *** raunak has joined #openstack-nova | 16:06 | |
| alaski | sdague: got it | 16:10 |
| *** ijw has joined #openstack-nova | 16:11 | |
| *** GB21 has joined #openstack-nova | 16:11 | |
| mriedem | alaski: claudiub: why are the user/tenant/roles config options for that CLI? why wouldn't they just be options for the CLI? | 16:11 |
| *** salv-orlando has joined #openstack-nova | 16:12 | |
| alaski | they are only CLI opts from what I see | 16:12 |
| alaski | only CONF.register_cli_opts was used | 16:13 |
| *** johnthetubaguy_w has quit IRC | 16:14 | |
| alaski | oh. yeah they could have been done with @cmd_common.args as well | 16:14 |
| *** annegentle has joined #openstack-nova | 16:17 | |
| *** bvanhav has joined #openstack-nova | 16:17 | |
| openstackgerrit | Mathieu Gagné proposed openstack/nova: Make scheduler less greedy when loading instances for tracking https://review.openstack.org/363944 | 16:18 |
| *** johnthetubaguy_w has joined #openstack-nova | 16:18 | |
| johnthetubaguy_w | sdague: just about still around, having crazy problems doing anything today :( | 16:19 |
| edleafe | johnthetubaguy_w: need to borrow my sledgehammer? | 16:19 |
| *** kevinz has quit IRC | 16:20 | |
| *** prometheanfire has joined #openstack-nova | 16:20 | |
| johnthetubaguy_w | edleafe: I have one, but otherwise, yes | 16:20 |
| prometheanfire | the new os-brick release (1.6.1) caused a fail http://logs.openstack.org/55/364255/1/check/gate-cross-nova-python27-db-ubuntu-xenial/26598da/testr_results.html.gz | 16:20 |
| prometheanfire | I think this is the release people wanted to use in newton | 16:20 |
| mriedem | alaski: yeah, the thing that trips me up is they aren't listed in the --help | 16:20 |
| prometheanfire | mriedem: ^ | 16:21 |
| *** BharatK_ has quit IRC | 16:21 | |
| *** lucasagomes is now known as lucas-dinner | 16:21 | |
| prometheanfire | cinder is failing too though | 16:21 |
| prometheanfire | https://review.openstack.org/364255 | 16:21 |
| openstackgerrit | Chris Dent proposed openstack/nova: WIP Add support for allocations in placement API https://review.openstack.org/339874 | 16:22 |
| openstackgerrit | Chris Dent proposed openstack/nova: Add create_all and delete_all for AllocationList https://review.openstack.org/363209 | 16:22 |
| cdent | okay dansmith that's good to go | 16:22 |
| cdent | the change on 363209 was just a typo fix in an exception | 16:22 |
| dansmith | cdent: I already pulled, what did you change? | 16:22 |
| dansmith | okay | 16:22 |
| prometheanfire | I bugged -cinder as well | 16:23 |
| mriedem | prometheanfire: jesus https://github.com/openstack/os-brick/compare/1.5.0...1.6.1 | 16:23 |
| *** hshiina has quit IRC | 16:23 | |
| *** zenoway has quit IRC | 16:23 | |
| *** BharatK has joined #openstack-nova | 16:23 | |
| cdent | dansmith:heh, above I said "that's _not_ it ^" | 16:23 |
| mriedem | prometheanfire: well this is the thing that breaks the test https://github.com/openstack/os-brick/commit/f8e4f3cdbef614367c5b3cfcb32e1584475946f8 | 16:23 |
| prometheanfire | mriedem: yep | 16:23 |
| cdent | the new version actually works, the other one did not (was calling the wrong method) | 16:23 |
| dansmith | cdent: heh, so you did | 16:23 |
| cdent | (as we kept changing it back and forth) | 16:23 |
| alaski | mriedem: I need to run out for a bit, but I'll post a followup to change the CLI args | 16:24 |
| mriedem | alaski: ok, we can bug fix it | 16:24 |
| mriedem | so no rush | 16:24 |
| cdent | but what's there now is passing tests and also addresses the problem that alex_xu found (with bad resource class strings) | 16:24 |
| *** coolsvap has joined #openstack-nova | 16:24 | |
| *** ManishD has joined #openstack-nova | 16:25 | |
| *** krtaylor has quit IRC | 16:25 | |
| sdague | cdent: did we figure out if the db query is what people want? | 16:28 |
| *** markus_z has quit IRC | 16:28 | |
| cdent | we had two conversation here that both seemed to reach the same conclusion and edleafe made some tweaks accordingly. we're still not 100% positive that everything is perfect, but we're pretty sure it is not damaging. In our usual way these past couple of days we are hoping jaypipes will show up to confirm or deny. | 16:29 |
| sdague | ok, looking at the most current patch | 16:29 |
| dansmith | sdague: IMHO we can't merge that until we have something that really uses it, which is what I'm furiously typing on right now | 16:30 |
| sdague | it changed the thing I was wondering about putting the _check_capacity inside the transaction | 16:30 |
| dansmith | I'm at the "swear at devstack" phase | 16:30 |
| sdague | dansmith: sure | 16:30 |
| ManishD | dansmith: Thanks for review of https://review.openstack.org/#/c/357131/ | 16:30 |
| *** jeblair_ is now known as jeblair | 16:30 | |
| dansmith | sdague: we already got from jay the word on making sure we fail the transaction if we would otherwise overcommit capacity, so that's answered | 16:30 |
| *** BharatK has quit IRC | 16:31 | |
| ManishD | dansmith: Problem is: Nova instance is shut down without any detaching/disconnection operations hence during startup it fails with scsi_id cmd which it runs against the block storage which are encrypted | 16:31 |
| dansmith | ManishD: no time to talk about this right now.. today is FF and we're working on priority stuff | 16:32 |
| *** johnthetubaguy_w has quit IRC | 16:32 | |
| *** zul has quit IRC | 16:32 | |
| jroll | is "swear at devstack" not the default phase? | 16:33 |
| sdague | zomg jay evil twin is freaking me out - https://twitter.com/jaypipes/status/771384326185488384 | 16:33 |
| jroll | lol | 16:33 |
| edleafe | sdague: _check_capacity is in the transaction only because we have to first delete existing allocations before we can see if the new will fit, and we need to be able to roll that back | 16:34 |
| sdague | edleafe: no, I'm glad it's in the transaction | 16:34 |
| *** tbachman has joined #openstack-nova | 16:34 | |
| sdague | it seemed weird before to check it outside the transaction then move forward | 16:34 |
| *** jaypipes has joined #openstack-nova | 16:34 | |
| edleafe | sdague: ah, I misread your comment | 16:34 |
| ManishD | dansmith: ok sure, pls have a look whenever you have some free time. Appreciate your help. | 16:35 |
| *** tanvirt has quit IRC | 16:35 | |
| edleafe | sdague: and who knew there was a jay clone? | 16:35 |
| dansmith | I did | 16:35 |
| dansmith | never seem him before now though | 16:35 |
| dansmith | it's fairly amusing | 16:35 |
| *** ssurana has joined #openstack-nova | 16:36 | |
| *** su_zhang has joined #openstack-nova | 16:36 | |
| cdent | freaky | 16:37 |
| *** karlamrhein has joined #openstack-nova | 16:37 | |
| sdague | ok, the inventory client and logging patches are now in the gate 28 mins from merge if all goes well | 16:38 |
| dansmith | eff yeah | 16:39 |
| jaypipes | dansmith, edleafe: he's the better pipes. | 16:39 |
| dansmith | OMG | 16:39 |
| dansmith | look who is here | 16:39 |
| *** Oku_OS is now known as Oku_OS-away | 16:39 | |
| dansmith | jaypipes: enough bw for a hangout? | 16:39 |
| *** Apoorva has joined #openstack-nova | 16:40 | |
| *** yamahata has quit IRC | 16:40 | |
| *** ijw has quit IRC | 16:40 | |
| *** daemontool has quit IRC | 16:40 | |
| *** su_zhang has quit IRC | 16:41 | |
| *** su_zhang has joined #openstack-nova | 16:41 | |
| * dansmith guesses not | 16:41 | |
| jaypipes | dansmith: I can try. one minute... lemme get to a quiet place | 16:41 |
| dansmith | cool | 16:41 |
| cdent | dansmith: I fixed the commit message on https://review.openstack.org/#/c/339874/ to remove the WIP, just FYI | 16:41 |
| *** woodster_ has joined #openstack-nova | 16:42 | |
| dansmith | cdent: okay, don't push yet though.. too many balls in the air :) | 16:42 |
| cdent | sorry, did it on gerrit, it pushed itself, but feel free to clobber it | 16:42 |
| dansmith | okay | 16:42 |
| openstackgerrit | Matt Riedemann proposed openstack/nova: Remove hacked test that fails with latest os-brick https://review.openstack.org/364454 | 16:44 |
| *** sahid has quit IRC | 16:45 | |
| *** akshai has quit IRC | 16:45 | |
| *** zul has joined #openstack-nova | 16:45 | |
| *** ljxiash has joined #openstack-nova | 16:47 | |
| *** akshai has joined #openstack-nova | 16:47 | |
| *** tbachman_ has joined #openstack-nova | 16:48 | |
| dansmith | https://hangouts.google.com/call/3uopugv745a5ndzzpuhvrh6rmie | 16:48 |
| openstackgerrit | Chris Dent proposed openstack/nova: Add {get_,set_}aggregates to objects.ResourceProvider https://review.openstack.org/355263 | 16:49 |
| openstackgerrit | Chris Dent proposed openstack/nova: Implement get and set aggregates in the placement API https://review.openstack.org/357993 | 16:49 |
| openstackgerrit | Chris Dent proposed openstack/nova: [placement] add a placement_aggregates table to api_db https://review.openstack.org/362863 | 16:49 |
| *** timello has quit IRC | 16:49 | |
| *** tbachman has quit IRC | 16:49 | |
| *** tbachman_ is now known as tbachman | 16:49 | |
| *** timello has joined #openstack-nova | 16:50 | |
| *** pradk has quit IRC | 16:51 | |
| *** amotoki has quit IRC | 16:51 | |
| *** trisq has quit IRC | 16:52 | |
| *** ljxiash has quit IRC | 16:53 | |
| *** unicell has quit IRC | 16:55 | |
| *** derekh has quit IRC | 16:56 | |
| *** sridharg has quit IRC | 16:57 | |
| *** su_zhang has quit IRC | 16:57 | |
| *** markvoelker has quit IRC | 16:58 | |
| *** su_zhang has joined #openstack-nova | 16:58 | |
| *** su_zhang has quit IRC | 16:58 | |
| * cdent finds dinner | 16:59 | |
| *** wanghao_ has quit IRC | 17:00 | |
| *** tesseract- has quit IRC | 17:01 | |
| *** yamahata has joined #openstack-nova | 17:01 | |
| mdbooth | kashyap: WRT https://review.openstack.org/#/c/290510/ bugfixes aren't subject to any of the various release freezes, are they? | 17:02 |
| kashyap | mdbooth: Good question, I think the answer is "it depends". This does strictly fall under a bug fix | 17:02 |
| kashyap | dansmith: When you have a minute, maybe you have a comment? ^ | 17:03 |
| *** ducttape_ has quit IRC | 17:03 | |
| *** cmartin has quit IRC | 17:04 | |
| *** ducttape_ has joined #openstack-nova | 17:04 | |
| openstackgerrit | Stephen Finucane proposed openstack/nova-specs: Reserve NUMA nodes with PCI devices attached https://review.openstack.org/364468 | 17:07 |
| *** mvk has quit IRC | 17:07 | |
| dansmith | kashyap: no freeze for bugs of course | 17:08 |
| kashyap | Cool | 17:08 |
| *** krtaylor has joined #openstack-nova | 17:08 | |
| *** jaypipes has quit IRC | 17:10 | |
| *** lpetrut has quit IRC | 17:12 | |
| *** dinobot_ has quit IRC | 17:13 | |
| *** ducttape_ has quit IRC | 17:15 | |
| *** tonytan4ever has quit IRC | 17:16 | |
| *** code-R_ has quit IRC | 17:17 | |
| *** pradk has joined #openstack-nova | 17:22 | |
| *** ducttape_ has joined #openstack-nova | 17:23 | |
| *** unicell has joined #openstack-nova | 17:24 | |
| *** inara has quit IRC | 17:26 | |
| *** avolkov` has joined #openstack-nova | 17:30 | |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: tests: verify cpu pinning with prefer policy https://review.openstack.org/355475 | 17:30 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Allow linear packing of cores https://review.openstack.org/342709 | 17:30 |
| mriedem | alaski: so far https://review.openstack.org/#/c/356138/ looks good, just need to review the tests | 17:31 |
| *** avolkov has quit IRC | 17:31 | |
| openstackgerrit | Andrew Laski proposed openstack/nova: Pull from cell0 and build_requests for instance list https://review.openstack.org/356138 | 17:31 |
| mriedem | kashyap: mdbooth: there is a freeze when we start building release candidates | 17:31 |
| alaski | mriedem: cool. just added the comment you mentioned | 17:31 |
| mriedem | after rc1 bugs can be fixed on master | 17:31 |
| mriedem | and backported | 17:32 |
| mriedem | but only stop ship stuff gets into the release candidates | 17:32 |
| *** akshai has quit IRC | 17:33 | |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: tests: verify cpu pinning with prefer policy https://review.openstack.org/355475 | 17:34 |
| openstackgerrit | Alberto Planas proposed openstack/nova: Add migrate_uri for invoking the migration https://review.openstack.org/356558 | 17:35 |
| *** ManishD has quit IRC | 17:35 | |
| *** ijw_ has joined #openstack-nova | 17:35 | |
| *** kaisers_ has joined #openstack-nova | 17:35 | |
| *** moshele has joined #openstack-nova | 17:40 | |
| *** kaisers_ has quit IRC | 17:40 | |
| mriedem | alaski: what difference does the services current vs not make in this https://review.openstack.org/#/c/356138/23/nova/tests/functional/wsgi/test_servers.py ? | 17:41 |
| mriedem | for listing instances with build requests? | 17:41 |
| alaski | functionally it makes no difference. The test is there to ensure that | 17:42 |
| *** unicell has quit IRC | 17:43 | |
| alaski | I guess you're wondering because there's no conditional using it? | 17:43 |
| *** unicell has joined #openstack-nova | 17:43 | |
| alaski | it's because there is a conditional in the build path, so this ensures the different build paths still end up with the same list response | 17:43 |
| mriedem | ok | 17:44 |
| mriedem | what about a test where an instance fails a build before it's scheduled to a cell and we list it out, and it's coming from the build request rather than the cell? | 17:44 |
| *** nic has joined #openstack-nova | 17:45 | |
| alaski | mriedem: fails in scheduling and goes to error? | 17:45 |
| melwitt | alaski: something I just thought of is, with the service version get minimum fix, does that mean everything acts like AllServicesCurrent now? | 17:45 |
| mriedem | alaski: something like that | 17:46 |
| mriedem | basically to make sure we get the instance from the build request | 17:46 |
| alaski | mriedem: right now any failures wipe out the buildrequest and rely on the instance record | 17:47 |
| alaski | we basically only use the buildrequest while booting and before scheduling | 17:47 |
| mriedem | oh right b/c of your new change under this one | 17:47 |
| alaski | any drift from the happy path relies on cell0, or an instance record, or the instance is deleted | 17:47 |
| mriedem | so what about the case of booting before scheduling, where there is no host for the instance? | 17:47 |
| *** dstepanenko_ has left #openstack-nova | 17:48 | |
| mriedem | in other words, is it possible to functionally test the build request part of this | 17:49 |
| mriedem | i see the unit test | 17:49 |
| dansmith | bauzas: FWIW, I don't get called with sign=-1 for a delete ever, so we'll have to find somewhere else to put that | 17:51 |
| *** ijw_ has quit IRC | 17:51 | |
| alaski | mriedem: this does functionally test the buildrequest because the buildrequest is deleted after scheduling during the boot process. These functional tests noop the conductor call so it gets to the end of compute/api where we have a buildrequest | 17:52 |
| *** unicell has quit IRC | 17:53 | |
| alaski | this stops the build process at a point where we have a buildrequest and then lists | 17:53 |
| mriedem | ah ok | 17:53 |
| *** mvk has joined #openstack-nova | 17:53 | |
| mriedem | heh, maybe i should have looked at the test name closer | 17:53 |
| *** inara has joined #openstack-nova | 17:53 | |
| alaski | heh | 17:53 |
| *** prometheanfire has left #openstack-nova | 17:54 | |
| mriedem | +W good buddy | 17:54 |
| alaski | melwitt: get_minimum_version defaults to 0 in the tests regardless of the lookup. except for certain functional tests that actually start a service | 17:55 |
| alaski | mriedem: \o/ | 17:55 |
| mriedem | so in ocata we start early on cell0 stuff i suppose | 17:55 |
| mriedem | and then multi-cell support | 17:55 |
| melwitt | alaski: okay, cool | 17:55 |
| *** lpetrut has joined #openstack-nova | 17:55 | |
| edleafe | dansmith: I only found sign=-1 from an aborted migration | 17:55 |
| edleafe | but I didn't look that deeply | 17:55 |
| mriedem | and then migrating quotas to the api db | 17:55 |
| alaski | ugh | 17:56 |
| alaski | but yeah | 17:56 |
| *** avolkov` has quit IRC | 17:57 | |
| *** harlowja_ has joined #openstack-nova | 17:58 | |
| *** harlowja has quit IRC | 18:00 | |
| *** ihrachys has quit IRC | 18:00 | |
| *** TravT_ has joined #openstack-nova | 18:04 | |
| *** TravT has quit IRC | 18:04 | |
| *** TravT_ has quit IRC | 18:05 | |
| *** claudiub has quit IRC | 18:05 | |
| *** salv-orl_ has joined #openstack-nova | 18:08 | |
| *** salv-orlando has quit IRC | 18:11 | |
| *** hoonetorg has quit IRC | 18:11 | |
| dansmith | edleafe: yeah seems like that's the case | 18:11 |
| * edleafe steps away from the computer for a couple of hours | 18:13 | |
| *** su_zhang has joined #openstack-nova | 18:13 | |
| sdague | ok, inventory client and logging are in | 18:14 |
| melwitt | alaski: I didn't find how get_minimum_version defaults to 0 anywhere so I just tried querying the min version in each test and it's 15 in both cases. I wonder if something changed | 18:14 |
| alaski | melwitt: hmm. perhaps I misunderstood something then | 18:15 |
| *** Jeffrey4l has quit IRC | 18:15 | |
| *** ijw has joined #openstack-nova | 18:16 | |
| *** hoonetorg has joined #openstack-nova | 18:16 | |
| *** tonytan4ever has joined #openstack-nova | 18:16 | |
| *** sudipto has joined #openstack-nova | 18:16 | |
| alaski | it may have been unit tests where I saw that behavior, since no services are started there | 18:17 |
| alaski | and assumed it carried over to functional | 18:17 |
| bauzas | dansmith: so, AFAIR, sign=-1 is when the claim is wrong | 18:17 |
| *** tonytan4ever has quit IRC | 18:17 | |
| *** sudipto has quit IRC | 18:17 | |
| *** sudipto has joined #openstack-nova | 18:17 | |
| *** sudipto_ has joined #openstack-nova | 18:17 | |
| *** openstackgerrit has quit IRC | 18:18 | |
| melwitt | alaski: yeah, the existence of AllServicesCurrent implies that normally things wouldn't be last version | 18:18 |
| *** cdent has quit IRC | 18:18 | |
| *** openstackgerrit has joined #openstack-nova | 18:19 | |
| bauzas | dansmith: like when the instance_claim_abort, it calls _update_usage_from_instance with is_removed which transforms to sign=-1 for _update_usage() | 18:19 |
| bauzas | dansmith: or when we migrate, the move abort calls directly _update_usage() with sign=-1 | 18:20 |
| bauzas | dansmith: that why I said to you that _update_usage_from_instance() is only called for boot requests, not for move operations, but fine by us | 18:20 |
| bauzas | given the heal will also call it | 18:21 |
| dansmith | bauzas: well, delete is the thing I really need | 18:21 |
| *** GB21 has quit IRC | 18:21 | |
| bauzas | dansmith: you mean when we delete an instance? | 18:21 |
| *** unicell has joined #openstack-nova | 18:22 | |
| dansmith | bauzas: yeah | 18:22 |
| bauzas | dansmith: that's because it calls another method, sec. | 18:22 |
| melwitt | alaski: you're right that it returns 0 for services that have not been started. the trouble with the test_servers.py is that they use the OSAPIFixture which starts the WSGIService which is what nova-osapi_compute queries | 18:22 |
| bauzas | dansmith: AFAIR, that's the compute_manager directly calling the RT for updating the instance | 18:23 |
| melwitt | alaski: if I query nova-compute, that returns 0 in the wsgi test_servers.py | 18:23 |
| bauzas | dansmith: trying to find the right LOC | 18:23 |
| dansmith | okay | 18:23 |
| dansmith | apparently there is no GET on allocations :( | 18:23 |
| sdague | ok, checking in, is there a thing I can help with over the next 30 minutes? | 18:24 |
| alaski | melwitt: okay. so I guess my fix did change the behavior here. I'm working on an inline service_version monkeypatch to return 0 | 18:24 |
| mriedem | sdague: don't think so | 18:24 |
| mriedem | bauzas: yeah the compute manager calls it from _complete_deletion | 18:25 |
| mriedem | and then the RT checks the instance.vm_state to set the is_removed_instance flag to True | 18:25 |
| melwitt | alaski: I think we can follow patch to fix it, I was just thinking about it | 18:25 |
| bauzas | mriedem: right | 18:25 |
| bauzas | mriedem: I was trying to look at code | 18:25 |
| melwitt | alaski: *follow up | 18:25 |
| bauzas | dansmith: as mriedem said https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L730 | 18:26 |
| *** sudipto has quit IRC | 18:26 | |
| *** sudipto_ has quit IRC | 18:26 | |
| *** mtanino__ has joined #openstack-nova | 18:26 | |
| alaski | melwitt: so it turns out there's nothing conditional in the path anyways, mriedem was right all along. The conditional in the boot path is in the conductor, and for instance show or delete. But not list | 18:27 |
| alaski | so I'll remove that test as a followup | 18:27 |
| melwitt | alaski: oh. heh | 18:27 |
| melwitt | even better! | 18:27 |
| *** mtanino has quit IRC | 18:27 | |
| alaski | yep. my mistake | 18:28 |
| dansmith | bauzas: that still gets me to _update_usage_from_instance() in the RT, which doesn't seem to be calling me for deleted instances | 18:28 |
| dansmith | bauzas: oh nevermind, it does | 18:29 |
| bauzas | dansmith: cool then, I would have been surprised if it doesn't :) | 18:29 |
| bauzas | dansmith: with the right sign=-1, right? | 18:30 |
| dansmith | bauzas: yes | 18:30 |
| bauzas | because of mriedem said again, the fact that we check vm_state in _update_usage_from_instance | 18:30 |
| dansmith | bauzas: it was failing before because I did something else wrong I think | 18:30 |
| bauzas | (gosh, that method name is super boring to write) | 18:30 |
| dansmith | I will have a WIP to look at in a sec | 18:30 |
| bauzas | dansmith: probably not yet going to the compute due to some delay? | 18:31 |
| bauzas | dansmith: I mean the instance deletion call | 18:31 |
| bauzas | anyway, that works now :) | 18:31 |
| bauzas | dansmith: FWIW, I'll have to sweat for a 5K run in the next 15 mins, but I'll look at that when I'm back | 18:31 |
| *** mtanino has joined #openstack-nova | 18:31 | |
| *** harlowja_ has quit IRC | 18:31 | |
| *** harlowja has joined #openstack-nova | 18:32 | |
| dansmith | woot, it works | 18:32 |
| dansmith | bauzas: sure | 18:32 |
| *** mtanino__ has quit IRC | 18:33 | |
| bauzas | dansmith: so, like I said, make a very clear comment in the commit msg that we won't yet provide allocations for migrations, but that it will be reconciliated by the periodic call and I'll +1 it | 18:34 |
| *** ijw has quit IRC | 18:35 | |
| dansmith | bauzas: eventually yeah | 18:35 |
| dansmith | incoming | 18:35 |
| openstackgerrit | Dan Smith proposed openstack/nova: Add support for allocations in placement API https://review.openstack.org/339874 | 18:35 |
| openstackgerrit | Dan Smith proposed openstack/nova: Add create_all and delete_all for AllocationList https://review.openstack.org/363209 | 18:35 |
| openstackgerrit | Dan Smith proposed openstack/nova: WIP Simple instance allocations https://review.openstack.org/364498 | 18:35 |
| dansmith | still has debug ^ | 18:35 |
| dansmith | but creates and deletes allocations for instance boot/delete | 18:35 |
| dansmith | mriedem: alaski sdague ^ | 18:35 |
| bauzas | dansmith: I think we could very easily add migrations by adding a call before https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L787 | 18:36 |
| *** unicell has quit IRC | 18:36 | |
| dansmith | okay, we can do that after this works | 18:36 |
| dansmith | we need this for FF I think | 18:36 |
| bauzas | agreed | 18:36 |
| bauzas | I could write a follow-up | 18:37 |
| bauzas | and maybe ask for a FFE :p | 18:37 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/363871 | 18:37 |
| *** akshai has joined #openstack-nova | 18:37 | |
| dansmith | or call it a bug which it is :) | 18:37 |
| sdague | dansmith: I love your DAN logs | 18:38 |
| *** irenab has joined #openstack-nova | 18:38 | |
| dansmith | sdague: easy grepping | 18:38 |
| dansmith | sdague: don't make fun | 18:38 |
| bauzas | heh | 18:39 |
| *** annegentle has quit IRC | 18:39 | |
| * bauzas goes away for (hopefully) 45 mins | 18:39 | |
| sdague | dansmith: :) | 18:39 |
| dansmith | a holdover from my printk("DAN: ..."); days | 18:39 |
| sdague | dansmith: I'm building a local version of that | 18:39 |
| bauzas | dansmith: first glance looks good :) | 18:39 |
| dansmith | cool | 18:39 |
| *** cdent has joined #openstack-nova | 18:39 | |
| *** xavvior has joined #openstack-nova | 18:40 | |
| *** unicell has joined #openstack-nova | 18:40 | |
| *** unicell has quit IRC | 18:41 | |
| dansmith | sdague: also, noted that your safe_connect *can* still return None if we're already disabled | 18:41 |
| dansmith | sdague: so need to handle that in places | 18:41 |
| sdague | dansmith: right, it can | 18:42 |
| sdague | we're never using it's value though, right? | 18:42 |
| *** salv-orl_ has quit IRC | 18:42 | |
| sdague | the can't return None part is the .get/.post/.put/.delete calls | 18:43 |
| dansmith | sdague: sure, we expect a response and just do response.status_code or response.text etc | 18:43 |
| sdague | no, we don't | 18:43 |
| dansmith | oh duh right | 18:43 |
| dansmith | sorry sorry | 18:43 |
| sdague | they are 2 different levels | 18:43 |
| dansmith | confusing my sheet | 18:43 |
| sdague | yep, now worries | 18:43 |
| dansmith | yeah, ignore me | 18:43 |
| sdague | yep, no worries, lots in flight | 18:43 |
| *** annegentle has joined #openstack-nova | 18:43 | |
| *** prateek has joined #openstack-nova | 18:46 | |
| *** parora has quit IRC | 18:46 | |
| alaski | dansmith: looks sane, commented on a couple of things. The volume handling could happen later since we don't handle it properly today either | 18:47 |
| dansmith | alaski: right | 18:47 |
| *** fragatin_ has quit IRC | 18:48 | |
| *** annegent_ has joined #openstack-nova | 18:49 | |
| *** david-lyle has quit IRC | 18:49 | |
| cdent | dansmith: are things ready to browse? I guess this is it: https://review.openstack.org/#/c/364498/ | 18:49 |
| *** ijw has joined #openstack-nova | 18:49 | |
| dansmith | cdent: cleaning up now | 18:49 |
| dansmith | cdent: but works yeah | 18:49 |
| *** su_zhang has quit IRC | 18:50 | |
| *** david-lyle has joined #openstack-nova | 18:50 | |
| *** tbachman has quit IRC | 18:50 | |
| mriedem | cool, nice work | 18:51 |
| mriedem | i have to run my car over to the shop to get the starter checked, just had to jump it. can't be w/o a car in case i've got jury duty tomorrow and laura is at work. i'll pick up from wifi there. | 18:51 |
| *** annegentle has quit IRC | 18:52 | |
| cdent | dansmith: looks sane. are you happy with how it is choosing when to delete, that part seems a bit vague | 18:54 |
| dansmith | cdent: why vague? it's already calculated in existing code | 18:54 |
| cdent | dansmith: the irc log made it sound like 'sign' was itself not a clear vague | 18:55 |
| *** mriedem has quit IRC | 18:56 | |
| cdent | s/vague/signal/ | 18:56 |
| *** divyakkonoor has quit IRC | 18:56 | |
| dansmith | cdent: I think it is, I was just unsure why I wasn't ever getting -1, but that was because I was failing on the initial allocation | 18:56 |
| *** dikonoor has joined #openstack-nova | 18:57 | |
| *** akshai has quit IRC | 18:57 | |
| *** yassou has quit IRC | 18:57 | |
| *** dikonoo has joined #openstack-nova | 18:57 | |
| cdent | ah, okay | 18:57 |
| dansmith | the way I tested this was devstack, | 19:00 |
| dansmith | before boot validate that nova_api.allocations is empty, | 19:00 |
| dansmith | boot an instance, | 19:00 |
| dansmith | validate that allocations has three for the instance, one of each type, | 19:01 |
| dansmith | then delete the instance and verify that the table is empty again | 19:01 |
| *** tbachman has joined #openstack-nova | 19:03 | |
| *** irenab has quit IRC | 19:03 | |
| *** irenab has joined #openstack-nova | 19:07 | |
| *** fragatina has joined #openstack-nova | 19:10 | |
| cdent | dansmith: that seems legit for a sanity check | 19:12 |
| openstackgerrit | Pushkar Umaranikar proposed openstack/nova: Removing duplicates from columns_to_join list https://review.openstack.org/342299 | 19:12 |
| *** harlowja_ has joined #openstack-nova | 19:13 | |
| *** prateek_ has joined #openstack-nova | 19:14 | |
| *** raunak has quit IRC | 19:14 | |
| * cdent starts swearing | 19:16 | |
| *** tonytan4ever has joined #openstack-nova | 19:16 | |
| sdague | dansmith: when are allocations supposed to be written? | 19:16 |
| *** tonytan4ever has quit IRC | 19:16 | |
| *** prateek has quit IRC | 19:16 | |
| dansmith | sdague: same as claims for the time being | 19:17 |
| dansmith | sdague: which is why putting it in the same path makes it easy | 19:17 |
| *** harlowja has quit IRC | 19:17 | |
| dansmith | cdent: does that mean you're starting devstack? | 19:17 |
| cdent | yes | 19:17 |
| dansmith | hah | 19:17 |
| sdague | oh, there I see it going now | 19:18 |
| *** mingyu has quit IRC | 19:18 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Simple instance allocations from resource tracker https://review.openstack.org/364498 | 19:18 |
| dansmith | cleaned up with unit tests ^ | 19:18 |
| sdague | I'm running tempest against it, placement api seems to be doing all the rights things | 19:18 |
| dansmith | sweet | 19:18 |
| sdague | the fact that you used ERROR for dan logs threw me :) | 19:19 |
| sdague | red in the console and all | 19:19 |
| *** lyarwood_ is now known as lyarwood | 19:20 | |
| *** su_zhang has joined #openstack-nova | 19:20 | |
| dansmith | heh | 19:20 |
| dansmith | fixed in the latest version | 19:20 |
| *** tbachman has quit IRC | 19:21 | |
| sdague | so, from the metric of "this does not horribly explode" it gets a +1 | 19:22 |
| dansmith | that is the only metric that matters on the afternoon of FF | 19:23 |
| *** salv-orlando has joined #openstack-nova | 19:23 | |
| cdent | anybody tried a resize or anything like that yet? | 19:23 |
| cdent | I guess tempest will do that at some point during sdague's run? | 19:23 |
| dansmith | I was going to in a sec | 19:23 |
| dansmith | pushing this back to my devstack machine | 19:24 |
| sdague | cdent: it will | 19:24 |
| dansmith | it will fail gracefully anyway, so some checking would be good | 19:24 |
| * cdent is still waiting on devstack | 19:24 | |
| sdague | yeh, it's mostly about just looking at the state of the logs after the successful runs | 19:24 |
| *** raunak has joined #openstack-nova | 19:24 | |
| *** raunak has quit IRC | 19:25 | |
| *** dikonoo has quit IRC | 19:25 | |
| *** dikonoor has quit IRC | 19:25 | |
| *** chrichip has joined #openstack-nova | 19:25 | |
| *** su_zhang has quit IRC | 19:25 | |
| sdague | it's a weird pattern that I'm seeing though where we put allocations, then we put inventories really quickly 3 times after | 19:26 |
| *** ssurana has quit IRC | 19:27 | |
| matias | should I ask for a recheck for the gate-tempest-dsvm-multinode-live-migration failure here? https://review.openstack.org/#/c/358599/ I'm still not familiar enough with the recheck procedure | 19:27 |
| *** ssurana has joined #openstack-nova | 19:27 | |
| sdague | matias: yeh, that was an overnight bug | 19:27 |
| sdague | that is now fixed | 19:27 |
| matias | sdague: thanks | 19:27 |
| dansmith | cdent: just tried it | 19:27 |
| dansmith | cdent: we basically have no new allocation during the resize, but when we confirm we allocate the new value | 19:28 |
| dansmith | cdent: so still some work to do there, but not *too* far off | 19:28 |
| dansmith | cdent: so there is no GET on allocations huh? | 19:28 |
| cdent | no, jay didn't want that info exposed | 19:29 |
| dansmith | I kinda need that in order to determine if we're doing allocations too often | 19:29 |
| dansmith | hrm | 19:29 |
| cdent | you can get usages on a resource_provider | 19:29 |
| dansmith | yeah, but that... doesn't make a lot of sense I don't think | 19:29 |
| dansmith | to determine if we need to update our allocation from that usage info | 19:29 |
| *** gabor_antal_ has joined #openstack-nova | 19:29 | |
| cdent | but why do you need/want to do that? | 19:30 |
| dansmith | cdent: right now I'm just updating the allocations for every instance every minute | 19:30 |
| dansmith | in addition to the inventory...every minute | 19:30 |
| * cdent nods | 19:31 | |
| cdent | but wouldn't you want to check a local cache not a remote cache (in the ideal world)? | 19:31 |
| *** ijw has quit IRC | 19:31 | |
| dansmith | sure | 19:32 |
| dansmith | although it seems like we *should* know that actually | 19:32 |
| *** ijw has joined #openstack-nova | 19:32 | |
| *** unicell has joined #openstack-nova | 19:32 | |
| *** annegent_ has quit IRC | 19:32 | |
| sdague | ok, taking off for the holiday weekend. Enjoy folks. Talk to you all on Tuesday. | 19:33 |
| dansmith | because we only call this if is_new_instance or is_removed_instance, yet we're clearly hitting it every minute | 19:33 |
| dansmith | sdague: o/ | 19:33 |
| cdent | the only person that can change the allocation is the same thing that set the allocation, right, because it is the consumer (in the compute node context) | 19:33 |
| cdent | s/person/entity/ | 19:34 |
| *** bvanhav has quit IRC | 19:34 | |
| cdent | dansmith: I'm guess that once an instance is 'is_new_instance' it never stops being so | 19:34 |
| cdent | until it is removed | 19:35 |
| dansmith | cdent: does the allocation process in the api retry if it fails due to a race? | 19:35 |
| dansmith | or is the client supposed to try to reallocate? | 19:35 |
| cdent | it doesn't, yet, but there's a todo (from jay) saying that it should | 19:36 |
| cdent | i thought perhaps it was better for the client to do it but alex responded that it would more efficient (since allocations happen alot) for the server to at least try | 19:36 |
| cdent | (which is probably right) | 19:36 |
| dansmith | ah, I see why | 19:36 |
| dansmith | cdent: okay I think it should try a few times | 19:36 |
| dansmith | cdent: the server I mean | 19:37 |
| dansmith | so we clear our tracked instances every once in a while and regenerate, because of theheal I guess | 19:37 |
| *** ijw has quit IRC | 19:37 | |
| *** sdague has quit IRC | 19:37 | |
| dansmith | cdent: which is why we shoud be able to GET the allocation.. so we can heal if it's wrong when we periodically check | 19:37 |
| melwitt | yeah, they are cleared out each call to update_available_resource | 19:38 |
| dansmith | so we can fix that later, but it's going to generate a lot of http and db traffic in its current form | 19:38 |
| dansmith | yeah | 19:38 |
| cdent | dansmith: I'm sorry, I'm still confused. Probably because there's a nearby tv making pretty nature tv pictures, but: | 19:39 |
| dansmith | alaski: sorry didn't do your things yet | 19:39 |
| cdent | you want to GET the allocation to compare with what you've got currently to decide if you want to send it again? | 19:39 |
| alaski | just that one comment should be removed | 19:40 |
| * cdent has done indepdent confirmation of allocation create and delete when server boot and delete. | 19:42 | |
| cdent | fun to see it real | 19:42 |
| dansmith | cdent: right | 19:42 |
| cdent | and there's no local record (yet) of what we did last time... | 19:43 |
| * cdent thinks | 19:43 | |
| *** FL1SK has quit IRC | 19:43 | |
| dansmith | cdent: it's distributed, so we always have to be able to reconcile and heal | 19:44 |
| *** tongli has quit IRC | 19:44 | |
| dansmith | alaski: thanks for that utils pointer, I just added that.. need to see how many thousands of unit tests it breaks :D | 19:44 |
| *** mnestratov has quit IRC | 19:45 | |
| alaski | that's always fun. I only have 145 broken tests on my latest cells patch | 19:46 |
| *** rmart04 has joined #openstack-nova | 19:46 | |
| cdent | the unique id of a single allocation (not the uri) is the rp_uuid and the consumer_uuid so how does the distributed nature of things apply? | 19:46 |
| cdent | (sorry if my questions are stupid, trying to unwind things in my head) | 19:46 |
| *** hemanthm is now known as hemanthm|bbl | 19:46 | |
| dansmith | cdent: if the compute node doesn't know if the placement db has the stuff it needs, and can't tell, it has no way to do anything other than push it again | 19:47 |
| *** trisq has joined #openstack-nova | 19:47 | |
| dansmith | cdent: and even if we could record that it happened last time we tried, that could get out of sync, databases can get rolled back, etc.. we always have to be able to heal ourselves | 19:47 |
| cdent | right, what I'm getting at is that a local cache is easy to do (if it is true that there can only be one resource tracker per compute node) | 19:48 |
| dansmith | cdent: we have a local cache now, we just periodically invalidate it so that we will notice if/when we need to heal | 19:48 |
| cdent | and every 10th heal or whatever, you flush the cache | 19:48 |
| dansmith | which is what I'm seeing | 19:48 |
| dansmith | right, it's doing that now | 19:48 |
| *** mriedem has joined #openstack-nova | 19:49 | |
| cdent | okay | 19:49 |
| mriedem | how's it looking? | 19:49 |
| dansmith | even still, pushing all our instances up to the placement db every ten minutes for no reason sucks | 19:49 |
| dansmith | mriedem: I have a patch up that seems to work for the simple cases we were shooting for | 19:49 |
| cdent | so the goal is to do an http request that needs to look in the database to avoid something doing an http requests that needs to write in the database? | 19:49 |
| dansmith | mriedem: making some edits, but cdent sdague and myself have tested locally | 19:49 |
| mriedem | ok | 19:50 |
| cdent | (I confirm that resize updates don't happen immediately but do happen eventually) | 19:50 |
| dansmith | cdent: sure, because the query doesn't cause other people to fail to update their rows with real business | 19:50 |
| mriedem | oh i see a major issue | 19:50 |
| cdent | yeah, just stating it all out explicitly | 19:50 |
| dansmith | cdent: if we're all writing all the time for no reason, we're just going to clash and have to retry a lot | 19:50 |
| cdent | I'm assuming that jay had some kind of idea on this. probably vast (and possibly misplaced) faith in the datastore. | 19:51 |
| dansmith | cdent: well, the server could just not do the actual write if it looks like the data is unchanged | 19:51 |
| dansmith | cdent: that simplifies the client, | 19:51 |
| *** raunak has joined #openstack-nova | 19:51 | |
| dansmith | and lets the future be different, since the client will always lag the server | 19:52 |
| openstackgerrit | Dan Smith proposed openstack/nova: Simple instance allocations from resource tracker https://review.openstack.org/364498 | 19:52 |
| dansmith | does the bfv check ^ | 19:52 |
| *** prateek_ has quit IRC | 19:52 | |
| dansmith | haven't tested that locally, but about to | 19:52 |
| dansmith | works for me locally | 19:54 |
| *** su_zhang has joined #openstack-nova | 19:54 | |
| dansmith | mriedem: fixed the bp link locally, let me know when you're done | 19:55 |
| dansmith | we need to let this sit long enough to get an actual tempest run on it | 19:55 |
| mriedem | yeah. i was going to check the spec against the allocation thing too. | 19:56 |
| dansmith | because that matters? :D | 19:56 |
| mriedem | it's deadly serious. | 19:57 |
| * dansmith quakes in his boots | 19:58 | |
| mriedem | the spec does point out some things we lose by not handling cleaning up in update_available_resource | 20:00 |
| mriedem | like local delete | 20:00 |
| dansmith | yeah, that's true | 20:01 |
| dansmith | we can try to do that in local_delete in a followup | 20:01 |
| dansmith | can't really test that for reals though | 20:01 |
| *** rmart04 has quit IRC | 20:02 | |
| mriedem | during the spec review i mentioned if we'd just try and do the allocation delete from the API and jay said the update_available_resource periodic in the compute would just clean those up - which it could, if we patch that in later. | 20:03 |
| cdent | I had some weirdness with usage when trying to resize down to a smaller size but I think maybe I have no idea what I'm doing. | 20:03 |
| mriedem | you can't resize down | 20:03 |
| cdent | well that would explain that then | 20:04 |
| mriedem | or shouldn't be able to i don't think | 20:04 |
| mriedem | i thought we caught that in the api though | 20:04 |
| cdent | it let me do it | 20:04 |
| cdent | but didn't actually change anything | 20:04 |
| *** ijw has joined #openstack-nova | 20:04 | |
| *** TravT has joined #openstack-nova | 20:04 | |
| dansmith | you can't resize disk down | 20:04 |
| * bauzas is back | 20:04 | |
| *** annegentle has joined #openstack-nova | 20:05 | |
| alaski | xenapi lets you do it, which is why the API doesn't check it | 20:05 |
| cdent | i chose a smaller flavor (m1.small to m1.tiny) | 20:05 |
| dansmith | mriedem: anything else on there or should I push with the updated link to start the clock? | 20:05 |
| dansmith | alaski: right | 20:05 |
| mriedem | dansmith: still working through | 20:05 |
| cdent | i have learned many things today | 20:05 |
| dansmith | okay | 20:06 |
| *** derekh has joined #openstack-nova | 20:06 | |
| cdent | I think rather than getting too far over quota I'm going to give myself an opportunity to sleep so I can play with this stuff in more earnest tomorrow. Are there particular needs? | 20:06 |
| *** prateek has joined #openstack-nova | 20:07 | |
| mriedem | not that i know of, but i just got back | 20:07 |
| *** ducttape_ has quit IRC | 20:07 | |
| dansmith | no | 20:07 |
| *** ducttape_ has joined #openstack-nova | 20:08 | |
| cdent | cool, I'll just mess with it | 20:08 |
| *** TravT has quit IRC | 20:08 | |
| cdent | thanks for the stalwart effort dansmith | 20:08 |
| cdent | 'night | 20:09 |
| dansmith | o/ | 20:09 |
| *** cdent has quit IRC | 20:09 | |
| openstackgerrit | Augustina Ragwitz proposed openstack/nova: Assume neutron port_binding extensions enabled https://review.openstack.org/357726 | 20:09 |
| bauzas | dansmith: mriedem: so what should we do with the N-3 tag ? | 20:09 |
| bauzas | given we have our meeting in less than 1h | 20:09 |
| *** rama_y has quit IRC | 20:09 | |
| *** ijw has quit IRC | 20:09 | |
| bauzas | mriedem: dansmith: should we try to merge https://review.openstack.org/#/c/363209/9 and the upper ones before cutting the milestone, or should we FFE that ? | 20:10 |
| dansmith | bauzas: he has until EOD to do it I think | 20:10 |
| bauzas | dansmith: looking at the gate | 20:10 |
| dansmith | gate looks healthy, should be fine | 20:11 |
| dansmith | 48/13 == fast | 20:11 |
| mriedem | dansmith: i'm a bit confused about what the spec says for calculating the disk | 20:11 |
| bauzas | yeah | 20:11 |
| dansmith | mriedem: link? | 20:11 |
| bauzas | mriedem: the spec says we should use the aggregate for that | 20:11 |
| *** tonytan4ever has joined #openstack-nova | 20:11 | |
| mriedem | http://specs.openstack.org/openstack/nova-specs/specs/newton/approved/resource-providers-allocations.html | 20:11 |
| bauzas | mriedem: which is something not yet done | 20:11 |
| mriedem | http://specs.openstack.org/openstack/nova-specs/specs/newton/approved/resource-providers-allocations.html#implementation | 20:11 |
| bauzas | mriedem: so here we feature parity | 20:11 |
| *** coolsvap has quit IRC | 20:12 | |
| mriedem | it doesn't say anything about aggregates | 20:12 |
| bauzas | mriedem: it says that for a shared storage, the disk resource would be shared | 20:12 |
| bauzas | with only one resource provider | 20:12 |
| mriedem | well, it's not the only confusing thing, because it's talking about needing a compute node uuid in the URL for the request | 20:13 |
| *** gyee has joined #openstack-nova | 20:13 | |
| mriedem | i.e. the consume (instance uuid) is scoped to the resource provider (compute node uuid) in the request | 20:13 |
| mriedem | but that doesn't appear to be how the API actually works | 20:13 |
| dansmith | mriedem: I don't think I agree with the spec | 20:13 |
| bauzas | mriedem: in order to do that, we need to map an aggregate that would get all the hosts sharing the disk with a resource provider where the inventory should be set by Cinder | 20:13 |
| dansmith | mriedem: it says volume space should be counted against the compute node, which is wrong | 20:13 |
| bauzas | I agree with dansmith | 20:14 |
| dansmith | the whole point of this work is to NOT do that :P | 20:14 |
| *** rmart04 has joined #openstack-nova | 20:14 | |
| *** rmart04 has quit IRC | 20:14 | |
| bauzas | yeah, that's the exact purpose of having a shared resource provider with DISK resource class for a list of computes | 20:15 |
| mriedem | gah laura is vacuuming and it's throwing me off, just like julie | 20:15 |
| dansmith | heh | 20:16 |
| *** tonytan_brb has joined #openstack-nova | 20:16 | |
| bauzas | heh, that's why I love working after 9pm | 20:16 |
| mriedem | whitespace issue in https://review.openstack.org/#/c/339874/38/nova/tests/functional/api/openstack/placement/gabbits/allocations.yaml not sure if that matters | 20:16 |
| bauzas | gnomes are sleeping and TV is working | 20:16 |
| *** TravT has joined #openstack-nova | 20:16 | |
| *** gabor_antal__ has joined #openstack-nova | 20:17 | |
| *** tonytan4ever has quit IRC | 20:17 | |
| dansmith | mriedem: I don't think we have any validation of that..I think we're fine and it's passing jenkins, so I think we should not reset it | 20:17 |
| *** mingyu has joined #openstack-nova | 20:18 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Simple instance allocations from resource tracker https://review.openstack.org/364498 | 20:19 |
| mriedem | gah | 20:19 |
| mriedem | some comments in ps3 | 20:19 |
| *** rmart04 has joined #openstack-nova | 20:19 | |
| dansmith | it's fine, just trying to get the clock going if there was nothing else | 20:19 |
| mriedem | nothing blocking | 20:19 |
| *** antal has joined #openstack-nova | 20:20 | |
| *** gabor_antal_ has quit IRC | 20:20 | |
| dansmith | mriedem: okay, so good to patch after the fact? | 20:21 |
| dansmith | mriedem: I can follow up with "fixes" for local delete I think.. my brain is about goo | 20:21 |
| openstackgerrit | Mathieu Gagné proposed openstack/nova: Make scheduler less greedy when loading instances for tracking https://review.openstack.org/363944 | 20:21 |
| *** gabor_antal_ has joined #openstack-nova | 20:21 | |
| *** tblakes has quit IRC | 20:22 | |
| mriedem | we could have a unit test for the RT actually calling the method in the reportclient... | 20:22 |
| mriedem | or stack that on top | 20:22 |
| *** gabor_antal__ has quit IRC | 20:23 | |
| mriedem | we don't have unit tests for the put and delete calls failing | 20:23 |
| dansmith | mriedem: I didn't add those because I need to figure out how to make a mock look falsey | 20:23 |
| *** mingyu has quit IRC | 20:23 | |
| *** rmart04 has quit IRC | 20:23 | |
| bauzas | mriedem: FWIW, the whitespace issue seems not a problem http://logs.openstack.org/74/339874/38/check/gate-nova-tox-db-functional/43297d5/console.html#_2016-09-01_18_46_13_851946 | 20:24 |
| mriedem | maybe you can create a MagicMock and override __bool__ | 20:24 |
| *** antal has quit IRC | 20:25 | |
| dansmith | it's nonzero I think | 20:25 |
| mriedem | MagicMock.__bool__.return_value = False | 20:25 |
| mriedem | something like that | 20:25 |
| *** gabor_antal__ has joined #openstack-nova | 20:25 | |
| *** tbachman has joined #openstack-nova | 20:26 | |
| dansmith | it's __nonzero__ | 20:26 |
| dansmith | I got it | 20:26 |
| *** ijw has joined #openstack-nova | 20:26 | |
| dansmith | mriedem: the resource tracker client call will take me longer | 20:26 |
| mriedem | ah yeah http://stackoverflow.com/questions/2233786/overriding-bool-for-custom-class | 20:26 |
| dansmith | mriedem: you want me to update this now or stack on top? | 20:27 |
| mriedem | dansmith: just stack on top | 20:27 |
| dansmith | you are the one that has to stay up late for the tag... | 20:27 |
| dansmith | okay :) | 20:27 |
| *** _ducttape_ has joined #openstack-nova | 20:28 | |
| mriedem | +2 on https://review.openstack.org/#/c/364498/ | 20:28 |
| mriedem | with todos marked inline | 20:28 |
| *** gabor_antal_ has quit IRC | 20:29 | |
| dansmith | mriedem: we do want a 404 to be a warning on delete | 20:29 |
| dansmith | mriedem: we won't have gotten here if placement is not running, and 404 means we leaked something somewhere | 20:30 |
| mriedem | i was going on the spec, which also assumed we had update_available_resource doing things, which would introduce a race | 20:30 |
| mriedem | and possible duplicate deletion and a 404 | 20:30 |
| dansmith | hmm I don't think we can get a double delete this way | 20:31 |
| dansmith | except for after a local delete maybe | 20:31 |
| mriedem | so yeah we can ignore that and just lump in with a general todo for update_available_resource | 20:31 |
| *** ijw has quit IRC | 20:31 | |
| dansmith | but that is probably fine | 20:31 |
| dansmith | okay | 20:31 |
| *** ducttape_ has quit IRC | 20:31 | |
| bauzas | mriedem: well, we share the same semaphore between claims and update_avail_res AFAIR | 20:32 |
| bauzas | mriedem: so, not sure where the race could be | 20:32 |
| * bauzas looking | 20:32 | |
| *** gouthamr has quit IRC | 20:34 | |
| bauzas | yeah we share COMPUTE_RESOURCE_SEMAPHORE | 20:34 |
| *** _ducttape_ has quit IRC | 20:34 | |
| mriedem | the pci device will all have to wait for another time too....maybe by pike | 20:35 |
| bauzas | but I'm rereading the spec | 20:35 |
| bauzas | mriedem: PCI devices were way off the bag for Newton | 20:35 |
| mriedem | bauzas: i think the 404 on delete came up when we were talking about local delete in the spec review | 20:35 |
| *** ducttape_ has joined #openstack-nova | 20:35 | |
| mriedem | so that's probably where it came from and we do'nt have to worry about it right now | 20:35 |
| mriedem | alaski: https://review.openstack.org/#/c/355602/ \o/ | 20:35 |
| bauzas | mriedem: but a local delete doesn't touch the ComputeNode table | 20:36 |
| bauzas | mriedem: that's only when the periodic task is run on the compute side that we reconcile that | 20:36 |
| bauzas | (for local deletes) | 20:36 |
| mriedem | the jobs change for cellsv2 is also approved https://review.openstack.org/#/c/363937/ | 20:36 |
| *** prometheanfire has joined #openstack-nova | 20:36 | |
| bauzas | so there would only be one call to the placement API for deletion | 20:37 |
| prometheanfire | mriedem: you know about the new neutronclient release? | 20:37 |
| mriedem | prometheanfire: gdi | 20:37 |
| mriedem | please | 20:37 |
| mriedem | no | 20:37 |
| prometheanfire | lololol | 20:37 |
| mriedem | prometheanfire: you know you're becoming like a bot of bad news for me right? | 20:37 |
| alaski | mriedem: woo | 20:37 |
| mriedem | bauzas: idk, maybe we just chalk these up to unanswered questions in the etherpad to discuss with jaypipes next week | 20:37 |
| mriedem | bauzas: let's start an unanswered questions thing in https://etherpad.openstack.org/p/placement-next | 20:38 |
| mriedem | we have a todo section at the bottom | 20:38 |
| mriedem | let's start a subsection for allocation things | 20:38 |
| prometheanfire | mriedem: I'm sorry, I'm the only one manning the requirements team during the day, so it's my job to be that guy :( | 20:38 |
| *** _ducttape_ has joined #openstack-nova | 20:38 | |
| prometheanfire | mriedem: if it helps they broke neutron and horizon gates too | 20:38 |
| prometheanfire | https://review.openstack.org/364504 | 20:39 |
| *** annegentle has quit IRC | 20:39 | |
| bauzas | mriedem: over, over. | 20:39 |
| mriedem | bauzas: i've added some notes in there at the bottom | 20:39 |
| mriedem | prometheanfire: yeah tonyb had me on horizon breaking last night for novaclient | 20:39 |
| mriedem | prometheanfire: http://logs.openstack.org/04/364504/1/check/gate-cross-nova-python27-db-ubuntu-xenial/76cdad7/console.html looks like a timing issue | 20:40 |
| prometheanfire | mriedem: it's that time of the release cycle | 20:40 |
| mriedem | as in the package isn't in the mirror yet | 20:40 |
| prometheanfire | oh, that's better | 20:40 |
| mriedem | No matching distribution found for python-neutronclient===6.0.0 (from -c /home/jenkins/workspace/gate-cross-nova-python27-db-ubuntu-xenial/upper-constraints.txt (line 298)) | 20:40 |
| smcginnis | mriedem: Did a patch go through for that os-brick logging failing test? | 20:40 |
| prometheanfire | I wonder if the bot needs a wait before it submits to our gate | 20:40 |
| openstackgerrit | Dan Smith proposed openstack/nova: Cleanup allocation todo items https://review.openstack.org/364547 | 20:40 |
| mriedem | smcginnis: no https://review.openstack.org/#/c/364454/ | 20:41 |
| mriedem | has a +@ | 20:41 |
| mriedem | +2 | 20:41 |
| *** ijw has joined #openstack-nova | 20:41 | |
| mriedem | dansmith: alaski: maybe one of you wants to hit https://review.openstack.org/#/c/364454/ | 20:41 |
| smcginnis | mriedem: Can we get some help to push that through? The Cinder test update is almost through gate. | 20:41 |
| mriedem | i've put out the nova-core signal | 20:41 |
| alaski | looking | 20:42 |
| *** ducttape_ has quit IRC | 20:42 | |
| mriedem | basically we have an old test that mocks internals of os-brick and unsurprisingly that hurt us | 20:42 |
| * smcginnis imagines a batman symbol in the sky with a big N | 20:42 | |
| prometheanfire | mriedem: ya, sorry, false alarm (for now) | 20:42 |
| mriedem | smcginnis: no it would have to be the mascot | 20:42 |
| smcginnis | mriedem: What did you guys end up with for a mascot? | 20:42 |
| mriedem | pile of poo | 20:43 |
| smcginnis | LOL | 20:43 |
| mriedem | or a star/supernova | 20:43 |
| mriedem | whichever comes first | 20:43 |
| smcginnis | ;O) | 20:43 |
| *** tblakes has joined #openstack-nova | 20:43 | |
| alaski | dansmith: you've seen https://review.openstack.org/#/c/339874/38 working locally while testing? | 20:45 |
| *** _ducttape_ has quit IRC | 20:45 | |
| dansmith | alaski: extensively, as has cdent and sdague | 20:45 |
| alaski | dansmith: great | 20:45 |
| dansmith | I just +2d | 20:45 |
| *** ducttape_ has joined #openstack-nova | 20:45 | |
| *** ayoung has quit IRC | 20:46 | |
| *** ijw has quit IRC | 20:46 | |
| *** derekh has quit IRC | 20:47 | |
| *** annegentle has joined #openstack-nova | 20:47 | |
| *** gabor_antal__ is now known as antal | 20:48 | |
| alaski | dansmith: I'm sure you're fried, but https://review.openstack.org/#/c/326906/ could use a quick review | 20:48 |
| alaski | heh, nvm | 20:48 |
| dansmith | mriedem: hmm, I put a bug in the happy path to test those __nonzero__ things, but .. hrm | 20:48 |
| mriedem | dansmith: i also thought the mock needed to be mock.__nonzero__.return_value = False | 20:49 |
| mriedem | you have it the other way around | 20:49 |
| dansmith | ah yeah | 20:50 |
| *** ljxiash has joined #openstack-nova | 20:50 | |
| mriedem | so status check, https://review.openstack.org/#/c/356138/ is #6 in the gate queue | 20:50 |
| dansmith | actually, I need more | 20:51 |
| dansmith | return_value.__nonzero__.return_value | 20:51 |
| dansmith | that's why | 20:51 |
| mriedem | discoverable policy cli is merged | 20:51 |
| *** ijw has joined #openstack-nova | 20:52 | |
| *** takashin has joined #openstack-nova | 20:52 | |
| *** xavvior has quit IRC | 20:52 | |
| *** baoli_ has quit IRC | 20:53 | |
| *** chrichip has quit IRC | 20:53 | |
| openstackgerrit | Takashi NATSUME proposed openstack/nova: Treat 'body_map' as a mapping of HTTP request bodies https://review.openstack.org/363457 | 20:54 |
| *** ljxiash has quit IRC | 20:54 | |
| *** chrichip has joined #openstack-nova | 20:55 | |
| mriedem | alaski: for https://blueprints.launchpad.net/nova/+spec/cells-cell0 i think we just mark it as partially complete and defer the rest to ocata right? | 20:55 |
| alaski | yes | 20:56 |
| alaski | and scheduling-interaction | 20:56 |
| alaski | but the buildrequest_obj one I would consider done | 20:56 |
| openstackgerrit | Dan Smith proposed openstack/nova: Cleanup allocation todo items https://review.openstack.org/364547 | 20:57 |
| *** Hosam has joined #openstack-nova | 20:57 | |
| bauzas | 3 mins before meeting? | 20:57 |
| dansmith | quick one I hope? | 20:58 |
| mriedem | yup | 20:58 |
| mriedem | yes | 20:58 |
| *** xavvior has joined #openstack-nova | 20:59 | |
| *** browne has quit IRC | 20:59 | |
| *** Gibi is now known as gibi | 21:00 | |
| *** rama_y has joined #openstack-nova | 21:01 | |
| *** moshele has quit IRC | 21:02 | |
| dansmith | alaski: cells failed on the allocations one: http://logs.openstack.org/98/364498/4/check/gate-tempest-dsvm-cells/85c557e/console.html#_2016-09-01_20_57_14_497815 | 21:02 |
| dansmith | look familiar? don't think it's related | 21:02 |
| dansmith | if it's clean I'll bump it to restart so we don't have to wait | 21:03 |
| *** prateek_ has joined #openstack-nova | 21:03 | |
| alaski | not familiar. And I agree it doesn't seem related | 21:03 |
| *** vladikr has quit IRC | 21:04 | |
| alaski | nothing stands out in the logs | 21:04 |
| openstackgerrit | Dan Smith proposed openstack/nova: Simple instance allocations from resource tracker https://review.openstack.org/364498 | 21:05 |
| dansmith | kicked ^ | 21:05 |
| *** prateek has quit IRC | 21:06 | |
| *** eharney has quit IRC | 21:10 | |
| *** kaisers_ has joined #openstack-nova | 21:14 | |
| tonyb | anyone know what sdqgue is doing with: https://review.openstack.org/#/c/364262/1 ? | 21:17 |
| alaski | dansmith: I looked through logs for a few more minutes, seems like a random timing thing with snapshot | 21:17 |
| dansmith | alaski: same | 21:17 |
| *** kaisers_ has quit IRC | 21:18 | |
| mriedem | i think i've seen that before | 21:19 |
| mriedem | melwitt's 326906 is #2 in the gate | 21:20 |
| *** burt has quit IRC | 21:21 | |
| *** gszasz has quit IRC | 21:22 | |
| dansmith | aye | 21:22 |
| * bauzas calls it a day \o | 21:23 | |
| *** xavvior has quit IRC | 21:24 | |
| dansmith | mriedem: the __nonzero__ thing doesn't work on py3 | 21:24 |
| *** rcernin has quit IRC | 21:24 | |
| mriedem | dansmith: i know | 21:25 |
| mriedem | you have to use __bool__ | 21:26 |
| dansmith | but will that work on py2? | 21:26 |
| mriedem | i guess i didn't think about that when i +2d | 21:26 |
| mriedem | yeah | 21:26 |
| *** thorst has quit IRC | 21:26 | |
| dansmith | okay I'll change | 21:26 |
| dansmith | have to rebase it anyway | 21:26 |
| mriedem | you can assign __bool__ = __nonzero__ in the mock to do both | 21:26 |
| openstackgerrit | Dan Smith proposed openstack/nova: Cleanup allocation todo items https://review.openstack.org/364547 | 21:26 |
| *** thorst has joined #openstack-nova | 21:27 | |
| dansmith | but bool works on both so why? | 21:27 |
| mriedem | it does? | 21:27 |
| dansmith | you said it does | 21:27 |
| bauzas | mriedem: dansmith: so you want to check if r is None, right? | 21:27 |
| dansmith | bauzas: no | 21:27 |
| mriedem | i thought nonzero was py2 and bool was py3 | 21:27 |
| mriedem | so idk | 21:27 |
| *** takashin has left #openstack-nova | 21:28 | |
| *** antal has quit IRC | 21:28 | |
| dansmith | mriedem: mock refuses __bool__ on py2 :D | 21:28 |
| mriedem | man | 21:28 |
| mriedem | well, we can six.py2 case that shit, or blacklist the test for py3 | 21:29 |
| mriedem | this isn't blocking the release anyway | 21:29 |
| mriedem | so no rush | 21:29 |
| *** lpetrut has quit IRC | 21:29 | |
| alaski | according to http://stackoverflow.com/questions/2233786/overriding-bool-for-custom-class it's __nonzero__ on py2 and __bool_ on py3 | 21:29 |
| *** takashin has joined #openstack-nova | 21:29 | |
| dansmith | I know | 21:29 |
| dansmith | alaski: right | 21:29 |
| *** rtheis has quit IRC | 21:29 | |
| *** cleong has quit IRC | 21:30 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Cleanup allocation todo items https://review.openstack.org/364547 | 21:30 |
| dansmith | I think this will work ^ | 21:30 |
| dansmith | if not, punt | 21:30 |
| *** thorst has quit IRC | 21:31 | |
| *** claudiub has joined #openstack-nova | 21:31 | |
| *** rama_y has quit IRC | 21:32 | |
| alaski | why does mock_put.return_value = False not work? | 21:32 |
| *** gouthamr has joined #openstack-nova | 21:32 | |
| dansmith | alaski: because we always get a response, but it's falsey if it's an HTTP failure | 21:33 |
| dansmith | which i what we want | 21:33 |
| dansmith | *is | 21:33 |
| bauzas | dansmith: not sure I got your problem but mock.patch('something'', new=None) could make the deal | 21:34 |
| dansmith | no | 21:34 |
| dansmith | really | 21:34 |
| alaski | ahh, I see. you need r.status_code to work | 21:34 |
| dansmith | alaski: right | 21:34 |
| bauzas | ooooh I see | 21:34 |
| dansmith | :) | 21:35 |
| alaski | you could create an inner class in the test with __bool__/__nonzero__/other attrs and return that. but if you can shorthand it that's certainly cleaner | 21:35 |
| dansmith | sure but it's just silly | 21:36 |
| alaski | yeah | 21:36 |
| alaski | well I'll check back later to see if anything needs a kick but I'm going to get some food in the meantime | 21:39 |
| * dansmith wishes he had food | 21:39 | |
| auggy | dansmith: http://www.coopscookies.com/ | 21:40 |
| *** mingyu has joined #openstack-nova | 21:40 | |
| alaski | dansmith: you're one internet order and some quick personal interaction away from that being a reality | 21:40 |
| * auggy just literally lol'ed | 21:41 | |
| *** priteau has quit IRC | 21:42 | |
| *** FL1SK has joined #openstack-nova | 21:43 | |
| mriedem | hmm https://review.openstack.org/#/c/356963/ | 21:43 |
| mriedem | so sean +2ed that, and it's the last thing in that series | 21:44 |
| *** ducttape_ has quit IRC | 21:44 | |
| *** gouthamr_ has joined #openstack-nova | 21:44 | |
| *** ijw has quit IRC | 21:44 | |
| edleafe | mriedem: yeah, I thought that was odd | 21:44 |
| *** mingyu has quit IRC | 21:45 | |
| mriedem | well i'm not sure if it has to get in today or can come later | 21:45 |
| *** ducttape_ has joined #openstack-nova | 21:45 | |
| edleafe | mriedem: with all those -2s on the earlier patches, it isn't getting in today :) | 21:45 |
| edleafe | mriedem: oh, nm | 21:46 |
| edleafe | I thought it was dependent on them | 21:46 |
| mriedem | no | 21:46 |
| dansmith | alaski: hah | 21:47 |
| mriedem | those were some other confusion that we cleared up a few weeks ago | 21:47 |
| *** gouthamr has quit IRC | 21:47 | |
| mriedem | https://review.openstack.org/#/c/326906 is merged | 21:49 |
| melwitt | \o/ | 21:50 |
| *** yamahata has quit IRC | 21:50 | |
| *** su_zhang has quit IRC | 21:51 | |
| *** su_zhang has joined #openstack-nova | 21:53 | |
| mriedem | alaski: you said https://blueprints.launchpad.net/nova/+spec/cells-scheduling-interaction gets moved to ocata right? | 21:54 |
| *** thorst has joined #openstack-nova | 21:57 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/363871 | 21:57 |
| *** haplo37__ has quit IRC | 21:59 | |
| *** thorst has quit IRC | 22:01 | |
| *** yamahata has joined #openstack-nova | 22:02 | |
| mriedem | i think because of https://review.openstack.org/#/c/319379/ | 22:02 |
| *** ijw has joined #openstack-nova | 22:04 | |
| *** Swami has quit IRC | 22:11 | |
| *** sameo has quit IRC | 22:11 | |
| *** john51 has joined #openstack-nova | 22:13 | |
| *** xyang1 has quit IRC | 22:14 | |
| *** john51_ has quit IRC | 22:15 | |
| openstackgerrit | Augustina Ragwitz proposed openstack/nova: Remove deprecated cinder options https://review.openstack.org/363803 | 22:16 |
| *** browne has joined #openstack-nova | 22:20 | |
| mriedem | dansmith: hmm, why don't i see any of the logging from your allocations change here? http://logs.openstack.org/98/364498/5/check/gate-tempest-dsvm-neutron-full-ubuntu-xenial/eb1cdb8/logs/screen-n-cpu.txt.gz | 22:20 |
| dansmith | allocations 2/3 in the gate | 22:20 |
| mriedem | like Submitted allocation for instance | 22:20 |
| dansmith | mriedem: does that have placement enabled? | 22:21 |
| mriedem | no | 22:21 |
| mriedem | good call | 22:21 |
| dansmith | so... | 22:21 |
| dansmith | heh | 22:21 |
| mriedem | yeah the placement job is still running | 22:21 |
| *** erhudy has quit IRC | 22:22 | |
| *** mdrabe has quit IRC | 22:24 | |
| mriedem | i have to get out and mow this damn yard at some point | 22:24 |
| *** mingyu has joined #openstack-nova | 22:25 | |
| alaski | mriedem: correct | 22:26 |
| *** ducttape_ has quit IRC | 22:27 | |
| alaski | it should be relatively easy to get that done early in O | 22:27 |
| *** sameo has joined #openstack-nova | 22:28 | |
| auggy | i wonder if there's a roomba version of a lawnmower | 22:28 |
| dansmith | there is | 22:28 |
| auggy | omg | 22:28 |
| dansmith | http://www.husqvarna.com/us/products/robotic-lawn-mowers/ | 22:28 |
| mriedem | placement job passed check | 22:28 |
| dansmith | mriedem: of course it did | 22:28 |
| mriedem | have you noticed the disk allocation is always 0? | 22:29 |
| mriedem | http://logs.openstack.org/98/364498/5/check/gate-tempest-dsvm-neutron-placement-full-ubuntu-xenial-nv/acb2458/logs/screen-n-cpu.txt.gz#_2016-09-01_21_40_41_104 | 22:29 |
| mriedem | oh, micro flavors? | 22:29 |
| mriedem | we don't use swap or ephemeral flavors in tempest | 22:29 |
| *** ayoung has joined #openstack-nova | 22:30 | |
| dansmith | right | 22:31 |
| mriedem | yeah disk is 0 on both of those | 22:31 |
| mriedem | ok | 22:31 |
| mriedem | whew | 22:31 |
| dansmith | stop trying to find reasons not to let this succeed matt | 22:31 |
| dansmith | be a team player | 22:31 |
| mriedem | http://logs.openstack.org/98/364498/5/check/gate-tempest-dsvm-neutron-placement-full-ubuntu-xenial-nv/acb2458/logs/screen-n-cpu.txt.gz#_2016-09-01_21_43_19_618 | 22:31 |
| mriedem | there is the 404 | 22:31 |
| mriedem | other failures on allocate: | 22:32 |
| mriedem | http://logs.openstack.org/98/364498/5/check/gate-tempest-dsvm-neutron-placement-full-ubuntu-xenial-nv/acb2458/logs/screen-n-cpu.txt.gz#_2016-09-01_22_08_53_412 | 22:32 |
| mriedem | Unable to allocate inventory: Unable to create allocation for 'VCPU' on resource provider 'b10cec9e-84fd-4417-a83d-27b709d65a0c'. The requested amount would exceed the capacity. ) | 22:32 |
| * dansmith waits an eternity for load | 22:33 | |
| dansmith | mriedem: yeah, so that over capacity thing means we have overcommitted resources legitimately | 22:33 |
| dansmith | and might be a config thing on our tests | 22:33 |
| dansmith | we can let jay and chris look at that next week | 22:34 |
| dansmith | like, we might not have our allocation ratios right | 22:34 |
| dansmith | and the current scheduler might not be doing the right thing | 22:34 |
| mriedem | alright | 22:34 |
| dansmith | mriedem: the 404 is a negative test, so maybe something that was intended not to start? | 22:34 |
| mriedem | i'll throw that into the etherpad | 22:34 |
| dansmith | yeah | 22:35 |
| mriedem | there are some tests in tempest where it deletes the instance before it's active | 22:35 |
| dansmith | ah | 22:35 |
| dansmith | given that it seems to always be vcpu, I think it's actually doing well | 22:35 |
| dansmith | the over-cap one | 22:35 |
| *** jaypipes has joined #openstack-nova | 22:36 | |
| mriedem | that's where melwitt's favorite local delete patch alwasy breaks the gate | 22:36 |
| dansmith | I'd bet we're just not gracious enough | 22:36 |
| dansmith | jaypipes: we're overcommiting VCPU in the gate: http://logs.openstack.org/98/364498/5/check/gate-tempest-dsvm-neutron-placement-full-ubuntu-xenial-nv/acb2458/logs/screen-n-cpu.txt.gz#_2016-09-01_22_16_35_936 | 22:36 |
| melwitt | :D | 22:36 |
| melwitt | local delete <3 | 22:36 |
| dansmith | jaypipes: next week we should look at that.. I'm guessing current scheduler does different math or something | 22:36 |
| mriedem | oh crap my family is home, i'd better get outside and start working. i'll check back on this in an hour or so. | 22:37 |
| dansmith | vool | 22:37 |
| dansmith | cool even | 22:37 |
| mriedem | btw we're also watching 356963 | 22:37 |
| dansmith | yeah, isn't corefilter disabled by default? | 22:38 |
| auggy | i feel like there's an "eat local" pun in there... "delete local!" | 22:38 |
| dansmith | if so, the scheduler is overcommitting us | 22:38 |
| dansmith | yu[ | 22:38 |
| *** zengine has quit IRC | 22:38 | |
| dansmith | no corefilter in the gate | 22:38 |
| dansmith | so we probably just need to up the allocation ratio to squelch that change | 22:38 |
| dansmith | er, message | 22:38 |
| dansmith | I will try a devstack change | 22:39 |
| *** tonytan_brb has quit IRC | 22:43 | |
| openstackgerrit | Ryan Rossiter proposed openstack/nova: WIP: Shared state host manager https://review.openstack.org/332463 | 22:44 |
| dansmith | https://review.openstack.org/364581 | 22:44 |
| dansmith | should quiet those warnings I think ^ | 22:44 |
| openstackgerrit | Ryan Rossiter proposed openstack/nova: Add configurable fake driver https://review.openstack.org/344887 | 22:45 |
| *** ducttape_ has joined #openstack-nova | 22:48 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Cleanup allocation todo items https://review.openstack.org/364547 | 22:48 |
| dansmith | ^ depends on that devstack change to run it with placement | 22:48 |
| *** marst has quit IRC | 22:49 | |
| *** BjoernT has quit IRC | 22:50 | |
| *** su_zhang has quit IRC | 22:50 | |
| *** tblakes has quit IRC | 22:51 | |
| dansmith | gah | 22:51 |
| *** Swami has joined #openstack-nova | 22:52 | |
| *** su_zhang has joined #openstack-nova | 22:52 | |
| *** Swami_ has joined #openstack-nova | 22:52 | |
| *** gouthamr_ has quit IRC | 22:53 | |
| *** ijw has quit IRC | 22:56 | |
| openstackgerrit | Augustina Ragwitz proposed openstack/nova: Use admin context to create neutron ports https://review.openstack.org/357540 | 22:58 |
| *** annegentle has quit IRC | 23:01 | |
| *** TravT has quit IRC | 23:01 | |
| *** TravT has joined #openstack-nova | 23:02 | |
| *** kaisers_ has joined #openstack-nova | 23:02 | |
| *** rrecio has quit IRC | 23:04 | |
| *** ijw has joined #openstack-nova | 23:04 | |
| *** marst has joined #openstack-nova | 23:04 | |
| *** ayoung has quit IRC | 23:05 | |
| *** dimtruck is now known as zz_dimtruck | 23:05 | |
| *** zz_dimtruck is now known as dimtruck | 23:05 | |
| *** kaisers_ has quit IRC | 23:07 | |
| *** hemna is now known as hemnafk | 23:07 | |
| *** TravT has quit IRC | 23:08 | |
| *** pwnall has quit IRC | 23:08 | |
| *** ijw has quit IRC | 23:09 | |
| *** gouthamr has joined #openstack-nova | 23:10 | |
| *** salv-orlando has quit IRC | 23:10 | |
| *** markvoelker has joined #openstack-nova | 23:10 | |
| alex_xu | \o/ everything merged before freeze | 23:13 |
| *** trisq has quit IRC | 23:13 | |
| *** gouthamr_ has joined #openstack-nova | 23:14 | |
| *** gouthamr has quit IRC | 23:14 | |
| *** dimtruck is now known as zz_dimtruck | 23:15 | |
| *** su_zhang has quit IRC | 23:16 | |
| *** sameo has quit IRC | 23:24 | |
| *** anush__ has quit IRC | 23:24 | |
| *** annegentle has joined #openstack-nova | 23:27 | |
| dansmith | hmm, I assume we don't have anything running grenade with placement... | 23:27 |
| *** vishwanathj has quit IRC | 23:30 | |
| *** gongysh has joined #openstack-nova | 23:35 | |
| *** nic has quit IRC | 23:35 | |
| *** Guest99973 has quit IRC | 23:43 | |
| *** markvoelker has quit IRC | 23:51 | |
| *** prometheanfire has left #openstack-nova | 23:52 | |
| *** markvoelker has joined #openstack-nova | 23:56 | |
| *** diogogmt has quit IRC | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!