| *** electrocucaracha has joined #openstack-nova | 00:00 | |
| vilobhmm1 | dansmith : https://review.openstack.org/#/c/138607/ is ready for review…whenever you are free.. | 00:08 |
|---|---|---|
| dansmith | dims___: done, will actually review it later | 00:08 |
| dims___ | dansmith: sounds good. thanks | 00:09 |
| *** gokrokve has quit IRC | 00:13 | |
| *** zzzeek has quit IRC | 00:14 | |
| *** gokrokve has joined #openstack-nova | 00:14 | |
| *** tjones1 has quit IRC | 00:14 | |
| *** sabari has quit IRC | 00:15 | |
| *** gokrokve has quit IRC | 00:18 | |
| *** patrickeast has quit IRC | 00:18 | |
| *** baoli has joined #openstack-nova | 00:21 | |
| openstackgerrit | Thomas Maddox proposed openstack/nova: Adding IPv6 static routes and static routes template https://review.openstack.org/187012 | 00:23 |
| *** sabari has joined #openstack-nova | 00:24 | |
| *** mtanino has quit IRC | 00:25 | |
| *** electrocucaracha has quit IRC | 00:27 | |
| *** electrocucaracha has joined #openstack-nova | 00:27 | |
| *** ijw has joined #openstack-nova | 00:28 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Ignore me - trying oslo.* master against Nova unit tests https://review.openstack.org/186413 | 00:30 |
| *** yamahata has quit IRC | 00:31 | |
| *** salv-orlando has joined #openstack-nova | 00:34 | |
| *** angdraug has quit IRC | 00:34 | |
| openstackgerrit | Thomas Maddox proposed openstack/nova: Adding IPv6 static routes and static routes template https://review.openstack.org/187012 | 00:36 |
| *** flwang1 has quit IRC | 00:37 | |
| *** salv-orlando has quit IRC | 00:41 | |
| dims___ | thomasem: ping | 00:42 |
| dims___ | thomasem: left a question in https://review.openstack.org/#/c/187012/ | 00:42 |
| *** kfox1111 has quit IRC | 00:45 | |
| *** sdake_ has joined #openstack-nova | 00:45 | |
| nic | @jbernard: with respect to tests, I might need a hand or three | 00:46 |
| nic | This little change: https://review.openstack.org/#/c/187395/1/nova/virt/libvirt/imagebackend.py | 00:47 |
| nic | …seems to have knocked over three dozen unit tests that all expect the code to work that way | 00:47 |
| nic | I figure something as innocuous-looking as that causing that much trouble means I should stop and ask someone who knows what they are doing what's going on | 00:48 |
| *** sdake has quit IRC | 00:49 | |
| jbernard | nic: certainly, ill do what i can | 00:50 |
| *** zhangtralon has joined #openstack-nova | 00:50 | |
| *** flwang has joined #openstack-nova | 00:51 | |
| nic | If the code change itself is uncontroversial, then it raises the question of why the unit tests went all explode-y over it | 00:52 |
| *** electrocucaracha has quit IRC | 00:52 | |
| *** ian_brown has joined #openstack-nova | 00:52 | |
| nic | But that much trouble would hint that there's something wrong with my approach | 00:53 |
| jbernard | it might be worth asking: do we need to snapshot the volume and delete/rollback soonafter? | 00:53 |
| *** hanzhang has quit IRC | 00:53 | |
| jbernard | could we get away without it? it would make the patchset smaller and avoid tripping over the image_create issue | 00:54 |
| *** promulo_ has joined #openstack-nova | 00:54 | |
| nic | Well, the image_create issue is the crux of why resizes don't work (but say they do) | 00:54 |
| nic | It hits that code path, and then no-ops because the image exists | 00:54 |
| nic | …instead of doing the size check and resizing, like it would if it entered the create_image method | 00:55 |
| *** zhangtralon has quit IRC | 00:55 | |
| nic | All the create_image implementations I looked at do their own existence checks, so it looks like somebody over-optimized it | 00:56 |
| nic | But again, there's A LOT of tests around it, so I figure I must be missing something | 00:56 |
| nic | (small words and purple crayon drawings would be a huge help here) | 00:57 |
| *** promulo has quit IRC | 00:57 | |
| jbernard | nic: lol | 00:57 |
| *** vilobhmm1 has quit IRC | 00:58 | |
| jbernard | nic: these paths are new to me as well, i'll need to look closer in the morning | 00:58 |
| *** shashankhegde has quit IRC | 00:59 | |
| nic | thanks | 00:59 |
| *** artom has quit IRC | 00:59 | |
| jroll | .v 129 | 01:00 |
| jroll | oops >.> | 01:00 |
| jbernard | nic: so things are broken before finish_migration() is called in the driver? | 01:01 |
| jbernard | nic: libvirt's _disk_resize() appears broken for rbd as well | 01:02 |
| jbernard | nic: size is passed as 0 | 01:02 |
| nic | As far as I can tell, yes | 01:03 |
| jbernard | nic: and the info dict looks odd as well | 01:03 |
| nic | But in my testing, resize and revert work with the patch set applied | 01:03 |
| nic | (and revert is way fast. whee!) | 01:03 |
| *** hanzhang has joined #openstack-nova | 01:05 | |
| jbernard | nic: at least there is that :) | 01:05 |
| nic | subunit disagrees with me as to the efficacy of my changes, but I swear, it works great with Ceph! :-) | 01:07 |
| nic | I've got to disconnect, so further discussion can happen in the bug and/or review | 01:09 |
| jbernard | nic: sounds good, thanks for working on this! | 01:09 |
| *** nic has quit IRC | 01:09 | |
| *** ijw_ has joined #openstack-nova | 01:12 | |
| *** shashankhegde has joined #openstack-nova | 01:13 | |
| *** ijw has quit IRC | 01:16 | |
| *** shashankhegde has quit IRC | 01:20 | |
| *** ijw has joined #openstack-nova | 01:23 | |
| *** Sukhdev has quit IRC | 01:24 | |
| *** Sukhdev has joined #openstack-nova | 01:25 | |
| *** ijw_ has quit IRC | 01:26 | |
| *** tjones1 has joined #openstack-nova | 01:33 | |
| *** sdake has joined #openstack-nova | 01:33 | |
| *** alanf-mc has quit IRC | 01:34 | |
| *** BadCub has quit IRC | 01:35 | |
| *** xylan_kong has quit IRC | 01:35 | |
| *** boris-42 has quit IRC | 01:35 | |
| *** wanghao has quit IRC | 01:35 | |
| *** aweeks has quit IRC | 01:35 | |
| *** DuncanT has quit IRC | 01:36 | |
| *** dougwig has quit IRC | 01:36 | |
| *** rex_lee has quit IRC | 01:36 | |
| *** briancurtin has quit IRC | 01:36 | |
| *** sdake_ has quit IRC | 01:37 | |
| *** BadCub has joined #openstack-nova | 01:38 | |
| openstackgerrit | Feng Xi Yan proposed openstack/nova: Transfer network info to metadata api when using config drive. https://review.openstack.org/183985 | 01:38 |
| *** xylan_kong has joined #openstack-nova | 01:41 | |
| *** aweeks has joined #openstack-nova | 01:42 | |
| *** wanghao has joined #openstack-nova | 01:43 | |
| *** briancurtin has joined #openstack-nova | 01:43 | |
| *** Sukhdev has quit IRC | 01:43 | |
| *** otter768 has joined #openstack-nova | 01:44 | |
| *** DuncanT has joined #openstack-nova | 01:45 | |
| *** shakamunyi_ has joined #openstack-nova | 01:46 | |
| *** shakamunyi_ has quit IRC | 01:46 | |
| *** ijw has quit IRC | 01:46 | |
| *** sdake has quit IRC | 01:48 | |
| *** otter768 has quit IRC | 01:49 | |
| *** sdake has joined #openstack-nova | 01:52 | |
| *** wwriverrat has quit IRC | 01:53 | |
| *** rex_lee has joined #openstack-nova | 01:55 | |
| *** dougwig has joined #openstack-nova | 01:55 | |
| *** boris-42 has joined #openstack-nova | 01:55 | |
| *** alexpilotti has quit IRC | 02:00 | |
| *** liusheng has quit IRC | 02:02 | |
| *** tobe has joined #openstack-nova | 02:03 | |
| *** zhenguo has joined #openstack-nova | 02:03 | |
| *** popw has joined #openstack-nova | 02:03 | |
| *** unicell1 has quit IRC | 02:07 | |
| *** pixelbeat has quit IRC | 02:08 | |
| *** baoli has quit IRC | 02:08 | |
| *** browne has quit IRC | 02:11 | |
| *** tsekiyama has quit IRC | 02:11 | |
| *** baoli has joined #openstack-nova | 02:11 | |
| *** ssurana has quit IRC | 02:12 | |
| *** yamahata has joined #openstack-nova | 02:12 | |
| *** bkopilov has quit IRC | 02:13 | |
| *** shashankhegde has joined #openstack-nova | 02:13 | |
| *** salv-orlando has joined #openstack-nova | 02:14 | |
| openstackgerrit | Rui Chen proposed openstack/nova: Set task_state=None when booting instance failed https://review.openstack.org/187145 | 02:15 |
| *** salv-orlando has quit IRC | 02:20 | |
| *** lyanchih has joined #openstack-nova | 02:20 | |
| *** xyang1 has quit IRC | 02:20 | |
| *** lyanchih has joined #openstack-nova | 02:20 | |
| *** lyanchih has left #openstack-nova | 02:20 | |
| *** lyanchih has joined #openstack-nova | 02:21 | |
| *** liusheng has joined #openstack-nova | 02:22 | |
| *** lyanchih has quit IRC | 02:23 | |
| *** wwriverrat has joined #openstack-nova | 02:25 | |
| *** shakamunyi has joined #openstack-nova | 02:26 | |
| *** mohitsharma has joined #openstack-nova | 02:27 | |
| *** lyanchih has joined #openstack-nova | 02:27 | |
| *** yamahata has quit IRC | 02:29 | |
| *** dboik has joined #openstack-nova | 02:31 | |
| *** shakamunyi has quit IRC | 02:33 | |
| openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova: Add and use raise_feature_not_supported() https://review.openstack.org/163697 | 02:37 |
| openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova: Add a hacking rule for consistent HTTP501 message https://review.openstack.org/163698 | 02:37 |
| *** browne has joined #openstack-nova | 02:42 | |
| *** baoli has quit IRC | 02:43 | |
| *** jogo has quit IRC | 02:44 | |
| *** VW has joined #openstack-nova | 02:45 | |
| *** dboik has quit IRC | 02:45 | |
| *** dims___ has quit IRC | 02:48 | |
| *** tjones1 has quit IRC | 02:48 | |
| *** VW has quit IRC | 02:50 | |
| *** yingjun has joined #openstack-nova | 02:51 | |
| openstackgerrit | Rui Chen proposed openstack/nova: CreateVM supports subnet specified https://review.openstack.org/182969 | 02:56 |
| *** VW has joined #openstack-nova | 02:56 | |
| *** lyanchih has quit IRC | 03:00 | |
| *** lyanchih has joined #openstack-nova | 03:12 | |
| *** shashankhegde has quit IRC | 03:13 | |
| *** lyanchih has quit IRC | 03:15 | |
| *** zhangtralon has joined #openstack-nova | 03:20 | |
| *** VW has quit IRC | 03:22 | |
| *** jogo has joined #openstack-nova | 03:22 | |
| *** zhangtralon has quit IRC | 03:26 | |
| *** yingjun has quit IRC | 03:33 | |
| *** VW has joined #openstack-nova | 03:38 | |
| *** yingjun has joined #openstack-nova | 03:39 | |
| *** yamahata has joined #openstack-nova | 03:43 | |
| *** otter768 has joined #openstack-nova | 03:45 | |
| *** ssurana has joined #openstack-nova | 03:45 | |
| *** VW has quit IRC | 03:46 | |
| *** VW has joined #openstack-nova | 03:47 | |
| *** VW has quit IRC | 03:48 | |
| *** mtanino has joined #openstack-nova | 03:48 | |
| *** dims_ has joined #openstack-nova | 03:48 | |
| openstackgerrit | wangxiyuan proposed openstack/nova: Add backup error information https://review.openstack.org/187445 | 03:49 |
| *** otter768 has quit IRC | 03:50 | |
| *** dims_ has quit IRC | 03:53 | |
| *** Nisha has joined #openstack-nova | 03:55 | |
| *** tjones1 has joined #openstack-nova | 03:59 | |
| *** deepthi has joined #openstack-nova | 04:03 | |
| *** yingjun has quit IRC | 04:07 | |
| *** unicell has joined #openstack-nova | 04:08 | |
| *** yamahata has quit IRC | 04:09 | |
| *** yamahata has joined #openstack-nova | 04:09 | |
| *** asettle has joined #openstack-nova | 04:10 | |
| *** garyk has joined #openstack-nova | 04:10 | |
| *** haigang has joined #openstack-nova | 04:19 | |
| *** fawadkhaliq has joined #openstack-nova | 04:21 | |
| *** Sukhdev has joined #openstack-nova | 04:21 | |
| *** bkopilov has joined #openstack-nova | 04:23 | |
| *** david-lyle has quit IRC | 04:24 | |
| *** david-lyle has joined #openstack-nova | 04:25 | |
| *** shashankhegde has joined #openstack-nova | 04:25 | |
| *** exploreshaifali has joined #openstack-nova | 04:26 | |
| *** mtanino has quit IRC | 04:27 | |
| *** tjones1 has quit IRC | 04:30 | |
| *** armax has quit IRC | 04:33 | |
| *** asettle has quit IRC | 04:34 | |
| *** Nisha has quit IRC | 04:38 | |
| *** Nisha has joined #openstack-nova | 04:43 | |
| *** salv-orlando has joined #openstack-nova | 04:46 | |
| *** pradipta has joined #openstack-nova | 04:50 | |
| *** exploreshaifali has quit IRC | 04:51 | |
| *** changbl_ has quit IRC | 04:52 | |
| *** salv-orlando has quit IRC | 04:53 | |
| *** shashankhegde has quit IRC | 04:55 | |
| *** Murali has joined #openstack-nova | 05:03 | |
| *** vilobhmm has joined #openstack-nova | 05:06 | |
| *** Longgeek has joined #openstack-nova | 05:07 | |
| *** sdake has quit IRC | 05:07 | |
| *** _gryf|afk has quit IRC | 05:08 | |
| *** tobe has quit IRC | 05:08 | |
| *** _gryf has joined #openstack-nova | 05:09 | |
| openstackgerrit | Earle F. Philhower, III proposed openstack/nova: Add support for HGST Solutions volumes to libvirt https://review.openstack.org/186594 | 05:12 |
| *** nosnos has joined #openstack-nova | 05:12 | |
| *** kiran-r has joined #openstack-nova | 05:13 | |
| *** yingjun has joined #openstack-nova | 05:13 | |
| *** Longgeek has quit IRC | 05:13 | |
| *** yingjun has quit IRC | 05:14 | |
| *** shashankhegde has joined #openstack-nova | 05:19 | |
| *** dave-mccowan has quit IRC | 05:19 | |
| *** Mike_D_laptop has joined #openstack-nova | 05:21 | |
| *** tobasco has joined #openstack-nova | 05:25 | |
| *** Mike_D_laptop has quit IRC | 05:26 | |
| *** sudipto has joined #openstack-nova | 05:29 | |
| *** gabor_ has joined #openstack-nova | 05:30 | |
| *** gabor_ is now known as Guest99527 | 05:30 | |
| *** achanda has joined #openstack-nova | 05:30 | |
| *** Guest99527 has quit IRC | 05:31 | |
| *** achanda_ has joined #openstack-nova | 05:37 | |
| *** nosnos has quit IRC | 05:37 | |
| *** zhenguo has quit IRC | 05:37 | |
| *** tobe has joined #openstack-nova | 05:37 | |
| *** achanda has quit IRC | 05:38 | |
| *** sgordon has quit IRC | 05:39 | |
| *** lpetrut has joined #openstack-nova | 05:39 | |
| *** sgordon has joined #openstack-nova | 05:39 | |
| *** otter768 has joined #openstack-nova | 05:46 | |
| *** rotbeard has joined #openstack-nova | 05:46 | |
| *** otter768 has quit IRC | 05:50 | |
| *** Longgeek has joined #openstack-nova | 05:52 | |
| openstackgerrit | Andrey Pavlov proposed openstack/nova: Extract helper method to get image metadata from volume https://review.openstack.org/187461 | 05:57 |
| *** yingjun has joined #openstack-nova | 06:04 | |
| openstackgerrit | Andrey Pavlov proposed openstack/nova: Extract helper method to get image metadata from volume https://review.openstack.org/187461 | 06:04 |
| *** Kennan2 has joined #openstack-nova | 06:06 | |
| *** Kennan has quit IRC | 06:06 | |
| *** achanda_ has quit IRC | 06:07 | |
| *** yuntong has joined #openstack-nova | 06:07 | |
| *** cfriesen has quit IRC | 06:10 | |
| *** salv-orlando has joined #openstack-nova | 06:13 | |
| *** tobe has quit IRC | 06:13 | |
| *** radek_ has joined #openstack-nova | 06:15 | |
| *** yingjun has quit IRC | 06:19 | |
| *** Sukhdev has quit IRC | 06:21 | |
| *** ajayaa has joined #openstack-nova | 06:23 | |
| *** salv-orlando has quit IRC | 06:25 | |
| *** ijw has joined #openstack-nova | 06:26 | |
| *** yingjun has joined #openstack-nova | 06:34 | |
| *** ildikov has quit IRC | 06:36 | |
| openstackgerrit | wangxiyuan proposed openstack/nova: Add specific error information when createBackup is not supported https://review.openstack.org/187445 | 06:36 |
| *** ian_brown has quit IRC | 06:37 | |
| *** dims_ has joined #openstack-nova | 06:37 | |
| *** tobe has joined #openstack-nova | 06:38 | |
| *** belmoreira has joined #openstack-nova | 06:38 | |
| *** ijw has quit IRC | 06:39 | |
| *** salv-orlando has joined #openstack-nova | 06:41 | |
| *** dims_ has quit IRC | 06:42 | |
| *** markus_z has joined #openstack-nova | 06:43 | |
| *** moshele has joined #openstack-nova | 06:43 | |
| *** masayukig_ has quit IRC | 06:47 | |
| *** arnaud____ has joined #openstack-nova | 06:47 | |
| *** jcoufal has joined #openstack-nova | 06:48 | |
| *** jcoufal has quit IRC | 06:50 | |
| *** jcoufal has joined #openstack-nova | 06:52 | |
| gilliard | Good $Morning Nova! | 06:53 |
| *** loquacity is now known as loquacities | 06:54 | |
| *** zhenguo has joined #openstack-nova | 06:54 | |
| *** qianli_ has joined #openstack-nova | 06:54 | |
| *** qianli has quit IRC | 06:55 | |
| *** masayukig_ has joined #openstack-nova | 06:59 | |
| *** paul-carlton has joined #openstack-nova | 07:02 | |
| *** gszasz has joined #openstack-nova | 07:05 | |
| *** ijw has joined #openstack-nova | 07:08 | |
| openstackgerrit | Andrey Pavlov proposed openstack/nova: Extract helper method to get image metadata from volume https://review.openstack.org/187461 | 07:09 |
| *** ijw_ has joined #openstack-nova | 07:10 | |
| *** ijw_ has quit IRC | 07:11 | |
| *** haigang has quit IRC | 07:11 | |
| *** ijw_ has joined #openstack-nova | 07:11 | |
| *** haigang has joined #openstack-nova | 07:12 | |
| *** ijw has quit IRC | 07:14 | |
| *** yassine has joined #openstack-nova | 07:14 | |
| *** yassine has quit IRC | 07:14 | |
| *** yassine has joined #openstack-nova | 07:17 | |
| *** yassine has quit IRC | 07:17 | |
| *** yassine has joined #openstack-nova | 07:18 | |
| *** moshele has quit IRC | 07:18 | |
| *** lpetrut has quit IRC | 07:19 | |
| *** sahid has joined #openstack-nova | 07:21 | |
| *** browne has quit IRC | 07:22 | |
| *** shashankhegde has quit IRC | 07:26 | |
| *** ildikov has joined #openstack-nova | 07:27 | |
| *** shashankhegde has joined #openstack-nova | 07:27 | |
| *** moshele has joined #openstack-nova | 07:28 | |
| *** dulek has joined #openstack-nova | 07:29 | |
| *** fawadkhaliq has quit IRC | 07:29 | |
| *** matrohon has joined #openstack-nova | 07:34 | |
| *** yamahata has quit IRC | 07:39 | |
| *** Kennan has joined #openstack-nova | 07:39 | |
| *** Kennan2 has quit IRC | 07:39 | |
| *** Nisha has quit IRC | 07:43 | |
| *** vilobhmm has quit IRC | 07:44 | |
| *** shashankhegde has quit IRC | 07:44 | |
| openstackgerrit | Zhengguang Ou proposed openstack/nova: libvirt: fix the race condition between detaching volume and creating image https://review.openstack.org/166152 | 07:46 |
| *** otter768 has joined #openstack-nova | 07:46 | |
| *** jlanoux has joined #openstack-nova | 07:48 | |
| *** alex_klimov has joined #openstack-nova | 07:48 | |
| *** jistr has joined #openstack-nova | 07:48 | |
| *** arnaud____ has quit IRC | 07:50 | |
| *** otter768 has quit IRC | 07:51 | |
| *** lpetrut has joined #openstack-nova | 07:54 | |
| *** oro has joined #openstack-nova | 07:54 | |
| bauzas | gilliard: "good" + random(set(['morning', afternoon', 'evening'])) | 07:56 |
| gilliard | :) | 07:57 |
| *** eantyshev has joined #openstack-nova | 08:08 | |
| *** mpavone has joined #openstack-nova | 08:10 | |
| *** lennyb has joined #openstack-nova | 08:10 | |
| *** zhangtralon has joined #openstack-nova | 08:11 | |
| *** danpb has joined #openstack-nova | 08:15 | |
| *** yingjun has quit IRC | 08:15 | |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Add a VolumeUsage object https://review.openstack.org/180442 | 08:15 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Replace conductor proxying calls with the new VolumeUsage object https://review.openstack.org/180443 | 08:15 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Remove and deprecate unused conductor method vol_usage_update() https://review.openstack.org/180444 | 08:15 |
| *** moshele_ has joined #openstack-nova | 08:15 | |
| *** moshele has quit IRC | 08:16 | |
| *** romainh has joined #openstack-nova | 08:18 | |
| *** xgsa has quit IRC | 08:18 | |
| *** lucasagomes has joined #openstack-nova | 08:18 | |
| *** derekh has joined #openstack-nova | 08:19 | |
| *** paul-carlton has quit IRC | 08:24 | |
| *** bjornar has joined #openstack-nova | 08:24 | |
| *** ndipanov has joined #openstack-nova | 08:24 | |
| bauzas | ndipanov: hi, FYI since I had to continue to work on RequestSpec, I had to provide a new PS for the change you -1 | 08:25 |
| bauzas | ndipanov: so I left a comment saying you were unhappy until we discuss with jaypipes and maybe more if possible today afternoon | 08:26 |
| ndipanov | bauzas, ok thanks very cool | 08:26 |
| ndipanov | I'll leave it until then | 08:26 |
| bauzas | ndipanov: if we can't do this, I'll write an email | 08:26 |
| ndipanov | async | 08:26 |
| bauzas | yup, that's the main problem we have, since the spec was merged this way | 08:27 |
| *** xgsa has joined #openstack-nova | 08:27 | |
| *** ZZelle has quit IRC | 08:28 | |
| *** ZZelle has joined #openstack-nova | 08:28 | |
| *** pradipta has quit IRC | 08:29 | |
| ndipanov | bauzas, well to me this is really expected - some things will not get figured out until there is code | 08:29 |
| ndipanov | especially when ti comes to refactoring of an established itnernal workflow | 08:29 |
| bauzas | ndipanov: yeah, I totally understand | 08:29 |
| ndipanov | we should not get nervous about it imho | 08:29 |
| bauzas | ndipanov: my point is, we have to find a way to discuss asyncly | 08:29 |
| ndipanov | ah yeah - email | 08:30 |
| ndipanov | :) | 08:30 |
| bauzas | ndipanov: honestly, I dislike mail threads because it's too much async and not serialized | 08:30 |
| openstackgerrit | Sudipta Biswas proposed openstack/nova-specs: Report host memory bandwidth as a metric in Nova https://review.openstack.org/180983 | 08:31 |
| bauzas | ndipanov: if the thread is opiniating, then you'll 100+ different discussions within the thread so we miss the consensus | 08:31 |
| *** paul-carlton has joined #openstack-nova | 08:31 | |
| ndipanov | yeah not ideal - though I would not necessarily got for consensus | 08:31 |
| ndipanov | more like - enough people think it's fine even if (nikola for example) doesn't | 08:31 |
| bauzas | ndipanov: okay, I thought about that, please provide a -1 again on my change, and I'll write an email | 08:31 |
| bauzas | ndipanov: asking for comments on your comment | 08:32 |
| ndipanov | kk let me look at ti | 08:32 |
| bauzas | ndipanov: and then I'll hassle the folks | 08:32 |
| * bauzas time for coffee | 08:32 | |
| openstackgerrit | sahid proposed openstack/nova: libvirt: make _create_domain to return a Guest object https://review.openstack.org/182185 | 08:33 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method poweroff to Guest object https://review.openstack.org/182194 | 08:33 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method resume to Guest object https://review.openstack.org/182195 | 08:33 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce a Guest to wrap around virConnect https://review.openstack.org/174313 | 08:33 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method get_guest to Host object https://review.openstack.org/181969 | 08:33 |
| sahid | ndipanov: can i ask you to take a look at this serie ^ | 08:34 |
| ndipanov | sahid, yes | 08:34 |
| sahid | ndipanov: thank you | 08:34 |
| *** popw has quit IRC | 08:35 | |
| sahid | danpb already gave a lot of reviews | 08:35 |
| moshele_ | hi johnthetubaguy | 08:36 |
| *** ssurana has quit IRC | 08:37 | |
| openstackgerrit | wangxiyuan proposed openstack/nova: Add specific error information when createBackup is not supported https://review.openstack.org/187445 | 08:39 |
| *** promulo_ has quit IRC | 08:40 | |
| *** lennyb has left #openstack-nova | 08:41 | |
| *** lennyb_ has joined #openstack-nova | 08:41 | |
| *** mohitsharma has quit IRC | 08:43 | |
| *** dguitarbite has joined #openstack-nova | 08:45 | |
| *** zhangtralon has quit IRC | 08:46 | |
| *** mleroy has quit IRC | 08:47 | |
| *** moshele_ has quit IRC | 08:47 | |
| *** wwriverrat has quit IRC | 08:48 | |
| *** moshele has joined #openstack-nova | 08:48 | |
| *** wwriverrat has joined #openstack-nova | 08:50 | |
| *** leopoldj has joined #openstack-nova | 08:53 | |
| *** kbyrne has joined #openstack-nova | 08:53 | |
| *** achanda has joined #openstack-nova | 08:54 | |
| *** zhangtralon has joined #openstack-nova | 08:54 | |
| *** wwriverrat has quit IRC | 08:55 | |
| openstackgerrit | sahid proposed openstack/nova-specs: console: convert consoles code to use objects framework https://review.openstack.org/165838 | 08:56 |
| *** achanda_ has joined #openstack-nova | 08:56 | |
| *** yingjun has joined #openstack-nova | 08:59 | |
| *** gszasz has quit IRC | 08:59 | |
| *** zhangtralon has quit IRC | 09:00 | |
| *** achanda has quit IRC | 09:00 | |
| *** zhangtralon has joined #openstack-nova | 09:01 | |
| *** aix has joined #openstack-nova | 09:01 | |
| *** claudiub has joined #openstack-nova | 09:03 | |
| johnthetubaguy | moshele: hi, bit late starting today, was travelling down to london | 09:05 |
| bauzas | ndipanov: oh man, I just discovered that the instance provided in the request spec can be a dict... | 09:05 |
| bauzas | *facepalm* | 09:06 |
| *** ociuhandu has joined #openstack-nova | 09:07 | |
| bauzas | ndipanov: https://github.com/openstack/nova/blob/master/nova/virt/hardware.py#L1113-L1116 | 09:08 |
| *** ssurana has joined #openstack-nova | 09:08 | |
| *** andreykurilin has joined #openstack-nova | 09:08 | |
| bauzas | ndipanov: loving your dirty ugly hacks you made in hardware.py | 09:08 |
| moshele | johnthetubaguy: lots of traffic in london | 09:09 |
| moshele | johnthetubaguy: I wanted to discuss Mellanox CI in nova | 09:09 |
| johnthetubaguy | moshele: I got the train, but its a long way from my house, so you know, worked on the train a bit | 09:10 |
| johnthetubaguy | moshele: ah, cool, hows it going? | 09:10 |
| moshele | johnthetubaguy: it seem we are commenting to patches we should be fillter | 09:12 |
| ndipanov | bauzas, yeah that was the quickest way to land numa without all of this | 09:13 |
| bauzas | ndipanov: yup I know :) | 09:13 |
| moshele | johnthetubaguy: this is why dansmith is angry regarding our CI | 09:13 |
| ndipanov | but if we can remove one case from that method | 09:13 |
| ndipanov | it's a win | 09:13 |
| ndipanov | we really don't want them at all | 09:13 |
| bauzas | ndipanov: so, I'll just make sure that if it's a dict, it will hydrate an object | 09:14 |
| johnthetubaguy | moshele: hmm, do you have an example he was upset about? | 09:14 |
| *** ssurana has quit IRC | 09:14 | |
| bauzas | ndipanov: or the fields since we haven't yet agreed | 09:14 |
| ndipanov | bauzas, that thing does that already I think | 09:14 |
| bauzas | ndipanov: I checked and we're just passing instance_props as it is | 09:15 |
| bauzas | ndipanov: so the filters are using the get method | 09:15 |
| johnthetubaguy | moshele: did you see the list of requirements for 3rd party CI: http://docs.openstack.org/infra/system-config/third_party.html | 09:15 |
| bauzas | ndipanov: which means they are not aware of it's a dict or an object, but that's fine | 09:15 |
| ndipanov | bauzas, the second bit of that function converts the dict to an object | 09:16 |
| *** ociuhandu has quit IRC | 09:16 | |
| ndipanov | if instance_numa_topology: | 09:16 |
| bauzas | ndipanov: I was talking about instance itself :) | 09:16 |
| bauzas | ndipanov: not the numa topology field :) | 09:17 |
| johnthetubaguy | moshele: we have a nova specific list of things as well: https://wiki.openstack.org/wiki/HypervisorSupportMatrix/DeprecationPlan#Specific_Requirements | 09:17 |
| *** ociuhandu has joined #openstack-nova | 09:17 | |
| bauzas | ndipanov: anyway, it was more for off the records | 09:17 |
| moshele | johnthetubaguy: I see | 09:19 |
| *** ijw_ has quit IRC | 09:19 | |
| moshele | johnthetubaguy: currently in the Mellanox CI we are filtering to be triggered only for pci code | 09:20 |
| *** achanda_ has quit IRC | 09:21 | |
| johnthetubaguy | moshele: oh, yeah, thats a hard thing, I would argue all patches except unit tests and docs might be a better filter | 09:21 |
| moshele | johnthetubaguy: but it seem zuul is still commenting on patches we are not triggered for like this https://review.openstack.org/#/c/187227 | 09:22 |
| *** fawadkhaliq has joined #openstack-nova | 09:22 | |
| johnthetubaguy | moshele: right, but its hard to prove what could break the PCI code, and running lots more tests shows up how more stable (or not) they system is | 09:23 |
| moshele | johnthetubaguy: is the recommendation not to do any file filtering? | 09:23 |
| johnthetubaguy | moshele: well, not quite, no need to vote on docs only and unit test only patches | 09:23 |
| johnthetubaguy | moshele: but ideally voting on anything that touches code would be good | 09:24 |
| johnthetubaguy | moshele: now, if thats not practical, thats OK, we can discuss that, but the ideal situation is you voting on all changes | 09:24 |
| johnthetubaguy | all code changes, I mean | 09:25 |
| lennyb_ | johnthetubaguy: the same issue of 'empty voing' will be on tests and docs as well. we cant 'decide' what to vote for and what not | 09:25 |
| *** dims_ has joined #openstack-nova | 09:26 | |
| johnthetubaguy | lennyb_: sorry, what do you mean by "empty voting"? I am clearly missing the context of the previous discussion here | 09:27 |
| *** toan-tran has joined #openstack-nova | 09:28 | |
| moshele | johnthetubaguy: it seems when you apply file filler rule on file in zuul, it will still comment on other files which are not in the filter | 09:29 |
| openstackgerrit | Deepak C Shetty proposed openstack/nova: Nova part of volume snapshot improvements https://review.openstack.org/172813 | 09:30 |
| moshele | johnthetubaguy: in this commit https://review.openstack.org/#/c/187227 we should have trigger because we had filler file in zuul but we still comment | 09:30 |
| moshele | johnthetubaguy: I mean we shouldn't | 09:31 |
| *** dims_ has quit IRC | 09:31 | |
| johnthetubaguy | moshele: oh, I would talk to infra about how they make that work, lots of the other CI system manage to not vote on docs, but vote on all other patches, but maybe they don't use zuul... not sure | 09:31 |
| johnthetubaguy | lennyb_: moshele: oh, I see what you mean by an empty vote now, that is bad | 09:32 |
| lennyb_ | johnthetubaguy: some of them not using zuul. | 09:32 |
| *** yingjun has quit IRC | 09:33 | |
| *** gszasz has joined #openstack-nova | 09:34 | |
| *** Nisha has joined #openstack-nova | 09:34 | |
| moshele | johnthetubaguy: that why dansmith was angry, we missed it because in jenkins we didn't see any job for that so we were not aware we are doing the empty voting | 09:34 |
| *** Murali has quit IRC | 09:37 | |
| *** aix has quit IRC | 09:45 | |
| openstackgerrit | Maxim Nestratov proposed openstack/nova: libvirt: don't add filesystem disk to parallels containers unconditionally https://review.openstack.org/186858 | 09:45 |
| *** aix has joined #openstack-nova | 09:45 | |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Remove/deprecate unused parts of the compute node object https://review.openstack.org/186757 | 09:46 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Make ComputeNode.service_id nullable to match db schema https://review.openstack.org/169284 | 09:46 |
| *** otter768 has joined #openstack-nova | 09:47 | |
| openstackgerrit | Sudipta Biswas proposed openstack/nova-specs: Report host memory bandwidth as a metric in Nova https://review.openstack.org/180983 | 09:49 |
| *** josecastroleon has joined #openstack-nova | 09:49 | |
| *** haigang has quit IRC | 09:50 | |
| openstackgerrit | sahid proposed openstack/nova-specs: libvirt: set admin root password https://review.openstack.org/187509 | 09:50 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Avoid querying for Service in resource tracker https://review.openstack.org/169288 | 09:51 |
| *** Nisha_away has joined #openstack-nova | 09:51 | |
| *** otter768 has quit IRC | 09:52 | |
| *** amotoki_ has joined #openstack-nova | 09:53 | |
| *** Nisha has quit IRC | 09:54 | |
| openstackgerrit | Nikola Dipanov proposed openstack/nova: bdm: Make sure that delete_on_termination is a boolean https://review.openstack.org/186910 | 09:54 |
| openstackgerrit | Nikola Dipanov proposed openstack/nova: snapshot: Copy some missing attrs to the snapshot bdms https://review.openstack.org/186911 | 09:54 |
| *** achanda has joined #openstack-nova | 10:00 | |
| *** Nisha_away has quit IRC | 10:00 | |
| *** Nisha has joined #openstack-nova | 10:04 | |
| johnthetubaguy | moshele: sorry, lost my connection, long story, but yeah, the empty voting would be good to stop | 10:05 |
| johnthetubaguy | moshele: I suspect infra are the best folks to help with that, I guess there is another config for skipping the vote all together, I know some folks are doing that, not sure if they are doing that with zuul | 10:05 |
| johnthetubaguy | moshele: testing everything would be a possibly "quick" fix I guess | 10:06 |
| openstackgerrit | Daniel Berrange proposed openstack/nova: compute: remove get_image_metadata method https://review.openstack.org/187251 | 10:06 |
| ndipanov | johnthetubaguy, do you remember what was the story behind this: https://review.openstack.org/#/c/176805/ | 10:08 |
| ndipanov | not linked to a bug/bp... | 10:08 |
| *** haigang has joined #openstack-nova | 10:08 | |
| Nisha | johnthetubaguy, hi ...wanted to discuss on the capabilities specs as in links https://review.openstack.org/182572 and https://review.openstack.org/182581 | 10:08 |
| *** gholler has joined #openstack-nova | 10:08 | |
| johnthetubaguy | ndipanov: docimpact seemed enough with that one | 10:08 |
| johnthetubaguy | ndipanov: there are quite a few bps tracking moving policy to more fine grained, it fits in with that I think | 10:09 |
| ndipanov | a-hah | 10:09 |
| ndipanov | no I was only wondering what inspired that particular patch | 10:09 |
| johnthetubaguy | ndipanov: I assume they want to restrict snapshots, I don't know the back story | 10:10 |
| Nisha | johnthetubaguy, the above specs are required for ironic or in other words ironic and nova work is interrelated for this. | 10:11 |
| *** ssurana has joined #openstack-nova | 10:11 | |
| *** zhangtralon has quit IRC | 10:11 | |
| openstackgerrit | Matthew Booth proposed openstack/nova: Remove unused variables from images api https://review.openstack.org/183475 | 10:11 |
| Nisha | johnthetubaguy, i havent seen any reviews on these specs so not sure what is the opinion within nova community for these | 10:11 |
| johnthetubaguy | Nisha: OK, not had chance to review many specs with all the pre/post summit work right now | 10:12 |
| johnthetubaguy | Nisha: nothing jumps out as really bad, but I would have to give it a full review before I can give more context I am afraid | 10:13 |
| Nisha | johnthetubaguy, i just wanted to bring that in notice...if you/other members can go through the specs we can discuss in IRC at the suitable time | 10:13 |
| johnthetubaguy | Nisha: hopefully will get to those this week, but afraid its hard to promise that at this point | 10:13 |
| Nisha | johnthetubaguy, that will help...thanks | 10:13 |
| johnthetubaguy | Nisha: its just a case of waiting for a review at this point, I am afraid, sorry thats not quicker | 10:14 |
| Nisha | its not just review...its about the opinion from nova community | 10:14 |
| *** gholler has quit IRC | 10:14 | |
| Nisha | the ironic changes are dependent on what nova agrees | 10:14 |
| Nisha | i have 4 specs in total, 2 in nova and 2 in ironic | 10:15 |
| johnthetubaguy | Nisha: yeah, I should find a way to raise the priority of those, but its tricky right now | 10:15 |
| Nisha | johnthetubaguy, thanks | 10:15 |
| *** ssurana has quit IRC | 10:15 | |
| *** ijw has joined #openstack-nova | 10:16 | |
| *** claudiub has quit IRC | 10:18 | |
| *** yassine has quit IRC | 10:19 | |
| *** ijw_ has joined #openstack-nova | 10:19 | |
| *** haigang has quit IRC | 10:22 | |
| *** ijw has quit IRC | 10:22 | |
| *** ijw_ has quit IRC | 10:24 | |
| *** moshele has quit IRC | 10:26 | |
| *** gholler has joined #openstack-nova | 10:26 | |
| *** jlanoux_ has joined #openstack-nova | 10:28 | |
| *** haigang has joined #openstack-nova | 10:28 | |
| *** jlanoux has quit IRC | 10:30 | |
| *** moshele has joined #openstack-nova | 10:34 | |
| *** jyuso1 has quit IRC | 10:34 | |
| openstackgerrit | John Garbutt proposed openstack/nova-specs: Add liberty priorities https://review.openstack.org/187272 | 10:36 |
| *** pixelbeat has joined #openstack-nova | 10:36 | |
| openstackgerrit | John Garbutt proposed openstack/nova-specs: Add liberty priorities https://review.openstack.org/187272 | 10:38 |
| *** zhangtralon has joined #openstack-nova | 10:41 | |
| *** samueldmq has joined #openstack-nova | 10:41 | |
| *** oro has quit IRC | 10:44 | |
| *** alexpilotti has joined #openstack-nova | 10:46 | |
| *** jlanoux has joined #openstack-nova | 10:53 | |
| *** dims_ has joined #openstack-nova | 10:55 | |
| *** jlanoux_ has quit IRC | 10:55 | |
| *** achanda_ has joined #openstack-nova | 10:56 | |
| *** fawadkhaliq has quit IRC | 10:56 | |
| *** achanda has quit IRC | 10:56 | |
| *** ian_brown has joined #openstack-nova | 10:57 | |
| *** Longgeek_ has joined #openstack-nova | 11:01 | |
| *** Longgeek has quit IRC | 11:01 | |
| jaypipes | johnthetubaguy: hey, you have the link to the etherpad that lists all the etherpads for nova at the summit? | 11:07 |
| *** ssurana has joined #openstack-nova | 11:12 | |
| lxsli | jaypipes: https://wiki.openstack.org/wiki/Design_Summit/Liberty/Etherpads#Nova ? | 11:13 |
| jaypipes | lxsli: ty sir! :) | 11:13 |
| lxsli | nw :) | 11:13 |
| *** ssurana has quit IRC | 11:17 | |
| *** tobasco has quit IRC | 11:17 | |
| *** ijw has joined #openstack-nova | 11:19 | |
| bauzas | jaypipes: hey | 11:21 |
| *** ian_brown has quit IRC | 11:21 | |
| bauzas | jaypipes: I discussed with ndipanov on the ReqSpec BP | 11:21 |
| jaypipes | bauzas: bonjour | 11:21 |
| *** rushiagr_away is now known as rushiagr | 11:22 | |
| ndipanov | jaypipes, hello\ | 11:22 |
| jaypipes | ndipanov: hi! :) | 11:22 |
| openstackgerrit | Zhenyu Zheng proposed openstack/nova: CONF.allow_resize_on_same_host should check only once in controller https://review.openstack.org/186236 | 11:22 |
| bauzas | jaypipes: and it seems we have different views | 11:22 |
| jaypipes | ndipanov: missed you in YVR :( | 11:22 |
| bauzas | jaypipes: aloha | 11:22 |
| bauzas | (note, my 9mo daughter is on my knees, don't expect me to discuss quickly) :) | 11:22 |
| bauzas | so | 11:22 |
| *** ijw_ has joined #openstack-nova | 11:23 | |
| ndipanov | yeah - couldn't really make it this time :( | 11:23 |
| bauzas | jaypipes: I would love your views on https://review.openstack.org/#/c/145528/13/nova/objects/request_spec.py,cm | 11:23 |
| bauzas | jaypipes: I will write an email | 11:23 |
| bauzas | jaypipes: so ndipanov and you could just reply | 11:23 |
| ndipanov | jaypipes, and if it's not clear what I meant on that review - pls ping me and I'll try to explain | 11:23 |
| jaypipes | ndipanov: will do. | 11:24 |
| jaypipes | guys, gimme some time to read and reply. | 11:24 |
| bauzas | jaypipes: tbc, here is the point https://review.openstack.org/#/c/145528/12/nova/objects/request_spec.py,cm | 11:24 |
| *** zhangtralon has quit IRC | 11:24 | |
| ndipanov | np - I'm here all day | 11:24 |
| jaypipes | kk | 11:24 |
| bauzas | jaypipes: sure thing, I'm not blocked | 11:24 |
| bauzas | jaypipes: since I'm working on the next changes | 11:24 |
| bauzas | folks, going out to lunch | 11:25 |
| jaypipes | bauzas: I have a weekly ops meeting from 11am EST - 12:30pm EST, every week. this means I always am missing the gantt/scheduler meeting now that daylight savings is in effect :( Any chance you can talk with n0ano about maybe moving that meeting? | 11:25 |
| bauzas | jaypipes: will do :) | 11:25 |
| *** ijw has quit IRC | 11:25 | |
| jaypipes | bauzas: merci | 11:25 |
| bauzas | jaypipes: we can just doodle it | 11:26 |
| jaypipes | kk | 11:26 |
| *** ijw_ has quit IRC | 11:27 | |
| *** mwagner_afk has quit IRC | 11:28 | |
| *** triggerz has joined #openstack-nova | 11:28 | |
| samueldmq | morning | 11:28 |
| samueldmq | could someone from nova take a look at bug #1458945 | 11:28 |
| openstack | bug 1458945 in murano "Use graduated oslo.policy instead of oslo-incubator code" [High,Confirmed] https://launchpad.net/bugs/1458945 - Assigned to Ekaterina Chernova (efedorova) | 11:28 |
| jaypipes | samueldmq: mornin. | 11:28 |
| *** achanda_ has quit IRC | 11:28 | |
| jaypipes | samueldmq: is this a Nova thing? | 11:28 |
| samueldmq | jaypipes, it's a policy thing .. incubated code that got graduated | 11:29 |
| jaypipes | samueldmq: but that bug is for Murano, not Nova...? | 11:29 |
| samueldmq | jaypipes, https://bugs.launchpad.net/nova/+bug/1458945 | 11:29 |
| openstack | Launchpad bug 1458945 in murano "Use graduated oslo.policy instead of oslo-incubator code" [High,Confirmed] - Assigned to Ekaterina Chernova (efedorova) | 11:29 |
| samueldmq | jaypipes, that affects several projects | 11:29 |
| jaypipes | samueldmq: ah, I see now.. | 11:29 |
| jaypipes | dims_: can you handle the nova patch for ^^? | 11:30 |
| samueldmq | jaypipes, hmm.. actually that's interesting, it said murano because it was the last update in there :) | 11:30 |
| jaypipes | samueldmq: yeah :) | 11:30 |
| samueldmq | jaypipes, great, if people are very busy with other tasks, let me know if you need me to do that | 11:31 |
| samueldmq | jaypipes, just to make sure we keep moving | 11:32 |
| samueldmq | jaypipes, thanks :) | 11:32 |
| *** zhangtralon has joined #openstack-nova | 11:36 | |
| *** yingjun has joined #openstack-nova | 11:37 | |
| *** Nisha has quit IRC | 11:40 | |
| *** ildikov has quit IRC | 11:41 | |
| jaypipes | samueldmq: of course, you are more than welcome to submit a patch for that to nova! | 11:42 |
| openstackgerrit | garyk proposed openstack/nova: Compute: replace incorrect instance object with dict https://review.openstack.org/187537 | 11:42 |
| *** deepthi has quit IRC | 11:42 | |
| garyk | dims_: ^ | 11:43 |
| *** lucasagomes is now known as lucas-hungry | 11:43 | |
| samueldmq | jaypipes, nice .. I will in the case dims_ can't, since I also have some other changes in the Dynamic Policy thing | 11:44 |
| dims_ | jaypipes: oslo.policy? there was a feature request from sdague for oslo.policy and he does not want it merged without it | 11:44 |
| samueldmq | jaypipes, hope you had heard about that in the summit :) | 11:44 |
| * dims_ trying to see if it's mentioned in any etherpad | 11:45 | |
| openstackgerrit | wangxiyuan proposed openstack/nova: Fix bdm image_meta when boot from volume https://review.openstack.org/186247 | 11:45 |
| kashyap | ndipanov: Hi, when you get a moment, re: https://bugs.launchpad.net/nova/+bug/1375868 | 11:46 |
| openstack | Launchpad bug 1375868 in OpenStack Compute (nova) "libvirt: race between hot unplug and XMLDesc in _get_instance_disk_info" [High,Confirmed] | 11:46 |
| ndipanov | yes | 11:47 |
| ndipanov | kashyap, what about it | 11:47 |
| ndipanov | ? | 11:47 |
| dims_ | samueldmq: jaypipes: https://etherpad.openstack.org/p/YVR-nova-contributor-meetup lines 154 onwards. specifically 160 | 11:47 |
| samueldmq | dims_, looking | 11:47 |
| kashyap | I see you're fixing bugs in that area (65d6eb0 - libvirt: make _get_instance_disk_info conservative) | 11:47 |
| *** dprince has joined #openstack-nova | 11:48 | |
| kashyap | ndipanov: Is that race still present? | 11:48 |
| sdague | right, it just doesn't seem sensible to do that merge until we get the new bits in | 11:48 |
| dims_ | sdague: ack. was someone on policy team going to do a spec? | 11:48 |
| ndipanov | kashyap, well it's not really a race | 11:48 |
| kashyap | ndipanov: Or as the bug you point to in that, which has a lot more detail, that says that libvirt's XML shouldn't be referred | 11:48 |
| *** otter768 has joined #openstack-nova | 11:48 | |
| ndipanov | kashyap, yes basically that's it | 11:49 |
| ndipanov | I looked into it briefly | 11:49 |
| *** sdake has joined #openstack-nova | 11:49 | |
| sdague | so, I think I need to write it up. Right now I'm writing up some english about the Nova microversions structure and solution as that seemed the most urgent | 11:49 |
| *** oro has joined #openstack-nova | 11:49 | |
| ndipanov | and I think it can be changed to not use the XML | 11:49 |
| sdague | jamielennox and I talked about it quite a bit | 11:49 |
| ndipanov | but it won't be a small change | 11:49 |
| samueldmq | sdague, hi .. what are the missing bits specifically ? | 11:49 |
| sdague | as well as with morganfainberg, it's on my list :) | 11:49 |
| kashyap | ndipanov: Okay, no rush. Just was looking through some bugs, and `git blame` said Nikola is fixing bugs in this area, so I pinged :-) | 11:49 |
| sdague | samueldmq: so the nova team would like base policy to be encoded in code, not a json file | 11:50 |
| ndipanov | kashyap, I think the race itslef was mitigated | 11:50 |
| sdague | the json files could be use to override | 11:50 |
| ndipanov | plus I did another horrible hack there to make it even less likely | 11:50 |
| sdague | and ideally some overrides would trigger warnings in the logs on startup as "terrible idea, might impact compatibility" | 11:50 |
| samueldmq | sdague, hmm, cool .. I will be glad to follow this, looks to be an interesting idea | 11:50 |
| ndipanov | but the root cauyse is still there | 11:50 |
| samueldmq | sdague, ++ | 11:51 |
| samueldmq | sdague, but for now ... we are only migrating the oslo-incubator version of policy.json to the graduated library | 11:51 |
| samueldmq | sdague, I don't see how this affects other changes, it just keep it as is | 11:51 |
| kashyap | ndipanov: Noted, thanks for the detail. You might want to update the bug with this context. | 11:52 |
| sdague | there are other things that emerged there right? Also the whole policy.d performance issues were never really addressed. | 11:52 |
| ndipanov | well all the context is in the linked bug really | 11:52 |
| ndipanov | or hmmm | 11:52 |
| sdague | it just seemed like making those changes now didn't really help anything | 11:53 |
| kashyap | ndipanov: Okay, if that's sufficient, please disregard me. | 11:53 |
| *** otter768 has quit IRC | 11:53 | |
| dims_ | samueldmq: there's no policy.json or py in oslo-incubator :) | 11:53 |
| ndipanov | well this is on my list of things to look at this month so will get to it for sure | 11:53 |
| samueldmq | dims_, policy.py | 11:53 |
| garyk | mdbooth: please check https://review.openstack.org/#/c/183475/ - silly pep8 stuff | 11:53 |
| samueldmq | dims_, it would only replace https://github.com/openstack/nova/blob/master/etc/nova/policy.json | 11:54 |
| samueldmq | dims_, by https://github.com/openstack/oslo.policy/blob/master/oslo_policy/policy.py, basically | 11:54 |
| dims_ | samueldmq: i mean we deletd policy.py from oslo-incubator https://github.com/openstack/oslo-incubator/tree/master/openstack/common | 11:54 |
| *** romainh has quit IRC | 11:54 | |
| mdbooth | garyk: Argh! pep8 failure due to merge conflict >:( | 11:54 |
| *** romainh1 has joined #openstack-nova | 11:54 | |
| mdbooth | Thanks | 11:54 |
| samueldmq | dims_, but sdague may be seeing things I am not :) | 11:54 |
| sdague | yeh, the incubator deleted it already. Honestly, part of the reason I want to have this done in this order is I think we need to make sure we've got buy in on direction before committing to the library, because we're going to have a hard minimum requirement here. | 11:55 |
| samueldmq | dims_, yes, and that's the reason we need to update nova to use oslo.policy, incubator code has gone | 11:55 |
| sdague | samueldmq: is there a specific bug you are trying to address? | 11:56 |
| *** sdake_ has joined #openstack-nova | 11:56 | |
| *** ildikov has joined #openstack-nova | 11:56 | |
| dims_ | samueldmq: as sdague said, if the nova team is not comfortable with oslo.policy, they can move policy.py in their tree to another location and own it | 11:56 |
| *** dave-mccowan has joined #openstack-nova | 11:56 | |
| samueldmq | sdague, only the bug for the migration itself https://bugs.launchpad.net/nova/+bug/1458945 | 11:56 |
| openstack | Launchpad bug 1458945 in murano "Use graduated oslo.policy instead of oslo-incubator code" [High,Confirmed] - Assigned to Ekaterina Chernova (efedorova) | 11:56 |
| samueldmq | sdague, which affects multiple projects | 11:56 |
| samueldmq | (yes, it says murano, but it's multiproject) | 11:57 |
| *** sdake has quit IRC | 11:57 | |
| sahid | jaypipes: hello, i wanted to ask you.. I like the idea you purpose during a design session about "ressource templates"" | 11:57 |
| sahid | https://etherpad.openstack.org/p/YVR-nova-liberty-priorities line 63 | 11:57 |
| *** zhenguo has quit IRC | 11:57 | |
| sahid | i guess a good use case could be fore NUMA | 11:57 |
| jaypipes | sahid: cool :) | 11:57 |
| jaypipes | sahid: yes, and Ironic stuffs. | 11:58 |
| sahid | so i would like to know if i can help here | 11:58 |
| dims_ | samueldmq: what is the CVE vulnerability itself? | 11:58 |
| sahid | yes for sure | 11:58 |
| sdague | samueldmq: right, so I feel like it would be better to get focus on supporting the policy changes we'd like first, then we'll happily jump. But the current transition just seems like a sideways bit of busy work. | 11:58 |
| samueldmq | dims_, ok I understand, I'll be following this work in here | 11:58 |
| sahid | did you think about to push a spec? or perhaps i can try to wor on a draft? | 11:58 |
| jaypipes | sahid: you certainly can. I will work on the nova spec for this today and will add you as a reviewer when I push it | 11:58 |
| samueldmq | sdague, ok I respect your decision in nova side :) | 11:58 |
| dims_ | samueldmq: happy to get help in oslo.policy for features needed by nova. so we can start there first? | 11:58 |
| samueldmq | sdague, I am working on the dynamic policy thing with ayoung | 11:59 |
| sahid | jaypipes: great, thank you | 11:59 |
| jaypipes | sahid: pas de probleme. | 11:59 |
| samueldmq | sdague, let me know if you need me for something related to policies | 11:59 |
| sahid | :) | 11:59 |
| *** radek_ has quit IRC | 11:59 | |
| * mdbooth really wishes we didn't use repeatable reads in MySQL. The context required to reason about it robustly is insane. | 11:59 | |
| dims_ | samueldmq: would love your help with oslo.policy features | 12:00 |
| sdague | samueldmq: right, this is going to intersect with that some as well. We probably need a discussion thread somewhere as well. jamielennox and I talked through some of the implications. I think the 2 are compatible, but they do impact each other | 12:00 |
| samueldmq | dims_, I need to understand your requirements first (sdague is going to write a spec right?) | 12:00 |
| sdague | samueldmq: yes, I will, I will try to get it out this week | 12:00 |
| sdague | which repo should it go in? | 12:00 |
| dims_ | samueldmq: not mine...nova team's | 12:00 |
| samueldmq | dims_, I will be happy to collaborate (although time is a bit thigh with the other dynamic policies changes) | 12:01 |
| dims_ | sdague: oslo-specs | 12:01 |
| sdague | dims_: ok | 12:01 |
| dims_ | samueldmq: ack thanks | 12:01 |
| samueldmq | dims_, yes, your = nova ;) | 12:01 |
| dims_ | samueldmq: i am wearing my oslo hat for this discussion :) | 12:01 |
| sdague | samueldmq: right, so I think the dynamic policy architecture needs to get exposed to more projects, because there were some assumptions there which I think don't match with the way projects think about things | 12:01 |
| * dims_ switching hats | 12:02 | |
| * mdbooth just wrote a 36 line comment explaining why the next 3 lines don't do the obvious thing | 12:02 | |
| sdague | it has far reaching consequences, and I'd hate to see us log jammed late on that one | 12:02 |
| dims_ | mdbooth: "Trust me, this works!" should be enough? :) | 12:02 |
| samueldmq | sdague, ++ I agree | 12:02 |
| mdbooth | dims_: For some cores, maybe ;) | 12:03 |
| samueldmq | sdague, I will sync up with ayoung (Adam Young), and once we have agreed on the overview spec internally (https://review.openstack.org/#/c/147651/) | 12:03 |
| *** radek_ has joined #openstack-nova | 12:03 | |
| samueldmq | sdague, we should be starting a ml thread, etc (what is needed ..) | 12:04 |
| samueldmq | morganfainberg, cc ^ | 12:04 |
| samueldmq | sdague, please feel free to ping me/add me to the spec review in gerrit | 12:04 |
| samueldmq | sdague, dims_ thanks | 12:05 |
| dims_ | samueldmq: thanks for bringing this up | 12:05 |
| samueldmq | dims_, np .. this way we make it as better as we can, together :) | 12:06 |
| *** haigang has quit IRC | 12:06 | |
| *** baoli has joined #openstack-nova | 12:07 | |
| *** haigang has joined #openstack-nova | 12:07 | |
| *** nihilifer has joined #openstack-nova | 12:07 | |
| *** tobe has quit IRC | 12:07 | |
| *** achanda has joined #openstack-nova | 12:09 | |
| openstackgerrit | wangxiyuan proposed openstack/nova: Add specific error information when createBackup is not supported https://review.openstack.org/187445 | 12:10 |
| openstackgerrit | wuhao proposed openstack/nova-specs: Add previous page links for Nova GET servers API https://review.openstack.org/165041 | 12:11 |
| *** thiagop has quit IRC | 12:11 | |
| *** ildikov has quit IRC | 12:12 | |
| *** ssurana has joined #openstack-nova | 12:13 | |
| *** ivasev has joined #openstack-nova | 12:14 | |
| *** sdake has joined #openstack-nova | 12:15 | |
| samueldmq | dims_, sdague I will mark the bug as confirmed on nova side, makes sense ? (even if it depends on other changes, it is valid) | 12:15 |
| sdague | samueldmq: yep, confirmed works for me | 12:16 |
| samueldmq | sdague, great, thanks | 12:16 |
| *** sdake has quit IRC | 12:16 | |
| *** yingjun has quit IRC | 12:16 | |
| *** rfolco has joined #openstack-nova | 12:16 | |
| *** claudiub has joined #openstack-nova | 12:17 | |
| *** jistr is now known as jistr|class | 12:17 | |
| *** sdake_ has quit IRC | 12:17 | |
| openstackgerrit | sahid proposed openstack/nova-specs: libvirt: set admin root password https://review.openstack.org/187509 | 12:17 |
| *** ssurana has quit IRC | 12:18 | |
| *** ijw has joined #openstack-nova | 12:18 | |
| *** radek_ has quit IRC | 12:20 | |
| *** achanda has quit IRC | 12:20 | |
| *** pask81 has joined #openstack-nova | 12:21 | |
| *** achanda has joined #openstack-nova | 12:21 | |
| *** jistr|class is now known as jistr | 12:22 | |
| *** ijw has quit IRC | 12:23 | |
| garyk | jogo: https://review.openstack.org/#/c/163043/ - this now has the VMware CI +1 | 12:24 |
| *** vladikr has joined #openstack-nova | 12:24 | |
| openstackgerrit | garyk proposed openstack/nova: Compute: improve logging using {} instead of dict https://review.openstack.org/187549 | 12:25 |
| *** achanda has quit IRC | 12:25 | |
| *** ildikov has joined #openstack-nova | 12:27 | |
| *** ajayaa has quit IRC | 12:29 | |
| *** aysyanne has joined #openstack-nova | 12:30 | |
| *** b3nt_pin is now known as beagles | 12:31 | |
| *** bkopilov has quit IRC | 12:32 | |
| *** sdake has joined #openstack-nova | 12:32 | |
| *** sdake has quit IRC | 12:34 | |
| RuiChen | any chance that who can please look at this spec https://review.openstack.org/#/c/169638/ | 12:35 |
| *** sdake has joined #openstack-nova | 12:40 | |
| lxsli | johnthetubaguy: mikal: is there a recommended hotel for the midcycle please? | 12:40 |
| lxsli | mriedem_away: ^^ | 12:46 |
| openstackgerrit | Adelina Tuvenie proposed openstack/nova: Adds Hyper-V vNic hot-plug implementation https://review.openstack.org/138255 | 12:51 |
| openstackgerrit | Adelina Tuvenie proposed openstack/nova: Moves OVS related code to a common location https://review.openstack.org/179727 | 12:51 |
| openstackgerrit | Adelina Tuvenie proposed openstack/nova: Adds Hyper-V OVS ViF driver https://review.openstack.org/140045 | 12:51 |
| garyk | claudiub: can you please check out my comments on https://review.openstack.org/#/c/169732. the test case is already covered | 12:52 |
| garyk | claudiub: this is a super importnat blocking bug… | 12:52 |
| *** sudipto has quit IRC | 12:54 | |
| *** ildikov has quit IRC | 12:55 | |
| *** ildikov has joined #openstack-nova | 12:56 | |
| *** kylek3h has joined #openstack-nova | 12:59 | |
| mriedem_away | lxsli: not really, there aren't any discount rates i'm aware of. there are quite a few hotels around the ibm campus though so if you want to go close and not too expensive that's probably available | 13:00 |
| *** oro has quit IRC | 13:01 | |
| mriedem_away | the more expensive hotels are downtown near the clinic/hospitals | 13:01 |
| *** yassine has joined #openstack-nova | 13:01 | |
| *** mriedem_away is now known as mriedem | 13:01 | |
| *** yassine has quit IRC | 13:01 | |
| *** mdrabe has joined #openstack-nova | 13:02 | |
| *** yassine has joined #openstack-nova | 13:02 | |
| lxsli | mriedem: OK, we were just hoping to stay in the same place as some others for ease of organising after work drinks + meals | 13:04 |
| sdague | johnthetubaguy: so I'm writing a lot of english right now about the Nova API changes | 13:05 |
| mriedem | lxsli: i haven't heard of any others booking a place yet | 13:05 |
| sdague | I've got a super raw buffer, which is how these things tend to work for me - https://www.dropbox.com/s/wb2lbwq63n1xg9p/NovaAPI.rst?dl=0 | 13:05 |
| sdague | while I think the history bits are long term useful somewhere, honestly, I'd skip them for now, because I mostly just had to get that out of my system before explaining what we had | 13:06 |
| sdague | Line 162 forward is the part that would be good to look at (starts with A New Beginning) | 13:06 |
| sdague | not complete yet, but feedback is probably useful at this point | 13:07 |
| openstackgerrit | Matthew Booth proposed openstack/nova: Implement compare-and-swap for instance update https://review.openstack.org/141115 | 13:07 |
| *** _gryf is now known as gryf|afk | 13:08 | |
| *** pask81 has quit IRC | 13:08 | |
| *** kiran-r has quit IRC | 13:09 | |
| openstackgerrit | Matthew Booth proposed openstack/nova: Remove unused variables from images api https://review.openstack.org/183475 | 13:09 |
| *** sdake_ has joined #openstack-nova | 13:09 | |
| *** emagana has joined #openstack-nova | 13:09 | |
| *** ajo has joined #openstack-nova | 13:09 | |
| openstackgerrit | Markus Zoeller (markus_z) proposed openstack/nova: docs: Add bug handling description to developer reference https://review.openstack.org/187571 | 13:10 |
| *** zhangtralon has quit IRC | 13:11 | |
| *** emagana has quit IRC | 13:12 | |
| *** emagana has joined #openstack-nova | 13:12 | |
| *** sdake has quit IRC | 13:13 | |
| garyk | sdague: if you are dealing with the api would it be possible that you also please look at https://review.openstack.org/#/c/177778/ | 13:15 |
| lxsli | mriedem: OK, I threw a dart and looks like we'll be in Homewood Suites | 13:15 |
| johnthetubaguy | lxsli: I know there wasn't one with a special rate, but yeah, a bit like palo alto | 13:16 |
| mriedem | lxsli: oh, the one with all of those murders a few years back | 13:16 |
| lxsli | sdague: ^^ | 13:16 |
| mriedem | good luck! | 13:16 |
| garyk | lxsli: i heard that people who do not find accomodation can crash in mriedem s garden | 13:16 |
| mriedem | go ahead, sprinklers start at 3am | 13:16 |
| garyk | :) | 13:16 |
| *** emagana has quit IRC | 13:16 | |
| lxsli | mriedem: we all live in castles over here of course so we're used to a few ghosts | 13:17 |
| johnthetubaguy | sdague: hey, hows that english going? | 13:17 |
| bauzas | folks, seriously considering to be remote for the midcycle this time :/ | 13:17 |
| claudiub | garyk: done | 13:17 |
| mriedem | homewood suites is probably nice, a bit on the spendy side since it's right across the street from the hospital | 13:17 |
| mriedem | but whatevs | 13:17 |
| *** mwagner_afk has joined #openstack-nova | 13:18 | |
| lxsli | apparently getting ill is the #1 thing to do in Rochester | 13:18 |
| *** sdake has joined #openstack-nova | 13:18 | |
| lxsli | #2 appears to be a graveyard | 13:18 |
| garyk | claudiub: gracias! | 13:18 |
| *** ijw has joined #openstack-nova | 13:18 | |
| mriedem | lxsli: i think getting cured is the idea | 13:18 |
| mriedem | most people here aren't from here, hence the bazillion hotels | 13:19 |
| mriedem | lxsli: anyway, you're within walking distance of a good breakfast place, so there's that at least | 13:19 |
| moshele | baoli: hi | 13:21 |
| baoli | moshele: Hi | 13:21 |
| *** sdake_ has quit IRC | 13:21 | |
| lxsli | mriedem: as long as there are pancakes <3 | 13:22 |
| bauzas | mriedem: I just looked at the air routes, but RST seems really hard | 13:22 |
| mriedem | bauzas: i know there was a direct delta flight from paris to MSP | 13:22 |
| bauzas | mriedem: do you know if there would be an alternative, ie. a bigger airport than Rochester, and then go by rail ? | 13:22 |
| lxsli | +1, we're flying LHR-MSP and driving for 1h25m | 13:23 |
| mriedem | bauzas: but i'm not sure it's worth it for a 3 day meetup, 8-9 hours of flying both ways | 13:23 |
| moshele | baoli: pci meeting, did I missed it? | 13:23 |
| bauzas | mriedem: luckily, I'm not a Parisian | 13:23 |
| mriedem | bauzas: i think most people will fly into MSP and then rent a car | 13:23 |
| bauzas | mriedem: how far is it ? | 13:23 |
| lxsli | 80 miles | 13:23 |
| mriedem | yeah, about an hour and a half drive or so | 13:23 |
| *** ijw has quit IRC | 13:23 | |
| bauzas | mriedem: oooh | 13:23 |
| bauzas | mriedem: good to know | 13:23 |
| mriedem | there are also shuttles that run between MSP and rochester | 13:23 |
| bauzas | because atm, the connections are pretty bad from my place | 13:24 |
| bauzas | and that's awfully expensive :/ | 13:24 |
| mriedem | http://www.gorochesterdirect.com/ | 13:24 |
| mriedem | ^ is the shuttle | 13:24 |
| bauzas | lemme check with MST | 13:24 |
| bauzas | (gosh the name is funny) | 13:24 |
| bauzas | don't find the French acronym | 13:24 |
| bauzas | oh MSP | 13:25 |
| bauzas | my bad :) | 13:25 |
| baoli | moshele, not really. I remember that during summit "P" said he was going to call a biweekly meeting | 13:25 |
| *** yingjun has joined #openstack-nova | 13:26 | |
| moshele | baoli: you still call him "P" :) , so it is next week | 13:26 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/186935 | 13:27 |
| *** Nisha has joined #openstack-nova | 13:27 | |
| baoli | moshele: still struggling to learn to pronounce it, and he didn't want to teach me | 13:27 |
| moshele | baoli: I hear you. I have the same problem | 13:28 |
| bauzas | so, MSP is actually also expensive and with 20 hours of travel (2 hops) | 13:28 |
| bauzas | gosh... | 13:28 |
| *** lucas-hungry is now known as lucasagomes | 13:29 | |
| mriedem | bauzas: you shouldn't feel required to come, it's a long haul for anyone outside the US | 13:29 |
| mriedem | bauzas: i probably won't be going to a meetup in europe if that's what happens in M | 13:29 |
| bauzas | mriedem: well, that's not all about having meetings, that's also a good opportunity for socializing :) | 13:30 |
| bauzas | mriedem: but yeah, sounds it's going to be the plan / | 13:30 |
| bauzas | :/ | 13:30 |
| *** VW has joined #openstack-nova | 13:30 | |
| *** nickchase has joined #openstack-nova | 13:30 | |
| moshele | pczesno: hi | 13:31 |
| bauzas | mriedem: I've been heard that johnthetubaguy was having a solid offer from HP Bristol :) | 13:31 |
| bauzas | for M midcycle I mean | 13:31 |
| dims_ | bauzas: MSP to rochester is a short drive as well, not sure i am helping | 13:32 |
| *** xyang1 has joined #openstack-nova | 13:32 | |
| bauzas | dims_: LYS/MSP is around 1800 USD and 20 hours travel, so I will probably pass | 13:32 |
| *** unicell has quit IRC | 13:32 | |
| dims_ | bauzas: we cans till try google hangouts + conf line when appropriate | 13:33 |
| bauzas | dims_: yeah, mikal made a good thing last time | 13:33 |
| * dims_ remembers mikal's special mike thingy | 13:34 | |
| bauzas | dims_: I would counteroffer the M midcycle to be in Grenoble for the winter season, but that would be harsh for PaulMurray :p | 13:34 |
| dims_ | thomasem: i am all set. thanks | 13:35 |
| *** ajayaa has joined #openstack-nova | 13:39 | |
| kashyap | mriedem: You're still testing this? https://review.openstack.org/#/c/147332/3/nova/utils.py | 13:40 |
| mriedem | kashyap: nope | 13:40 |
| *** ErickHeinrich has joined #openstack-nova | 13:41 | |
| *** ErickHeinrich has quit IRC | 13:41 | |
| *** rushiagr is now known as rushiagr_away | 13:41 | |
| *** marun has joined #openstack-nova | 13:45 | |
| *** yingjun has quit IRC | 13:46 | |
| *** yamahata has joined #openstack-nova | 13:48 | |
| claudiub | mriedem: I gots a question, if you don't mind | 13:49 |
| *** burt has joined #openstack-nova | 13:49 | |
| claudiub | mriedem: is this valid? https://review.openstack.org/#/c/184339/ | 13:49 |
| *** otter768 has joined #openstack-nova | 13:49 | |
| claudiub | mriedem: in requirements.txt, oslo.messaging is >=1.4.0,<1.5.0 | 13:49 |
| mriedem | claudiub: so, that came up at the ops session on rabbitmq in vancouver | 13:49 |
| mriedem | claudiub: b/c some big operators using cells were using newer oslo.messaging for the heartbeat patch | 13:50 |
| sdague | johnthetubaguy: https://www.dropbox.com/s/wb2lbwq63n1xg9p/NovaAPI.rst?dl=0 | 13:50 |
| claudiub | and the bug it is addressed in the commit message is targeted to 1.50 | 13:50 |
| mriedem | claudiub: yeah, i know the oslo.messaging versions don't line up with g-r | 13:50 |
| *** dboik has joined #openstack-nova | 13:50 | |
| *** dims_ has quit IRC | 13:51 | |
| claudiub | mriedem: I see, so operators are using newer olso.messaging than the recommended one. Ok, ty for the info. :) | 13:51 |
| *** dims_ has joined #openstack-nova | 13:51 | |
| *** mtanino has joined #openstack-nova | 13:51 | |
| mriedem | claudiub: yeah, for that heartbeat patch for rabbit | 13:52 |
| *** jecarey has joined #openstack-nova | 13:52 | |
| mriedem | so they are on juno w/ oslo.messaging 1.8.0 + patches i think, and patches for cells | 13:52 |
| mriedem | like that one ^ | 13:52 |
| claudiub | k, I see. | 13:53 |
| mriedem | updated the meetup wiki with more details for anyone looking at this today: https://wiki.openstack.org/wiki/Sprints/NovaLibertySprint | 13:53 |
| *** armax has joined #openstack-nova | 13:54 | |
| *** otter768 has quit IRC | 13:54 | |
| *** artom has joined #openstack-nova | 13:57 | |
| *** artom has quit IRC | 13:57 | |
| *** artom has joined #openstack-nova | 13:58 | |
| *** marun has quit IRC | 13:58 | |
| *** marun has joined #openstack-nova | 13:59 | |
| *** annegentle has joined #openstack-nova | 13:59 | |
| *** sdake_ has joined #openstack-nova | 14:00 | |
| *** eharney has joined #openstack-nova | 14:01 | |
| openstackgerrit | Thomas Maddox proposed openstack/nova: Adding IPv6 static routes and static routes template https://review.openstack.org/187012 | 14:02 |
| thomasem | dims_: Great! I went ahead and removed the preserve_routes stuff | 14:02 |
| thomasem | thanks! :) | 14:02 |
| *** sdake has quit IRC | 14:03 | |
| sahid | jaypipes: thanks for your reviews, i have replied to the first one and since i would like to address your comments.. - can you take a look? | 14:04 |
| sahid | https://review.openstack.org/#/c/174313/16 | 14:04 |
| jaypipes | sahid: yup, got it. | 14:05 |
| sahid | cool | 14:05 |
| *** moshele has quit IRC | 14:05 | |
| *** sdake_ has quit IRC | 14:07 | |
| *** sdake has joined #openstack-nova | 14:08 | |
| *** annegentle has quit IRC | 14:08 | |
| dims_ | mriedem: want me to throw in this to the mid-cycle wiki? http://www-03.ibm.com/systems/resources/Driving_parking.pdf | 14:09 |
| *** gokrokve has joined #openstack-nova | 14:09 | |
| dims_ | mriedem: there's one more - http://www-03.ibm.com/systems/resources/TravelDirections2012.pdf | 14:09 |
| *** bkopilov has joined #openstack-nova | 14:11 | |
| *** tonytan4ever has joined #openstack-nova | 14:12 | |
| *** mohitsharma has joined #openstack-nova | 14:14 | |
| bauzas | dims_: so you hacked IBM ? :p | 14:14 |
| * bauzas tries to joke | 14:15 | |
| dims_ | bauzas: used to go often there to build systems :) those pdf(s) helped | 14:15 |
| *** annegentle has joined #openstack-nova | 14:15 | |
| *** jistr is now known as jistr|mtg | 14:16 | |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: scheduler: weight nodes based on their cached images https://review.openstack.org/173304 | 14:18 |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: Add new image cache monitor in compute nodes https://review.openstack.org/173303 | 14:18 |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: compute monitors: set update delta from base class https://review.openstack.org/173987 | 14:18 |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: vmware: add image_cache_manager property https://review.openstack.org/187602 | 14:18 |
| *** boris-42 has quit IRC | 14:18 | |
| *** ijw has joined #openstack-nova | 14:18 | |
| openstackgerrit | Daniel Berrange proposed openstack/nova: libvirt: convert imagebackend to support nova.virt.image.model classes https://review.openstack.org/132024 | 14:19 |
| openstackgerrit | Daniel Berrange proposed openstack/nova: virt: convert VFS API to use nova.virt.image.model https://review.openstack.org/132022 | 14:19 |
| openstackgerrit | Daniel Berrange proposed openstack/nova: virt: convert disk API over to use nova.virt.image.model https://review.openstack.org/132023 | 14:19 |
| openstackgerrit | Daniel Berrange proposed openstack/nova: virt: introduce model for describing local image metadata https://review.openstack.org/132020 | 14:19 |
| openstackgerrit | Daniel Berrange proposed openstack/nova: virt: convert disk mount API to use nova.virt.image.model https://review.openstack.org/132021 | 14:19 |
| *** radek_ has joined #openstack-nova | 14:19 | |
| *** nkrinner has quit IRC | 14:20 | |
| *** ajayaa has quit IRC | 14:21 | |
| *** gszasz has quit IRC | 14:21 | |
| openstackgerrit | Baodong (Robert) Li proposed openstack/nova: Handle port delete event https://review.openstack.org/179390 | 14:21 |
| *** mdrabe has quit IRC | 14:22 | |
| *** ijw has quit IRC | 14:23 | |
| *** gszasz has joined #openstack-nova | 14:24 | |
| dims_ | garyk: easy oslo.vmware review when you get a chance - https://review.openstack.org/#/c/183654/ | 14:25 |
| bauzas | jaypipes: ndipanov: dansmith: alaski: FYI http://lists.openstack.org/pipermail/openstack-dev/2015-June/065463.html | 14:25 |
| garyk | dims_: thanks. looking now | 14:26 |
| openstackgerrit | Markus Zoeller (markus_z) proposed openstack/nova: docs: Add bug handling description to developer reference https://review.openstack.org/187571 | 14:26 |
| *** fawadkhaliq has joined #openstack-nova | 14:26 | |
| garyk | dims_: minor nit with import section. other than that looks great. Can you update if you have a sec | 14:27 |
| dims_ | garyk: ack | 14:27 |
| garyk | gracias | 14:27 |
| jaypipes | bauzas: yeah, I saw that... need to read it through and provide you feedback. | 14:27 |
| bauzas | jaypipes: np | 14:28 |
| bauzas | jaypipes: I tried to discuss on the high-level | 14:28 |
| alaski | bauzas: yeah, just saw that | 14:28 |
| bauzas | jaypipes: ie. should we use the request spec for more than what's used now ? | 14:28 |
| dims_ | garyk: done. thanks | 14:29 |
| *** READ10 has joined #openstack-nova | 14:29 | |
| garyk | dims_: done. thanks! | 14:30 |
| *** emagana has joined #openstack-nova | 14:30 | |
| *** jistr|mtg is now known as jistr | 14:30 | |
| garyk | sorry for taking a while to getting round to this. too many rebases as of late | 14:30 |
| alaski | bauzas: just thinking out loud for a second, but what about turning this around? Why does this duplicated info need to be stored on the instance object? | 14:32 |
| alaski | I guess that's sort of what ndipanov is saying in the second part of his review comment | 14:32 |
| luv | hey. i see there have been changes before juno and current git master in how security groups are handled | 14:33 |
| luv | in juno, secgroups listed by nova secgroup-list and neutron security-group-list match | 14:33 |
| luv | in current git master the security groups listed in nova use integers as ids | 14:33 |
| *** thangp has joined #openstack-nova | 14:33 | |
| ndipanov | alaski, it does not but imho | 14:33 |
| ndipanov | this will make us change a lot of assumptions in a lot of places | 14:34 |
| ndipanov | assumption being that instance is what gets passed into everything on the compute host | 14:34 |
| luv | im running in an issue regarding this. If I do "nova secgroup-add-rule default tcp 22 22 0.0.0.0/0" the rule is not ignored by neutron openvswitch agent but if i do "neutron security-group-rule-create --port-range-min 22 --port-range-max 22 --ethertype ipv4 --protocol tcp a73604dc-bde4-46cd-802a-7ea3be3fde2d" the rule works fine | 14:34 |
| ndipanov | alaski, there was a parallel discussion about the fact that we need to save the _new and _old for "complicated" resources like NUMA when doing migrates | 14:35 |
| ndipanov | and the question is where to stick it | 14:35 |
| ndipanov | (this is for a not completely unrelated bug :) ) | 14:36 |
| ndipanov | and everyone seems to have been more comfortable with this being part of instance_extra than anything else... | 14:36 |
| ndipanov | I am unsure if this is the write answer in the grand scheme of things :) | 14:37 |
| alaski | ndipanov: very true re: compute manager apis needing a huge change to accomadate a change like that | 14:37 |
| ndipanov | but I think making things require a spec object for data we asumed will be on the instance will make for a lot of refactoring | 14:37 |
| ndipanov | yes | 14:37 |
| ndipanov | right* not write answer :) | 14:38 |
| luv | http://pastebin.blesmrt.net/prxnwhbbq#line-1 ... Im really confused here why i see "id=1" but in the docs http://docs.openstack.org/kilo/install-guide/install/apt/content/launch-instance-neutron.html the ID is an identifier like a neutron secgroup id | 14:38 |
| alaski | right now we have pretty convoluted models for things, and I'm trying to think about what it might look like in a more ideal situation | 14:38 |
| luv | but if i do "describe security_groups;" in nova database id is int(11) | 14:38 |
| mdbooth | My Jenkins job has failed, all subtasks complete, but it's still sat in the gate | 14:38 |
| ndipanov | alaski, my somwhat pragmatic suggestion was | 14:38 |
| mdbooth | anybody know why that might be? | 14:38 |
| ndipanov | make a nice clear API on the instance object | 14:39 |
| bauzas | alaski: ndipanov: sorry, I have been diverted for 10 mins by a 9mo goblin | 14:39 |
| luv | have there been changes between kilo and current git in this respect? | 14:39 |
| luv | im also noticed following in the logs: http://pastebin.blesmrt.net/psghyiunw | 14:39 |
| mdbooth | Ah, never mind. Browser snafu by the looks of it. Manual refresh shows unfinished job. | 14:39 |
| *** cfriesen has joined #openstack-nova | 14:40 | |
| openstackgerrit | garyk proposed openstack/nova: VMware: detach cinder volume when instance destroyed https://review.openstack.org/169732 | 14:40 |
| *** baoli has quit IRC | 14:40 | |
| jaypipes | luv: probably best to write a ML post to get answers on that... | 14:40 |
| *** mdrabe has joined #openstack-nova | 14:40 | |
| ndipanov | bauzas, good for you my 3wo goblin is asleep so guess when he will be wide awake... | 14:40 |
| bauzas | alaski: ndipanov: so, the question I would say is "do we make sure that we need more than what's currently in the spec", and ndipanov had a nice insight with the flavor stuff | 14:41 |
| *** n0ano has joined #openstack-nova | 14:41 | |
| bauzas | ndipanov: eh | 14:41 |
| alaski | ndipanov: that sounds like a good approach to me. I'm trying to sort out what the reason is for certain things, like a lot of stuff in the request spec is only there for movement of an instance. but something like numa_topology may have more meaning than that, and I'm not sure how to conceptualize it properly | 14:41 |
| bauzas | ndipanov: at least, I'm making sure not to food her after midnight | 14:41 |
| bauzas | ndipanov: alaski: which kind of nice "API" would you see for that ? | 14:42 |
| *** Hosam has joined #openstack-nova | 14:42 | |
| ndipanov | bauzas, I think I had an example | 14:42 |
| openstackgerrit | garyk proposed openstack/nova: VMware: remove redundant check for block devices https://review.openstack.org/170641 | 14:42 |
| bauzas | ndipanov: oh it would be welcomed | 14:42 |
| ndipanov | something like instance.current_flavor(request_spec) | 14:43 |
| *** baoli has joined #openstack-nova | 14:43 | |
| *** zz_jgrimm is now known as jgrimm | 14:43 | |
| ndipanov | which would also check for migration objects created etc | 14:43 |
| bauzas | ndipanov: and then the spec would have the whole instance object so this helper would basically check if that's the new or the old flavor ? | 14:43 |
| *** armax has quit IRC | 14:44 | |
| bauzas | attached to the instance stored in the spec, I mean | 14:44 |
| ndipanov | bauzas, well I haven't really thought it out :) | 14:44 |
| *** armax has joined #openstack-nova | 14:44 | |
| ndipanov | I assumed only the reference to it | 14:44 |
| dansmith | +1 for the reference if we can get away with it | 14:44 |
| *** Nisha has quit IRC | 14:45 | |
| dansmith | we really only should nest them if we need to pass the parent and child together over rpc to save a db hit later | 14:45 |
| bauzas | ndipanov: you mean the reference to the instance is stored in the spec object ? | 14:45 |
| *** coolsvap|afk has quit IRC | 14:45 | |
| ndipanov | the idea was to use the Instance object to store the logic (and thus design API's in that fashion) | 14:45 |
| ndipanov | since this is really waht request object is about - what's happening to an instance | 14:46 |
| ndipanov | however flipping it on it's head | 14:46 |
| ndipanov | and making the request object central | 14:46 |
| ndipanov | might be too much work | 14:46 |
| bauzas | that said, a request can be for multiple instances | 14:46 |
| luv | jaypipes: do i have to register to openstack ML or can I just email? (different open source MLs have different rules ...) | 14:47 |
| bauzas | at least until we get rid of min_count/max_count | 14:47 |
| dansmith | bauzas: but we would persist one request spec for each instance we create, right? | 14:47 |
| bauzas | dansmith: I would assume a N:1 relationship yup | 14:47 |
| bauzas | dansmith: ie. multiple instances can have the same spec reference | 14:48 |
| *** henrique_ has quit IRC | 14:48 | |
| dansmith | no, see what I meant was, | 14:48 |
| jaypipes | luv: you need to register: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev | 14:48 |
| dansmith | we would have an initial request spec that we dupe and save per instance in the multi-create case | 14:48 |
| *** htruta has joined #openstack-nova | 14:48 | |
| luv | umm, i better dig in the code and git blame then *punk smiley* | 14:48 |
| *** coolsvap|afk has joined #openstack-nova | 14:48 | |
| *** VW has quit IRC | 14:49 | |
| bauzas | dansmith: oic | 14:49 |
| ndipanov | one other thing that occurred to me when reviewing that spec is our famous DB races | 14:49 |
| ndipanov | so from that POV I'd prefer the model where | 14:49 |
| bauzas | for the reference, the current approved model for persisting is http://specs.openstack.org/openstack/nova-specs/specs/liberty/approved/persist-request-spec.html | 14:49 |
| ndipanov | transact stuff to an instance | 14:49 |
| ndipanov | and use request object just to talk to the scheduler | 14:49 |
| bauzas | ndipanov: in that case, you don't need the whole Instance | 14:50 |
| bauzas | ndipanov: and I prefer to only keep what's checked by the filters and consume_from_instance() | 14:50 |
| alaski | ndipanov: if I'm understanding you correcty, that's what I would prefer as well. But we can have a requestspec before we have an instance, so the ownership is hard to model that way | 14:51 |
| dansmith | bauzas: I don't think that spec prescribes a behavior for multi-create, does it? | 14:51 |
| *** exploreshaifali has joined #openstack-nova | 14:51 | |
| bauzas | dansmith: it doesn't | 14:51 |
| bauzas | dansmith: but it implicitely accepts to duplicate | 14:51 |
| bauzas | dansmith: since it's based on instance_extra | 14:51 |
| ndipanov | alaski, yeah.... | 14:51 |
| ndipanov | alaski, the more I think about it having 1 thing (Instance) and no request_spec would be very useful | 14:52 |
| ndipanov | except for a lot of RPC overhead | 14:52 |
| *** exploreshaifali has quit IRC | 14:52 | |
| bauzas | ndipanov: RequestSpec was originally designed for being used solely by the Scheduler | 14:53 |
| bauzas | ndipanov: and just replace that dict | 14:53 |
| bauzas | including filter_properties | 14:53 |
| openstackgerrit | Petrut Lucian proposed openstack/nova: Libvirt: Fix SMBFS volume attach race condition https://review.openstack.org/187619 | 14:53 |
| dansmith | ndipanov: you mean just storing the request details with the instance, yes? | 14:53 |
| ndipanov | bauzas, well - the thing is that it is also kind of used for communicating retries | 14:53 |
| bauzas | which means that it keeps more than just an instance | 14:53 |
| alaski | ndipanov: what I really want is a way to get an object to send to the scheduler, I'm less concerned with where that data is persisted(on instance or duped in a request_spec table) | 14:53 |
| bauzas | ndipanov: agreed, that's why I proposed to model it using the Retry object | 14:54 |
| mdbooth | My jenkins job has a sub-job, check-tempest-dsvm-cells, which is still running after 1hr 46mins | 14:54 |
| dansmith | alaski: how about we persist the request spec in the api database during build and then stash it in instance_extra once the instance is built? | 14:54 |
| mdbooth | Looking at another, successful job, it completed in 18 minutes | 14:54 |
| *** jaypipes has quit IRC | 14:54 | |
| mdbooth | Is there any way to kill it? | 14:54 |
| alaski | ndipanov: but I need a way to get that scheduler object before an instance has been created, or to rethink instances a bit | 14:54 |
| dansmith | mdbooth: you can try pushing a new rev | 14:54 |
| mdbooth | Preferrably with fire, as it's non-voting anyway | 14:54 |
| alaski | dansmith: I'm good with that, I think I had that in one of my spec revisions | 14:55 |
| dansmith | that makes the most sense to me | 14:55 |
| mdbooth | dansmith: Well I have 2 other failures which I'd genuinely like to investigate. They're probably bogus, but not necessarily. | 14:55 |
| ndipanov | dansmith, that seems like what I'm saying | 14:55 |
| alaski | dansmith: it complicates rehydrating the object a bit, but seems doable | 14:55 |
| dansmith | alaski: why? | 14:55 |
| dansmith | mdbooth: ah, then I think you need to wait for the job timeout | 14:56 |
| bauzas | ok, initially the question was not about how to persist it, but what kind of information to persist :) | 14:56 |
| alaski | dansmith: because it can come from the api db in one format, or instance extra in another. Since I'm assuming we may only store info in instance_extra that's not already on instance | 14:56 |
| dansmith | alaski: I don't think that makes it hard at all | 14:56 |
| dansmith | alaski: we store flavors in the flavors table, and in instance_extra | 14:56 |
| dansmith | pulling them out of either location is different, but not hard | 14:56 |
| alaski | dansmith: no, not hard. just more complicated than what has been done to this point | 14:56 |
| dansmith | alaski: okay | 14:57 |
| *** VW has joined #openstack-nova | 14:57 | |
| alaski | I'll propose a spec update with that | 14:58 |
| ndipanov | I'm really mostly concerned with the migration case | 14:58 |
| bauzas | dansmith: alaski: ndipanov: I have to run for the scheduler meeting | 14:58 |
| bauzas | alaski: feel free to CC me on your spec change | 14:58 |
| alaski | bauzas: can do | 14:59 |
| ndipanov | we are moving an instance around - what is a) the data scheduler needs b) the data both resource trackers need and how to have it accessible through a not horrible API | 14:59 |
| bauzas | alaski: but I think we haven't agreed on what to store *within* the spec :) | 14:59 |
| ndipanov | for the record '_new' is a horrible API :) | 14:59 |
| ndipanov | bauzas, I don't think we have it figured out | 15:01 |
| bauzas | ndipanov: yeah hence my point :) | 15:01 |
| bauzas | ndipanov: that said, that discussion was interesting because we told about some new things about how to persist it | 15:01 |
| ndipanov | currently what alaski said made me think... what if request_spec is what is being done to the instance at this very moment... | 15:02 |
| bauzas | and the relationship with an instancze | 15:02 |
| ndipanov | so then we have on boot a blank instance, and a request spec | 15:02 |
| dansmith | ndipanov: you mean having request and new_request for a resize is a horrible API? | 15:02 |
| *** hemnafk is now known as hemna | 15:02 | |
| ndipanov | dansmith, more like naming is horrible and the fact that we pass a _new string | 15:02 |
| *** patrickeast has joined #openstack-nova | 15:02 | |
| ndipanov | not the very existance of it | 15:03 |
| dansmith | ndipanov: is this specified somewhere in a spec or something? | 15:03 |
| ndipanov | dansmith, what? | 15:03 |
| dansmith | ndipanov: I feel like you're talking about some specific thing someone has suggested doing, or that we're currently doing.. the "_new" string bit | 15:04 |
| bauzas | ndipanov: honestly, what we need to pass to the scheduler is like a flavor saying 'please decrement your usage by this' | 15:04 |
| dansmith | or do you mean the flavor stashing namespace example/ | 15:04 |
| *** jcoufal has quit IRC | 15:04 | |
| bauzas | ndipanov: what I still wonder is why we need to have more than a flavor and some extra fields like AZ | 15:04 |
| openstackgerrit | Deepak C Shetty proposed openstack/nova: Nova part of volume snapshot improvements https://review.openstack.org/172813 | 15:05 |
| alaski | dansmith: this conversation spawned from https://review.openstack.org/#/c/145528/12 and a ML post from bauzas | 15:05 |
| lxsli | bauzas: I'm looking (well, rewriting) Jay's resource-objects spec right now | 15:05 |
| ndipanov | dansmith, right so - I think: having flavor_new and flavor_old is very bad from a semantic POV | 15:05 |
| mriedem | dims_: sure, i'd have to review to see if they are still accurate | 15:05 |
| alaski | dansmith: the _new thing wasn't proposed anywhere that I saw | 15:05 |
| dansmith | alaski: right, I missed .. ah okay | 15:05 |
| ndipanov | rather we have sched asking - what flavor do I need to consider now | 15:06 |
| ndipanov | and we have 2 RTs asking the same question for tracking | 15:06 |
| ndipanov | and the answer is different and _new and _old gives us no clear way of mapping the data to the answer | 15:06 |
| ndipanov | without knowing a bunch of other things | 15:07 |
| dansmith | hmm, not sure I get it | 15:07 |
| *** arnaud____ has joined #openstack-nova | 15:07 | |
| dansmith | if we have a request_spec and we want to change that to ask the scheduler for a different thing, feels like we need to store a request_spec_new as well | 15:07 |
| dansmith | bauzas: I think the reason to have an actual request_spec and not just a flavor+AZ is if we go forward with something like jaypipes was suggesting, a request may be a flavor+option1+option7+option3 | 15:08 |
| ndipanov | well we really only ever have one request spec and a series of obsolete ones | 15:08 |
| bauzas | technically a resize is another request | 15:08 |
| ndipanov | one or 0 that is | 15:08 |
| ndipanov | if instance is just staying where it is and no requests are in flight | 15:08 |
| edleafe | dansmith: +1 | 15:09 |
| dansmith | ndipanov: I dunno, splitting hairs I guess | 15:09 |
| edleafe | ndipanov: if an instance was created and then moved, why do we need the original spec? | 15:10 |
| bauzas | dansmith: I agree, that's more than just a flavor, but looking at the consumers of the spec, it's just redunding data | 15:10 |
| ndipanov | well I'm not sure about that since there are several bugs that have been in nova forever that are exactly the consequence of not having a clear model | 15:10 |
| ndipanov | edleafe, well we don't once the request is successful (resize is confirmed for example) | 15:10 |
| dansmith | edleafe: once it's moved we don't, but since moving it is a long process, we do need two at times | 15:10 |
| *** zzzeek has joined #openstack-nova | 15:10 | |
| ndipanov | but up until then you need all the data doubled | 15:10 |
| ndipanov | and I guess what we are trying to solve here is which data needs to go to the scheduler and which doesn't | 15:11 |
| bauzas | ndipanov: I was speaking of https://github.com/openstack/nova/blob/master/nova/scheduler/filter_scheduler.py#L172 | 15:11 |
| ndipanov | and where is the best place to keep them | 15:11 |
| bauzas | ndipanov: which is why we need to pass an instance-ish object | 15:11 |
| alaski | I think we have a pretty good view of what data the scheduler needs, just not where to store it all | 15:12 |
| edleafe | ndipanov: sure - I thought you wanted to persist N specs forever | 15:12 |
| bauzas | ndipanov: plus some filters that are consuming the instance properties, like the NUMA one | 15:12 |
| ndipanov | edleafe, well we kind of already do that with migration objects | 15:12 |
| ndipanov | there's a ton of overlap between the 2 actually as well | 15:12 |
| edleafe | ndipanov: even if we do that now, *should* we be doing that? Is it important? | 15:13 |
| ndipanov | edleafe, no no - but orthogonal to this discussion I think | 15:14 |
| bauzas | okay, that's now 40 mins that we're discussing about it, could we try to find a direction for http://lists.openstack.org/pipermail/openstack-dev/2015-June/065463.html ? | 15:15 |
| edleafe | ndipanov: well, unless you were thinking that because we do that with migration objects, we should do it here | 15:15 |
| bauzas | ie. what's needed by the scheduler is part of the existing spec object | 15:16 |
| ndipanov | edleafe, I mean keeping stuff forever is orthogonal to this discussion - because soft deletes are what they are | 15:16 |
| *** agireud has joined #openstack-nova | 15:16 | |
| bauzas | if we consider to add more than that, then it would mean that we change how we access the information | 15:16 |
| ndipanov | bauzas, right | 15:16 |
| bauzas | ndipanov: yeah, agreed, we should not focus on how to persist, but *what* to persist | 15:17 |
| bauzas | there is a spec for persisting, and alaski told us that he'll provide an update about it | 15:17 |
| bauzas | so the "how" is covered | 15:17 |
| bauzas | but the "what in" isn't | 15:17 |
| *** ijw has joined #openstack-nova | 15:18 | |
| bauzas | so, unless there is a clear win on storing more than what the scheduler needs, I'm in favor of leaving what's in, until someone comes up with a plan to use ReqSpec for more than what's it's used today | 15:19 |
| openstackgerrit | jichenjc proposed openstack/nova: Add common function for v2.1 API flavor_get. https://review.openstack.org/136568 | 15:19 |
| openstackgerrit | jichenjc proposed openstack/nova: Add common function for v2 API flavor_get. https://review.openstack.org/136567 | 15:19 |
| bauzas | saying that if we persist, that's only for the sake of giving it back to the scheduler in case of a later reschedule | 15:19 |
| ndipanov | bauzas, imho - we should flip this question around | 15:19 |
| ndipanov | we should ask ourselves | 15:19 |
| ndipanov | what data does a) RT thread need b) claim thread need c) scheduler service need | 15:20 |
| *** changbl_ has joined #openstack-nova | 15:20 | |
| bauzas | ndipanov: well, RT and claims are being covered by resource-objects | 15:21 |
| bauzas | ndipanov: https://blueprints.launchpad.net/nova/+spec/resource-objects | 15:21 |
| alaski | I definitely think we should figure that out. this is just focused on c) atm | 15:21 |
| bauzas | alaski: fair enough, but atm we at least to do c) | 15:21 |
| bauzas | at least *need | 15:22 |
| bauzas | evezn | 15:22 |
| * bauzas has gloves today | 15:22 | |
| *** markus_z has quit IRC | 15:22 | |
| ndipanov | so if it's just focused on c) - maybe just dupe it on the request_spec as you did bauzas | 15:22 |
| bauzas | ndipanov: for the moment, the contract was on the head of the scheduler | 15:23 |
| ndipanov | and be clear that none of that data is to be used on the compute node, i.e needs to be recalculated | 15:23 |
| bauzas | ndipanov: http://specs.openstack.org/openstack/nova-specs/specs/kilo/approved/resource-objects.html | 15:23 |
| bauzas | ndipanov: ^ is what you ask | 15:23 |
| ndipanov | bauzas, not sure what "on the head of the scheduler" means | 15:23 |
| *** ijw has quit IRC | 15:24 | |
| bauzas | ndipanov: sorry, it was a metaphor about guns | 15:24 |
| bauzas | and bandits | 15:24 |
| bauzas | anyway | 15:24 |
| openstackgerrit | garyk proposed openstack/nova: Compute: replace incorrect instance object with dict https://review.openstack.org/187537 | 15:24 |
| bauzas | so, the plan is to have c) covered by reqspec, and a) and b) covered by resource-objects | 15:25 |
| ndipanov | yeah I guess I'm not an armed robbery enthusiast so it was lost on me - sorry | 15:25 |
| bauzas | well, if you're reading it like a git thing, then yeah :) | 15:26 |
| bauzas | git HEAD | 15:26 |
| bauzas | anyway | 15:26 |
| ndipanov | bauzas, tbh I don't think so - resource objects are just about having a unified API for dealing with resources | 15:26 |
| bauzas | ndipanov: isn't what claims are ? | 15:26 |
| ndipanov | they don't answer how we store them and when duplicated which ones to consider | 15:26 |
| bauzas | ndipanov: because I think that's another step :) | 15:27 |
| bauzas | ndipanov: for the moment, we have big blobs having redundant data | 15:27 |
| *** krtaylor has quit IRC | 15:27 | |
| bauzas | ndipanov: anyone can just add the key he wants | 15:27 |
| ndipanov | bauzas, ok ok | 15:27 |
| ndipanov | got your point | 15:27 |
| bauzas | so I need to stop that :) | 15:27 |
| ndipanov | yeah I don't know - I feel that there is stuff ehre that needs to be thought through a bit better | 15:29 |
| bauzas | for example, I'm not a fan of https://review.openstack.org/#/c/118010/ | 15:29 |
| bauzas | ndipanov: yeah I know | 15:29 |
| bauzas | ndipanov: and I don't disagree with you :) | 15:29 |
| bauzas | ndipanov: what is still unclear in my mind is how ReqSpec can help more than the scheduler | 15:30 |
| *** andreykurilin has quit IRC | 15:30 | |
| bauzas | well, there is the cells effort, for sure | 15:30 |
| ndipanov | bauzas, so is this BP related to persisting that information to have later access to affinity info? | 15:32 |
| ndipanov | or is that "the next step" | 15:32 |
| bauzas | ndipanov: you mean request-spec-object ? | 15:32 |
| ndipanov | yes | 15:32 |
| bauzas | ndipanov: yeah | 15:32 |
| bauzas | ndipanov: since the hints will be stored as well | 15:32 |
| bauzas | ndipanov: we're merging the filter properties and the request spec | 15:33 |
| ndipanov | so in that case - we definitely need to rethink this since that stuff is needed on the instance when migrating | 15:33 |
| bauzas | ndipanov: the general use case was that in case of a migration or an evacuation, the scheduler was called with no filter properties | 15:34 |
| *** andreykurilin has joined #openstack-nova | 15:34 | |
| bauzas | ndipanov: so the scheduler was having zero way to know what the user asked at boot time | 15:34 |
| ndipanov | bauzas, ok so we solve that by creating a whole new table to store it | 15:35 |
| ndipanov | I mean | 15:35 |
| bauzas | ndipanov: what is ran by the compute manager once the scheduler elected it is not covered by this spec | 15:35 |
| bauzas | s/ran/run (dammit != | 15:35 |
| *** barra204 has joined #openstack-nova | 15:35 | |
| *** shakamunyi has joined #openstack-nova | 15:35 | |
| ndipanov | I think this kind of design work is exactly where the whole spec process hopelessly fails | 15:35 |
| bauzas | ndipanov: but I'm not against thinking about it, for sure | 15:35 |
| *** cbader has quit IRC | 15:35 | |
| bauzas | ndipanov: lol | 15:36 |
| ndipanov | no I'm serious | 15:36 |
| bauzas | ndipanov: yeah take it as an ironic laugh | 15:36 |
| alaski | ndipanov: in the past what I've done is create a high level spec to work through this stuff, then split it out into work specs | 15:36 |
| bauzas | sooooo | 15:36 |
| ndipanov | alaski, that seems like the better approach here | 15:36 |
| ndipanov | we really need a high level : there are the problems we want to solve, this is a series of steps that we'll take | 15:37 |
| bauzas | at least we need to make sure that we reach the number of people reviewing the spec so that we're sure that we covered the whole lifecycle | 15:37 |
| *** beagles is now known as beagles_brb | 15:37 | |
| *** boris-42 has joined #openstack-nova | 15:37 | |
| claudiub | alaski: | 15:38 |
| ndipanov | we are talking about at least 4 different issues here | 15:38 |
| claudiub | alaski: I heard you are the database master expert, right? :) | 15:38 |
| ndipanov | that are very much inter-dependent as they deal with the same data model | 15:38 |
| alaski | claudiub: I certainly wouldn't call myself that. but I know a few things | 15:39 |
| claudiub | alaski: well, I hope my problem is among those things. :) So, any idea how to add a new element to an Enum? I have a commit on this, https://review.openstack.org/#/c/185962/ but it's not a proper solution, at least not for postgresql. | 15:40 |
| bauzas | ndipanov: okay, so what are we doing ? | 15:40 |
| bauzas | ndipanov: while I probably missed the compute side of the spec | 15:40 |
| claudiub | alaski: I was hoping for some advice on that. | 15:40 |
| bauzas | ndipanov: shall we just leave that out and continue or just kill the existing progress and discuss ? | 15:40 |
| *** lucasagomes has quit IRC | 15:40 | |
| bauzas | ndipanov: because that spec is around for now 2 cycles and I really would like to land something in the right direction | 15:41 |
| ndipanov | well I don't know | 15:41 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: make _create_domain return a Guest object https://review.openstack.org/182185 | 15:42 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method poweroff to Guest object https://review.openstack.org/182194 | 15:42 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method resume to Guest object https://review.openstack.org/182195 | 15:42 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce a Guest to wrap around virDomain https://review.openstack.org/174313 | 15:42 |
| openstackgerrit | sahid proposed openstack/nova: libvirt: introduce method get_guest to Host object https://review.openstack.org/181969 | 15:42 |
| ndipanov | I didn't know some of this stuff 2 cycles ago | 15:42 |
| openstackgerrit | jichenjc proposed openstack/nova: Add DiskNotFound and VolumeNotFound test https://review.openstack.org/181455 | 15:42 |
| openstackgerrit | jichenjc proposed openstack/nova: Ignore Cinder error when shutdown instance https://review.openstack.org/181454 | 15:42 |
| bauzas | but before doing anything, I just want to make sure that my spec is not basically hurting something bad | 15:42 |
| ndipanov | now we know them so we need to adjust | 15:42 |
| alaski | my opinion is that we shouldn't stop what's in progress, but it would be good to have the big picture laid out for discussion, and understanding, somewhere | 15:42 |
| *** lucasagomes has joined #openstack-nova | 15:42 | |
| *** VirtualPCGuy has joined #openstack-nova | 15:42 | |
| bauzas | so the question is turned into 'is the current design flexible enough for being adapted in the future ?' | 15:42 |
| ndipanov | bauzas, in this case - probably reducing the scope of the spec to just objectifying stuff would not block it | 15:42 |
| *** openstackgerrit has quit IRC | 15:42 | |
| bauzas | ndipanov: that's exactly the rationale :) | 15:43 |
| ndipanov | while we can accept changes to the object later | 15:43 |
| alaski | claudiub: is your question about the bd migration in the previous patch, or the object change? | 15:43 |
| *** Nisha has joined #openstack-nova | 15:43 | |
| *** openstackgerrit has joined #openstack-nova | 15:43 | |
| cfriesen | bauzas: ndipanov: alaski: somewhere in all this we should be factoring in the fact we need to do proper resource tracking over migrate/live-migrate/resize/rebuild/evacuate | 15:43 |
| bauzas | ndipanov: ie. define a contract between the conductor and the scheduler | 15:43 |
| ndipanov | persistence part is imho in need of further discussion | 15:43 |
| ndipanov | cfriesen, yes | 15:43 |
| ndipanov | that's why I'm complaining mostly :) | 15:44 |
| *** agireud has quit IRC | 15:44 | |
| bauzas | ndipanov: if we're just storing that spec information in a separate table and read it only when the conductor is calling the scheduler, then it doesn't really hurt any model change we could do | 15:45 |
| claudiub | alaski: oups, my bad. db migration in the previous patch | 15:45 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair index api call https://review.openstack.org/70485 | 15:45 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair show api call https://review.openstack.org/183643 | 15:45 |
| claudiub | alaski: my bad, the commit message titles are similar, I misread. | 15:45 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: tests: functional tests for keypairs create in API microversion 2.4 https://review.openstack.org/183645 | 15:45 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair create api call https://review.openstack.org/183644 | 15:45 |
| alaski | bauzas: what I'm getting from this is that it's hard to reason about the changes in isolation because it's not clear how, if at all, it affects other related areas that people care about | 15:45 |
| ndipanov | bauzas, idk - that really seems like a big hatchet to me | 15:45 |
| ndipanov | I mean - it seems fine to do that just to get somewhere | 15:46 |
| bauzas | ndipanov: tbc, I'm no more opiniated on splitting the scheduler since we made good progress in Nova | 15:46 |
| *** wznoinsk has joined #openstack-nova | 15:46 | |
| ndipanov | but we know what (some of) the downsides are | 15:46 |
| ndipanov | it seems silly to pretend we don't | 15:46 |
| bauzas | ndipanov: so defining clear APIs for the scheduler is cool, but the main point is how we can make sure that we replay the information for each reschedule | 15:46 |
| *** yamahata has quit IRC | 15:46 | |
| *** tsekiyama has joined #openstack-nova | 15:47 | |
| bauzas | ndipanov: so, I think we actually have 2 problems : 1/ a design problem, and 2/ a process problem since we are post-Summit and the next discussions will be either in Rochester, MN or at the Summit | 15:48 |
| bauzas | ndipanov: I'm sincerely not sure we can kick-off some nice effort about redesign how we can manage our global lifecycle for VMs (about resources) | 15:48 |
| bauzas | now | 15:48 |
| ndipanov | idk - by having just a wiki that clearly lists the problems we want to solve isntead of having 15 useless specs that talk about testing and security impact | 15:49 |
| ndipanov | we might have something to go on | 15:49 |
| bauzas | ndipanov: we now have a new process called backlog spec | 15:50 |
| *** otter768 has joined #openstack-nova | 15:50 | |
| bauzas | ndipanov: that's basically a manifesto saying that we have a problem | 15:50 |
| bauzas | ndipanov: and that we need to agree on a plan | 15:50 |
| ndipanov | yes that would be useful | 15:50 |
| bauzas | ndipanov: at least, it describes the problem | 15:50 |
| ndipanov | except it's several bugs here an there in this case that require us to remodel things | 15:51 |
| bauzas | ndipanov: happy with filling a backlog spec at least ? | 15:51 |
| ndipanov | 1) we need scheduler hints for later | 15:51 |
| bauzas | hints are a pure scheduler information | 15:51 |
| bauzas | they are not used for claiming right ? | 15:51 |
| ndipanov | bauzas, well IDK - you could say they are - what happens if I move a host to a different aggregate | 15:52 |
| *** sdake_ has joined #openstack-nova | 15:52 | |
| *** cbader has joined #openstack-nova | 15:52 | |
| ndipanov | they don't need to be | 15:52 |
| ndipanov | 2) we need to store information for an ongoing live-migration for resource tracking | 15:53 |
| ndipanov | 3) we need to do the same for rebuilds | 15:53 |
| ndipanov | 4) we need to store additional information than what is currently available in the migration object | 15:53 |
| ndipanov | 5) do we need the migration object if we persist RequestSpec | 15:54 |
| openstackgerrit | jichenjc proposed openstack/nova: Add os_shutdown_timeout support https://review.openstack.org/177217 | 15:54 |
| ndipanov | 6) (probably) we need a nicer API than just flipping a bunch of instance attributes inline when doing/reverting migrations | 15:55 |
| *** otter768 has quit IRC | 15:55 | |
| ndipanov | 7) WE need to structure request spec | 15:55 |
| ndipanov | 8) with previous 6 in mind, also decide what goes in it | 15:55 |
| *** agireud has joined #openstack-nova | 15:55 | |
| ndipanov | so yeah a high level document that focuses on this stuff would be very useful | 15:55 |
| *** sdake has quit IRC | 15:56 | |
| ndipanov | 9) also probably some stuff about resource tracking and what is a resource | 15:56 |
| bauzas | mmm | 15:56 |
| bauzas | mmm | 15:56 |
| bauzas | mmm | 15:56 |
| bauzas | (does it show how I'm torn ?) | 15:56 |
| alaski | claudiub: sorry, got pulled away. This isn't something I'm really knowledgeable on, but I might suggest zzzeek as a good resource | 15:57 |
| ndipanov | 10) also a bunch of stuff dansmith found with restarting flapping services | 15:57 |
| bauzas | ndipanov: what I know is that we have multiple items in place for being written or currently being written | 15:57 |
| bauzas | ndipanov: 1/ we have the ReqSpec stuff | 15:57 |
| claudiub | alaski: I see. Thanks for the pointer. :) | 15:57 |
| *** yamahata has joined #openstack-nova | 15:57 | |
| bauzas | 2/ we have the resource-objects BP | 15:57 |
| bauzas | 3/ we have the robustify-evacuation which will allow to keep track of evacuations | 15:58 |
| ndipanov | but they also don't take into account a ton of stuff | 15:58 |
| bauzas | ndipanov: what you ask is a more high-level view | 15:58 |
| ndipanov | I'm all for the agile approach | 15:58 |
| ndipanov | but it seems silly to simply disregard issues about stuff we already know | 15:58 |
| bauzas | ndipanov: ssssshhhht :) | 15:59 |
| bauzas | eh | 15:59 |
| bauzas | honestly, I'm all for discussion | 15:59 |
| bauzas | ndipanov: the problem I have is that I only know certain pieces of the puzzle | 15:59 |
| bauzas | ndipanov: which is one of the reasons I'm no longer only focusing on the scheduler, since it's just silly to do so | 16:00 |
| bauzas | but still, you have your own experience | 16:00 |
| *** READ10 has quit IRC | 16:00 | |
| bauzas | ndipanov: so the point is, how we can make sure that our mutual feedback are helpful ? | 16:00 |
| zzzeek | claudiub: that isn’t the database model, if nova uses the PG ENUM type, that would require an ALTER TYPE directive. at the very least if the sqlalchemy.types.Enum type is being used that directive would have to change. | 16:00 |
| bauzas | ndipanov: saying that, I just want to figure out an action plan if possible and then iterate while we discuss on the bigger problem | 16:01 |
| bauzas | ndipanov: but at least including johnthetubaguy is worth it :) | 16:01 |
| *** jlanoux has quit IRC | 16:03 | |
| *** krtaylor has joined #openstack-nova | 16:04 | |
| ndipanov | bauzas, I think these kind of issues that require a more in depth design effort just won't be easily solved with "iteration" | 16:04 |
| ndipanov | at least not with our constraints | 16:04 |
| *** unicell has joined #openstack-nova | 16:04 | |
| bauzas | ndipanov: hence johnthetubaguy in the loop and probably a backlog spec for describing the problem :) | 16:05 |
| zzzeek | claudiub: yeah that is also in models.py and also I dont know that that is upgradable at the moment | 16:05 |
| *** jistr has quit IRC | 16:05 | |
| *** oro has joined #openstack-nova | 16:05 | |
| *** Mike_D_laptop has joined #openstack-nova | 16:05 | |
| *** amotoki_ has quit IRC | 16:05 | |
| claudiub | zzzeek: so, I did not find any way to alter through sqlalchemy. The only way I know now would be through raw sql, but i don't know how acceptable that is as a commit. | 16:05 |
| zzzeek | claudiub: nova uses sqlalchemy-migrate for DB migrations but they are moving to a new system that I dont think yet supports this | 16:06 |
| ndipanov | bauzas, not sure what johnthetubaguy has to do with it - as I said: let's try to come up with a list of known issues | 16:06 |
| ndipanov | that we know are related to this whole modeling exercise | 16:06 |
| ndipanov | and then go from there - see which ones we can maybe put off for later | 16:07 |
| ndipanov | and see how we can come up with an approach that lets us cover some of the existing oens | 16:07 |
| *** agireud has quit IRC | 16:07 | |
| claudiub | zzzeek: I see. That is a bit problematic then. | 16:07 |
| ndipanov | and then iterate | 16:07 |
| zzzeek | claudiub: ill ask :) | 16:07 |
| ndipanov | bauzas, if you think you need more info on the migration stuff | 16:08 |
| claudiub | zzzeek: thanks, let me know if you find out anything. :) | 16:08 |
| ndipanov | feel free to ping us | 16:08 |
| ndipanov | er me | 16:08 |
| bauzas | ndipanov: what I need to understand are the corner cases that you identify | 16:08 |
| dansmith | zzzeek: claudiub: right now, migrations still have to be written with sqlalchemy-migrate.. the "other thing" is experimental to find things like this | 16:09 |
| *** mleroy has joined #openstack-nova | 16:09 | |
| dansmith | zzzeek: claudiub: so sqla-migrate can't describe an enum change? | 16:09 |
| ndipanov | bauzas, yes - that's exaclty the "design" process we keep skipping | 16:09 |
| zzzeek | dansmith: this is a thing that the “other thing” cannot do right now | 16:09 |
| bauzas | ndipanov: in the mean time, I need to dig into the compute manager code and the virt code since it seems I'm missing something crucial to your eyes | 16:09 |
| zzzeek | dansmith: migrate can, b.c. at worst you just write the “ALTER” directly. it is very different on MySQL vs. Postgresql | 16:10 |
| zzzeek | dansmith: not sure if migrate has a built-in directive | 16:10 |
| *** browne has joined #openstack-nova | 16:10 | |
| claudiub | dansmith: I don't know of any direct / nice way to do that. Changing the enum is only problematic for postgresql, from what I can see. | 16:10 |
| dansmith | zzzeek: okay, well, that's all claudiub needs to worry about | 16:10 |
| zzzeek | dansmith: truth be told I don’t use native enums in any case for this reason, adding elements is a total pita | 16:10 |
| openstackgerrit | jichenjc proposed openstack/nova: Handle MessageTimeout to MigrationPreCheckError https://review.openstack.org/168635 | 16:10 |
| dansmith | zzzeek: we have enums in our upper layers, so I'd be fine with saying "don't use native enums" if we know they're hard | 16:11 |
| *** ildikov has quit IRC | 16:11 | |
| zzzeek | dansmith: if this is merged into L as a migrate-only change then…we can’t do an “expand” unless we add support | 16:11 |
| zzzeek | dansmith: OK but you alreay have three native enums in your model | 16:11 |
| ndipanov | bauzas, let me try to write some of these down in the response to that email pls | 16:11 |
| *** shashankhegde has joined #openstack-nova | 16:11 | |
| dansmith | zzzeek: yep, understand | 16:11 |
| bauzas | ndipanov: fair enougfh | 16:11 |
| *** leopoldj has quit IRC | 16:11 | |
| *** fawadkhaliq has quit IRC | 16:12 | |
| *** belmoreira has quit IRC | 16:12 | |
| dansmith | zzzeek: sqla-migrate is the only thing that has to work. The point of having the online stuff in-tree alongside migrate is specifically to find things like this that we need to do that it doesn't support, so that | 16:12 |
| dansmith | is fine | 16:12 |
| zzzeek | dansmith: OK but you want online to support this eventually right | 16:12 |
| dansmith | but if native enums are problematic, we should probably just nuke them first | 16:12 |
| dansmith | zzzeek: if we just make a call not to use native enums, then I don't car4e | 16:13 |
| dansmith | if it's reasonable to add it to online, then sure | 16:13 |
| dansmith | claudiub: are you asking if this is okay to do? https://review.openstack.org/#/c/185961/1/nova/db/sqlalchemy/migrate_repo/versions/294_add_live_resize_migration_type.py,cm | 16:14 |
| *** gokrokve_ has joined #openstack-nova | 16:14 | |
| *** alex_klimov has quit IRC | 16:14 | |
| zzzeek | dansmith: just noting, my alternative approach does not have this problem | 16:14 |
| claudiub | dansmith: yeah. It only fails for postgresql. | 16:14 |
| dansmith | claudiub: oh, I see | 16:15 |
| dansmith | claudiub: this is a tangent, but two things: (a) I thought we weren't going to do live resize in L, and (b) I'm not sure we need a migration record for something like that if we're not moving the instance | 16:15 |
| *** sahid has quit IRC | 16:16 | |
| *** gokrokve has quit IRC | 16:17 | |
| claudiub | dansmith: so, it's not yet been decided, if we do live resize or not. The question was if there are enough people that would want this feature or not. I've had some reply / replies on the spec, saying that that would be something they would like. | 16:18 |
| dansmith | claudiub: I don't think that was the decision point (demand) :) | 16:18 |
| *** dprince has quit IRC | 16:18 | |
| *** ijw has joined #openstack-nova | 16:18 | |
| claudiub | dansmith: as for (b), well, it's a bit complicated. In the first phase, it will live resize on the same host, Second phase, live resize on a different host. | 16:18 |
| mdbooth | Is there a strict upgrade ordering for api and compute, by any chance? | 16:19 |
| dansmith | I think it was about discoverability of policy and/or support of the feature, which is an api impacting thing we're not going to do in L (IIRC) | 16:19 |
| mdbooth | Specifically, is it ever permissible to have an api service older than your compute nodes? | 16:19 |
| dansmith | mdbooth: no | 16:19 |
| mdbooth | dansmith: That makes life simpler :) Thanks. | 16:19 |
| edleafe | bauzas: jaypipes: n0ano: lxsli: PaulMurray: here's the doodle for a new meeting time: http://doodle.com/akuv4b4ftv68q3me | 16:19 |
| mdbooth | dansmith: Actually, let me just check what that no referred to | 16:20 |
| mdbooth | No strict ordering | 16:20 |
| mdbooth | Or no api older than compute | 16:20 |
| dansmith | mdbooth: let me clarify then | 16:20 |
| claudiub | dansmith: I don't think I got what you are saying. | 16:20 |
| dansmith | mdbooth: api and conductor (and scheduler) have to match the database schema because they speak directly to it | 16:20 |
| *** arnaud____ has quit IRC | 16:21 | |
| dansmith | mdbooth: so the way we prescribe that an upgrade happens is api first then computes | 16:21 |
| *** mohitsharma has quit IRC | 16:21 | |
| dansmith | mdbooth: because api is mostly stateless and fast | 16:21 |
| mdbooth | dansmith: Awesome, thanks. | 16:21 |
| dansmith | mdbooth: so, that's the arrangement we try to ensure works.. newer computes with newer conductors and an older api might kinda work, but it's not something we test or aim for | 16:22 |
| *** garyk has quit IRC | 16:22 | |
| *** arnaud____ has joined #openstack-nova | 16:22 | |
| dansmith | sdague: I recall we're punting on live inplace resize until we have discoverable policy for disabling such a thing, which won't happen in L. Is that you recollection of that discussion as well? | 16:23 |
| *** ijw has quit IRC | 16:24 | |
| sdague | dansmith: that was my preference | 16:24 |
| *** nelsnelson has joined #openstack-nova | 16:24 | |
| *** jaypipes has joined #openstack-nova | 16:24 | |
| sdague | and I think the folks in the room were generally on board with that | 16:25 |
| dansmith | agreed | 16:25 |
| dansmith | claudiub: ^ | 16:25 |
| *** danpb has quit IRC | 16:25 | |
| *** agireud has joined #openstack-nova | 16:25 | |
| claudiub | dansmith: well, can't it just be simply disabled through existing ways, if there are people that do not want it? | 16:26 |
| sdague | dims_, samueldmq - http://lists.openstack.org/pipermail/openstack-dev/2015-June/065496.html I'm starting that as an ML thread before writing up specs on it | 16:26 |
| sdague | claudiub: the problem is interop | 16:26 |
| sdague | adding features that are basically going to be turned off all the time to the API with no discovery mechanism of that fact is taking a bad situation and making it worse | 16:26 |
| *** arnaud____ has quit IRC | 16:27 | |
| dims_ | sdague: ack | 16:27 |
| *** shashankhegde has quit IRC | 16:27 | |
| samueldmq | sdague, great, thanks! | 16:27 |
| openstackgerrit | Jeffrey Zhang proposed openstack/nova: Support all_tenants search_opts for neutron https://review.openstack.org/187657 | 16:27 |
| *** amotoki_ has joined #openstack-nova | 16:29 | |
| *** annegentle has quit IRC | 16:29 | |
| *** sdake has joined #openstack-nova | 16:29 | |
| *** baoli has quit IRC | 16:29 | |
| *** yassine has quit IRC | 16:29 | |
| *** jogo has quit IRC | 16:30 | |
| *** toan-tran has quit IRC | 16:30 | |
| *** baoli has joined #openstack-nova | 16:30 | |
| luv | jaypipes: alright. so i figured it out (by following the source and pdb'ing around :) ) .... I had security_group_api = nova in my nova.conf instead of security_group_api = neutron | 16:31 |
| jaypipes | makes sense, luv | 16:31 |
| lxsli | edleafe: I see times 2pm - 6pm, those have been localised for me right? | 16:31 |
| *** sdake__ has joined #openstack-nova | 16:32 | |
| luv | yup | 16:32 |
| *** sdake_ has quit IRC | 16:32 | |
| *** liusheng has quit IRC | 16:33 | |
| *** Nisha has quit IRC | 16:33 | |
| *** liusheng has joined #openstack-nova | 16:33 | |
| *** rpodolyaka2 has joined #openstack-nova | 16:34 | |
| edleafe | lxsli: yes. It should note the time zone it's showing on the page | 16:34 |
| edleafe | I see 8am - noon | 16:34 |
| lxsli | edleafe: awesome, thanks for confirming | 16:35 |
| claudiub | sdague: that discovery, isn't / can't it done through extensions? | 16:35 |
| sdague | claudiub: extensions are deprecated | 16:36 |
| *** sdake has quit IRC | 16:36 | |
| sdague | because they were a giant interop problem | 16:36 |
| *** sdake has joined #openstack-nova | 16:36 | |
| *** dprince has joined #openstack-nova | 16:36 | |
| *** jogo has joined #openstack-nova | 16:37 | |
| claudiub | I get that, but it is a bit strange to deprecate something and offer nothing in its stead.. | 16:37 |
| *** shashankhegde has joined #openstack-nova | 16:38 | |
| sdague | the goal is actually much less optionality | 16:38 |
| sdague | so it's not that weird, we really want the nova api to mean one thing, not just be a random pass through layer | 16:39 |
| dansmith | right, it's not so much deprecating extensions as making a conscious decision not to allow pluggability | 16:39 |
| sdague | exactly | 16:39 |
| *** sdake__ has quit IRC | 16:40 | |
| *** nelsnels_ has joined #openstack-nova | 16:40 | |
| *** nelsnelson has quit IRC | 16:40 | |
| cfriesen | sdague: it would be good to define for vendors what the expected method of doing extensions is though, since most are going to want to do some sort of "special sauce" to differentiate themselves rather than compete purely on price/value | 16:41 |
| dansmith | there is no method | 16:41 |
| dansmith | that's the point | 16:42 |
| *** nic has joined #openstack-nova | 16:42 | |
| cfriesen | dansmith: I've heard a few people say "use a separate endpoint" | 16:42 |
| cfriesen | dansmith: which I think would be reasonable....you'd have the vanilla openstack endpoint and then the vendor-specific one for people that want to access special features | 16:42 |
| *** gokrokve_ has quit IRC | 16:43 | |
| dansmith | if it's an endpoint of just vendor things, then maybe, but not if the separate endpoint is nova+vendor | 16:43 |
| *** gokrokve has joined #openstack-nova | 16:43 | |
| dansmith | but the former is really outside the scope of nova entirely, so us providing a recipe for doing that would be kinda silly, IMHO | 16:44 |
| cfriesen | dansmith: so what's the expectation if a vendor wants to add a parameter to an instance that is never going to make it upstream (for vertical scaling, for example) | 16:44 |
| dansmith | cfriesen: not doing it, or not calling themselves openstack | 16:44 |
| *** beagles_brb is now known as beagles | 16:44 | |
| cfriesen | dansmith: not sure how we can restrict them from doing a nova+vendor endpoint | 16:45 |
| dansmith | I don't know that defcore would know or care about a nova+vendor endpoint if the nova one exists | 16:45 |
| cfriesen | dansmith: and not sure if we want to, as long as they also have a pure nova endpoint | 16:45 |
| *** gokrokve has quit IRC | 16:45 | |
| dansmith | but that one would not be the nova api | 16:45 |
| *** gokrokve_ has joined #openstack-nova | 16:45 | |
| dansmith | we can't prevent people from doing it, of course | 16:46 |
| dansmith | but we can not concern ourselves with making it easy/pluggable, which is all I think we've decided to do, yes? | 16:46 |
| *** mpavone has quit IRC | 16:46 | |
| lxsli | by forcing it to go on a separate endpoint you clearly demarcate the OpenStack (interoperable) API from the proprietary API, which allows cloud users to choose whether to rely on the proprietary API or not | 16:47 |
| cfriesen | dansmith: I'm actually thinking that it would be better to be pragmatic and say "you need a pure nova endpoint, but if you want to add extensions then here is how you add a new nova+changes endpoint efficiently". Clearly there is demand for extensions (witness the number that were added) so I think it makes sense to plan for it somehow. | 16:47 |
| dansmith | cfriesen: everything was an extension before, so measuring the demand by the count is an error. | 16:48 |
| openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Fix SMBFS volume attach race condition https://review.openstack.org/187673 | 16:48 |
| cfriesen | dansmith: okay, fair enough | 16:48 |
| dansmith | cfriesen: I don't doubt the demand of vendors wanting to customize the api to their heart's content | 16:48 |
| openstackgerrit | Nicolas Simonds proposed openstack/nova: Fix/implement resize and revert for RBD-backed images https://review.openstack.org/187395 | 16:49 |
| cfriesen | dansmith: my worry is simply that if we say "we don't support extensions", then they'll all go off and implement them in different ways. If we say "we support extensions in this way as long as you *also* have a pure nova API", then we exercise a bit of control. | 16:50 |
| dansmith | cfriesen: do you think it's okay for vendors to define their own syscalls in their linux distros? | 16:51 |
| cfriesen | And then end-users using the pure API get interop, while ones that really want the extra functionality can make an informed decision to use non-interoperable calls. | 16:51 |
| *** thangp has quit IRC | 16:51 | |
| cfriesen | dansmith: sure. I've done it, because our customers wanted a way to do something efficiently that couldn't be done any other way. | 16:51 |
| dansmith | cfriesen: end users will use the thing that gives them the thing they want right now. That locks them in retroactively | 16:51 |
| dansmith | cfriesen: so you break ABI compatibility when you have to move that syscall number? | 16:52 |
| cfriesen | dansmith: yep, because it was for an embedded device that got built from source each time | 16:52 |
| openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Implement nova rescue https://review.openstack.org/127159 | 16:52 |
| dansmith | heh, okay, so if you're exposing a non-public interface where you control the client and server at all times, then you just add a microversion on top (or don't if it suits you) and then just replace both each time you upgrade, right? | 16:53 |
| dansmith | if you care about public-facing things where you don't have that control, | 16:53 |
| dansmith | then you can't do it, just like you can't do it in your kernel and expect binary compatibility with the rest of the world | 16:53 |
| *** Nisha has joined #openstack-nova | 16:54 | |
| *** IanGovett has joined #openstack-nova | 16:54 | |
| *** diegolp has quit IRC | 16:55 | |
| *** pixelbeat has quit IRC | 16:57 | |
| *** yamahata has quit IRC | 16:57 | |
| *** emagana has quit IRC | 16:57 | |
| *** thangp has joined #openstack-nova | 16:57 | |
| *** lpetrut has quit IRC | 16:58 | |
| cfriesen | dansmith: right...this is different than the kernel case in that we do want to preserve the stock nova endpoint. But it coming from a project that used to support extensions, it now seems harsh to say "if you want anything other than stock upstream API then you have to patch the source". | 16:58 |
| *** baoli has quit IRC | 16:58 | |
| *** matrohon has quit IRC | 16:58 | |
| cfriesen | dansmith: anyways, not the most important thing to work on right now. :) | 16:58 |
| *** emagana has joined #openstack-nova | 16:59 | |
| dansmith | it seems completely undifferent from the kernel case, but yeah, very happy to stop discussing it :) | 16:59 |
| *** IanGovett1 has joined #openstack-nova | 16:59 | |
| *** IanGovett has quit IRC | 17:01 | |
| mriedem | so...i want my own nova v2.1.1 api, that's cool right? | 17:01 |
| mriedem | sorry, hello friends | 17:01 |
| *** baoli has joined #openstack-nova | 17:01 | |
| mriedem | i meant to preface with 'hello friends' | 17:01 |
| *** nelsnels_ has quit IRC | 17:02 | |
| *** nelsnelson has joined #openstack-nova | 17:02 | |
| dansmith | mine is nova v2.1.37-dan.el7-321extrasuff~3 | 17:02 |
| mriedem | v2.1.1 is sure to be much glorious api #1 | 17:02 |
| *** nelsnelson has quit IRC | 17:02 | |
| *** nelsnelson has joined #openstack-nova | 17:04 | |
| *** nelsnelson has quit IRC | 17:04 | |
| *** ildikov has joined #openstack-nova | 17:05 | |
| *** derekh has quit IRC | 17:06 | |
| mriedem | i'd like to propose an api that allows you to dynamically create flavors when booting or resizing an instance.... | 17:06 |
| dansmith | mriedem: lol :) | 17:06 |
| *** agireud has quit IRC | 17:07 | |
| *** achanda has joined #openstack-nova | 17:08 | |
| *** IanGovett has joined #openstack-nova | 17:09 | |
| *** IanGovett has quit IRC | 17:09 | |
| *** pbandzi has quit IRC | 17:09 | |
| *** vilobhmm has joined #openstack-nova | 17:09 | |
| *** dulek has quit IRC | 17:10 | |
| *** alanf-mc has joined #openstack-nova | 17:10 | |
| *** gszasz_ has joined #openstack-nova | 17:11 | |
| *** gszasz_ has quit IRC | 17:11 | |
| *** IanGovett1 has quit IRC | 17:11 | |
| *** baoli has quit IRC | 17:12 | |
| *** gszasz has quit IRC | 17:14 | |
| *** jdurgin has joined #openstack-nova | 17:15 | |
| *** angdraug has joined #openstack-nova | 17:16 | |
| *** rushiagr_away is now known as rushiagr | 17:17 | |
| *** baoli has joined #openstack-nova | 17:18 | |
| *** lpetrut has joined #openstack-nova | 17:18 | |
| *** ijw has joined #openstack-nova | 17:18 | |
| *** lucasagomes is now known as lucas-dinner | 17:19 | |
| *** shashankhegde has quit IRC | 17:22 | |
| *** ijw has quit IRC | 17:24 | |
| *** yamahata has joined #openstack-nova | 17:24 | |
| *** gholler has quit IRC | 17:24 | |
| *** zhangtralon has joined #openstack-nova | 17:27 | |
| *** tonytan4ever has quit IRC | 17:28 | |
| *** Hosam has quit IRC | 17:28 | |
| *** baoli has quit IRC | 17:28 | |
| *** tonytan4ever has joined #openstack-nova | 17:28 | |
| *** baoli has joined #openstack-nova | 17:29 | |
| *** emagana has quit IRC | 17:32 | |
| *** mtanino has quit IRC | 17:33 | |
| *** paul-carlton has quit IRC | 17:33 | |
| *** tonytan4ever has quit IRC | 17:33 | |
| *** zhangtralon has quit IRC | 17:33 | |
| *** amotoki_ has quit IRC | 17:34 | |
| *** romainh1 has left #openstack-nova | 17:37 | |
| melwitt | dansmith: I've been investigating an intermittent problem while I test my "send Instance object to cells" patch where an instance.destroy() fails because "host changed". however when I look at "host" on the instance, it's None still and with sqlalchemy debugging turned all the way up, I see a "ROLLBACK" after the UPDATE that's supposed to delete the row. do you know what the ROLLBACK means? | 17:38 |
| dansmith | melwitt: well, it means it undoes everything in the transaction | 17:39 |
| dansmith | melwitt: is the rollback the thing that is changing host, or the thing that fails because host is changed? | 17:39 |
| *** josecastroleon has quit IRC | 17:39 | |
| melwitt | dansmith: the latter. it's the update to soft delete the instance | 17:40 |
| *** agireud has joined #openstack-nova | 17:40 | |
| dansmith | melwitt: so there is something in instance_destroy() that checks instance.host? | 17:40 |
| *** baoli has quit IRC | 17:41 | |
| *** lpetrut has quit IRC | 17:41 | |
| dansmith | melwitt: ah, is it the constraint thing? | 17:41 |
| melwitt | dansmith: sorry I'm confused about how this stuff works. yes, the constraint, if host isn't set on the instance, it will do the db.instance_destroy with a constraint that host must be null | 17:41 |
| *** ildikov has quit IRC | 17:42 | |
| *** mriedem has quit IRC | 17:42 | |
| dansmith | melwitt: okay, so the raise because of the failed constraint is why the ROLLBACK is happening, I think | 17:42 |
| dansmith | so that's expected | 17:42 |
| dansmith | the reason for the constraint being unsatisfied is what you need to chase I think | 17:42 |
| *** ociuhandu has quit IRC | 17:43 | |
| dansmith | I think maybe alaski had some thoughts about this, relating to the instance being updated in one cell (and gaining a host or something) before this delete happens... | 17:43 |
| melwitt | yeah, it's just it seems like the constraint was satisfied, at least host is null after it's all done and failed. and recently alaski added a catch ObjectActionError to retry delete if cell_name is set, which it seems to be, but that doesn't fire either | 17:44 |
| *** tonytan4ever has joined #openstack-nova | 17:44 | |
| *** baoli has joined #openstack-nova | 17:44 | |
| *** ssurana has joined #openstack-nova | 17:44 | |
| *** lpetrut has joined #openstack-nova | 17:44 | |
| alaski | yeah, I added a retry in one spot for this | 17:45 |
| alaski | melwitt: are you sure host is null in the db when it fails, not just on the object? | 17:45 |
| dansmith | so the constraint is that if we don't know what our host is, or it's set to None, make sure we don't delete the instance if the DB has host != None, yes? | 17:45 |
| melwitt | alaski: yeah, in looking at the code, I don't see how host could change without cell_name being set | 17:45 |
| melwitt | but that's what I'm seeing. I just don't get it | 17:45 |
| melwitt | dansmith: yes | 17:46 |
| alaski | melwitt: I agree, I don't see how host could change without cell_name being set | 17:46 |
| alaski | but there could be a difference between the db, and whats on your current object | 17:47 |
| melwitt | alaski: I'm not sure. I've got all this sqlalchemy debug output that doesn't seem to tell me as much as I thought it would. I thought it would show what the fields in the instance row are at the time the soft delete fails | 17:47 |
| *** shashankhegde has joined #openstack-nova | 17:47 | |
| dansmith | alaski: the constraint is only set if the current object is unset or =None, so it has to be non-None in the database to fail, right? | 17:47 |
| alaski | dansmith: right | 17:48 |
| alaski | melwitt: can you try an instance.refresh and then see what's set? | 17:48 |
| melwitt | when I check the db manually after this happens, host is NULL | 17:48 |
| alaski | that's odd | 17:49 |
| melwitt | so I was wondering if it was at all possible the constraint wasn't querying right, but I see with the debug output it's UPDATE instances SET updated_at=updated_at, deleted_at=%s, deleted=id WHERE instances.deleted = %s AND instances.uuid = %s AND instances.host IS NULL | 17:49 |
| melwitt | which should be right | 17:49 |
| *** Nisha has quit IRC | 17:50 | |
| *** otter768 has joined #openstack-nova | 17:51 | |
| alaski | looking at instance_destroy in sqlalchemy/api.py it seems that the constraint failure can be raised if no rows are updated | 17:52 |
| alaski | so it's possible the instances.deleted = %s is failing | 17:52 |
| *** zhangtralon has joined #openstack-nova | 17:53 | |
| melwitt | alaski: right. that's what I was trying to figure out, if there was some other way it could fail since host seems to be null | 17:53 |
| alaski | if deleted != 0 it could fail I think | 17:54 |
| melwitt | doesn't instances.deleted mean not deleted? that is, | 17:54 |
| *** baoli has quit IRC | 17:54 | |
| *** ildikov has joined #openstack-nova | 17:55 | |
| *** nickchase has quit IRC | 17:55 | |
| *** otter768 has quit IRC | 17:55 | |
| alaski | if we assume it passes in 0 there, yeah. it won't try to redelete something that's deleted | 17:55 |
| melwitt | the delete path before scheduled involves a broadcast to all cells to delete, followed by a local delete at the top | 17:55 |
| *** mriedem has joined #openstack-nova | 17:57 | |
| melwitt | so it might be possible for an instance_destroy_at_top to happen before the local delete, if there was indeed an instance found in the child. I need to check that | 17:57 |
| *** ociuhandu has joined #openstack-nova | 17:57 | |
| *** jlk has joined #openstack-nova | 17:57 | |
| *** aix has quit IRC | 17:57 | |
| jlk | Hey all, fighting a resize issue, trying to figure out where Nova resolves hostname to IP address in order to talk to other compute hosts to set up the transfer over ssh. Can anybody help? | 17:58 |
| *** zhangtralon has quit IRC | 17:59 | |
| melwitt | alaski: you're right, it's because it was already deleted. now I will find out how that happened | 17:59 |
| alaski | melwitt: ahh, cool. seems likely that it's the race you described | 18:00 |
| alaski | I mean, cool that I'm right, not that it's happening :) | 18:01 |
| *** baoli has joined #openstack-nova | 18:01 | |
| melwitt | alaski: hehe. yeah, while my patch "solves" the race where instances can be "undeleted" I found we just move to a different race, the opposite double delete | 18:02 |
| dansmith | jlk: there is this thing | 18:02 |
| dansmith | jlk: it does the translation | 18:02 |
| dansmith | jlk: it's called DNS | 18:02 |
| * dansmith ducks | 18:02 | |
| * jlk tweets nasty things | 18:02 | |
| dansmith | backfired! | 18:03 |
| jlk | okay, looks like migrate_disk_and_power_off gets a dest from somewhere | 18:03 |
| dansmith | from the scheduler, if you didn't select a destination | 18:04 |
| jlk | alright, and scheduler does the resolution? | 18:04 |
| jlk | I ask because nova-compute is logging the resolved IP address | 18:04 |
| dansmith | so the hypervisor hostname is what nova will try to use for the ssh command I think | 18:04 |
| *** jwcroppe has joined #openstack-nova | 18:04 | |
| dansmith | it is? | 18:04 |
| *** zhangtralon has joined #openstack-nova | 18:04 | |
| *** rotbeard has quit IRC | 18:04 | |
| jlk | error while running command. Command: ssh 173.247.104.36 mkdir -p /var/lib/nova/instances/195e844c- | 18:05 |
| jlk | but my host-list is names like ds1351.sea03 | 18:05 |
| dansmith | jlk: hmm, does nova show on the instance show that the hypervisor hostname is an IP? | 18:05 |
| jlk | so somewhere along the way this host-list is getting resolved into an IP address (wrongly) | 18:05 |
| jlk | no, it shows the host-list name, ds1352.sea03 | 18:05 |
| dansmith | I could be wrong, but I don't think it's pre-resolving it | 18:06 |
| jlk | then it would be rootwrap doing it? | 18:06 |
| jlk | | OS-EXT-SRV-ATTR:hypervisor_hostname | ds1352.sea03 | 18:06 |
| dansmith | no, I don't think so | 18:07 |
| jroll | jlk: you haven't learned by now to always blame dns? :) | 18:07 |
| jlk | I tried, by stuffing values in /etc/hosts | 18:07 |
| jlk | but either I'm doing that wrong, or... | 18:07 |
| jlk | ah hah | 18:08 |
| jlk | nova/compute/resource_tracker.py: migration.dest_host = self.driver.get_host_ip_addr() | 18:08 |
| dansmith | huh | 18:09 |
| jlk | returns CONF.my_ip | 18:09 |
| jlk | for libvirt | 18:09 |
| * jlk shaves the yak some more | 18:09 | |
| dansmith | oh, on the destination call? | 18:09 |
| *** openstackgerrit has quit IRC | 18:09 | |
| jlk | yeah when it's determining a destination, it does the IP look up | 18:10 |
| *** openstackgerrit has joined #openstack-nova | 18:10 | |
| jlk | so if my_ip isn't defined, goes to nova.netconf | 18:10 |
| dansmith | no, I think that's after it has determined the host, | 18:10 |
| jlk | sure | 18:10 |
| dansmith | and when the source makes the call to the destination, it's saying "so, what IP should I use?" | 18:11 |
| jlk | picks a host, and then translates that host into an IP, to pass off to the migration calls | 18:11 |
| dansmith | I think it's asking the host itself to do the translation (which is really just "tell me what your configured IP is") | 18:11 |
| dansmith | but yeah | 18:11 |
| *** ociuhandu has quit IRC | 18:11 | |
| jlk | oh, and that does a socket.socket.getsockname() | 18:12 |
| jlk | but which service is this, shoudl be the nova-compute service, ya? | 18:13 |
| dansmith | yeah | 18:13 |
| jlk | I'm betting this doesn't take /etc/hosts into account dos it? | 18:13 |
| dansmith | if it's just looking at CONF then no | 18:13 |
| jlk | there is nothing in my conf for my_ip | 18:14 |
| *** promulo has joined #openstack-nova | 18:14 | |
| dansmith | oh you mean in the fallback path | 18:14 |
| *** mtanino has joined #openstack-nova | 18:14 | |
| * dansmith shrugs | 18:14 | |
| jlk | yeah | 18:14 |
| jlk | oh dear | 18:17 |
| jlk | I see what this is doing | 18:17 |
| jlk | it's checking to see what address it would connect /out/ as to the Internet. Which in our case is odd. | 18:17 |
| jlk | yeah that's not going to work at all, guess we have to configure my_ip | 18:17 |
| *** annegentle has joined #openstack-nova | 18:21 | |
| jlk | success! | 18:21 |
| vilobhmm | dansmith, melwitt : can you please review https://review.openstack.org/#/c/138607/ when you are free? | 18:21 |
| dansmith | vilobhmm: you don't need to ask me every day.. I will re-review it when I'm free.. honest. | 18:22 |
| *** gokrokve_ has quit IRC | 18:23 | |
| vilobhmm | cool | 18:23 |
| *** thiagop has joined #openstack-nova | 18:26 | |
| *** amotoki_ has joined #openstack-nova | 18:35 | |
| *** tralon_ has joined #openstack-nova | 18:36 | |
| *** vilobhmm has quit IRC | 18:38 | |
| *** zhangtralon has quit IRC | 18:39 | |
| *** amotoki_ has quit IRC | 18:40 | |
| *** tjones1 has joined #openstack-nova | 18:40 | |
| tjones1 | @jaypipes if you need any help with https://review.openstack.org/#/c/184534/1 i'd love to be involved | 18:41 |
| *** nelsnelson has joined #openstack-nova | 18:41 | |
| ndipanov | hey does anyone by any chance remember how to display instance faults with the cli? | 18:41 |
| jaypipes | tjones1: sweet! thanks! I will certainly keep that in mind | 18:41 |
| dansmith | ndipanov: nova show, if the instance is in error state | 18:41 |
| *** nelsnelson has quit IRC | 18:41 | |
| *** erhudy has quit IRC | 18:42 | |
| *** gokrokve has joined #openstack-nova | 18:42 | |
| ndipanov | but if not then nothing... | 18:42 |
| ndipanov | I see | 18:42 |
| *** nelsnelson has joined #openstack-nova | 18:42 | |
| *** tralon_ has quit IRC | 18:43 | |
| *** jwcroppe has quit IRC | 18:43 | |
| melwitt | dansmith: another thing I learned is behavior has changed from what cells expects when broadcasting a delete to all cells (when host is None). there are several places in compute where InstanceNotFound is caught and ignored for cells, as it expects the exception to be raised when attempting instance.save(). this isn't necessarily true when instance goes to _save_flavor() and does db.instance_extra_update_by_uuid before the db.instance_update_and_ge | 18:43 |
| melwitt | t_original. in that case, if the instance doesn't exist, the FK constraint will fail trying to do the instance_extra update | 18:43 |
| melwitt | oops | 18:43 |
| *** gokrokve has quit IRC | 18:44 | |
| dansmith | ah | 18:44 |
| *** gokrokve has joined #openstack-nova | 18:44 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Begin the transition to an explicit object registry https://review.openstack.org/187718 | 18:44 |
| openstackgerrit | Dan Smith proposed openstack/nova: Convert test objects to new field formats https://review.openstack.org/187719 | 18:44 |
| openstackgerrit | Dan Smith proposed openstack/nova: Use fields from oslo.versionedobjects https://review.openstack.org/187720 | 18:44 |
| dansmith | melwitt: we really need to make instance_update do all of that more atomically and in a single transaction anyway | 18:44 |
| *** achanda has quit IRC | 18:45 | |
| melwitt | dansmith: yeah. I was about to ask what you suggest :) | 18:45 |
| dansmith | melwitt: is it hurting anything (i.e. failing) or just doing extra stuff in the DB? | 18:47 |
| melwitt | dansmith: well, for deleting an instance that hasn't been scheduled, the broadcast to delete in all cells will fail in cells that don't have the instance. DBReferenceError is raised instead of the expected and caught InstanceNotFound which rolls back quotas | 18:49 |
| *** vilobhmm has joined #openstack-nova | 18:49 | |
| dansmith | ah, hmm | 18:50 |
| mriedem | dansmith: when is your birthday? http://www.amazon.com/Am-Better-Than-Your-Kids/dp/1439182868/ref=asap_bc?ie=UTF8 | 18:50 |
| *** kiran-r has joined #openstack-nova | 18:50 | |
| dansmith | mriedem: heh | 18:50 |
| *** cbader has quit IRC | 18:51 | |
| dansmith | mriedem: if you're going to get me something, a poster of this would be better: http://photos1.meetupstatic.com/photos/event/b/3/4/e/highres_323505902.jpeg | 18:51 |
| *** cbader has joined #openstack-nova | 18:51 | |
| mriedem | no | 18:51 |
| mriedem | book or nothing | 18:51 |
| melwitt | dansmith: like this https://github.com/openstack/nova/blob/master/nova/compute/api.py#L1615-L1618 thought it says "race condition" which I thought it wouldn't be in the case of the broadcast for non-scheduled instance | 18:51 |
| *** cbader has quit IRC | 18:51 | |
| *** cbader has joined #openstack-nova | 18:52 | |
| melwitt | dansmith: and this is where the DBReferenceError gets raised if the instance doesn't exist https://github.com/openstack/nova/blob/master/nova/compute/api.py#L1538 | 18:52 |
| *** alexpilotti has quit IRC | 18:52 | |
| dansmith | melwitt: just so I'm clear, it raises DBReferenceError when you try to save something like the flavor, yes? | 18:53 |
| dansmith | as part of the instance save | 18:53 |
| melwitt | dansmith: yes. I saw it was trying to save flavor in there | 18:54 |
| dansmith | right, so could we make things like _save_flavor() catch DBReferenceError and raise InstanceNotFound? I would think "instance doesn't exist" is the only way we can hit that anyway, so that should be accurate, right? | 18:54 |
| melwitt | during the broadcast to instance_delete_everywhere for all cells | 18:54 |
| melwitt | dansmith: yeah, I think that would work | 18:55 |
| *** asselin has quit IRC | 18:55 | |
| *** artom has quit IRC | 18:55 | |
| *** cbader has quit IRC | 18:55 | |
| *** artom has joined #openstack-nova | 18:55 | |
| *** cbader has joined #openstack-nova | 18:55 | |
| *** asselin has joined #openstack-nova | 18:55 | |
| dansmith | should be able to check it even in unit tests, by hand-creating an instance object with a dirty flavor and trying to run .save() on it | 18:55 |
| *** asselin has quit IRC | 18:56 | |
| *** cbader has quit IRC | 18:56 | |
| *** openstackgerrit has quit IRC | 18:56 | |
| *** kaisers1 has quit IRC | 18:56 | |
| *** asselin has joined #openstack-nova | 18:56 | |
| *** cbader has joined #openstack-nova | 18:56 | |
| *** cbader has quit IRC | 18:56 | |
| *** asselin has quit IRC | 18:56 | |
| *** openstackgerrit has joined #openstack-nova | 18:56 | |
| *** kaisers has quit IRC | 18:57 | |
| *** asselin has joined #openstack-nova | 18:57 | |
| *** cbader has joined #openstack-nova | 18:57 | |
| *** asselin has quit IRC | 18:57 | |
| *** cbader has quit IRC | 18:57 | |
| *** kaisers has joined #openstack-nova | 18:57 | |
| *** asselin has joined #openstack-nova | 18:57 | |
| *** cbader has joined #openstack-nova | 18:58 | |
| *** cbader has quit IRC | 18:58 | |
| *** cbader has joined #openstack-nova | 18:58 | |
| *** kaisers1 has joined #openstack-nova | 18:59 | |
| melwitt | dansmith: cool, I'll cook it! thanks | 18:59 |
| *** asselin has quit IRC | 19:00 | |
| *** asselin has joined #openstack-nova | 19:00 | |
| *** rushiagr is now known as rushiagr_away | 19:01 | |
| dansmith | melwitt: cool :) | 19:01 |
| *** asselin has quit IRC | 19:01 | |
| *** asselin has joined #openstack-nova | 19:01 | |
| *** asselin has quit IRC | 19:01 | |
| *** asselin has joined #openstack-nova | 19:02 | |
| *** asselin has quit IRC | 19:02 | |
| *** cbader has quit IRC | 19:03 | |
| *** asselin has joined #openstack-nova | 19:03 | |
| *** cbader has joined #openstack-nova | 19:03 | |
| *** achanda has joined #openstack-nova | 19:03 | |
| *** asselin has quit IRC | 19:04 | |
| *** cbader has quit IRC | 19:04 | |
| *** cbader has joined #openstack-nova | 19:04 | |
| *** asselin has joined #openstack-nova | 19:05 | |
| *** asselin has quit IRC | 19:05 | |
| *** cbader has quit IRC | 19:05 | |
| *** cbader has joined #openstack-nova | 19:06 | |
| *** asselin has joined #openstack-nova | 19:06 | |
| *** asselin has quit IRC | 19:06 | |
| *** asselin has joined #openstack-nova | 19:07 | |
| *** vladikr has quit IRC | 19:07 | |
| *** asselin has quit IRC | 19:08 | |
| *** asselin has joined #openstack-nova | 19:08 | |
| *** asselin has quit IRC | 19:09 | |
| *** cbader has quit IRC | 19:09 | |
| *** cbader has joined #openstack-nova | 19:09 | |
| *** asselin has joined #openstack-nova | 19:09 | |
| *** cbader has quit IRC | 19:10 | |
| *** asselin has quit IRC | 19:10 | |
| *** asselin has joined #openstack-nova | 19:10 | |
| *** cbader has joined #openstack-nova | 19:10 | |
| mriedem | claudiub: want to take another look at this? i fixed the test failures and squashed the other commit that it was pulling in so it's a bit more explicit https://review.openstack.org/#/c/174779/ | 19:11 |
| *** cbader has quit IRC | 19:12 | |
| *** cbader has joined #openstack-nova | 19:12 | |
| *** cbader has quit IRC | 19:14 | |
| samueldmq | sdague, dims_ Dynamic Policy for Access Control Subteam Meeting http://lists.openstack.org/pipermail/openstack-dev/2015-June/065524.html | 19:14 |
| samueldmq | sdague, dims_ you guys might be interested on this | 19:14 |
| *** cbader has joined #openstack-nova | 19:15 | |
| *** tellesnobrega_ has joined #openstack-nova | 19:16 | |
| *** baoli has quit IRC | 19:17 | |
| *** tonytan4ever has quit IRC | 19:17 | |
| dansmith | mriedem: actually that book looks pretty funny | 19:17 |
| *** baoli_ has joined #openstack-nova | 19:18 | |
| *** gholler has joined #openstack-nova | 19:18 | |
| *** iamjarvo has joined #openstack-nova | 19:19 | |
| *** belmoreira has joined #openstack-nova | 19:20 | |
| *** kaisers has quit IRC | 19:21 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Use fields from oslo.versionedobjects https://review.openstack.org/187720 | 19:23 |
| *** vladikr has joined #openstack-nova | 19:23 | |
| dansmith | dims_: you don't need to waste your time reviewing (unless you want), but just for progress reporting ^^ | 19:23 |
| *** kaisers has joined #openstack-nova | 19:23 | |
| *** achanda has quit IRC | 19:26 | |
| *** VW has quit IRC | 19:27 | |
| *** tonytan4ever has joined #openstack-nova | 19:28 | |
| *** ajo has quit IRC | 19:29 | |
| dims_ | dansmith: +1'ed the g-r bump | 19:31 |
| dansmith | dims_: thanks | 19:31 |
| openstackgerrit | Ed Leafe proposed openstack/nova-specs: Improve reporting when Scheduler returns NoValidHost https://review.openstack.org/187739 | 19:31 |
| *** mtanino has quit IRC | 19:33 | |
| *** kaisers1 has quit IRC | 19:35 | |
| dansmith | man I hate gerrit's people searchg | 19:35 |
| *** annegentle has quit IRC | 19:35 | |
| *** annegentle has joined #openstack-nova | 19:36 | |
| claudiub | mriedem: done. will +A once Jenkins says he's happy. | 19:36 |
| *** kaisers1 has joined #openstack-nova | 19:36 | |
| mriedem | claudiub: grenade failed so i'll check that out | 19:37 |
| *** VW has joined #openstack-nova | 19:37 | |
| openstackgerrit | Anand Shanmugam proposed openstack/nova: Changing defult value of description from None to empty string https://review.openstack.org/187741 | 19:37 |
| *** ajo has joined #openstack-nova | 19:38 | |
| melwitt | dansmith: me too. the best I could figure out is owner:"First Last" with the exact way their full name is | 19:39 |
| dansmith | yeah, but quotes don't seem to work in the add-to-this-review box | 19:39 |
| mriedem | alaski: check-tempest-dsvm-cells should probably not be running on stable/juno right? | 19:40 |
| mriedem | it's non-voting, but is there even a point in running it? | 19:40 |
| openstackgerrit | Anand Shanmugam proposed openstack/nova: Changing defult value of description from None to empty string https://review.openstack.org/187741 | 19:40 |
| openstackgerrit | Anand Shanmugam proposed openstack/nova: Changing default value of description from None to empty string https://review.openstack.org/187741 | 19:40 |
| melwitt | dansmith: oh yeah, the add to review box. I haven't figured that out | 19:40 |
| dansmith | melwitt: I tried swearing like a sailor at it, but that didn't help either :/ | 19:41 |
| melwitt | dansmith: :) I do the primitive "start typing their email" and hope it autocompletes and try that repeatedly | 19:42 |
| *** krak has joined #openstack-nova | 19:43 | |
| *** kaisers has quit IRC | 19:43 | |
| *** qianli_ has quit IRC | 19:44 | |
| *** moshele has joined #openstack-nova | 19:44 | |
| *** qianli_ has joined #openstack-nova | 19:44 | |
| krak | Always reproducible, can someone please have a look? http://paste.openstack.org/show/257173/ | 19:44 |
| mriedem | krak: see channel topic please | 19:45 |
| *** kaisers has joined #openstack-nova | 19:45 | |
| krak | @mriedem : sorry didnt get you ? | 19:46 |
| mriedem | krak: this is a development channel, not a support channel | 19:46 |
| krak | aah okay | 19:46 |
| mriedem | support questions are for the #openstack channel or ask.openstack.org | 19:46 |
| krak | sure . thanks mriedem | 19:47 |
| *** annegent_ has joined #openstack-nova | 19:47 | |
| mriedem | the formatting of your paste is hard to read btw | 19:47 |
| krak | umm, you need to read it in raw. | 19:47 |
| *** kiran-r has quit IRC | 19:48 | |
| krak | mriedem: It's something related to VNC, so I thought someone must have worked on it here. | 19:48 |
| *** mtanino has joined #openstack-nova | 19:48 | |
| *** ivasev has quit IRC | 19:49 | |
| *** belmoreira has quit IRC | 19:49 | |
| *** annegentle has quit IRC | 19:50 | |
| *** otter768 has joined #openstack-nova | 19:52 | |
| *** ddaskal has joined #openstack-nova | 19:54 | |
| *** alanf-mc has quit IRC | 19:54 | |
| *** otter768 has quit IRC | 19:56 | |
| *** tjones1 has quit IRC | 19:57 | |
| *** matrohon has joined #openstack-nova | 19:57 | |
| dansmith | I've seen a several ironic pxe job fails in the last couple hours | 19:58 |
| dansmith | is that a thing? | 19:58 |
| dims_ | dansmith: oslo.messaging tripped up ironic conductor we are trying to merge a fix for it | 19:59 |
| mriedem | dansmith: link? | 19:59 |
| alaski | mriedem: right, the tempest-cells job is never going to pass on juno unless we backport a lot of stuff, which isn't likely to happen | 19:59 |
| mriedem | dims_: is there a bug for it? | 19:59 |
| dansmith | dims_: okay | 19:59 |
| dansmith | mriedem: http://logs.openstack.org/18/187718/1/check/check-tempest-dsvm-ironic-pxe_ssh/2bfa220/console.html | 19:59 |
| *** sdake_ has joined #openstack-nova | 19:59 | |
| mriedem | ha http://logs.openstack.org/18/187718/1/check/check-tempest-dsvm-ironic-pxe_ssh/2bfa220/logs/screen-ir-cond.txt.gz?level=TRACE | 19:59 |
| dims_ | mriedem: yes https://review.openstack.org/#/c/187722/ | 20:00 |
| dansmith | not in e-r yet that I see | 20:00 |
| mriedem | e-r is behind | 20:00 |
| alaski | dansmith: melwitt it's not foolproof, but searching on lp and using their lp username often works | 20:00 |
| mriedem | jobs were hung since sundsay | 20:00 |
| mriedem | *sunday | 20:00 |
| dansmith | alaski: yeah, but that doesn't seem to always work, as you say | 20:01 |
| dansmith | mriedem: I'm seeing quick response from e-r this morning | 20:01 |
| dansmith | mriedem: but I mean I don't see it on the e-r page | 20:01 |
| mriedem | dansmith: that's why i asked for the bug number | 20:01 |
| dansmith | okay | 20:01 |
| *** whenry has joined #openstack-nova | 20:01 | |
| mriedem | i'll have it up shortly | 20:02 |
| *** tjones1 has joined #openstack-nova | 20:02 | |
| dansmith | seems like we should make a point of putting any gate-blocking thing into e-r immediately :) | 20:02 |
| mriedem | yeah | 20:02 |
| dansmith | to stop people like myself from asking dumb questions and feeling stupid :) | 20:02 |
| mriedem | have to know there is an issue first | 20:02 |
| dansmith | well, sounds like the oslo people knew | 20:02 |
| *** sdake has quit IRC | 20:03 | |
| *** Mike_D_laptop has quit IRC | 20:03 | |
| mriedem | i knew there were a bunch of oslo releases scheduled for today | 20:03 |
| dims_ | mriedem: all of them were done this morning | 20:03 |
| *** samueldmq has quit IRC | 20:05 | |
| dims_ | dansmith: bringing it up | 20:06 |
| *** pixelbeat has joined #openstack-nova | 20:06 | |
| bauzas | alaski: hey, do we have tomorrow's meeting for cells at 1700 UTC ? | 20:07 |
| bauzas | alaski: because I could maybe be off | 20:07 |
| alaski | bauzas: that's what I have on my calendar | 20:07 |
| bauzas | alaski: okay | 20:07 |
| *** ddaskal has quit IRC | 20:08 | |
| bauzas | alaski: so did you get the discussion with ndipanov ? | 20:08 |
| bauzas | alaski: sounds like he would be unhappy with persisting the ReqSpecd | 20:08 |
| alaski | I think I got most of it | 20:08 |
| ndipanov | well not exactly unhappy | 20:08 |
| *** Raildo has joined #openstack-nova | 20:08 | |
| ndipanov | I'd just like to see more thought put into it | 20:08 |
| alaski | it's the duplication of data that's a big concern, right? | 20:08 |
| ndipanov | I may be wrong and persisting is the right thing to do | 20:08 |
| ndipanov | or it's not that big of a deal | 20:09 |
| dansmith | well, | 20:09 |
| alaski | and being clear which object is used for which purpose | 20:09 |
| ndipanov | alaski, yes that for sure | 20:09 |
| dansmith | we *have* to persist it until we have scheduled | 20:09 |
| dansmith | alaski: right.? | 20:09 |
| dansmith | alaski: else we can't answer queries for the instance | 20:09 |
| alaski | dansmith: yes | 20:09 |
| alaski | but once the instance is created, we can hydrate the object from data on the instance and not need to dupe it | 20:09 |
| alaski | which I think addresses some concern | 20:10 |
| dansmith | with the exception of the things we don't store currently | 20:10 |
| *** openstackgerrit has quit IRC | 20:10 | |
| alaski | right, which can go to instance_extra | 20:10 |
| bauzas | that's what alaski propopsed with BuildRequest right ? | 20:10 |
| *** openstackgerrit has joined #openstack-nova | 20:10 | |
| ndipanov | alaski, right | 20:10 |
| bauzas | https://review.openstack.org/#/c/169901/3/specs/liberty/approved/add-buildrequest-obj.rst,cm | 20:11 |
| alaski | bauzas: sort of, buildrequest is sort of an empty instance | 20:11 |
| bauzas | okay, here are my thoughts | 20:11 |
| bauzas | 1/ we know that ReqSpec can't be helping all the issues we have | 20:11 |
| *** ddaskal has joined #openstack-nova | 20:12 | |
| bauzas | 2/ we need to sit down and discuss on the high-level issues that we have, and ndipanov will provide a view for that | 20:12 |
| *** alex_klimov has joined #openstack-nova | 20:12 | |
| bauzas | 3/ in the meantime, we can move on and work on implementing the ReqSpec objectification and try to see how we can persist it | 20:12 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair index api call https://review.openstack.org/70485 | 20:13 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair show api call https://review.openstack.org/183643 | 20:13 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: tests: functional tests for keypairs create in API microversion 2.4 https://review.openstack.org/183645 | 20:13 |
| openstackgerrit | Vladik Romanovsky proposed openstack/nova: Adding user_id handling to keypair create api call https://review.openstack.org/183644 | 20:13 |
| bauzas | 4/ if some new usage for ReqSpec has to be done, then we need to discuss on how it could be done | 20:13 |
| bauzas | guys ? | 20:13 |
| bauzas | (trying to figure out actions by 11pm here) | 20:13 |
| bauzas | 10.20pm actually | 20:13 |
| alaski | 5/ give up and go shopping | 20:14 |
| dansmith | yes | 20:14 |
| dansmith | that | 20:14 |
| bauzas | alaski: yey | 20:14 |
| bauzas | alaski: I'll be in Disneyland next week | 20:14 |
| bauzas | alaski: so I'll ask Cinderall | 20:14 |
| bauzas | Cinderella | 20:14 |
| *** iamjarvo has quit IRC | 20:14 | |
| alaski | my take is that the requestspec object you have proposed is an interface for what we send to the scheduler, and that's good. the issue of persistence is where the data is stored so it's not duped, or a total mess | 20:15 |
| bauzas | alaski: okay I think we agree then | 20:15 |
| alaski | and overall we have a glob of data that we need to slice up for use by the scheduler/compute/resource tracker/whatever with as little duplication as possible | 20:16 |
| *** tellesnobrega_ has quit IRC | 20:16 | |
| bauzas | alaski: fair enough, will continue to work on the implementation | 20:17 |
| alaski | and the important thing is getting the APIs right, and then sorting out persistence as best we can given what we have now | 20:17 |
| bauzas | alaski: oh +1 | 20:17 |
| *** salv-orlando has quit IRC | 20:18 | |
| ndipanov | aI completely agree with alaski fwiw | 20:18 |
| *** ppal has joined #openstack-nova | 20:18 | |
| *** nihilifer has quit IRC | 20:19 | |
| bauzas | okay, hands-on then :) | 20:19 |
| bauzas | thanks for the clarification folks, greatly appreciated | 20:20 |
| ndipanov | bauzas, so fleshing out that series more to the point where it is clear how the persisted data is accessed is where I think we would really see more informed oppinions forming | 20:20 |
| alaski | cool. And this should be written up somewhere so we can talk things through as we flesh it out | 20:20 |
| bauzas | ndipanov: that's on my path :) | 20:21 |
| ndipanov | alaski, I will respond to the email bauzas sent with some of the details I poitned out | 20:21 |
| bauzas | alaski: you mean amending the persistence spec ? | 20:22 |
| *** apmelton_away is now known as apmelton | 20:22 | |
| bauzas | nvm | 20:22 |
| bauzas | ndipanov: thanks for that | 20:22 |
| *** dprince has quit IRC | 20:22 | |
| bauzas | okay, I didn't want initially to discuss on the ReqSpec object, but rather on cells :) | 20:22 |
| alaski | ndipanov: great, that will be helpful | 20:23 |
| bauzas | alaski: I've been heard that melwitt is chasing the last 2 outstanding bugs that we have for the v1 job ? | 20:23 |
| ndipanov | ok folks I'm off | 20:23 |
| *** ppal has quit IRC | 20:23 | |
| alaski | o/ | 20:24 |
| *** ndipanov has quit IRC | 20:24 | |
| bauzas | \o and thanks | 20:24 |
| alaski | bauzas: she's chasing one that I know of | 20:24 |
| alaski | the big one that's causing the unexpected task state failures | 20:24 |
| *** amotoki_ has joined #openstack-nova | 20:24 | |
| bauzas | alaski: okay, I need fresh flesh for changing my mind from that reqspec bp | 20:24 |
| alaski | there's still the list instances negative test failure, I don't know if it's on her list | 20:25 |
| bauzas | alaski: yeah, saw that one | 20:25 |
| bauzas | alaski: okay, let's sort that out tomorrow | 20:26 |
| bauzas | alaski: I'll try to attend the meeting hopefully | 20:26 |
| bauzas | alaski: and see how I can help | 20:26 |
| bauzas | (for the negative test I mean) | 20:26 |
| alaski | cool | 20:26 |
| bauzas | alaski: gotta go, bye | 20:27 |
| bauzas | alaski: will review the cells specs by tomorrow before the meeting | 20:27 |
| alaski | bye | 20:27 |
| *** ddaskal has quit IRC | 20:28 | |
| alaski | hmm, that means I should probably update those | 20:28 |
| *** amotoki_ has quit IRC | 20:29 | |
| *** alanf-mc has joined #openstack-nova | 20:29 | |
| bauzas | alaski: based on the Summit feedback ? eh, if you have time | 20:30 |
| bauzas | alaski: I said that because I have a awful review backlog and I need to refresh my mind before the meeting | 20:31 |
| alaski | I need to scrape the etherpads and remind myself what I signed up for. But I really do need to update those specs and push up a few more | 20:32 |
| bauzas | alaski: good tip https://etherpad.openstack.org/p/YVR-nova-liberty-summit-action-items | 20:32 |
| alaski | well that's handy | 20:33 |
| bauzas | alaski: that's WIP | 20:33 |
| bauzas | alaski: at the bottom are the etherpads processed by johnthetubaguy's brain engine | 20:34 |
| *** Raildo_ has joined #openstack-nova | 20:35 | |
| melwitt | bauzas alaski: I think the unexpected task state isn't really the problem, it's the "sync from cells can undelete an instance" that's the issue, I opened this bug to track https://launchpad.net/bugs/1460350 | 20:38 |
| openstack | Launchpad bug 1460350 in OpenStack Compute (nova) "Cells: Race deleting instance can lead to instances "undeleted" at the top" [Low,In progress] - Assigned to melanie witt (melwitt) | 20:38 |
| *** operator99 is now known as gyee | 20:39 | |
| bauzas | oh my | 20:39 |
| *** Raildo has quit IRC | 20:39 | |
| melwitt | I think it gets solved by my patch that sends object to cells update_at_top, since that can't undelete anything. right now I'm running my patch through the failing test in a loop and exposing other race conditions | 20:40 |
| bauzas | melwitt: fair point | 20:40 |
| alaski | melwitt: nice | 20:40 |
| *** ociuhandu has joined #openstack-nova | 20:40 | |
| bauzas | melwitt: you mean https://review.openstack.org/#/c/176518/ ? | 20:41 |
| melwitt | the metadata key delete failure is another race, where the metadata delete request hits before the metadata update that adds the key hits, I think. I have to look at that one more | 20:41 |
| *** radek_ has quit IRC | 20:41 | |
| *** emagana has joined #openstack-nova | 20:41 | |
| alaski | an issue that I looked at internally just before the summit is instances that are deleted, but terminated_at isn't set. Which I think is a similar thing, out of order messages causing updates to fail | 20:41 |
| bauzas | melwitt: could you please update the old testing etherpad we have so I could dig into it ? | 20:41 |
| melwitt | bauzas: yes that's the patch. and do you mean update it about the metadata key thing? | 20:42 |
| bauzas | alaski: which is a bit worrying me since we will also have 2 databases for cells v2 | 20:42 |
| *** iamjarvo has joined #openstack-nova | 20:42 | |
| bauzas | melwitt: yeah, I need to refresh my mind | 20:42 |
| alaski | bauzas: but not two databases with the same info that needs to be synced | 20:43 |
| bauzas | melwitt: that's too long that I didn't help the cells job | 20:43 |
| bauzas | alaski: right | 20:43 |
| *** emagana has quit IRC | 20:43 | |
| alaski | bauzas: in general I don't think info is going to propagate up in cellsv2 | 20:43 |
| *** emagana has joined #openstack-nova | 20:44 | |
| bauzas | alaski: I remember some discussion we had about a possible usecase | 20:44 |
| bauzas | alaski: but I don't figure out which one | 20:44 |
| *** krtaylor has quit IRC | 20:44 | |
| bauzas | alaski: oh right, I remember now | 20:44 |
| bauzas | alaski: but we agreed on not doing it now | 20:45 |
| *** ddaskal has joined #openstack-nova | 20:45 | |
| bauzas | alaski: ie. migrations or evacuations out of child cells | 20:45 |
| *** moshele has quit IRC | 20:46 | |
| bauzas | okay, this time leaving | 20:47 |
| bauzas | bye | 20:47 |
| *** shashankhegde has quit IRC | 20:51 | |
| *** gyee has quit IRC | 20:53 | |
| *** gyee has joined #openstack-nova | 20:53 | |
| *** salv-orlando has joined #openstack-nova | 20:54 | |
| *** shashankhegde has joined #openstack-nova | 20:56 | |
| *** emagana has quit IRC | 20:58 | |
| *** shakamunyi_ has joined #openstack-nova | 21:03 | |
| *** vilobhmm has quit IRC | 21:03 | |
| *** vilobhmm has joined #openstack-nova | 21:03 | |
| *** Raildo__ has joined #openstack-nova | 21:03 | |
| *** aysyanne has quit IRC | 21:04 | |
| *** thangp has quit IRC | 21:04 | |
| *** shakamunyi__ has joined #openstack-nova | 21:04 | |
| *** lpetrut has quit IRC | 21:05 | |
| *** shakamunyi__ is now known as tylerdurden | 21:05 | |
| *** tylerdurden has quit IRC | 21:08 | |
| *** shakamunyi_ has quit IRC | 21:08 | |
| *** Raildo_ has quit IRC | 21:08 | |
| *** baoli_ has quit IRC | 21:08 | |
| *** VW_ has joined #openstack-nova | 21:09 | |
| *** krtaylor has joined #openstack-nova | 21:11 | |
| *** shashankhegde has quit IRC | 21:11 | |
| *** VW has quit IRC | 21:12 | |
| *** VW_ has quit IRC | 21:14 | |
| *** yamahata has quit IRC | 21:14 | |
| *** eharney has quit IRC | 21:15 | |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Add documentation on command deprecation process https://review.openstack.org/185260 | 21:15 |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Remove unused novaclient.tests.unit.v2.utils module https://review.openstack.org/185261 | 21:15 |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Deprecate volume/volume-type/volume-snapshot CRUD CLIs/APIs https://review.openstack.org/185141 | 21:15 |
| *** burt has quit IRC | 21:16 | |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Add documentation on command deprecation process https://review.openstack.org/185260 | 21:20 |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Remove unused novaclient.tests.unit.v2.utils module https://review.openstack.org/185261 | 21:20 |
| openstackgerrit | Matt Riedemann proposed openstack/python-novaclient: Deprecate volume/volume-type/volume-snapshot CRUD CLIs/APIs https://review.openstack.org/185141 | 21:20 |
| *** dboik has quit IRC | 21:21 | |
| *** Raildo__ has quit IRC | 21:22 | |
| *** jecarey has quit IRC | 21:22 | |
| *** shashankhegde has joined #openstack-nova | 21:24 | |
| *** iamjarvo has quit IRC | 21:25 | |
| *** matrohon has quit IRC | 21:27 | |
| *** annegent_ has quit IRC | 21:29 | |
| *** annegentle has joined #openstack-nova | 21:31 | |
| *** thiagop has quit IRC | 21:32 | |
| *** mriedem is now known as mriedem_away | 21:37 | |
| *** krak has quit IRC | 21:41 | |
| *** claudiub has quit IRC | 21:45 | |
| *** annegentle has quit IRC | 21:48 | |
| *** aix has joined #openstack-nova | 21:50 | |
| *** otter768 has joined #openstack-nova | 21:53 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Enable more python34 tests https://review.openstack.org/187785 | 21:53 |
| *** Raildo has joined #openstack-nova | 21:53 | |
| *** sdake has joined #openstack-nova | 21:54 | |
| *** ZZelle_ has joined #openstack-nova | 21:55 | |
| *** armax has quit IRC | 21:55 | |
| *** mwagner_afk has quit IRC | 21:57 | |
| *** fawadkhaliq has joined #openstack-nova | 21:57 | |
| *** otter768 has quit IRC | 21:58 | |
| *** sdake_ has quit IRC | 21:58 | |
| *** apuimedo has joined #openstack-nova | 21:58 | |
| *** wwriverrat has joined #openstack-nova | 21:58 | |
| *** lucas-dinner has quit IRC | 22:00 | |
| *** jlk has quit IRC | 22:01 | |
| *** jgrimm is now known as zz_jgrimm | 22:01 | |
| cfriesen | if anyone's on still...is there a way to ensure that an object member has been loaded from the DB? I'm hitting a case in build_request_spec() where it just assigns "instance" to "instance_properties", but the instance.pci_requests member isn't loaded | 22:03 |
| *** agireud has quit IRC | 22:03 | |
| cfriesen | I can trace back to where it's initially loaded from the uuid and make sure it's loaded there, but what about code paths where we don't know initially what we might need to load? | 22:04 |
| *** kun_huang has quit IRC | 22:04 | |
| *** tonytan4ever has quit IRC | 22:06 | |
| *** tonytan4ever has joined #openstack-nova | 22:07 | |
| *** aweeks has quit IRC | 22:08 | |
| *** vilobhmm has quit IRC | 22:08 | |
| *** vilobhmm has joined #openstack-nova | 22:08 | |
| *** tonytan4ever has quit IRC | 22:11 | |
| *** annegentle has joined #openstack-nova | 22:12 | |
| *** ZZelle_ has quit IRC | 22:14 | |
| openstackgerrit | Ed Leafe proposed openstack/nova-specs: Improve reporting when Scheduler returns NoValidHost https://review.openstack.org/187739 | 22:15 |
| *** BadCub has quit IRC | 22:21 | |
| *** annegentle has quit IRC | 22:29 | |
| *** artom has quit IRC | 22:29 | |
| *** artom has joined #openstack-nova | 22:30 | |
| *** dims__ has joined #openstack-nova | 22:31 | |
| *** baoli has joined #openstack-nova | 22:33 | |
| *** Erming_ has joined #openstack-nova | 22:33 | |
| *** kylek3h has quit IRC | 22:34 | |
| *** mdrabe has quit IRC | 22:34 | |
| Erming_ | Hi, I am wondering if it's possible to let VM use a specified physical partition? For example, let ephemeral disk use a separated physical partition? | 22:34 |
| *** dims_ has quit IRC | 22:34 | |
| *** Sukhdev has joined #openstack-nova | 22:35 | |
| *** kun_huang has joined #openstack-nova | 22:35 | |
| *** slong has joined #openstack-nova | 22:38 | |
| Erming_ | in default it only uses the root partition which often is tight on space. | 22:42 |
| Erming_ | I mean on the hypervisor node | 22:42 |
| *** baoli has quit IRC | 22:43 | |
| *** baoli has joined #openstack-nova | 22:46 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Enable more python34 tests https://review.openstack.org/187785 | 22:47 |
| *** ddaskal has quit IRC | 22:50 | |
| *** aweeks has joined #openstack-nova | 22:50 | |
| *** ian_brown has joined #openstack-nova | 22:52 | |
| *** whenry has quit IRC | 22:52 | |
| *** diegows has joined #openstack-nova | 22:53 | |
| *** mwagner_afk has joined #openstack-nova | 22:54 | |
| *** liusheng has quit IRC | 22:55 | |
| *** baoli has quit IRC | 22:56 | |
| *** liusheng has joined #openstack-nova | 22:56 | |
| *** marun has quit IRC | 22:57 | |
| *** apmelton is now known as apmelton_away | 22:59 | |
| *** baoli has joined #openstack-nova | 22:59 | |
| *** yamahata has joined #openstack-nova | 23:00 | |
| *** ian_brown has quit IRC | 23:03 | |
| *** zzzeek has quit IRC | 23:03 | |
| *** BadCub has joined #openstack-nova | 23:05 | |
| *** ian_brown has joined #openstack-nova | 23:06 | |
| *** BadCub has quit IRC | 23:09 | |
| *** aweeks has quit IRC | 23:09 | |
| *** baoli has quit IRC | 23:10 | |
| *** baoli has joined #openstack-nova | 23:15 | |
| *** BadCub has joined #openstack-nova | 23:16 | |
| *** Alexandra_ has joined #openstack-nova | 23:17 | |
| *** markvoelker_ has quit IRC | 23:18 | |
| *** annegentle has joined #openstack-nova | 23:18 | |
| *** aweeks has joined #openstack-nova | 23:19 | |
| *** annegentle has quit IRC | 23:20 | |
| *** nelsnelson has quit IRC | 23:22 | |
| *** armax has joined #openstack-nova | 23:27 | |
| *** baoli has quit IRC | 23:29 | |
| *** changbl_ has quit IRC | 23:30 | |
| mikal | . | 23:31 |
| openstackgerrit | Zhang Ni proposed openstack/nova-specs: Add volume type to create server API https://review.openstack.org/187812 | 23:32 |
| *** ian_brown has quit IRC | 23:32 | |
| openstackgerrit | Zhang Ni proposed openstack/nova-specs: Add volume type to create server API https://review.openstack.org/187812 | 23:34 |
| openstackgerrit | Puneet proposed openstack/nova: nova response code 403 on block device quota error https://review.openstack.org/187813 | 23:35 |
| *** baoli has joined #openstack-nova | 23:37 | |
| *** markvoelker has joined #openstack-nova | 23:40 | |
| *** alex_klimov has quit IRC | 23:42 | |
| *** bradjones has quit IRC | 23:43 | |
| *** nelsnelson has joined #openstack-nova | 23:43 | |
| *** bradjones has joined #openstack-nova | 23:45 | |
| *** hshiina has joined #openstack-nova | 23:47 | |
| *** salv-orlando has quit IRC | 23:47 | |
| *** otter768 has joined #openstack-nova | 23:54 | |
| *** ijw has joined #openstack-nova | 23:54 | |
| *** diegows has quit IRC | 23:56 | |
| *** barra204 has quit IRC | 23:57 | |
| *** shakamunyi has quit IRC | 23:57 | |
| *** browne has quit IRC | 23:57 | |
| *** otter768 has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
