| jogo | dansmith: o/ | 00:00 |
|---|---|---|
| *** salv-orlando has joined #openstack-nova | 00:00 | |
| cfriesen | gate-nova-python27 just failed with the following: testtools.matchers._impl.MismatchError: '5.5.5.5-g9ec3421' != '2015.2.0-g9ec3421' | 00:01 |
| cfriesen | is that a known thing? | 00:01 |
| cfriesen | failed testcase is test_version_string_with_package_is_good | 00:01 |
| jogo | cfriesen: sounds like maybe a PBR release? | 00:02 |
| jogo | lifeless: ^ | 00:02 |
| *** IanGovett has quit IRC | 00:04 | |
| jogo | cfriesen: testing locally | 00:04 |
| *** salv-orlando has quit IRC | 00:05 | |
| cfriesen | failed logs are at http://logs.openstack.org/27/169827/8/check/gate-nova-python27/2009c78/ | 00:05 |
| cfriesen | bunch of other jenkins tests failed too | 00:06 |
| dims | melwitt: you forgot co-authored-by tag for yourself in https://review.openstack.org/#/c/159626/? | 00:06 |
| alex_xu | cfriesen: thanks for your review on https://review.openstack.org/170397 :) | 00:06 |
| *** Alexandra_ has quit IRC | 00:06 | |
| melwitt | dims: no need, only took a few minutes :) | 00:06 |
| dims | but, but, you should! | 00:07 |
| melwitt | I just moved some stuff around a little :P | 00:07 |
| dansmith | dims: you can fix that in the web ui pretty easy :D | 00:07 |
| dims | dansmith: y, doing that right now | 00:07 |
| jogo | cfriesen: I think I have the fix, will be a few minutes though | 00:08 |
| *** emagana has quit IRC | 00:08 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Fix disconnect_volume issue when find_multipath_device returns None https://review.openstack.org/159626 | 00:09 |
| eliqiao1 | hi dansmith: I put a comments on https://review.openstack.org/#/c/160075/ ,can you help to take a look at it? | 00:10 |
| melwitt | heh, thanks dims | 00:10 |
| jogo | mtreinish: ping | 00:10 |
| mtreinish | jogo: pong | 00:11 |
| dims | :) | 00:11 |
| lifeless | cfriesen: jogo: here, looking | 00:11 |
| jogo | mtreinish: nova is hitting https://review.openstack.org/#/c/179278/1 as well | 00:11 |
| jogo | http://logs.openstack.org/27/169827/8/check/gate-nova-python27/2009c78/console.html#_2015-04-30_23_29_19_402 | 00:11 |
| jogo | but we don | 00:11 |
| jogo | but I don't think we want to move to version 5.5.5.5 | 00:11 |
| openstackgerrit | Chris Friesen proposed openstack/nova: ix "down" nova-compute service spuriously marked as "up" https://review.openstack.org/163060 | 00:12 |
| lifeless | 2015 > 5 anyway | 00:12 |
| jogo | the test looks for a exact match it seems | 00:13 |
| mtreinish | jogo: yeah that looks like something with pbr or the test | 00:13 |
| lifeless | jogo: thats mocked out anyway | 00:13 |
| *** Marga_ has joined #openstack-nova | 00:14 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 00:14 |
| lifeless | jogo: at a high level it looks like nova has some version stuff that isn't inside pbr for some reason | 00:15 |
| jogo | lifeless: so this is blocking nova | 00:15 |
| lifeless | jogo: and the mocking hasn't actually worked | 00:15 |
| lifeless | jogo: yes, and I'm here to help :) | 00:15 |
| lifeless | so it mocks version_info | 00:15 |
| lifeless | it doesn't mock package_string | 00:16 |
| *** alanf-mc has quit IRC | 00:16 | |
| lifeless | jogo: and no, this isn't https://review.openstack.org/#/c/179278/1 | 00:16 |
| lifeless | jogo: its not an enforcement thing, its just a unit test impacted by library changes in pbr | 00:17 |
| jogo | lifeless: yeah I see that now | 00:17 |
| lifeless | so the 1.2.3.4-gx665656 stuff | 00:17 |
| lifeless | isn't what pbr returns now | 00:17 |
| jogo | lifeless: can we just drop that test? | 00:17 |
| *** ZZelle_ has quit IRC | 00:17 | |
| *** alanf-mc has joined #openstack-nova | 00:18 | |
| lifeless | I don't know | 00:19 |
| lifeless | like | 00:19 |
| lifeless | you've got custom code there | 00:19 |
| lifeless | the stub is setting version on the VersionInfo object | 00:20 |
| lifeless | what I think you actually want to stub out | 00:20 |
| lifeless | is the version_string call | 00:20 |
| lifeless | because thats the interface nova is depending on | 00:20 |
| lifeless | I'd like to see this code moved into pbr btw | 00:20 |
| lifeless | e.g. you register your config value with the VersionInfo object and get appropriately munged strings back | 00:21 |
| lifeless | so that other servers can do it and we can have it be consistent | 00:21 |
| lifeless | but for now, changing the stub to stub out version.version_info 'version_string', lambda:'5.5.5' | 00:21 |
| lifeless | or however its spelt in Mox | 00:21 |
| lifeless | should do it | 00:22 |
| jogo | lifeless: testing that now | 00:23 |
| *** eliqiao1 has quit IRC | 00:23 | |
| *** ssurana has quit IRC | 00:23 | |
| *** ssurana has joined #openstack-nova | 00:24 | |
| jogo | cfriesen: can you file a bug for this | 00:25 |
| openstackgerrit | Joe Gordon proposed openstack/nova: Make test_version_string_with_package_is_good work with pbr 0.11 https://review.openstack.org/179307 | 00:26 |
| *** joefides has joined #openstack-nova | 00:27 | |
| *** ssurana has quit IRC | 00:28 | |
| jogo | I'll file on then | 00:29 |
| *** mrda is now known as cafepressbot | 00:33 | |
| jogo | lifeless: thanks for the help | 00:33 |
| *** cafepressbot is now known as mrda | 00:33 | |
| openstackgerrit | Joe Gordon proposed openstack/nova: Make test_version_string_with_package_is_good work with pbr 0.11 https://review.openstack.org/179307 | 00:33 |
| jogo | dansmith: ^ | 00:34 |
| jogo | any nova cores around? | 00:34 |
| jogo | unbreak the gate patc | 00:34 |
| lifeless | you'll probably want to backport that too | 00:36 |
| lifeless | to unbreak stable | 00:36 |
| lifeless | since stable doesn't have (and can't sanely) a pbr pin | 00:36 |
| *** annegentle has joined #openstack-nova | 00:37 | |
| jogo | already have the kilo patch up | 00:37 |
| lifeless | juno too :) | 00:37 |
| lifeless | or should that be :/ | 00:37 |
| jogo | lifeless: checking juno now | 00:38 |
| *** igordcard has quit IRC | 00:39 | |
| *** promulo has joined #openstack-nova | 00:40 | |
| jogo | backporting to juno as well | 00:43 |
| *** annegentle has quit IRC | 00:43 | |
| lifeless | is i still supported? | 00:43 |
| jogo | yup | 00:43 |
| lifeless | https://review.openstack.org/#/c/179278/1 has merged. yay. | 00:43 |
| *** promulo__ has quit IRC | 00:44 | |
| jogo | https://review.openstack.org/#/q/I4887b8000c9943c91f8add56fcc54fa18e78d683,n,z | 00:44 |
| jogo | on to icehouse ... | 00:45 |
| *** davideagnello has quit IRC | 00:47 | |
| jogo | mikal: you around? | 00:49 |
| *** harlowja has quit IRC | 00:50 | |
| jogo | alaski: ^ | 00:50 |
| openstackgerrit | Eli Qiao proposed openstack/nova-specs: Validate the service state before deleting it https://review.openstack.org/163274 | 00:50 |
| jogo | looking for a core review to help unbreak master | 00:50 |
| jogo | ahh pep8! | 00:52 |
| *** Sukhdev_ has quit IRC | 00:53 | |
| openstackgerrit | Joe Gordon proposed openstack/nova: Make test_version_string_with_package_is_good work with pbr 0.11 https://review.openstack.org/179307 | 00:54 |
| *** annegentle has joined #openstack-nova | 00:54 | |
| jogo | mriedem_away: thanks | 01:00 |
| *** harlowja has joined #openstack-nova | 01:00 | |
| *** mriedem has joined #openstack-nova | 01:01 | |
| mriedem | jogo: you're not going to have a pep8 failure here? https://review.openstack.org/#/c/179313/ | 01:02 |
| jogo | mriedem: just fixed it | 01:02 |
| mriedem | hawt | 01:03 |
| mriedem | we should maybe be more aggressive with capping pbr in the future :) | 01:03 |
| mriedem | or per should release a 1.0 | 01:03 |
| mriedem | *br | 01:04 |
| mriedem | damn | 01:04 |
| mriedem | *pbr | 01:04 |
| mriedem | jogo: how do you feel about getting the requirements sync job working again? https://review.openstack.org/#/c/179254/1 | 01:05 |
| *** dsanders has quit IRC | 01:05 | |
| jogo | mriedem: maybe, this was a nova bug and easy to fix | 01:06 |
| mriedem | yeah i know | 01:06 |
| *** angdraug has quit IRC | 01:06 | |
| jogo | requirements sync? how is your patch related? | 01:06 |
| jogo | ohh I see | 01:06 |
| jogo | mriedem: heading out for a bit o/ | 01:07 |
| jogo | want to babysit my patches? | 01:07 |
| *** angdraug has joined #openstack-nova | 01:08 | |
| *** browne has quit IRC | 01:08 | |
| mriedem | jogo: sure | 01:08 |
| *** krak has quit IRC | 01:09 | |
| *** otter768 has joined #openstack-nova | 01:17 | |
| *** melwitt has quit IRC | 01:17 | |
| *** armax has joined #openstack-nova | 01:22 | |
| mriedem | dansmith: re: previous discussion on wtf is going on with the neutron multi node job, looks like linuxbridge was made the neutron plugin by default in devstack http://lists.openstack.org/pipermail/openstack-dev/2015-April/061077.html | 01:24 |
| mriedem | so i'm not sure what that means for the event stuff or how it's still sending events w/o nova admin auth | 01:24 |
| mriedem | if i remember tomorrow i'll get in the neutron channel and ask around | 01:25 |
| openstackgerrit | Brooks Kaminski proposed openstack/nova: Xenapi: Change generate-configdrive to use cdrom https://review.openstack.org/179135 | 01:26 |
| *** annegentle has quit IRC | 01:26 | |
| *** mwagner_lap has joined #openstack-nova | 01:30 | |
| *** irenab has joined #openstack-nova | 01:34 | |
| *** kaufer has joined #openstack-nova | 01:34 | |
| openstackgerrit | Fei Long Wang proposed openstack/nova: Fix nova backup for volume-backed instance https://review.openstack.org/164494 | 01:36 |
| *** Marga_ has quit IRC | 01:37 | |
| *** jecarey has joined #openstack-nova | 01:37 | |
| *** Marga_ has joined #openstack-nova | 01:37 | |
| *** ferest has joined #openstack-nova | 01:37 | |
| *** irenab has quit IRC | 01:39 | |
| *** joefides_ has joined #openstack-nova | 01:41 | |
| *** joefides has quit IRC | 01:41 | |
| *** mriedem has quit IRC | 01:42 | |
| *** Marga_ has quit IRC | 01:42 | |
| *** alanf-mc has quit IRC | 01:46 | |
| *** vilobhmm1 has quit IRC | 01:48 | |
| *** kaufer has quit IRC | 01:54 | |
| *** patrickeast has quit IRC | 01:57 | |
| jogo | mriedem_away: multinode neutron is dvr now btw | 02:01 |
| *** otter768 has quit IRC | 02:02 | |
| jogo | mtreinish: have a whole set of stable maint patches for nova that need a +W https://review.openstack.org/#/q/I4887b8000c9943c91f8add56fcc54fa18e78d683,n,z to unbreak the world | 02:07 |
| *** yamahata has quit IRC | 02:07 | |
| *** annegentle has joined #openstack-nova | 02:13 | |
| *** whenry has quit IRC | 02:14 | |
| *** gyee has quit IRC | 02:15 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Remove version string from the setup.cfg https://review.openstack.org/179322 | 02:19 |
| *** dave-mccowan has quit IRC | 02:20 | |
| *** ferest_ has joined #openstack-nova | 02:20 | |
| *** ferest has quit IRC | 02:21 | |
| dims | jogo: shouldn't we remove version from setup.cfg? ^^^ | 02:21 |
| *** VW_ has joined #openstack-nova | 02:22 | |
| *** ferest has joined #openstack-nova | 02:26 | |
| *** ferest_ has quit IRC | 02:27 | |
| *** annegentle has quit IRC | 02:27 | |
| *** baoli has joined #openstack-nova | 02:29 | |
| *** joefides_ has quit IRC | 02:32 | |
| *** unicell has quit IRC | 02:33 | |
| *** annegentle has joined #openstack-nova | 02:35 | |
| *** whenry has joined #openstack-nova | 02:35 | |
| *** joefides has joined #openstack-nova | 02:38 | |
| *** dsanders has joined #openstack-nova | 02:44 | |
| *** baoli has quit IRC | 02:50 | |
| jogo | dims: AFAIK no | 02:54 |
| jogo | we get the wrong version without it | 02:54 |
| dims | jogo: with 0.11.0 it will work fine (get from git). if you see tempest just got rid of version from their setup.cfg after 0.11.0 was published | 02:56 |
| alex_xu | how can I know whether a meeting channel free at specified time | 02:58 |
| jogo | on master 'nova-api --version' gives 2015.1.0 with your patch | 02:59 |
| *** busterswt has joined #openstack-nova | 03:00 | |
| jogo | and nova-api --version | 03:00 |
| jogo | 2015.2.0 | 03:00 |
| jogo | without your patch | 03:00 |
| *** busterswt has quit IRC | 03:04 | |
| dims | jogo: interesting. thanks | 03:05 |
| jogo | dims: I already tried that myself :) | 03:05 |
| dims | hehe | 03:05 |
| *** busterswt has joined #openstack-nova | 03:06 | |
| *** achanda has quit IRC | 03:10 | |
| *** unicell has joined #openstack-nova | 03:11 | |
| *** unicell has quit IRC | 03:15 | |
| *** unicell has joined #openstack-nova | 03:16 | |
| *** isd has quit IRC | 03:17 | |
| *** annegent_ has joined #openstack-nova | 03:17 | |
| alex_xu | gilliard: I just send nova api meeting reminder out | 03:19 |
| *** annegentle has quit IRC | 03:21 | |
| *** harlowja has quit IRC | 03:21 | |
| *** irenab has joined #openstack-nova | 03:23 | |
| *** dsanders has quit IRC | 03:25 | |
| *** fifieldt has joined #openstack-nova | 03:25 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: fix "down" nova-compute service spuriously marked as "up" https://review.openstack.org/163060 | 03:27 |
| *** irenab has quit IRC | 03:28 | |
| *** busterswt has quit IRC | 03:31 | |
| *** zzzeek has quit IRC | 03:32 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: fix "down" nova-compute service spuriously marked as "up" https://review.openstack.org/163060 | 03:35 |
| *** angdraug has quit IRC | 03:36 | |
| *** josecastroleon has joined #openstack-nova | 03:36 | |
| dansmith | mriedem_away: I didn't think that neutron would send events with linuxbridge, so that's interesting, but the events *are* coming in.. I'm so confused | 03:38 |
| *** dims has quit IRC | 03:38 | |
| *** vilobhmm1 has joined #openstack-nova | 03:39 | |
| *** josecastroleon has quit IRC | 03:39 | |
| *** VW_ has quit IRC | 03:43 | |
| *** VW_ has joined #openstack-nova | 03:44 | |
| *** yamahata has joined #openstack-nova | 03:44 | |
| *** marun has quit IRC | 03:46 | |
| *** whenry has quit IRC | 03:50 | |
| *** baoli has joined #openstack-nova | 03:51 | |
| *** baoli has quit IRC | 03:56 | |
| *** fifieldt has quit IRC | 03:59 | |
| *** isd has joined #openstack-nova | 04:02 | |
| *** mmedvede has quit IRC | 04:04 | |
| *** mmedvede has joined #openstack-nova | 04:08 | |
| *** VW_ has quit IRC | 04:10 | |
| *** achanda has joined #openstack-nova | 04:10 | |
| *** VW_ has joined #openstack-nova | 04:13 | |
| *** achanda has quit IRC | 04:15 | |
| *** annegent_ has quit IRC | 04:18 | |
| *** VW_ has quit IRC | 04:21 | |
| *** marun has joined #openstack-nova | 04:22 | |
| *** busterswt has joined #openstack-nova | 04:27 | |
| *** marun has quit IRC | 04:27 | |
| *** cfriesen has quit IRC | 04:28 | |
| *** achanda has joined #openstack-nova | 04:30 | |
| *** otter768 has joined #openstack-nova | 04:30 | |
| *** busterswt has quit IRC | 04:32 | |
| *** gtt116__ has joined #openstack-nova | 04:35 | |
| *** busterswt has joined #openstack-nova | 04:36 | |
| *** gtt116_ has quit IRC | 04:37 | |
| *** melwitt has joined #openstack-nova | 04:37 | |
| *** dims has joined #openstack-nova | 04:39 | |
| *** axpBen has joined #openstack-nova | 04:39 | |
| *** otter768 has quit IRC | 04:41 | |
| *** dims has quit IRC | 04:45 | |
| *** pixelb has quit IRC | 04:46 | |
| jogo | dansmith: we use DVR now AFAIK wonder what that does | 04:47 |
| *** vilobhmm1 has quit IRC | 04:47 | |
| *** lpetrut has joined #openstack-nova | 04:52 | |
| *** xyang1 has quit IRC | 04:52 | |
| *** axpBen has quit IRC | 04:53 | |
| *** garyk has quit IRC | 04:55 | |
| *** busterswt has quit IRC | 05:01 | |
| *** flwang1 has quit IRC | 05:01 | |
| *** dsanders has joined #openstack-nova | 05:02 | |
| *** bkopilov has quit IRC | 05:08 | |
| *** ildikov has quit IRC | 05:19 | |
| *** isd has quit IRC | 05:20 | |
| *** emagana has joined #openstack-nova | 05:31 | |
| *** ildikov has joined #openstack-nova | 05:32 | |
| *** yamahata has quit IRC | 05:43 | |
| *** markvoelker has quit IRC | 05:48 | |
| *** emagana has quit IRC | 05:48 | |
| *** browne has joined #openstack-nova | 05:58 | |
| *** emagana has joined #openstack-nova | 06:03 | |
| *** yamahata has joined #openstack-nova | 06:04 | |
| *** emagana has quit IRC | 06:09 | |
| *** emagana has joined #openstack-nova | 06:10 | |
| *** emagana has quit IRC | 06:14 | |
| *** heyongli has quit IRC | 06:18 | |
| *** heyongli has joined #openstack-nova | 06:18 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Imported Translations from Transifex https://review.openstack.org/178055 | 06:19 |
| *** Longgeek_ has joined #openstack-nova | 06:24 | |
| *** Longgeek has quit IRC | 06:27 | |
| *** irenab has joined #openstack-nova | 06:27 | |
| *** alanf-mc has joined #openstack-nova | 06:28 | |
| *** irenab has quit IRC | 06:32 | |
| *** lpetrut has quit IRC | 06:34 | |
| *** hightall has joined #openstack-nova | 06:36 | |
| *** otter768 has joined #openstack-nova | 06:42 | |
| *** zul has joined #openstack-nova | 06:43 | |
| *** hightall has quit IRC | 06:43 | |
| *** otter768 has quit IRC | 06:46 | |
| *** melwitt has quit IRC | 06:50 | |
| *** alanf-mc_ has joined #openstack-nova | 06:54 | |
| *** alanf-mc has quit IRC | 06:58 | |
| *** achanda has quit IRC | 06:58 | |
| *** alanf-mc_ has quit IRC | 07:05 | |
| *** alanf-mc has joined #openstack-nova | 07:06 | |
| *** achanda has joined #openstack-nova | 07:08 | |
| *** dsanders has quit IRC | 07:12 | |
| gilliard | alex_xu: Thanks! | 07:17 |
| gilliard | Good morning Nova | 07:18 |
| *** bkopilov has joined #openstack-nova | 07:18 | |
| *** achanda has quit IRC | 07:23 | |
| *** zul has quit IRC | 07:25 | |
| *** zul has joined #openstack-nova | 07:25 | |
| *** oro has joined #openstack-nova | 07:34 | |
| *** browne has quit IRC | 07:34 | |
| *** alanf-mc has quit IRC | 07:35 | |
| *** jlanoux has joined #openstack-nova | 07:40 | |
| *** ildikov has quit IRC | 07:46 | |
| *** salv-orlando has joined #openstack-nova | 07:47 | |
| * gilliard checks calendar | 07:50 | |
| *** gilliard is now known as gilllliard | 07:50 | |
| *** Longgeek_ has quit IRC | 07:52 | |
| *** Longgeek has joined #openstack-nova | 07:56 | |
| *** arnaud____ has joined #openstack-nova | 07:56 | |
| *** sahid has joined #openstack-nova | 07:57 | |
| *** zul has quit IRC | 08:02 | |
| *** doron_afk has joined #openstack-nova | 08:04 | |
| *** romainh has joined #openstack-nova | 08:05 | |
| *** zul has joined #openstack-nova | 08:09 | |
| gilllliard | What does 'nova resume' do? Different from 'unpause' or 'start' ? | 08:10 |
| *** lucasagomes has joined #openstack-nova | 08:11 | |
| *** irenab has joined #openstack-nova | 08:13 | |
| *** boris-42 has joined #openstack-nova | 08:13 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/179340 | 08:14 |
| alex_xu | gilllliard: np, good morning | 08:15 |
| gilllliard | Hi alex_xu - thanks for the reminder email about the meeting. | 08:15 |
| alex_xu | gilllliard: np, I can't find Ken'ichi, I'm afraid reminder email is too late, so I decide send it first :) | 08:16 |
| *** irenab has quit IRC | 08:18 | |
| *** sahid has quit IRC | 08:23 | |
| *** arnaud____ has quit IRC | 08:25 | |
| *** derekh has joined #openstack-nova | 08:28 | |
| *** rook has quit IRC | 08:30 | |
| *** ferest has quit IRC | 08:31 | |
| *** zul has quit IRC | 08:31 | |
| *** otter768 has joined #openstack-nova | 08:42 | |
| *** Longgeek has quit IRC | 08:46 | |
| *** dsanders has joined #openstack-nova | 08:47 | |
| *** otter768 has quit IRC | 08:47 | |
| *** igordcard has joined #openstack-nova | 08:52 | |
| *** arnaud____ has joined #openstack-nova | 09:00 | |
| *** ndipanov has joined #openstack-nova | 09:02 | |
| johnthetubaguy | gilllliard: alex_xu: what time is the meeting again, I would like to join you, if possible, 1200UTC? | 09:03 |
| alex_xu | johnthetubaguy: yes, it is 1200UTC | 09:03 |
| johnthetubaguy | ah, reads email, got it | 09:03 |
| johnthetubaguy | alex_xu: gilllliard: did you find a meeting room thats free, doesn't seems to have specified it there | 09:03 |
| johnthetubaguy | actually usual place looks good | 09:04 |
| alex_xu | johnthetubaguy: I checked at here, https://wiki.openstack.org/wiki/Meetings there isn't other meeting at the same time, is it correct way booking the meeting room? | 09:04 |
| johnthetubaguy | alex_xu: its the best way I know, there is an iCAL feed, but I don't remember where that is now | 09:05 |
| *** arnaud____ has quit IRC | 09:05 | |
| alex_xu | johnthetubaguy: ok | 09:05 |
| *** zul has joined #openstack-nova | 09:07 | |
| *** gtt116_ has joined #openstack-nova | 09:08 | |
| *** ociuhandu has joined #openstack-nova | 09:08 | |
| *** gtt116__ has quit IRC | 09:12 | |
| *** ociuhandu has quit IRC | 09:12 | |
| gilllliard | johnthetubaguy: Yeah there's no conflicts that I saw. IDK how to update the iCal though. It's at https://www.google.com/calendar/ical/bj05mroquq28jhud58esggqmh4@group.calendar.google.com/public/basic.ics | 09:13 |
| *** rotbeard has joined #openstack-nova | 09:14 | |
| *** ferest has joined #openstack-nova | 09:20 | |
| *** Murali has joined #openstack-nova | 09:24 | |
| *** Murali_ has joined #openstack-nova | 09:29 | |
| *** markvoelker has joined #openstack-nova | 09:29 | |
| *** lucasagomes has quit IRC | 09:29 | |
| *** Murali has quit IRC | 09:32 | |
| *** Murali_ is now known as Murali | 09:32 | |
| *** dsanders has quit IRC | 09:43 | |
| openstackgerrit | Matthew Gilliard proposed openstack/nova-specs: Allow more instance operations during live migration https://review.openstack.org/179346 | 09:46 |
| *** lucasagomes has joined #openstack-nova | 09:48 | |
| openstackgerrit | John Garbutt proposed openstack/nova: devref: add information to clarify nova scope https://review.openstack.org/178623 | 09:55 |
| *** sdake_ has joined #openstack-nova | 09:55 | |
| *** IanGovett has joined #openstack-nova | 09:58 | |
| *** sdake_ has quit IRC | 09:58 | |
| *** sdake has quit IRC | 09:59 | |
| *** irenab has joined #openstack-nova | 10:02 | |
| *** IanGovett has quit IRC | 10:06 | |
| *** irenab has quit IRC | 10:06 | |
| *** josecastroleon has joined #openstack-nova | 10:09 | |
| *** Murali has quit IRC | 10:10 | |
| *** exploreshaifali has joined #openstack-nova | 10:14 | |
| *** exploreshaifali has quit IRC | 10:17 | |
| *** romainh has left #openstack-nova | 10:18 | |
| alex_xu | gilllliard: it may need mac? I don't have mac :( | 10:24 |
| gilllliard | The ICS feed? You can add it to your google calendar if you have one. | 10:24 |
| gilllliard | Other Calendars > Add by URL | 10:24 |
| alex_xu | gilllliard: oh, thanks let me try | 10:25 |
| *** Murali has joined #openstack-nova | 10:28 | |
| *** IanGovett has joined #openstack-nova | 10:28 | |
| openstackgerrit | Daniel Berrange proposed openstack/nova: tests: make API signature test also check static function https://review.openstack.org/155707 | 10:35 |
| *** josecastroleon has quit IRC | 10:37 | |
| *** otter768 has joined #openstack-nova | 10:43 | |
| *** pixelb has joined #openstack-nova | 10:44 | |
| *** zul has quit IRC | 10:44 | |
| *** otter768 has quit IRC | 10:48 | |
| *** arnaud____ has joined #openstack-nova | 10:50 | |
| *** arnaud____ has quit IRC | 10:54 | |
| *** Hosam has joined #openstack-nova | 10:55 | |
| *** irenab has joined #openstack-nova | 11:03 | |
| *** Murali has quit IRC | 11:03 | |
| lxsli | danpb: do you know how libvirt could return memory_total = 0 here please? https://review.openstack.org/#/c/171079/ | 11:03 |
| gilllliard | eliqiao: ^^ ? | 11:04 |
| *** rotbeard has quit IRC | 11:05 | |
| *** irenab has quit IRC | 11:07 | |
| *** Hosam has quit IRC | 11:18 | |
| *** dave-mccowan has joined #openstack-nova | 11:22 | |
| *** dprince has joined #openstack-nova | 11:30 | |
| *** markvoelker has quit IRC | 11:30 | |
| *** kaufer has joined #openstack-nova | 11:31 | |
| *** lucasagomes is now known as lucas-hungry | 11:38 | |
| *** arnaud____ has joined #openstack-nova | 11:51 | |
| *** dprince has quit IRC | 11:52 | |
| gilllliard | Nova API meeting in 5 minutes in #openstack-meeting | 11:53 |
| *** arnaud____ has quit IRC | 11:55 | |
| alex_xu | and 3 minutes left... | 11:57 |
| gilllliard | johnthetubaguy jaypipes oomich alex_xu ^^ | 11:58 |
| *** oomichi has joined #openstack-nova | 11:59 | |
| *** markvoelker has joined #openstack-nova | 12:01 | |
| *** eliqiao1 has joined #openstack-nova | 12:03 | |
| openstackgerrit | John Garbutt proposed openstack/nova: devref: add information to clarify nova scope https://review.openstack.org/178623 | 12:03 |
| *** markvoelker has quit IRC | 12:06 | |
| *** Sukhdev has joined #openstack-nova | 12:07 | |
| *** markvoelker has joined #openstack-nova | 12:12 | |
| *** irenab has joined #openstack-nova | 12:18 | |
| *** jaypipes is now known as leakypipes | 12:21 | |
| *** irenab has quit IRC | 12:23 | |
| *** joefides_ has joined #openstack-nova | 12:27 | |
| *** zul has joined #openstack-nova | 12:30 | |
| *** joefides has quit IRC | 12:31 | |
| *** hightall has joined #openstack-nova | 12:32 | |
| *** hightall has quit IRC | 12:33 | |
| *** bwensley has joined #openstack-nova | 12:34 | |
| *** zul has quit IRC | 12:35 | |
| *** dprince has joined #openstack-nova | 12:39 | |
| *** bwensley has quit IRC | 12:39 | |
| *** otter768 has joined #openstack-nova | 12:44 | |
| *** baoli has joined #openstack-nova | 12:45 | |
| *** baoli has quit IRC | 12:47 | |
| *** baoli has joined #openstack-nova | 12:48 | |
| *** cbader has quit IRC | 12:49 | |
| *** otter768 has quit IRC | 12:49 | |
| *** lucas-hungry is now known as lucasagomes | 12:49 | |
| *** zzzeek has joined #openstack-nova | 12:53 | |
| *** annegentle has joined #openstack-nova | 12:54 | |
| *** ivasev has joined #openstack-nova | 12:58 | |
| leakypipes | bauzas, gilllliard, edleafe, sandywalsh: for your Friday amusement: http://bit.ly/existential-excuses | 12:59 |
| *** dboik_ has quit IRC | 12:59 | |
| *** rfolco has joined #openstack-nova | 12:59 | |
| *** xyang1 has joined #openstack-nova | 13:03 | |
| *** moshele has joined #openstack-nova | 13:04 | |
| sdague | oh, hey, new time. I'll have to remember that one | 13:05 |
| leakypipes | sdague: mornin. | 13:05 |
| sdague | moring | 13:05 |
| sdague | gilllliard: the google calendar ics is updated by ttx manually based on the wiki | 13:05 |
| *** alaski is now known as lascii | 13:06 | |
| gilllliard | sdague: Thanks | 13:06 |
| * gilllliard goes to double-check the wiki | 13:06 | |
| *** READ10 has joined #openstack-nova | 13:06 | |
| *** vladikr has joined #openstack-nova | 13:06 | |
| *** iamjarvo_ has joined #openstack-nova | 13:07 | |
| *** oomichi has quit IRC | 13:07 | |
| *** burt has joined #openstack-nova | 13:10 | |
| *** artom has joined #openstack-nova | 13:11 | |
| *** iamjarvo_ has quit IRC | 13:11 | |
| *** unicell1 has joined #openstack-nova | 13:14 | |
| *** edleafe is now known as figleaf | 13:16 | |
| *** unicell has quit IRC | 13:16 | |
| lxsli | Dearest cores, I added 20 trivial bugs here: https://etherpad.openstack.org/p/liberty-nova-priorities-tracking | 13:16 |
| *** eliqiao1 has quit IRC | 13:17 | |
| *** dboik has joined #openstack-nova | 13:18 | |
| *** dboik has quit IRC | 13:18 | |
| *** dboik has joined #openstack-nova | 13:19 | |
| *** irenab has joined #openstack-nova | 13:19 | |
| *** neelashah has joined #openstack-nova | 13:19 | |
| *** sdake has joined #openstack-nova | 13:24 | |
| *** irenab has quit IRC | 13:24 | |
| *** VW_ has joined #openstack-nova | 13:25 | |
| figleaf | leakypipes: I enjoyed that, but my life drifts on unchanged. | 13:26 |
| dansmith | jogo: oh, hmm | 13:28 |
| *** sdake has quit IRC | 13:29 | |
| *** joefides_ has quit IRC | 13:29 | |
| *** joefides has joined #openstack-nova | 13:29 | |
| *** busterswt has joined #openstack-nova | 13:29 | |
| *** dims has joined #openstack-nova | 13:30 | |
| *** kaufer has quit IRC | 13:31 | |
| johnthetubaguy | jogo: dansmith: as promised, been attempting to document what we don't want nova to do here, reviews welcome: https://review.openstack.org/#/c/178623/ | 13:32 |
| dansmith | johnthetubaguy: cool | 13:33 |
| dansmith | johnthetubaguy: on the api stuff I just missed, the reason we dropped v3 was that it wasn't worth making most of the changes, which were just cleanliness | 13:34 |
| dansmith | johnthetubaguy: the worst offenders were the renames, capitalization changes, - to _, etc | 13:34 |
| dansmith | johnthetubaguy: I understand leakypipes' argument, and think that some such changes are definitely worth it | 13:35 |
| dansmith | johnthetubaguy: this lock one doesn't seem like it is to me, based on the behavior and what a client would really be doing in either case | 13:35 |
| dansmith | but that's all, just opinion | 13:35 |
| *** cfriesen has joined #openstack-nova | 13:36 | |
| johnthetubaguy | dansmith: I am with you there | 13:37 |
| johnthetubaguy | dansmith: although if we agree a standard that says we should do 409s like that, then I wouldn't stop people adding that I guess, assuming they did it everywhere for consistency, but I do have a slight preference for the current approach myself | 13:38 |
| *** packet has joined #openstack-nova | 13:39 | |
| * gilllliard just writing an api-wg patch for the 409 case. | 13:39 | |
| johnthetubaguy | dansmith: I mean if you snapshot a deleted instance, thats a 409, and indeed snapshot a snapshotting instance is also 409, but for lock, I duno, seems like locking a locked being a success is useful | 13:39 |
| gilllliard | I would also prefer locking a locked instance to be a success. You asked for it to be locked and it is locked. | 13:39 |
| *** arnaud____ has joined #openstack-nova | 13:40 | |
| dansmith | johnthetubaguy: it seems idempotent to me, yeah | 13:40 |
| dansmith | the question is whether locking an instance that was locked by an admin needs to be a 409 | 13:40 |
| johnthetubaguy | gilllliard: right, its borderline I guess, but I kinda prefer the current behaviour | 13:40 |
| dansmith | and to me, if I want it locked, I just want to know that it's locked | 13:40 |
| dansmith | I might care who locked it when I unlock, but that's different | 13:40 |
| dansmith | and I get an indication of that on unlock | 13:41 |
| johnthetubaguy | dansmith: yeah | 13:41 |
| johnthetubaguy | dansmith: unlocking and already unlocked thing, meh, similar really | 13:41 |
| dansmith | yeah | 13:41 |
| dansmith | so again, this goes to "is it worth it" IMHO, and the behavior here might not be ideal, | 13:42 |
| johnthetubaguy | dansmith: frankly my biggest issue is spending our time debating this | 13:42 |
| johnthetubaguy | right | 13:42 |
| dansmith | but changing it for people already used to one just doesn't seem like a huge deal | 13:42 |
| dansmith | johnthetubaguy: yep, I'm happy to drop it | 13:42 |
| *** zzzeek has quit IRC | 13:42 | |
| dansmith | just 'splainin' my feelins' :) | 13:42 |
| johnthetubaguy | dansmith: yeah, sorry I wasn't really meaning that | 13:43 |
| dansmith | suuuure :P | 13:43 |
| johnthetubaguy | dansmith: I mean its more, I don't want to spend the whole of liberty discuss small API tweaks :) | 13:43 |
| dansmith | heh, yeah | 13:43 |
| *** kaufer has joined #openstack-nova | 13:44 | |
| *** arnaud____ has quit IRC | 13:45 | |
| *** lpetrut has joined #openstack-nova | 13:48 | |
| *** packet has quit IRC | 13:48 | |
| *** thangp has joined #openstack-nova | 13:49 | |
| *** packet has joined #openstack-nova | 13:49 | |
| *** openstackgerrit has quit IRC | 13:51 | |
| *** openstackgerrit has joined #openstack-nova | 13:52 | |
| leakypipes | figleaf: yes, I had a chuckle. and then realized how pointless laughing is, and decided to stop. | 13:55 |
| *** dansmith is now known as superdan | 13:57 | |
| *** dboik_ has joined #openstack-nova | 13:59 | |
| openstackgerrit | Sean Dague proposed openstack/nova: Add ability to inject routes in interfaces.template https://review.openstack.org/115409 | 13:59 |
| openstackgerrit | Dan Smith proposed openstack/nova-specs: Add admin-query-any-keypair.rst https://review.openstack.org/175579 | 13:59 |
| mtreinish | jogo: +A | 14:01 |
| openstackgerrit | Baodong (Robert) Li proposed openstack/nova: Handle port delete event https://review.openstack.org/179390 | 14:02 |
| *** dboik has quit IRC | 14:03 | |
| *** iamjarvo has joined #openstack-nova | 14:04 | |
| *** iamjarvo has quit IRC | 14:05 | |
| *** iamjarvo has joined #openstack-nova | 14:05 | |
| *** lpetrut has quit IRC | 14:06 | |
| superdan | leakypipes: gilllliard: do the two changes proposed against the api-wg repo actually apply to what we have here? | 14:07 |
| *** Sukhdev has quit IRC | 14:07 | |
| superdan | or rather, the 409 clarification | 14:07 |
| leakypipes | superdan: just that one, not the 500 one | 14:07 |
| superdan | the 200 one does and I think we're all on board with that | 14:07 |
| superdan | leakypipes: the 500 on/ | 14:08 |
| superdan | I see one about 200 and one about 409 | 14:08 |
| superdan | https://review.openstack.org/#/c/179386/2/guidelines/http.rst,cm | 14:08 |
| *** ndipanov is now known as ndipanoff | 14:08 | |
| leakypipes | superdan: the 500 one is a dep of the 409.. | 14:08 |
| *** mriedem_away is now known as mriedem | 14:08 | |
| superdan | oh okay | 14:08 |
| leakypipes | superdan: but it does not need to be. | 14:08 |
| leakypipes | I blame gilllliard :) | 14:09 |
| superdan | so the 409 one doesn't seem to apply to the locking situation, or is too vaguely related | 14:09 |
| superdan | because reading that, I'd still think the current behavior is right | 14:09 |
| leakypipes | superdan: the current behaviour returns a 404 not a 409. | 14:09 |
| superdan | leakypipes: I don't think it does, did you see my comments? | 14:10 |
| * leakypipes goes back for the fifth time... | 14:10 | |
| superdan | sorry man, feel free to ignore me | 14:10 |
| *** markvoelker has quit IRC | 14:10 | |
| leakypipes | superdan: no, 99% of the time you are right and I am wrong. | 14:10 |
| superdan | reading the code, I don't see how it can yield a 404 | 14:10 |
| superdan | Eli's analysis after that comment of mine seems to indicate that it's not a 404, if I'm reading the broke-ass gerrit formatting right | 14:11 |
| *** oro has quit IRC | 14:12 | |
| superdan | I mean, 404 if the instance is deleted, but that's correct I think | 14:12 |
| leakypipes | superdan: are you looking at the v3 plugin or the v2 contrib/admin_actions.py module? | 14:12 |
| superdan | leakypipes: it's mostly compute/api that matters I think, but let me pull it back up | 14:13 |
| superdan | I'm sure I was looking at the v3 plugin because v2 is frozen and irrelevant now :) | 14:13 |
| leakypipes | superdan, eliqiao: well, this: https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/contrib/admin_actions.py#L215 is in direct violation of this: http://git.openstack.org/cgit/openstack/api-wg/tree/guidelines/http.rst#n47 | 14:13 |
| superdan | yeah, v3/plugins/lock_server.py | 14:14 |
| *** markvoelker_ has joined #openstack-nova | 14:14 | |
| superdan | leakypipes: but that code is frozen, not changing, slated for deletion | 14:14 |
| superdan | and not what this change is talking about at all | 14:14 |
| leakypipes | superdan: even though it correctly returns a 202 Accepted :( | 14:14 |
| superdan | leakypipes: it's going to be deleted! :) | 14:14 |
| superdan | the 202->200 change is fine with me, I'm talking about the 409 here | 14:15 |
| leakypipes | ya | 14:15 |
| sdague | leakypipes: so, someone is going to get rid of that 405 thing right? | 14:15 |
| *** oro has joined #openstack-nova | 14:15 | |
| leakypipes | sdague: apparently that code is being deleted? | 14:15 |
| *** markvoel_ has joined #openstack-nova | 14:15 | |
| sdague | well, it's the same wrong reading of "method" in an http context | 14:15 |
| sdague | I was just looking at the http guidline | 14:15 |
| leakypipes | superdan: k, looking at compute api now... sorry for distraction | 14:15 |
| superdan | https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/plugins/v3/lock_server.py#L34 | 14:16 |
| superdan | returns 202 right now, should be 200, totally agreed | 14:16 |
| superdan | I don't think we need a microversion to make that change | 14:16 |
| sdague | leakypipes: if there isn't a patch up, I can post one | 14:16 |
| leakypipes | superdan: I disagree. it's a change that is backwards incompatible. plus, microversion bumps are cheap now. | 14:17 |
| johnthetubaguy | superdan: mind you, micro versions are almost free, so we could bump for that, if we wanted I guess? Although I suppose that means keeping all old requests as 202, which is a bit silly, there is devref patch to review about this: https://review.openstack.org/#/c/177778/ | 14:18 |
| leakypipes | superdan: w.r.t. the 409, that would be an additional change that could be done in the same microversion bump, IMO | 14:18 |
| leakypipes | Changing 202 -> 200 is a backwards incompatible change. | 14:18 |
| superdan | leakypipes: why? | 14:19 |
| *** signed8bit has joined #openstack-nova | 14:19 | |
| *** markvoelker_ has quit IRC | 14:19 | |
| leakypipes | That's not the same thing as fixing something that erroneously returns a 500 to be the correct response code. | 14:19 |
| superdan | leakypipes: I think that we specifically said that such things don't need microversions | 14:19 |
| superdan | https://github.com/openstack/nova/blob/master/doc/source/devref/api_microversions.rst | 14:19 |
| superdan | leakypipes: we *could* make the 409 change, but I don't think we should | 14:19 |
| leakypipes | superdan: because clients check for the 202 now. if that changes, with NO version bump, then clients will not view the 200 as a successful return. | 14:19 |
| superdan | leakypipes: when/where do we return a 500? | 14:19 |
| superdan | leakypipes: when we were determining what needed to be versioned, I feel pretty confident we said that things like 200->202 were standard http behaviors and if you consider 200 successful and 202 failure (or the other way around) then your client is broken | 14:20 |
| leakypipes | superdan: sorry, the 500 is a separate issue that came up this morning at the Nova API meeting. I was explaining that there is a difference between not needing to bump the microversion if fixing a bug where the server is returning a 500 erroneously due to a user-initiated action. | 14:20 |
| sdague | oh, never mind, the 405 thing is right there | 14:21 |
| sdague | I was reading wrong | 14:21 |
| superdan | leakypipes: that's a api-wg change, nothing about this lock code, right? | 14:21 |
| leakypipes | superdan: I know that's what was in the API Change Guidelines. But that is wrong, IMO | 14:21 |
| leakypipes | superdan: it's a Nova bug and an API WG change (the 500 thing), yes. unrelated to this bug (the 409/202 thing) | 14:21 |
| superdan | well, fine, allocate a microversion for the 200 change then | 14:22 |
| leakypipes | superdan: I bring the 500 thing up because it's the only occasion where I believe it is fine to change the return code without doing a bump. | 14:22 |
| *** claudiub has joined #openstack-nova | 14:22 | |
| superdan | so, aren't we free to do or queue anything we want? | 14:23 |
| sdague | leakypipes: so... we've got that API compat guideline wiki page | 14:23 |
| superdan | like, if we can't do your thing right away (200) can't we queue (202) it at our discretion? | 14:23 |
| leakypipes | superdan: I was advocating doing *both* the 200 and 409 changes in the same microversion bump because they are in the same API call... | 14:23 |
| leakypipes | sdague: that part is and always has been wrong, IMO. | 14:23 |
| sdague | is that going to come back into the api-wg repo? | 14:23 |
| superdan | leakypipes: right, but I think the 409 change is not necessary, and TBH, it goes against the 200 advice in that api-wg change | 14:23 |
| superdan | leakypipes: which says if it's already locked, return 2xx | 14:24 |
| leakypipes | sdague: about changing success codes is a "backwards compatible" thing. It's definitely not. | 14:24 |
| sdague | leakypipes: hey, no complaints from me | 14:24 |
| sdague | I was against that being an allowed change, tempest actually trips up on it | 14:24 |
| sdague | because we check for specific codes that are in the docs | 14:24 |
| leakypipes | superdan: Dan... | 14:25 |
| superdan | sdague: so you know that means that if we have to change something from being sync to async in the code, we can't honor our own API anymore, right? | 14:25 |
| sdague | but... it seems like that change guideline should be put in the api-wg repo instead of random wiki | 14:25 |
| leakypipes | superdan: it does not say 2XX. It says 200. https://review.openstack.org/#/c/179386/2/guidelines/http.rst | 14:25 |
| *** ferest has quit IRC | 14:25 | |
| superdan | leakypipes: Jay, I know man, I'm saying 2xx instead of 409 | 14:25 |
| sdague | superdan: if we've changed the behavior of an endpoint from sync to async, I think that's an API change | 14:25 |
| *** ferest has joined #openstack-nova | 14:26 | |
| sdague | because it's definitely going to trip people up with existing code that assumes things are good | 14:26 |
| sdague | and moves onto the next action | 14:26 |
| leakypipes | superdan: ah, sorry, yes, I see what you're saying now about the conflicting bullet in https://review.openstack.org/#/c/163275/5/specs/liberty/approved/improve-lock-api.rst, line 19 | 14:26 |
| sdague | when they are now not | 14:26 |
| *** BadCub_Hiding is now known as BadCub | 14:26 | |
| superdan | sdague: well, I get that, but it seems like the kind of thing that has to apply within context | 14:27 |
| leakypipes | superdan: w.r.t. "so you know that means that if we have to change something from being sync to async in the code, we can't honor our own API anymore, right?" <-- yes, as sdague says, a semantic behavior change like that *should* trigger a bump, IMO | 14:27 |
| superdan | sdague: if we return 200, people don't assume the instance is booted, they assume it's recorded.. if we return 202, they need to maybe check for it again if they care | 14:28 |
| *** pkoniszewski has joined #openstack-nova | 14:28 | |
| superdan | whatever, I really REALLY don't care about the 200 thing here, I care about the 409 thing | 14:28 |
| superdan | the thing I'm trying to get at with the 200 thing I guess is that if they request v2.1.0 we're not going to make the locking async, so we're just lying | 14:30 |
| *** ferest has quit IRC | 14:30 | |
| superdan | but if the microversion is the doc penalty we have to take, then that's fine | 14:30 |
| *** zz_jgrimm is now known as jgrimm | 14:31 | |
| *** oro has quit IRC | 14:31 | |
| johnthetubaguy | I think we should all review this one: https://review.openstack.org/#/c/177778/ | 14:31 |
| johnthetubaguy | if its the wrong place, lets move it, but I think we need to write this down | 14:31 |
| leakypipes | superdan: I added a comment about https://review.openstack.org/#/c/163275/ being non-aligned with the guidance on 409 in https://review.openstack.org/#/c/179386/2/guidelines/http.rst | 14:32 |
| johnthetubaguy | open to other ideas, as always | 14:32 |
| superdan | leakypipes: thanks, I'll STFU now | 14:32 |
| leakypipes | superdan: no, as I mentioned above, you are right 99% of the time and I am wrong. as was mostly the case here. | 14:32 |
| *** oro has joined #openstack-nova | 14:33 | |
| * leakypipes self-removes from nova-core and api-wg | 14:33 | |
| superdan | pfft | 14:33 |
| * leakypipes soothes superdan with soothing voice and demeanor. | 14:34 | |
| sdague | so, I think the interesting point that superdan raises is about changing the v2.1 behavior inconsistently before this bump | 14:34 |
| leakypipes | now now superdan | 14:34 |
| * superdan soothes self by smashing things | 14:34 | |
| leakypipes | rofl | 14:34 |
| sdague | it's a code green! | 14:34 |
| * sdague watched new avengers last night | 14:35 | |
| leakypipes | sdague: any good? | 14:35 |
| sdague | if you are into marvel super hero movies, yes | 14:36 |
| leakypipes | sdague: I am, so good :) | 14:36 |
| *** achanda has joined #openstack-nova | 14:37 | |
| *** cfriesen is now known as cfriesen_away | 14:41 | |
| *** joefides_ has joined #openstack-nova | 14:42 | |
| *** achanda has quit IRC | 14:43 | |
| *** joefides_ has quit IRC | 14:45 | |
| *** otter768 has joined #openstack-nova | 14:45 | |
| *** joefides has quit IRC | 14:45 | |
| *** otter768 has quit IRC | 14:50 | |
| *** tonytan4ever has joined #openstack-nova | 14:51 | |
| *** cbader has joined #openstack-nova | 14:53 | |
| superdan | johnthetubaguy: I was extra picky about grammar in that doc, just because I think it's likely to be read by a lot of folks, just FYI :) | 14:57 |
| dims | johnthetubaguy: can you please restore the docker spec? https://review.openstack.org/#/c/128753/ when you get a chance | 14:57 |
| *** dims is now known as dimsum__ | 14:58 | |
| openstackgerrit | George Shuklin proposed openstack/nova: If rescue failed set instance to ERROR If rescue operation failed in the middle set to ERROR state instead leave it broken and 'ACTIVE'. https://review.openstack.org/155120 | 14:58 |
| openstackgerrit | George Shuklin proposed openstack/nova: If rescue failed set instance to ERROR https://review.openstack.org/155120 | 14:59 |
| johnthetubaguy | dims: submitter should be able to restore, but I can do that if you want to update it :) | 14:59 |
| dimsum__ | johnthetubaguy: yes, please | 14:59 |
| dimsum__ | i have a change ready to go | 15:00 |
| johnthetubaguy | dimsum__: I am hungry now I saw your nick | 15:00 |
| dimsum__ | haha | 15:00 |
| *** sdake has joined #openstack-nova | 15:01 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova-specs: Add the docker hypervisor plugin to Nova https://review.openstack.org/128753 | 15:01 |
| johnthetubaguy | dimsum__: how are the tests going now? | 15:02 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova-specs: Add the docker hypervisor plugin to Nova https://review.openstack.org/128753 | 15:03 |
| johnthetubaguy | superdan: my grammar is terrible, appreciate the help with that! I will take a look at how thats going | 15:03 |
| superdan | cool | 15:04 |
| dimsum__ | johnthetubaguy: warning - docker ci jobs don't run on nova changes. so just for the nova-docker reviews, the jobs seem to be fine | 15:04 |
| mriedem | superdan: you might enjoy this https://review.openstack.org/#/c/179406/ | 15:04 |
| *** dboik_ has quit IRC | 15:05 | |
| dimsum__ | johnthetubaguy: occasional hiccups with the CI changes, but nothing out of the ordinary | 15:05 |
| leakypipes | sdague: how do I see who's on the horizon core team? | 15:05 |
| superdan | mriedem: hah, nice | 15:05 |
| *** dboik has joined #openstack-nova | 15:05 | |
| dimsum__ | leakypipes: https://review.openstack.org/#/admin/groups/43,members | 15:05 |
| *** emagana has joined #openstack-nova | 15:05 | |
| leakypipes | thx dimsum__ | 15:06 |
| johnthetubaguy | dimsum__: can we try running them on all Nova jobs for a bit, non voting I guess, just to show stability? maybe thats silly, not sure | 15:06 |
| mriedem | leakypipes: do you need a horizon person to yell at? | 15:06 |
| mriedem | i work with one | 15:06 |
| dimsum__ | johnthetubaguy: i'd be very happy to | 15:06 |
| leakypipes | mriedem: no, just wondereing. | 15:06 |
| superdan | sigh, are we really doing the docker thing again? | 15:06 |
| mriedem | tell nova-docker to get their own translation domain and i18n module | 15:07 |
| dimsum__ | mriedem: not needed if folded back into nova. right? | 15:07 |
| *** irenab has joined #openstack-nova | 15:08 | |
| figleaf | mriedem: if your sensibilities are really that delicate, you should quote the id string in that line. | 15:08 |
| *** nelsnelson has joined #openstack-nova | 15:08 | |
| * figleaf has even more delicate grammar sensibilities | 15:08 | |
| *** emagana has quit IRC | 15:09 | |
| *** hemna_ has joined #openstack-nova | 15:12 | |
| *** irenab has quit IRC | 15:12 | |
| johnthetubaguy | leakypipes: didn't see we had code up for reivew on this, can we get this BP reapproved for you? https://blueprints.launchpad.net/nova/+spec/lock-free-quota-management | 15:12 |
| openstackgerrit | George Shuklin proposed openstack/nova: If rescue failed set instance to ERROR https://review.openstack.org/155120 | 15:14 |
| leakypipes | johnthetubaguy: I'm afraid that particular work won't be possible to complete in liberty. I still need to do benchmarking of it, and the colleague who was assigned it has been doing other things... | 15:14 |
| *** moshele has quit IRC | 15:16 | |
| johnthetubaguy | leakypipes: ah, OK, we have a summit session scheduler to discuss pushing this forward at the moment, maybe its time to get more help, if that works? Although given we have such a concrete plan, its maybe not worth a summit session | 15:16 |
| johnthetubaguy | leakypipes: well, I guess it was only ever a side note in that session, but anyways | 15:17 |
| leakypipes | johnthetubaguy: yeah, I don't recommend taking up a sesion on it. | 15:18 |
| *** READ10 has quit IRC | 15:19 | |
| johnthetubaguy | leakypipes: cool, it was a quota next steps and database next steps rolled into one, I need to go back and make sure I know whats left is worth bothering about! | 15:20 |
| *** dprince has quit IRC | 15:20 | |
| *** yamahata has quit IRC | 15:24 | |
| *** pkoniszewski has quit IRC | 15:27 | |
| *** arnaud____ has joined #openstack-nova | 15:28 | |
| *** jbernard has quit IRC | 15:29 | |
| *** dimsum__ has quit IRC | 15:31 | |
| *** jbernard has joined #openstack-nova | 15:31 | |
| *** dimsum__ has joined #openstack-nova | 15:31 | |
| *** dsanders has joined #openstack-nova | 15:31 | |
| *** arnaud____ has quit IRC | 15:33 | |
| *** hemna_ has quit IRC | 15:36 | |
| *** iamjarvo has quit IRC | 15:38 | |
| *** jbernard has quit IRC | 15:39 | |
| *** hemna_ has joined #openstack-nova | 15:40 | |
| *** jbernard has joined #openstack-nova | 15:42 | |
| *** jogo is now known as flashgordon | 15:45 | |
| *** nelsnels_ has joined #openstack-nova | 15:47 | |
| *** yamahata has joined #openstack-nova | 15:47 | |
| *** nelsnelson has quit IRC | 15:47 | |
| *** nelsnels_ has quit IRC | 15:47 | |
| *** nelsnelson has joined #openstack-nova | 15:48 | |
| *** bnemec has quit IRC | 15:48 | |
| *** iamjarvo has joined #openstack-nova | 15:48 | |
| *** xyang1 has quit IRC | 15:49 | |
| *** bnemec has joined #openstack-nova | 15:51 | |
| *** ferest has joined #openstack-nova | 15:52 | |
| *** tonytan4ever has quit IRC | 15:53 | |
| *** dsanders has quit IRC | 15:57 | |
| *** zzzeek has joined #openstack-nova | 15:58 | |
| *** rfolco has quit IRC | 16:00 | |
| *** angdraug has joined #openstack-nova | 16:01 | |
| *** Nic has joined #openstack-nova | 16:01 | |
| *** annegentle has quit IRC | 16:04 | |
| *** browne has joined #openstack-nova | 16:04 | |
| *** busterswt has quit IRC | 16:04 | |
| *** jlanoux has quit IRC | 16:05 | |
| *** arnaud____ has joined #openstack-nova | 16:09 | |
| *** Marga_ has joined #openstack-nova | 16:09 | |
| *** busterswt has joined #openstack-nova | 16:09 | |
| *** dave-mccowan has quit IRC | 16:13 | |
| *** angdraug has quit IRC | 16:15 | |
| *** unicell1 has quit IRC | 16:16 | |
| *** IanGovett1 has joined #openstack-nova | 16:19 | |
| *** IanGovett has quit IRC | 16:21 | |
| *** Marga_ has quit IRC | 16:22 | |
| *** Marga_ has joined #openstack-nova | 16:22 | |
| *** iamjarvo has quit IRC | 16:23 | |
| *** hemna_ has quit IRC | 16:23 | |
| *** claudiub has quit IRC | 16:24 | |
| *** hemna_ has joined #openstack-nova | 16:24 | |
| *** gokrokve has joined #openstack-nova | 16:25 | |
| *** arnaud____ has quit IRC | 16:26 | |
| *** gokrokve has quit IRC | 16:26 | |
| *** gokrokve has joined #openstack-nova | 16:27 | |
| *** derekh has quit IRC | 16:28 | |
| *** tjones2 has joined #openstack-nova | 16:31 | |
| *** hemna_ has quit IRC | 16:33 | |
| *** Nic has quit IRC | 16:34 | |
| *** hemna_ has joined #openstack-nova | 16:34 | |
| *** busterswt has quit IRC | 16:36 | |
| *** Nic has joined #openstack-nova | 16:36 | |
| *** marun has joined #openstack-nova | 16:37 | |
| *** unicell has joined #openstack-nova | 16:41 | |
| *** melwitt has joined #openstack-nova | 16:41 | |
| superdan | melwitt: mind leaving a few +1s for the rest of us? kthx | 16:43 |
| *** ajo has joined #openstack-nova | 16:45 | |
| melwitt | superdan: ... :) | 16:45 |
| superdan | heh | 16:45 |
| *** otter768 has joined #openstack-nova | 16:46 | |
| *** iamjarvo has joined #openstack-nova | 16:47 | |
| *** alanf-mc has joined #openstack-nova | 16:48 | |
| *** markmcclain has quit IRC | 16:49 | |
| jroll | I'm trying to get network info in a virt driver method, what's the best way to do that, just instantiate network.API() and use it? | 16:51 |
| *** otter768 has quit IRC | 16:51 | |
| *** gokrokve has quit IRC | 16:51 | |
| superdan | jroll: for an instance? | 16:51 |
| jroll | superdan: yes | 16:52 |
| superdan | instance.info_cache.network_info should have what you want | 16:52 |
| jroll | ah, perfect | 16:52 |
| superdan | you definitely should not instantiate a network.API :) | 16:52 |
| *** dave-mccowan has joined #openstack-nova | 16:52 | |
| jroll | I should have known that, thank you :) | 16:52 |
| *** exploreshaifali has joined #openstack-nova | 16:52 | |
| superdan | np | 16:52 |
| *** HenryG is now known as floccinaucinihil | 16:54 | |
| *** lucasagomes is now known as lucas-beer | 16:54 | |
| *** ssurana has joined #openstack-nova | 16:55 | |
| *** patrickeast has joined #openstack-nova | 16:55 | |
| *** floccinaucinihil is now known as HenryThe8th | 16:55 | |
| *** jecarey has quit IRC | 16:56 | |
| *** irenab has joined #openstack-nova | 16:57 | |
| *** emagana has joined #openstack-nova | 16:59 | |
| *** vilobhmm1 has joined #openstack-nova | 16:59 | |
| *** markvoel_ has quit IRC | 16:59 | |
| *** vilobhmm1 has quit IRC | 16:59 | |
| *** vilobhmm1 has joined #openstack-nova | 17:00 | |
| *** markvoelker has joined #openstack-nova | 17:00 | |
| *** irenab has quit IRC | 17:01 | |
| *** harlowja has joined #openstack-nova | 17:02 | |
| *** cfriesen_away is now known as cfriesen | 17:03 | |
| *** markmcclain has joined #openstack-nova | 17:03 | |
| *** annegentle has joined #openstack-nova | 17:04 | |
| *** josecastroleon has joined #openstack-nova | 17:06 | |
| *** Marga_ has quit IRC | 17:07 | |
| *** achanda has joined #openstack-nova | 17:07 | |
| *** Marga_ has joined #openstack-nova | 17:07 | |
| *** salv-orlando has quit IRC | 17:10 | |
| *** josecastroleon has quit IRC | 17:12 | |
| *** Marga_ has quit IRC | 17:13 | |
| *** Marga_ has joined #openstack-nova | 17:13 | |
| *** dsanders has joined #openstack-nova | 17:16 | |
| *** gokrokve has joined #openstack-nova | 17:18 | |
| *** claudiub has joined #openstack-nova | 17:19 | |
| *** gokrokve_ has joined #openstack-nova | 17:20 | |
| *** subscope has joined #openstack-nova | 17:21 | |
| *** gokrokve has quit IRC | 17:22 | |
| *** gokrokv__ has joined #openstack-nova | 17:22 | |
| *** gokrokve_ has quit IRC | 17:22 | |
| johnthetubaguy | sdague: lascii and I just had a good conversation with Vek about quotas, I think we identified quite a few things we can do to simplify things, quite exciting, will can roll that into the summit session we have scheduled on that | 17:22 |
| *** busterswt has joined #openstack-nova | 17:23 | |
| flashgordon | johnthetubaguy: woot any solution that involves simplifying quotas sounds like a step in the right direction | 17:25 |
| johnthetubaguy | flashgordon: :) | 17:25 |
| flashgordon | johnthetubaguy: for too long the default reaction was: they don't work lets complicate things | 17:26 |
| *** welldannit has quit IRC | 17:27 | |
| vilobhmm1 | johnthetubaguy : ping | 17:28 |
| vilobhmm1 | hey john | 17:28 |
| johnthetubaguy | vilobhmm1: hi | 17:28 |
| vilobhmm1 | must be late at your end will you be around for some time ? | 17:28 |
| * jlk is back again to talk resize | 17:28 | |
| johnthetubaguy | vilobhmm1: its half six, just finishin up, but happy to help if its quick | 17:29 |
| vilobhmm1 | johnthetubaguy : so regarding https://review.openstack.org/#/c/138607/ | 17:29 |
| johnthetubaguy | jlk: ah, resize, I like all things resize | 17:29 |
| jlk | so a resize scenario. I've issued a resize to take a server from flavor 2 to 3. It's now in VERIFY_RESIZE, should the IP of the instance be reachable at that point? | 17:29 |
| vilobhmm1 | you wanted a precsise plan | 17:29 |
| jlk | it was reachable before the resize attempt | 17:29 |
| vilobhmm1 | so here i what i am experimenting with | 17:29 |
| superdan | jlk: yes | 17:29 |
| jlk | well.... poo. | 17:29 |
| vilobhmm1 | 1. For non-db drivers zc.py : change tooz to make sure the data format understood by nova servicegroup driver and tooz zk driver are same for seemless migration | 17:30 |
| jlk | it's not reachable while in verify, but is reachable once again after revert. | 17:30 |
| johnthetubaguy | vilobhmm1: hi, really just need pick one of the options you already described, it seemed to have most things covered | 17:30 |
| dimsum__ | johnthetubaguy: re: quotas, we should ping Salvatore (https://review.openstack.org/#/c/132127/1/specs/kilo/review_quota_module.rst,unified) | 17:30 |
| *** thangp has quit IRC | 17:31 | |
| vilobhmm1 | 2. For mc.py…its tough as it keeps info at 2 places heartbeat in memcache whereas services info in mysql db…so may be come with a migration script to move service info from mysql db to memcache to make sure that tooz mc can start using it | 17:31 |
| *** marun has quit IRC | 17:32 | |
| johnthetubaguy | dimsum__: ah, I wondered who was driving that effort, thanks for the heads up | 17:32 |
| *** marun has joined #openstack-nova | 17:34 | |
| vilobhmm1 | johnthetubaguy : still need to unify the admin interface and the API interface as the admin interface still acts on the db (services table) | 17:34 |
| johnthetubaguy | vilobhmm1: disabled is different to up vs down, we might not be too bad in there, but yeah, need to check | 17:39 |
| johnthetubaguy | vilobhmm1: I like the approaches though, but honestly it might be overkill given the state of the current drivers, but I think its worth keeping that plan listed as a possible extension idea to help us remove the old drivers quicker | 17:40 |
| *** burt has quit IRC | 17:40 | |
| johnthetubaguy | vilobhmm1: thank you for pushing on this by the way, I think its a good idea, and I support it | 17:40 |
| johnthetubaguy | vilobhmm1: the other bit is just making it clear we keep the db driver when reading through the spec | 17:41 |
| vilobhmm1 | johnthetubaguy : thank you ! given the response on the mailing list looks like no one is using mc,zk driver but yeah can't completely rely on it | 17:41 |
| vilobhmm1 | so will have a migration plan in place | 17:42 |
| vilobhmm1 | but again will verify the admin path | 17:42 |
| vilobhmm1 | bcz its critical for making sure that mc and zk driver function properly | 17:42 |
| johnthetubaguy | vilobhmm1: yeah, lets document a plan for a migration, I mean just have a non-live upgrade work might be good enough for now, then lets get operator feedback on that to see if that good enough | 17:42 |
| vilobhmm1 | dont want to end up in a situation where we have 2 sources of truth | 17:42 |
| jlk | oh, no, it does eventually ping, just took a while for ARP to clear out I guess | 17:43 |
| johnthetubaguy | vilobhmm1: totally agreed | 17:43 |
| johnthetubaguy | vilobhmm1: lets keep it simple (switch over from one to the other across a global service restart) | 17:43 |
| vilobhmm1 | johnthetubaguy : i have proposed a session on this in liberty-nova-design session….will it be a good idea to get feedback in the summit on this topic let me know what you and other cores think about it | 17:43 |
| johnthetubaguy | vilobhmm1: then see how it goes after that | 17:43 |
| vilobhmm1 | so that we have a concrete plan | 17:44 |
| vilobhmm1 | sure | 17:44 |
| vilobhmm1 | ok | 17:44 |
| johnthetubaguy | vilobhmm1: cools, we can give you a 10 min unconference if we don't get the spec merged, does that sound OK? | 17:44 |
| *** dprince has joined #openstack-nova | 17:44 | |
| vilobhmm1 | unconference ? what does that mean ? | 17:44 |
| johnthetubaguy | vilobhmm1: but please remind me I promised you that while I can still follow through on that :) | 17:44 |
| johnthetubaguy | vilobhmm1: it just means we have an etherpad of 10 min slots really, let me get you the link | 17:45 |
| johnthetubaguy | vilobhmm1: http://libertydesignsummit.sched.org/event/a065a336855290d06a8949bac5d9f8be | 17:45 |
| harlowja | sup | 17:45 |
| johnthetubaguy | vilobhmm1: sorry, I have to run how, I hope that helps | 17:45 |
| johnthetubaguy | harlowja: sorry, I have to run, ill be honest, I am going to the cinema and if I don't go seen I don't get to eat desert! so its like really important :) | 17:46 |
| vilobhmm1 | ok…have a good weekend | 17:46 |
| johnthetubaguy | harlowja: I am keen on the hole tooz thing, just need to get the plan clear in the spec, we are almost there | 17:46 |
| johnthetubaguy | vilobhmm1: and you when it happens! | 17:46 |
| *** ajo has quit IRC | 17:47 | |
| superdan | vilobhmm1: unconf sessions are where you can just sign up to talk for ten minutes on a thing | 17:47 |
| superdan | that doesn't require a full session | 17:47 |
| harlowja | johnthetubaguy sweet | 17:47 |
| superdan | and is good for getting people on board with a thing like this | 17:47 |
| harlowja | super vilobhmm1 weill have it under control | 17:47 |
| harlowja | *will | 17:48 |
| jlk | superdan: so I did a manual test of the resize with your patch. Resize up, after a period of time, networking works on the resized instance (on a new host). Revert the resize and almost immediately networking works again back on the original host. | 17:48 |
| superdan | jlk: are you looking for dansmith? he doesn't work on fridays | 17:48 |
| tjones2 | haha | 17:48 |
| jlk | lol | 17:48 |
| *** penick has joined #openstack-nova | 17:49 | |
| jlk | you should change /whois info if you're going to play that game | 17:49 |
| superdan | jlk: after a period of time, networking works on the new host? | 17:49 |
| superdan | heh | 17:49 |
| jlk | superdan: yeah it's not right away. I'm guessing it's an ARP timeout at play | 17:49 |
| superdan | yeah | 17:49 |
| vilobhmm1 | harlowja : lets see..aw…focus back to migration script now … | 17:49 |
| harlowja | k | 17:49 |
| jlk | the vif gets plugged but arping just can't find it from the router network namespace | 17:50 |
| cfriesen | jlk: shouldn't something be doing GARP? | 17:50 |
| superdan | jlk: probably a question for some neutron person that understands your network driver | 17:50 |
| vilobhmm1 | thanks for all the help harlowja. much appreciated | 17:50 |
| jlk | resolves fairly quick. | 17:50 |
| *** annegentle has quit IRC | 17:50 | |
| harlowja | vilobhmm1 and mess around more of that old-zk (run at same time as tooz-new zk) | 17:50 |
| jlk | cfriesen: maybe? I vaguely know what that term means, but this is black neutron magic, so I don't know for sure | 17:50 |
| superdan | cfriesen: with nova-net we do I think, but that's why I'm saying: ask a neutron person | 17:50 |
| vilobhmm1 | harlowja : sure..trying it out | 17:50 |
| superdan | jlk: either poisoning the arp cache, or sending a gratuitous update for the new thing | 17:50 |
| jlk | superdan: I don't have a question actually. I'm providing data, in that I think the patch is good and does the right thing | 17:51 |
| superdan | jlk: i.e. just broadcasting an ARP packet from the new location | 17:51 |
| superdan | jlk: ah, okay | 17:51 |
| jlk | tempest doesn't do anything to verify that the networking is functional, so I did it manually | 17:51 |
| superdan | I see, I thought you were concerned about the delay | 17:51 |
| jlk | nah. frankly, that networking works at all is amazing to me | 17:51 |
| jlk | but I'll save that observation for a later tweet ;) | 17:51 |
| superdan | are you on the same L2 as the instance or going through a switch/router? | 17:52 |
| superdan | heh | 17:52 |
| superdan | because if you are, you're probably seeing a greater delay than would be expected | 17:52 |
| *** ZZelle has quit IRC | 17:52 | |
| jlk | Same L2. I'm arpinging from the controller node where neutron services are running, the router and dhcp agents | 17:52 |
| superdan | general-purpose operating systems have some weird arp caching algorithms I think | 17:52 |
| *** ZZelle has joined #openstack-nova | 17:52 | |
| jlk | I should try giving it a floating IP and see how that all works out | 17:53 |
| superdan | in the neutron case, I'm really not sure who is responsible and for what on this kind of stuff | 17:53 |
| jlk | "it depends" | 17:53 |
| superdan | I would think the OVS could be told what is happening so that there is no delay and no need for gARP | 17:53 |
| jlk | depends on which driver you've thrown your money at. | 17:54 |
| superdan | right | 17:54 |
| superdan | well, I know *that* :) | 17:54 |
| superdan | I figured you were using normal OVS | 17:54 |
| jlk | in this cluster, it's linux-bridge and VLAN | 17:54 |
| superdan | oh, so, | 17:54 |
| superdan | linux-bridge is wholly unsupported by neutron and not recommended, as I understand it | 17:54 |
| superdan | like "don't use it at all" unsupported | 17:54 |
| jlk | the ml2? | 17:54 |
| jlk | I thought that was The Future | 17:54 |
| *** marun has quit IRC | 17:55 | |
| superdan | I think "the future" is ovs, no? | 17:55 |
| jlk | I thought we were all running away from the OVS black box | 17:55 |
| jlk | but maybe I'm wrong. | 17:55 |
| jlk | all our stuff has dropped OVS, we're all ml2 and either vlan or vxlan | 17:55 |
| superdan | I'm clearly showing my ignorance | 17:55 |
| jlk | me too | 17:55 |
| superdan | when people say "linux bridge" in neutron, I think that's usually the bad one I thought | 17:55 |
| cfriesen | jlk: isn't ml2 able to work with either ovs or linuxbridge | 17:55 |
| superdan | but who knows, it's all bad to me :) | 17:55 |
| jlk | cfriesen: yes, that's correct | 17:56 |
| superdan | mestery: can you save us? | 17:56 |
| mriedem | superdan: except it's the default in devstack now | 17:56 |
| superdan | mriedem: for specific reasons right? | 17:57 |
| cfriesen | jlk: I had thought that linuxbridge was being proposed as a simple starting point, but OVS was necessary for "real" stuff | 17:57 |
| mriedem | to get people over from nova-network? | 17:57 |
| jlk | I guess it depends on your definition of "real" | 17:57 |
| * mestery tries to catchup to save you guys | 17:57 | |
| *** isd has joined #openstack-nova | 17:57 | |
| jlk | for giant clusters, maybe. OVS or some other SDN | 17:57 |
| *** irenab has joined #openstack-nova | 17:57 | |
| jlk | for 3~20 node clusters...? | 17:57 |
| superdan | mestery: I thought ml2 + linuxbridge was not recommended for anyone in prod | 17:57 |
| mestery | LB is only unsupported in that nothing we gate on uses it. PEr sc68cal, it does actually work now, last I heard. | 17:57 |
| superdan | okay, I thought it was also a thing you didn't want to support | 17:58 |
| mestery | superdan: That's a pretty valid statement right now, though things may work. We just don't gate on it, so it's current state isn't fully known. | 17:58 |
| *** dprince has quit IRC | 17:58 | |
| mestery | superdan: I think we have to support it, we've heard from enough ops they want to use it. | 17:58 |
| superdan | mestery: okay excellent, I agree with *that* :) | 17:58 |
| mestery | lol | 17:58 |
| superdan | mestery: so is linuxbridge maybe missing some gARP bits to make migrations work seamlessly | 17:59 |
| mestery | superdan: sc68cal has done some work (with others) to revivie support recently | 17:59 |
| superdan | without arp delays? | 17:59 |
| superdan | mestery: excellent | 17:59 |
| mestery | superdan: That's entirally possible, yes. | 17:59 |
| superdan | jlk: ^ | 17:59 |
| mriedem | mestery: so what does this mean? | 17:59 |
| mriedem | http://logs.openstack.org/06/179406/1/check/check-tempest-dsvm-neutron-full/2f7d58a/logs/etc/neutron/plugins/ml2/ml2_conf.ini.txt.gz | 17:59 |
| *** packet has quit IRC | 17:59 | |
| mriedem | mechanism_drivers = openvswitch,linuxbridge | 17:59 |
| *** dprince has joined #openstack-nova | 17:59 | |
| *** dprince has quit IRC | 17:59 | |
| mriedem | firewall_driver = neutron.agent.linux.iptables_firewall.OVSHybridIptablesFirewallDriver | 18:00 |
| mestery | mriedem: That means the ML2 plugin has the logic for LB and OVS enabled | 18:00 |
| mriedem | well ignore firewall | 18:00 |
| * jlk will make it a point to ask all the ops show-and-tell folks to tell us what network driver they're using. | 18:00 | |
| *** dprince has joined #openstack-nova | 18:00 | |
| sc68cal | having both openvswitch and linuxbridge in the mechanism_drivers setting is misleading, as I have learned recently | 18:00 |
| mestery | sc68cal: ++ | 18:00 |
| sc68cal | since you can only run either the OVS agent or LB agent | 18:00 |
| *** kaufer has quit IRC | 18:01 | |
| mestery | sc68cal: Well, per host, but you can run different agents on different hosts with ML2. | 18:01 |
| superdan | sc68cal: if you read back far enough, you'll see a question from jlk that sounds like he's not getting gARP after a migration with linuxbridge | 18:01 |
| superdan | sc68cal: which he seems altogether unfazed about, but thought I'd mention it :) | 18:02 |
| *** irenab has quit IRC | 18:02 | |
| mriedem | mestery: sc68cal: was also wondering how in here http://logs.openstack.org/06/179406/1/check/check-tempest-dsvm-neutron-full/2f7d58a/logs/etc/neutron/neutron.conf.txt.gz - none of the nova_admin* options are set but notify_nova_on_port_status_changes = True and events are showing up in nova-compute, | 18:02 |
| *** busterswt has quit IRC | 18:02 | |
| jlk | heh | 18:02 |
| mriedem | how does that work if neutron doesn't know the credentials to talk to nova api? | 18:02 |
| superdan | yeah, I wanna know that too :) | 18:02 |
| * mestery looks | 18:02 | |
| * sc68cal needs coffee if he's going to solve mysteries | 18:02 | |
| mriedem | scooby snacks? | 18:03 |
| sc68cal | :) | 18:03 |
| jlk | keystone_authtoken | 18:03 |
| jlk | that's where the data is | 18:03 |
| isd | Hey, so I'm a newbie trying to get familiar with the nova codebase, and I was going through the developer guide. The bit about adding a method to the API references something called ApiRouter, but it's not where it says it should be, and grepping the source tree for it only yields the line in that document. I poked through the history, and I'm not sure this was *ever* correct -- even in the commit that added that document, it isn't there. | 18:03 |
| jlk | wait, not quite | 18:03 |
| superdan | jlk: no | 18:03 |
| superdan | jlk: unless something has changed | 18:03 |
| * sc68cal tries to think of the apropriate channel to talk about sane linux bridge defaults | 18:05 | |
| jlk | yeah I'm wrong, it's in the [DEFAULT] block | 18:05 |
| *** busterswt has joined #openstack-nova | 18:05 | |
| *** igordcard has quit IRC | 18:05 | |
| *** ajo has joined #openstack-nova | 18:05 | |
| *** packet has joined #openstack-nova | 18:05 | |
| *** dprince has quit IRC | 18:05 | |
| *** dprince has joined #openstack-nova | 18:06 | |
| *** igordcard has joined #openstack-nova | 18:06 | |
| * mriedem has to reboot | 18:06 | |
| *** mriedem has quit IRC | 18:06 | |
| *** annegentle has joined #openstack-nova | 18:07 | |
| melwitt | isd: there is APIRouter (different capitalization), which might be what it's talking about | 18:07 |
| superdan | lascii: replied | 18:08 |
| *** isd has quit IRC | 18:08 | |
| *** lucas-beer has quit IRC | 18:08 | |
| superdan | lascii: oh damn, that notification got delayed | 18:09 |
| *** hemna_ has quit IRC | 18:09 | |
| superdan | lascii: I was trying to keep up the impression that I had your system rooted.. oh well :) | 18:09 |
| *** sdake_ has joined #openstack-nova | 18:09 | |
| *** marun has joined #openstack-nova | 18:09 | |
| *** hemna_ has joined #openstack-nova | 18:10 | |
| *** annegentle has quit IRC | 18:10 | |
| *** annegentle has joined #openstack-nova | 18:11 | |
| *** salv-orlando has joined #openstack-nova | 18:11 | |
| *** mriedem has joined #openstack-nova | 18:12 | |
| *** sdake has quit IRC | 18:13 | |
| ndipanoff | superdan, bauzas https://review.openstack.org/#/c/163440/ | 18:14 |
| ndipanoff | I think Bart is actually right there | 18:14 |
| ndipanoff | though it's a subtle matter | 18:14 |
| *** neelashah has quit IRC | 18:15 | |
| *** sdake has joined #openstack-nova | 18:15 | |
| *** mriedem has quit IRC | 18:16 | |
| *** mriedem has joined #openstack-nova | 18:16 | |
| superdan | ndipanoff: I don't understand that at all | 18:16 |
| cfriesen | ndipanoff: I work with Bart...we've got a version of what he's proposing working internally, needed it to get CPU pinning working properly over migration/evac/resize/rebuild/etc. | 18:17 |
| superdan | ndipanoff: the state is persisted in the instance object's numa_topology, right? | 18:17 |
| ndipanoff | well the state for the compute host it's currently on | 18:17 |
| ndipanoff | but not the state for the "target" | 18:17 |
| ndipanoff | superdan, in theory this could be persisted on the instance similar to the old_ and new_ flavors | 18:19 |
| superdan | I guess I don't see why they're different | 18:19 |
| *** salv-orlando has quit IRC | 18:19 | |
| *** sdake_ has quit IRC | 18:19 | |
| superdan | ndipanoff: I *hate* putting it on the migration object, just FYI | 18:19 |
| ndipanoff | so it does not _have_ to be on the migration | 18:19 |
| cfriesen | superdan: for pci and CPU pinning, we need to know the exact cpus/devices that were claimed | 18:19 |
| *** dprince_ has joined #openstack-nova | 18:19 | |
| ndipanoff | right | 18:19 |
| ndipanoff | cfriesen, | 18:19 |
| ndipanoff | this is what I tried to explain on the last comment | 18:19 |
| *** dprince has quit IRC | 18:19 | |
| cfriesen | ndipanoff: yeah, just reading that. | 18:20 |
| ndipanoff | where it goes in the DB is a different matter | 18:20 |
| ndipanoff | the point is it has to go somewhere | 18:20 |
| cfriesen | ndipanoff: something similar is needed for live migration too..have to update the xml of the instance being migrated with the new pinning/numa information | 18:20 |
| ndipanoff | cfriesen, right - but that is a bigger problem since we don't even have claims | 18:21 |
| cfriesen | superdan: the nice thing about putting it on the migration object is that it's preserved if we abort the migration | 18:21 |
| superdan | ndipanoff: you mean we have to save both copies just so that if we succeed, we can start reporting that those resources are taken, and if we revert, the same? | 18:21 |
| *** dprince_ has quit IRC | 18:21 | |
| lascii | superdan: hah. you were 18 minutes delayed | 18:22 |
| cfriesen | superdan: the scheduler thinks those resources are taken, so we need to actually use the specific resources that were reserved | 18:22 |
| superdan | lascii: not me, the notification :) | 18:22 |
| lascii | superdan: heh | 18:22 |
| ndipanoff | superdan, once there is a migration going on - we report usage on both hosts | 18:22 |
| *** isd has joined #openstack-nova | 18:23 | |
| ndipanoff | but we need to report _that exact_ usage that was claimed | 18:23 |
| superdan | ndipanoff: so, is this just the cpu_topology element of instance_numa_topology? | 18:23 |
| *** isd has quit IRC | 18:23 | |
| ndipanoff | lemme check, but yes | 18:23 |
| ndipanoff | only the dict that maps to host cpus is what changes | 18:23 |
| *** packet has quit IRC | 18:23 | |
| ndipanoff | weeeelll | 18:23 |
| cfriesen | superdan: ndipanoff: wouldn't pci devices be included as well? | 18:23 |
| ndipanoff | actually we could have a completely different topology | 18:23 |
| ndipanoff | more cells etc | 18:23 |
| ndipanoff | based on the flavor | 18:24 |
| superdan | cfriesen: I'm talking the numa bits now | 18:24 |
| ndipanoff | so if it's migrate then yes | 18:24 |
| ndipanoff | if it's resize - then no | 18:24 |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 18:25 |
| ndipanoff | fwiw - when cfriesen and bart sent the email to the list | 18:25 |
| ndipanoff | I also complained about it being on the migration object/table | 18:25 |
| * ndipanoff read the email again | 18:25 | |
| ndipanoff | superdan, so - no it does not actually have to be on the migration | 18:25 |
| *** bwensley has joined #openstack-nova | 18:27 | |
| superdan | we're talking about some actual stuff here now, so saving it forever on the migration object doesn't seem ideal to me | 18:28 |
| superdan | and, I'll say, keeping two copies at all doesn't seem great either | 18:28 |
| ndipanoff | well at some point we need both of them | 18:29 |
| superdan | it's just frustrating that this stuff has gaps this large :/ | 18:29 |
| cfriesen | superdan: so what's the proposal? where would we keep the information about the old/new numa topology (and pci devices) to allow for reverting a resize/migration | 18:29 |
| ndipanoff | superdan, lol - you haven't read the scheduler code in a while then if you think these are gaps | 18:29 |
| *** neelashah has joined #openstack-nova | 18:29 | |
| ndipanoff | cfriesen, fwiw - pci devices don't have this same problem | 18:30 |
| superdan | cfriesen: we keep those on the instance right now | 18:30 |
| cfriesen | ndipanoff: when you can't migrate a pinned instance, I think that's a gap. :) | 18:30 |
| *** neelashah1 has joined #openstack-nova | 18:30 | |
| ndipanoff | because | 18:30 |
| superdan | cfriesen: it would have been nice if the numa object had an old_set and new_set so that we were set up for this, is kinda what I'm getting at | 18:30 |
| ndipanoff | pci devices are already stateful objects | 18:30 |
| superdan | I think the pci_requests object has a flag for "next" | 18:30 |
| superdan | for this reason, but I could be wrong | 18:31 |
| leakypipes | gotta love when you go to get a link to something in nova and discover two bugs in glance. :( | 18:31 |
| *** Marga_ has quit IRC | 18:31 | |
| ndipanoff | i.e. every pci device is a row in the db with a state | 18:31 |
| ndipanoff | no such thing for cpus | 18:31 |
| *** kaufer has joined #openstack-nova | 18:31 | |
| cfriesen | superdan: I think we'd need duplicate cpu_pinning_raw then | 18:31 |
| ndipanoff | (maybe that would have been a better design?) | 18:31 |
| *** Marga_ has joined #openstack-nova | 18:31 | |
| cfriesen | superdan: and as ndipanoff said for resize we need new everything | 18:31 |
| ndipanoff | cfriesen, superdan the problem with that is the whole topology can change | 18:31 |
| mriedem | mestery: sc68cal: did you figure out the mystery of the missing nova api credentials and events thing while i was rebooting? | 18:31 |
| ndipanoff | i.e. resize from 1 to 2 cell instance | 18:32 |
| superdan | cfriesen: ndipanoff: well, then I guess we need current and new_ or something directly on instance | 18:32 |
| superdan | ndipanoff: yeah | 18:32 |
| ndipanoff | that would work yes | 18:32 |
| superdan | ndipanoff: I just don't want to duplicate any higher than we need | 18:32 |
| cfriesen | superdan: for resize you need new InstanceNUMATopology so I think it makes sense to reuse that for migration too | 18:33 |
| sc68cal | mriedem: nope - still a mystery me | 18:33 |
| superdan | cfriesen: obviously they'd both use the same temp storage yeah | 18:33 |
| superdan | so, | 18:33 |
| ndipanoff | IMHO 2 things that we want to consider - what the python api looks like, and that there is consistency | 18:33 |
| ndipanoff | from consistency pov - it should be old_ and new_ | 18:34 |
| superdan | well, | 18:34 |
| ndipanoff | like flavor really | 18:34 |
| superdan | do there need to be three? | 18:34 |
| *** neelashah has quit IRC | 18:34 | |
| cfriesen | ndipanoff: what about the one currently in use? | 18:34 |
| superdan | because it's terrible that flavor has three, but that's how it is | 18:34 |
| ndipanoff | it's old_ | 18:34 |
| superdan | should be current and _temp or something | 18:34 |
| ndipanoff | ok cool | 18:34 |
| cfriesen | superdan: agreed | 18:34 |
| superdan | at one point _temp is the one you will roll to, and then _temp becomes what you roll back to | 18:34 |
| ndipanoff | sounds good | 18:35 |
| superdan | so, I know this sounds just like using the migration object, but: | 18:35 |
| superdan | I wonder if what we really want is a MigrationContext object, | 18:35 |
| superdan | that holds a spare flavor, a numa, a pcithing, etc | 18:35 |
| superdan | and when we're doing a migration, that holds the things we want, | 18:35 |
| superdan | and when we're done it all goes away, no waste | 18:35 |
| ndipanoff | that does sound like the migration object :) | 18:36 |
| superdan | the difference is it's not a big join to a thing that never gets deleted | 18:36 |
| superdan | it's a column in instancextra that we null out when we're done | 18:36 |
| *** Mike_D_laptop has joined #openstack-nova | 18:36 | |
| ndipanoff | sounds good | 18:36 |
| superdan | if we're going to get to actually deleting stuff, we have to start doing that kind of thing | 18:36 |
| *** iamjarvo has quit IRC | 18:37 | |
| ndipanoff | cfriesen, ^ makes sense to you? | 18:38 |
| bwensley | Do we care that the resource audit on compute nodes is going to be doing more DB lookups? | 18:38 |
| ndipanoff | bwensley, it won't it will get it with the instance | 18:38 |
| bwensley | It already has the migration objects, but now will be digging around in instance extra as well. | 18:38 |
| *** Marga_ has quit IRC | 18:38 | |
| ndipanoff | and we join on instance extra already | 18:38 |
| superdan | right, using extra is much better for overhead | 18:38 |
| *** Marga_ has joined #openstack-nova | 18:38 | |
| bwensley | OK - I'm not keen on the current and temp naming. | 18:39 |
| bwensley | Using source and dest makes the code a LOT easier to read. | 18:39 |
| superdan | bwensley: you'll say Instance.get_by_uuid( ... , expected_attrs=[migrationstuffs]) | 18:39 |
| ndipanoff | bwensley, well if we have migration context | 18:39 |
| bwensley | If we're deleting these anyways after the migration is done... | 18:39 |
| ndipanoff | it can just be same as on instance | 18:39 |
| ndipanoff | so mig_ctxt.numa_topology | 18:39 |
| superdan | right | 18:39 |
| superdan | nice symmetry there | 18:40 |
| superdan | because at least with flavors, | 18:40 |
| ndipanoff | yes I like this | 18:40 |
| superdan | you need to switch the new to the current at some point | 18:40 |
| ndipanoff | well same here | 18:40 |
| superdan | and then the temp stash becomes the old one if you need to rollback | 18:40 |
| cfriesen | presumably we'd delete the hypothetical migration context when doing the confirm migrate/resize? | 18:41 |
| superdan | ndipanoff: so let me ask | 18:41 |
| superdan | cfriesen: right | 18:42 |
| superdan | ndipanoff: it's not that we need to be able to rebuild the old topo if we revert, but that the old host needs to continue to report the old topo for minutes while the migration is going on, right? | 18:42 |
| superdan | s/not/not just/ | 18:43 |
| cfriesen | superdan: we don't want to rebuild it, we want to be able to revert back to it if desired | 18:43 |
| ndipanoff | so it definitely needs to report it while the mig is happening | 18:43 |
| cfriesen | superdan: but yeah, the resources are in use until migration/resize is confirmed | 18:43 |
| *** salv-orlando has joined #openstack-nova | 18:43 | |
| superdan | ndipanoff: how does that work with flavors? does the old compute host know to start reporting instance.old_flavor magically? | 18:43 |
| ndipanoff | and revert also uses it no? | 18:43 |
| ndipanoff | superdan, look at usage_from_migration | 18:44 |
| ndipanoff | in instance tracker | 18:44 |
| ndipanoff | er | 18:44 |
| ndipanoff | resource_tracker | 18:44 |
| *** ajo has quit IRC | 18:44 | |
| ndipanoff | or rather _update_usage_from_migration | 18:44 |
| ndipanoff | it's nicely commented | 18:44 |
| ndipanoff | even | 18:44 |
| superdan | okay | 18:45 |
| superdan | I see, yeah | 18:45 |
| *** burt has joined #openstack-nova | 18:45 | |
| ndipanoff | so we need to hold both so that we can revert | 18:46 |
| superdan | so wait | 18:46 |
| superdan | oh right | 18:46 |
| superdan | nevermind | 18:46 |
| *** salv-orlando has quit IRC | 18:46 | |
| *** leakypipes has quit IRC | 18:46 | |
| *** lpetrut has joined #openstack-nova | 18:46 | |
| ndipanoff | basically once a claim is made on the target host | 18:47 |
| ndipanoff | we are double booking the migrating instance | 18:47 |
| *** Mike_D_laptop has quit IRC | 18:47 | |
| *** otter768 has joined #openstack-nova | 18:47 | |
| superdan | so the scheduler doesn't ever look at instance.numa_topology, right? It just looks at the usage that we send from the compute node, right? | 18:47 |
| ndipanoff | yes | 18:47 |
| ndipanoff | well | 18:47 |
| ndipanoff | it tries to place the instance it is scheduling | 18:47 |
| superdan | so it seems to me like maybe it's a flawed design in the first place to store the topo that we assigned like we do, such that we have this state we have to sync, | 18:47 |
| ndipanoff | ok let me parse that again | 18:48 |
| superdan | and instead let the hypervisor driver build up the details we need to report | 18:48 |
| ndipanoff | well that's not how RT works | 18:48 |
| superdan | because all the scheduler cares about is what cpus and nodes are used/free right? | 18:48 |
| superdan | "how RT works" is a really weak argument IMHO :) | 18:48 |
| ndipanoff | RT loops on instances from the db | 18:48 |
| ndipanoff | and accumulates the usage | 18:49 |
| bwensley | The hypervisor may not have up-to-date info | 18:49 |
| superdan | bwensley: WAT | 18:49 |
| ndipanoff | superdan, but yeah | 18:49 |
| *** hemna_ has quit IRC | 18:49 | |
| bwensley | If several things are migrating to a node and haven't actually spawned yet. | 18:49 |
| superdan | bwensley: for the instances actually running it better :) | 18:49 |
| bwensley | agreed | 18:49 |
| bwensley | But not for things scheduled but not yet running. | 18:49 |
| cfriesen | the scheduler technically only needs to know how many cpus are available on which numa nodes, not which specific CPUs are in use. | 18:49 |
| ndipanoff | cfriesen, well | 18:49 |
| *** iamjarvo has joined #openstack-nova | 18:50 | |
| *** ivasev has quit IRC | 18:50 | |
| ndipanoff | now - yes - but if we add the sibling preference feature that we cut from the orriginal BP then no | 18:50 |
| superdan | so this seems like it would be massively simpler if we didn't store anything about the topo we assigned, build that from the HV's knowledge of reality, and then just support placing a claim on the resources of the incoming compute node | 18:50 |
| *** iamjarvo has joined #openstack-nova | 18:50 | |
| ndipanoff | superdan, well that kind of completely invalidates the premise of how RT works now | 18:51 |
| cfriesen | superdan: but once we've claimed those resources we need to preserve that claim. | 18:51 |
| ndipanoff | which is - Nova DB is the truth | 18:51 |
| superdan | cfriesen: we could do that in memory or in the db if desired | 18:51 |
| ndipanoff | everything else is lies damn lies | 18:51 |
| superdan | ndipanoff: well, it does that for instance usage information maybe not not for host usage yes? | 18:51 |
| *** otter768 has quit IRC | 18:51 | |
| ndipanoff | ? | 18:52 |
| superdan | ndipanoff: even still, we could run a single thing before we do an update cycle that built the numa context, and just update that as we go along | 18:52 |
| superdan | ndipanoff: we report actual host memory usage, disk usage, etc, not just what we think we've used based on the instances | 18:52 |
| ndipanoff | nope | 18:52 |
| ndipanoff | we only log it | 18:52 |
| superdan | well, another good example of using RT's current behavior as a counterexample :) | 18:53 |
| bwensley | superdan: what does using the hypervisor info buy us? | 18:53 |
| *** salv-orlando has joined #openstack-nova | 18:53 | |
| superdan | bwensley: not having to store it, sync it, etc | 18:53 |
| bwensley | We still need the DB info for instances not yet launched (or migrating) | 18:53 |
| superdan | bwensley: we have that, that's the point | 18:53 |
| superdan | bwensley: we're keeping this info in a lot of places | 18:54 |
| superdan | bwensley: only one of them matters | 18:54 |
| *** marun has quit IRC | 18:54 | |
| superdan | bwensley: we're *going* to store the request spec, which has it, and the hypervisor has it, and we store the assigned version of it, and we're going to now store the next/planned version of it | 18:54 |
| bwensley | The only duplication there is between the hypervisor and the assigned version in the instance. | 18:56 |
| bwensley | The request spec is not the same as the assigned topology. | 18:56 |
| superdan | right, I get that the four copies are different in very tiny ways :) | 18:57 |
| bwensley | :) | 18:57 |
| bwensley | I like the idea that nova is in charge - it should be telling the hypervisor what it should be running - not relying on the hypervisor to tell it what is running. | 18:58 |
| *** irenab has joined #openstack-nova | 18:58 | |
| superdan | well, that's the opposite approach we take for things like instance state | 18:58 |
| superdan | they're different things for sure | 18:58 |
| superdan | but the HV state is the only one that matters, and storing it doesn't really make any sense to me. it's not like the HV can avoid the duplication of storing it | 18:59 |
| *** signed8bit is now known as signed8bit_ZZZzz | 19:00 | |
| ndipanoff | I kinda agree with bwensley here - nova's job is to do these things | 19:00 |
| ndipanoff | it's the "management" software | 19:00 |
| bwensley | And if the instance dies in the hypervisor, nova needs to be able to re-create it. | 19:00 |
| ndipanoff | it should manage stuff | 19:01 |
| ndipanoff | :) | 19:01 |
| superdan | anyway, it sounds like the migration context is the shortest path here, which I think everyone was okay with | 19:01 |
| *** achanda has quit IRC | 19:01 | |
| *** penick has quit IRC | 19:02 | |
| superdan | this also reminds me that at one point there was a suggestion for each compute node to have some local persistent storage for things like this | 19:02 |
| *** irenab has quit IRC | 19:02 | |
| *** sdake_ has joined #openstack-nova | 19:04 | |
| *** subscope has quit IRC | 19:04 | |
| *** sdake__ has joined #openstack-nova | 19:05 | |
| *** sdake has quit IRC | 19:05 | |
| *** ijw has joined #openstack-nova | 19:07 | |
| *** busterswt has quit IRC | 19:08 | |
| *** Marga_ has quit IRC | 19:09 | |
| *** Marga_ has joined #openstack-nova | 19:09 | |
| *** sdake_ has quit IRC | 19:09 | |
| *** emagana has quit IRC | 19:10 | |
| cfriesen | superdan: on the topic of "cpu model per flavor", you had mentioned letting users define their own "cpu flavor". How would that work without requiring detailed knowledge of what features are supported by what hypervisor? I'm envisioning a setup where a user makes a flavor but only one compute node in the entire cluster can actually run it. | 19:11 |
| *** pixelb has quit IRC | 19:11 | |
| cfriesen | superdan: would we need a "how many compute nodes can run this flavor" query? | 19:11 |
| cfriesen | cpu flavor, rather | 19:12 |
| superdan | cfriesen: I wasn't completely serious about us doing that, but more for the exercise | 19:12 |
| *** signed8bit_ZZZzz is now known as signed8bit | 19:12 | |
| superdan | cfriesen: you'd create it, and it would succeed if it didn't violate policy, and then yeah, it might seriously limit what you can boot on, | 19:12 |
| superdan | cfriesen: but so will requesting a very specific set of flags through your current approach | 19:12 |
| superdan | cfriesen: right now providers mask out flags for reasons I mentioned, | 19:13 |
| cfriesen | superdan: our "actual" current approach just allows specifying one of a half-dozen cpu models. :) | 19:13 |
| superdan | cfriesen: so if you want a consistent cpu model in your guest, it's really hard to know whether you can get that or what something that is compatible actually looks like | 19:13 |
| *** emagana has joined #openstack-nova | 19:13 | |
| superdan | cfriesen: yes, and those models differ per provider even though they seem like they should be the same | 19:14 |
| *** isd has joined #openstack-nova | 19:15 | |
| ndipanoff | superdan, re: local storage on compute nodes - interesting idea | 19:15 |
| *** marun has joined #openstack-nova | 19:15 | |
| cfriesen | superdan: the main issue is that I want end-users to be able to build with "gcc -march=<model>" and be able to specify something equivalent. And maybe it'll need a bit of tweaking. I don't know that it's realistic to expect it to work across all hypervisors or all providers. | 19:15 |
| superdan | ndipanoff: I forget what the original reason for it was, but there were some interesting things we could do better if we had it | 19:15 |
| isd | ack, connection dies right after I ask a question... | 19:16 |
| superdan | cfriesen: I know that that's what you want | 19:16 |
| superdan | cfriesen: the point of our API is to get us to the goal of being as provider and hypervisor-agnostic as possible | 19:16 |
| superdan | cfriesen: since these flags are not fluid things, we shouldn't punt here, IMHO | 19:16 |
| cfriesen | superdan: how can we possibly do that if different providers mask arbitrary features? Or do we do like Alex' | 19:17 |
| cfriesen | Alex | 19:17 |
| cfriesen | grr, can't type | 19:17 |
| cfriesen | Alex's suggestion and have each hypervisor report supported model/features. | 19:18 |
| superdan | cfriesen: well, the point of the cpu flavor thing was that we could check whether a specific set of flags was possibly supported on a cloud | 19:18 |
| superdan | cfriesen: and of course, there would be default flavors they provide | 19:18 |
| ndipanoff | superdan, maybe woudl be cool to respond on https://review.openstack.org/#/c/163440/ and -2->-1 it | 19:19 |
| *** isd has quit IRC | 19:19 | |
| superdan | cfriesen: so maybe your application says "I have three clouds, and I need to run a thing that does transcoding which needs a foobar cpu, so that means two of the three are candidates" | 19:19 |
| *** marun has quit IRC | 19:21 | |
| *** bkopilov has quit IRC | 19:21 | |
| *** Marga_ has quit IRC | 19:22 | |
| *** Marga_ has joined #openstack-nova | 19:23 | |
| cfriesen | superdan: yeah, so it sounds like you're literally talking about nova using a whole new "cpu flavor" object that corresponds to a set of low-level features (since model names aren't precise enough) | 19:23 |
| cfriesen | superdan: and then each compute node would export what features it supports? | 19:24 |
| *** marun has joined #openstack-nova | 19:24 | |
| cfriesen | superdan: when booting an instance the hypervisor driver could map a bunch of features to a model if it wants, or just use a feature list | 19:24 |
| *** isd has joined #openstack-nova | 19:24 | |
| *** VW_ has quit IRC | 19:25 | |
| superdan | cfriesen: well, the model mapping would just be a sample flavor | 19:25 |
| superdan | like the regular flavors we have today | 19:25 |
| cfriesen | superdan: yeah, but if we're trying to be generic then the nova "model" wouldn't necessarily map 1:1 to a hypervisor "model" | 19:26 |
| superdan | the thing is, encoding this all in a 255 char metadata string really sucks | 19:26 |
| superdan | cfriesen: right | 19:26 |
| *** arnaud____ has joined #openstack-nova | 19:27 | |
| superdan | cfriesen: again I'm not really suggesting we do this as my preference for solving the problem, | 19:27 |
| cfriesen | superdan: stick it in the DB and query it from the compute node? | 19:27 |
| *** marun has left #openstack-nova | 19:27 | |
| superdan | I'm sure people like jaypipes will freak if they read my comment | 19:27 |
| superdan | but encoding all of intel's history in 255 chars is dumb, IMHO :) | 19:28 |
| cfriesen | superdan: it seems like either we need something like your cpu flavor idea, or else we need to accept that a model name will give different results on different hypervisors | 19:28 |
| superdan | well, I don't think that the latter is doable | 19:29 |
| superdan | even danpb didn't like that idea, IIRC :) | 19:29 |
| superdan | and when danpb and I agree on something.. holy crap, look out :) | 19:29 |
| * superdan employs hyperbole | 19:29 | |
| cfriesen | danpb just suggested having all the other hypervisors use the libvirt model names, so that's cheating a bit. :) | 19:30 |
| *** Marga_ has quit IRC | 19:30 | |
| superdan | well, it is because libvirt ain't the only game in town | 19:30 |
| *** Marga_ has joined #openstack-nova | 19:30 | |
| *** VW_ has joined #openstack-nova | 19:31 | |
| superdan | IIRC, xen even had some fictitious cpu flags it used to expose hypervisor characteristics | 19:31 |
| *** arnaud____ has quit IRC | 19:32 | |
| *** bkopilov has joined #openstack-nova | 19:32 | |
| cfriesen | lots of different ideas here...I'm a bit worried that we'll end up with something so generic (but complicated) that it takes three releases to get done | 19:34 |
| *** hemna_ has joined #openstack-nova | 19:34 | |
| *** isd has quit IRC | 19:38 | |
| *** bkopilov has quit IRC | 19:39 | |
| *** annegentle has quit IRC | 19:43 | |
| superdan | cfriesen: on the other hand, it's extremely likely we'll end up with something so naive that it's not really useful, yet it's codified in our API forever | 19:43 |
| superdan | so, you know, pick your poison | 19:43 |
| *** alanf-mc_ has joined #openstack-nova | 19:48 | |
| *** penick has joined #openstack-nova | 19:49 | |
| *** HenryThe8th has quit IRC | 19:50 | |
| *** ajo has joined #openstack-nova | 19:50 | |
| *** bkopilov has joined #openstack-nova | 19:51 | |
| *** alanf-mc has quit IRC | 19:51 | |
| *** ajo has quit IRC | 19:52 | |
| *** ajo has joined #openstack-nova | 19:52 | |
| *** HenryG has joined #openstack-nova | 19:54 | |
| *** HenryG has quit IRC | 19:55 | |
| *** HenryG has joined #openstack-nova | 19:56 | |
| cfriesen | superdan: yeah, fair enough. maybe we need operator input. :) | 19:56 |
| *** cfriesen is now known as cfriesen_away | 19:56 | |
| superdan | oh, that'll help for sure :) | 19:56 |
| * superdan runs | 19:56 | |
| *** bkopilov has quit IRC | 19:57 | |
| *** irenab has joined #openstack-nova | 19:59 | |
| *** penick has quit IRC | 19:59 | |
| *** achanda has joined #openstack-nova | 19:59 | |
| *** dave-mccowan has quit IRC | 19:59 | |
| *** isd has joined #openstack-nova | 20:00 | |
| *** jecarey has joined #openstack-nova | 20:02 | |
| *** xyang1 has joined #openstack-nova | 20:03 | |
| *** irenab has quit IRC | 20:03 | |
| *** whenry has joined #openstack-nova | 20:05 | |
| *** bwensley has quit IRC | 20:06 | |
| *** VW_ has quit IRC | 20:06 | |
| *** yamahata has quit IRC | 20:07 | |
| *** lpetrut has quit IRC | 20:10 | |
| *** annegentle has joined #openstack-nova | 20:10 | |
| openstackgerrit | Josh Gachnang proposed openstack/nova: Reschedules sometimes do not allocate networks https://review.openstack.org/177470 | 20:11 |
| *** baoli has quit IRC | 20:12 | |
| *** baoli has joined #openstack-nova | 20:13 | |
| *** krak has joined #openstack-nova | 20:13 | |
| *** alanf-mc has joined #openstack-nova | 20:14 | |
| *** isd has quit IRC | 20:16 | |
| *** HenryG has quit IRC | 20:16 | |
| *** penick has joined #openstack-nova | 20:17 | |
| *** alanf-mc_ has quit IRC | 20:19 | |
| *** emagana has quit IRC | 20:21 | |
| *** HenryG has joined #openstack-nova | 20:21 | |
| *** dave-mccowan has joined #openstack-nova | 20:21 | |
| *** isd has joined #openstack-nova | 20:22 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/179340 | 20:22 |
| *** isd has quit IRC | 20:22 | |
| *** emagana has joined #openstack-nova | 20:23 | |
| *** mwagner_lap has quit IRC | 20:23 | |
| *** isd has joined #openstack-nova | 20:24 | |
| *** isd has quit IRC | 20:25 | |
| *** isd has joined #openstack-nova | 20:26 | |
| *** isd has quit IRC | 20:26 | |
| *** igordcard has quit IRC | 20:26 | |
| *** sdake__ is now known as sdake | 20:27 | |
| *** isd has joined #openstack-nova | 20:29 | |
| *** isd has quit IRC | 20:29 | |
| *** isd has joined #openstack-nova | 20:31 | |
| *** emagana has quit IRC | 20:32 | |
| *** isd has quit IRC | 20:32 | |
| *** isd has joined #openstack-nova | 20:35 | |
| *** isd has quit IRC | 20:35 | |
| *** jgrimm is now known as zz_jgrimm | 20:37 | |
| *** exploreshaifali has quit IRC | 20:38 | |
| *** mriedem has quit IRC | 20:38 | |
| *** mriedem has joined #openstack-nova | 20:39 | |
| *** isd has joined #openstack-nova | 20:41 | |
| *** isd has quit IRC | 20:41 | |
| *** isd has joined #openstack-nova | 20:44 | |
| *** isd has quit IRC | 20:45 | |
| *** isd has joined #openstack-nova | 20:46 | |
| *** isd has quit IRC | 20:46 | |
| *** otter768 has joined #openstack-nova | 20:48 | |
| *** burt has quit IRC | 20:48 | |
| *** claudiub has quit IRC | 20:49 | |
| *** burt has joined #openstack-nova | 20:51 | |
| *** isd has joined #openstack-nova | 20:51 | |
| *** isd has quit IRC | 20:52 | |
| *** otter768 has quit IRC | 20:52 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 20:52 |
| *** isd has joined #openstack-nova | 20:53 | |
| *** isd has quit IRC | 20:53 | |
| *** salv-orlando has quit IRC | 20:59 | |
| *** IanGovett1 has quit IRC | 21:00 | |
| *** ndipanoff has quit IRC | 21:01 | |
| *** mriedem has quit IRC | 21:03 | |
| *** mriedem has joined #openstack-nova | 21:03 | |
| *** penick has quit IRC | 21:04 | |
| *** emagana has joined #openstack-nova | 21:06 | |
| *** ijw has quit IRC | 21:06 | |
| *** ajayaa has joined #openstack-nova | 21:07 | |
| *** ijw_ has joined #openstack-nova | 21:08 | |
| *** neelashah1 has quit IRC | 21:09 | |
| *** ajo has quit IRC | 21:09 | |
| *** salv-orlando has joined #openstack-nova | 21:09 | |
| *** packet has joined #openstack-nova | 21:13 | |
| *** ajo has joined #openstack-nova | 21:15 | |
| *** iamjarvo has quit IRC | 21:15 | |
| *** ajo has quit IRC | 21:18 | |
| *** boris-42 has quit IRC | 21:18 | |
| *** burt has quit IRC | 21:18 | |
| *** achanda has quit IRC | 21:19 | |
| *** burt has joined #openstack-nova | 21:19 | |
| *** achanda has joined #openstack-nova | 21:20 | |
| *** browne has quit IRC | 21:21 | |
| *** iamjarvo has joined #openstack-nova | 21:21 | |
| *** iamjarvo has quit IRC | 21:21 | |
| *** iamjarvo has joined #openstack-nova | 21:22 | |
| *** oro has quit IRC | 21:23 | |
| *** packet has quit IRC | 21:23 | |
| *** emagana has quit IRC | 21:28 | |
| *** iamjarvo has quit IRC | 21:28 | |
| *** cbader has quit IRC | 21:28 | |
| *** emagana has joined #openstack-nova | 21:31 | |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: introduce libosinfo library to set hardware policy https://review.openstack.org/149625 | 21:33 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: use osinfo when configuring network model https://review.openstack.org/149627 | 21:33 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: adding libosinfo configuration https://review.openstack.org/149626 | 21:33 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: use osinfo when configuring the disk bus https://review.openstack.org/149628 | 21:33 |
| *** tjones2 has quit IRC | 21:41 | |
| *** tjones2 has joined #openstack-nova | 21:41 | |
| *** melwitt_ has joined #openstack-nova | 21:43 | |
| *** melwitt has quit IRC | 21:46 | |
| *** mriedem has quit IRC | 21:46 | |
| *** melwitt_ is now known as melwitt | 21:46 | |
| *** baoli has quit IRC | 21:47 | |
| *** baoli has joined #openstack-nova | 21:48 | |
| *** irenab has joined #openstack-nova | 21:48 | |
| *** kaufer has quit IRC | 21:49 | |
| *** irenab has quit IRC | 21:53 | |
| superdan | JoshNang: does this mean that it worked? | 21:54 |
| *** dimsum__ has quit IRC | 21:56 | |
| *** penick has joined #openstack-nova | 21:56 | |
| JoshNang | superdan: oh shoot forgot to put that in my comments. yup as far as i can tell! even after adding a 50% artifical fail rate in the virt driver to cause more reschedules | 21:56 |
| superdan | JoshNang: that is...excellent news :) | 21:56 |
| *** dimsum__ has joined #openstack-nova | 21:56 | |
| JoshNang | thanks for the help! | 21:56 |
| superdan | np! | 21:57 |
| *** dboik has quit IRC | 21:59 | |
| *** nelsnelson has quit IRC | 22:02 | |
| *** bkopilov has joined #openstack-nova | 22:08 | |
| *** bkopilov has quit IRC | 22:13 | |
| *** bkopilov has joined #openstack-nova | 22:14 | |
| *** patrickeast has quit IRC | 22:16 | |
| *** emagana has quit IRC | 22:21 | |
| *** bkopilov has quit IRC | 22:21 | |
| *** Marga_ has quit IRC | 22:21 | |
| *** bkopilov has joined #openstack-nova | 22:21 | |
| *** Marga_ has joined #openstack-nova | 22:22 | |
| *** ijw has joined #openstack-nova | 22:24 | |
| *** artom has quit IRC | 22:25 | |
| *** artom has joined #openstack-nova | 22:26 | |
| flashgordon | mikal: you around? | 22:27 |
| flashgordon | johnthetubaguy: ping | 22:27 |
| flashgordon | re: nova-ci want to stop the hpyerv from voting as its broken | 22:27 |
| *** ijw_ has quit IRC | 22:28 | |
| *** ajayaa has quit IRC | 22:30 | |
| *** Marga_ has quit IRC | 22:31 | |
| *** Marga_ has joined #openstack-nova | 22:32 | |
| morganfainberg | johnthetubaguy: have a question re: stable maint nova group if you have a few moments (specifically do you have the ability to edit people in the nova-stable-maint group)? | 22:32 |
| *** bkopilov has quit IRC | 22:33 | |
| flashgordon | morganfainberg: I am 99% johnthetubaguy just forgot to become zz_johnthetubaguy | 22:34 |
| morganfainberg | flashgordon: haha | 22:34 |
| flashgordon | I don't think he is online at 10PM on a friday night | 22:35 |
| *** ijw has quit IRC | 22:36 | |
| *** annegentle has quit IRC | 22:36 | |
| *** Marga_ has quit IRC | 22:36 | |
| *** Marga_ has joined #openstack-nova | 22:36 | |
| morganfainberg | fair enough | 22:37 |
| morganfainberg | UK? | 22:37 |
| *** dboik has joined #openstack-nova | 22:37 | |
| *** emagana has joined #openstack-nova | 22:38 | |
| *** dboik_ has joined #openstack-nova | 22:38 | |
| flashgordon | yup | 22:39 |
| *** Marga_ has quit IRC | 22:40 | |
| *** bkopilov has joined #openstack-nova | 22:40 | |
| *** dboik has quit IRC | 22:42 | |
| *** Marga_ has joined #openstack-nova | 22:44 | |
| flashgordon | superdan: you ever see this: http://logs.openstack.org/98/179398/2/experimental/check-tempest-dsvm-multinode-full/d44f762/logs/screen-n-cpu.txt.gz?level=TRACE | 22:46 |
| *** bkopilov has quit IRC | 22:46 | |
| *** cfriesen_away is now known as cfriesen | 22:47 | |
| *** bkopilov has joined #openstack-nova | 22:47 | |
| *** packet has joined #openstack-nova | 22:47 | |
| cfriesen | so I'm working on https://review.openstack.org/#/c/168418/ and I'm getting a weird error in the functional tests: http://logs.openstack.org/18/168418/9/check/gate-nova-tox-functional/81b0de3/testr_results.html.gz | 22:48 |
| cfriesen | anyone have any ideas where this is coming from? | 22:48 |
| *** markvoelker has quit IRC | 22:48 | |
| *** otter768 has joined #openstack-nova | 22:49 | |
| *** Marga_ has quit IRC | 22:49 | |
| *** Marga_ has joined #openstack-nova | 22:50 | |
| *** artom has quit IRC | 22:51 | |
| *** otter768 has quit IRC | 22:53 | |
| *** Marga_ has quit IRC | 22:54 | |
| *** packet has quit IRC | 22:56 | |
| *** gokrokv__ has quit IRC | 22:58 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 22:59 |
| *** dave-mccowan has quit IRC | 22:59 | |
| *** penick has quit IRC | 23:00 | |
| *** signed8b_ has joined #openstack-nova | 23:01 | |
| *** bkopilov has quit IRC | 23:04 | |
| *** signed8bit has quit IRC | 23:04 | |
| *** Marga_ has joined #openstack-nova | 23:08 | |
| *** bkopilov has joined #openstack-nova | 23:11 | |
| *** hemna_ has quit IRC | 23:11 | |
| *** IanGovett has joined #openstack-nova | 23:12 | |
| *** boris-42 has joined #openstack-nova | 23:18 | |
| *** signed8bit has joined #openstack-nova | 23:19 | |
| *** IanGovett has quit IRC | 23:20 | |
| *** achanda_ has joined #openstack-nova | 23:21 | |
| *** signed8b_ has quit IRC | 23:22 | |
| *** achanda has quit IRC | 23:25 | |
| *** achanda_ has quit IRC | 23:26 | |
| *** figleaf is now known as edleafe | 23:29 | |
| *** browne has joined #openstack-nova | 23:34 | |
| *** sdake_ has joined #openstack-nova | 23:36 | |
| *** irenab has joined #openstack-nova | 23:38 | |
| *** sdake has quit IRC | 23:39 | |
| *** sdake has joined #openstack-nova | 23:39 | |
| *** josecastroleon has joined #openstack-nova | 23:41 | |
| *** sdake_ has quit IRC | 23:42 | |
| *** irenab has quit IRC | 23:43 | |
| *** josecastroleon has quit IRC | 23:44 | |
| alex_xu | cfriesen: looks like you already fixed https://review.openstack.org/168418? tested it works in my local | 23:45 |
| alex_xu | cfriesen: ok, you fixed it. new patchset updated the version api sample file | 23:46 |
| *** emagana has quit IRC | 23:52 | |
| *** pixelb has joined #openstack-nova | 23:52 | |
| *** flwang has quit IRC | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!