| *** achanda has quit IRC | 00:00 | |
| mriedem | i don't see any recent changes in nova.virt.disk | 00:03 |
|---|---|---|
| mriedem | boris-42: did anything change in rally around that time? | 00:04 |
| *** baoli has joined #openstack-nova | 00:04 | |
| boris-42 | mriedem: I even don't know what to say, why changes in rally are producing errors in nova logs?) | 00:04 |
| boris-42 | I mean how they can* | 00:04 |
| *** baoli has quit IRC | 00:04 | |
| boris-42 | mriedem: as well we didn't merge too much at this period https://github.com/stackforge/rally/commits/master | 00:05 |
| *** salv-orl_ has quit IRC | 00:06 | |
| *** armax has quit IRC | 00:07 | |
| boris-42 | mriedem: let me try to reset nova to march 26 =) | 00:11 |
| boris-42 | and make depends-on patch in rally | 00:12 |
| *** VW_ has quit IRC | 00:16 | |
| openstackgerrit | Boris Pavlovic proposed openstack/nova: do not merge https://review.openstack.org/168747 | 00:18 |
| *** VW_ has joined #openstack-nova | 00:18 | |
| mriedem | boris-42: well, i'm thinking config/deployment of rally | 00:19 |
| *** promulo__ has joined #openstack-nova | 00:19 | |
| mriedem | like a change to devstack or something, i don't know how rally is setup | 00:19 |
| mriedem | i'd think if there were huge timing issues going on in nova, we'd be having timeouts in the gate jobs with tempest | 00:19 |
| boris-42 | mriedem: so nothing changed I mean rally uses standard dsvm job (like tempest) | 00:20 |
| boris-42 | mriedem: and we didn't change integration between rally-dsvm-gates for a long period of time | 00:20 |
| boris-42 | mriedem: so this is going to test nova https://review.openstack.org/#/c/168748/ | 00:22 |
| *** VW_ has quit IRC | 00:22 | |
| boris-42 | mriedem: that was on the end of 25 March | 00:23 |
| *** promulo has quit IRC | 00:23 | |
| *** ferest has joined #openstack-nova | 00:24 | |
| *** annegentle has quit IRC | 00:26 | |
| *** sdake has joined #openstack-nova | 00:26 | |
| *** yingjun has joined #openstack-nova | 00:28 | |
| *** annegentle has joined #openstack-nova | 00:29 | |
| *** ferest has quit IRC | 00:29 | |
| *** shaohe_feng has quit IRC | 00:36 | |
| *** haomaiwang has quit IRC | 00:36 | |
| *** arnaud__ has joined #openstack-nova | 00:38 | |
| *** yingjun has quit IRC | 00:39 | |
| *** gokrokve has joined #openstack-nova | 00:39 | |
| *** gokrokve has quit IRC | 00:44 | |
| *** smccully has quit IRC | 00:48 | |
| *** jhesketh has quit IRC | 00:50 | |
| *** jhesketh has joined #openstack-nova | 00:53 | |
| *** yingjun has joined #openstack-nova | 00:53 | |
| *** hdd has quit IRC | 00:59 | |
| *** achanda has joined #openstack-nova | 01:01 | |
| boris-42 | mriedem: it's strange that I can't reporduce this in local run.. | 01:04 |
| *** eharney has quit IRC | 01:04 | |
| *** achanda has quit IRC | 01:06 | |
| *** salv-orlando has joined #openstack-nova | 01:07 | |
| *** shakamunyi has joined #openstack-nova | 01:07 | |
| *** shakamunyi_ has joined #openstack-nova | 01:07 | |
| *** VW_ has joined #openstack-nova | 01:09 | |
| *** salv-orlando has quit IRC | 01:12 | |
| *** yingjun has quit IRC | 01:13 | |
| *** yingjun has joined #openstack-nova | 01:14 | |
| *** VW_ has quit IRC | 01:15 | |
| *** Longgeek has joined #openstack-nova | 01:15 | |
| *** annegentle has quit IRC | 01:19 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Avoid MODULEPATH environment var in config generator https://review.openstack.org/168745 | 01:23 |
| *** eliqiao has joined #openstack-nova | 01:23 | |
| *** casusbelli has joined #openstack-nova | 01:30 | |
| *** shaohe_feng has joined #openstack-nova | 01:30 | |
| *** zhangtralon has joined #openstack-nova | 01:32 | |
| *** casusbel_ has quit IRC | 01:33 | |
| *** kaisers1 has joined #openstack-nova | 01:35 | |
| *** arnaud__ has quit IRC | 01:35 | |
| *** kaisers has quit IRC | 01:35 | |
| *** annashen has joined #openstack-nova | 01:37 | |
| *** baoli has joined #openstack-nova | 01:37 | |
| *** shaohe_robot has quit IRC | 01:38 | |
| *** gokrokve has joined #openstack-nova | 01:39 | |
| *** annashen has quit IRC | 01:41 | |
| *** gokrokve has quit IRC | 01:44 | |
| *** mhorban has quit IRC | 01:49 | |
| *** mhorban has joined #openstack-nova | 01:49 | |
| *** erkules has joined #openstack-nova | 01:50 | |
| *** sdake has quit IRC | 01:51 | |
| *** VW_ has joined #openstack-nova | 01:53 | |
| *** erkules_ has quit IRC | 01:53 | |
| *** salv-orlando has joined #openstack-nova | 01:53 | |
| *** igordcard_ has quit IRC | 01:54 | |
| *** salv-orlando has quit IRC | 01:58 | |
| openstackgerrit | Eli Qiao proposed openstack/nova: Libvirt: show real vCpuCompareResult instead of a url link https://review.openstack.org/165671 | 01:59 |
| *** baoli has quit IRC | 02:01 | |
| openstackgerrit | p-draigbrady proposed openstack/nova: support getting quota usage for all tenants https://review.openstack.org/168644 | 02:03 |
| *** yasu_ has joined #openstack-nova | 02:04 | |
| *** salv-orlando has joined #openstack-nova | 02:10 | |
| *** VW_ has quit IRC | 02:12 | |
| *** salv-orlando has quit IRC | 02:12 | |
| *** Longgeek has quit IRC | 02:12 | |
| *** VW_ has joined #openstack-nova | 02:13 | |
| *** haomaiwang has joined #openstack-nova | 02:13 | |
| *** Longgeek has joined #openstack-nova | 02:16 | |
| *** yingjun has quit IRC | 02:17 | |
| *** VW_ has quit IRC | 02:18 | |
| *** popw has joined #openstack-nova | 02:18 | |
| *** Longgeek has quit IRC | 02:26 | |
| *** Longgeek has joined #openstack-nova | 02:26 | |
| *** yingjun has joined #openstack-nova | 02:29 | |
| *** baoli has joined #openstack-nova | 02:29 | |
| *** mriedem1 has joined #openstack-nova | 02:37 | |
| *** annashen has joined #openstack-nova | 02:38 | |
| *** mriedem has quit IRC | 02:39 | |
| *** gokrokve has joined #openstack-nova | 02:39 | |
| *** baoli has quit IRC | 02:42 | |
| *** annashen has quit IRC | 02:42 | |
| *** gokrokve has quit IRC | 02:44 | |
| *** wverdugo500 has quit IRC | 02:44 | |
| *** baoli has joined #openstack-nova | 02:45 | |
| *** annegentle has joined #openstack-nova | 02:45 | |
| *** annegentle has quit IRC | 02:51 | |
| *** shakamunyi__ has joined #openstack-nova | 02:51 | |
| *** david-ly_ is now known as david-lyle | 02:52 | |
| *** promulo__ has quit IRC | 02:52 | |
| *** shakamunyi has quit IRC | 02:54 | |
| *** shakamunyi_ has quit IRC | 02:54 | |
| *** shakamunyi has joined #openstack-nova | 02:54 | |
| *** mriedem1 has quit IRC | 02:58 | |
| *** dims has quit IRC | 02:59 | |
| *** dims has joined #openstack-nova | 02:59 | |
| *** dims has quit IRC | 03:00 | |
| *** wverdugo500 has joined #openstack-nova | 03:02 | |
| *** yingjun has quit IRC | 03:03 | |
| *** haigang has joined #openstack-nova | 03:04 | |
| *** Longgeek_ has joined #openstack-nova | 03:07 | |
| *** yingjun has joined #openstack-nova | 03:08 | |
| *** Longgeek has quit IRC | 03:10 | |
| *** iamjarvo has joined #openstack-nova | 03:16 | |
| *** iamjarvo has quit IRC | 03:16 | |
| *** arnaud__ has joined #openstack-nova | 03:24 | |
| *** armax has joined #openstack-nova | 03:24 | |
| *** iamjarvo has joined #openstack-nova | 03:24 | |
| *** salv-orlando has joined #openstack-nova | 03:25 | |
| *** armax has quit IRC | 03:25 | |
| *** sandywalsh has quit IRC | 03:28 | |
| *** arnaud__ has quit IRC | 03:28 | |
| *** salv-orlando has quit IRC | 03:30 | |
| *** dave-mcc_ has joined #openstack-nova | 03:34 | |
| *** sandywalsh has joined #openstack-nova | 03:37 | |
| *** dave-mccowan has quit IRC | 03:37 | |
| *** otter768 has quit IRC | 03:38 | |
| *** gokrokve has joined #openstack-nova | 03:39 | |
| *** wverdugo500 has quit IRC | 03:40 | |
| *** hdd has joined #openstack-nova | 03:42 | |
| *** gokrokve has quit IRC | 03:43 | |
| *** wverdugo500 has joined #openstack-nova | 03:45 | |
| *** iamjarvo has quit IRC | 03:47 | |
| *** baoli has quit IRC | 03:49 | |
| *** sdake has joined #openstack-nova | 03:49 | |
| *** casusbelli has quit IRC | 03:51 | |
| *** deepthi has joined #openstack-nova | 04:01 | |
| *** yingjun has quit IRC | 04:05 | |
| *** shaohe_feng has quit IRC | 04:08 | |
| *** takedakn has joined #openstack-nova | 04:12 | |
| *** dims has joined #openstack-nova | 04:15 | |
| *** dims has quit IRC | 04:21 | |
| *** neelashah has joined #openstack-nova | 04:22 | |
| *** pixelb has quit IRC | 04:23 | |
| *** neelashah has quit IRC | 04:24 | |
| *** achanda has joined #openstack-nova | 04:24 | |
| *** arnaud__ has joined #openstack-nova | 04:24 | |
| *** takedakn has quit IRC | 04:26 | |
| *** jecarey has quit IRC | 04:27 | |
| *** arnaud__ has quit IRC | 04:29 | |
| *** arnaud__ has joined #openstack-nova | 04:31 | |
| *** annashen has joined #openstack-nova | 04:39 | |
| *** shakamunyi_ has joined #openstack-nova | 04:39 | |
| *** gokrokve has joined #openstack-nova | 04:39 | |
| *** shakamunyi__ has quit IRC | 04:41 | |
| *** shakamunyi has quit IRC | 04:41 | |
| *** shakamunyi has joined #openstack-nova | 04:41 | |
| *** zhangtralon has quit IRC | 04:42 | |
| *** annashen has quit IRC | 04:44 | |
| *** gokrokve has quit IRC | 04:44 | |
| *** lpetrut has joined #openstack-nova | 04:46 | |
| *** welldannit has joined #openstack-nova | 04:46 | |
| *** sdake has quit IRC | 04:49 | |
| *** takedakn has joined #openstack-nova | 04:50 | |
| *** garyk has joined #openstack-nova | 04:53 | |
| *** pixelb has joined #openstack-nova | 04:54 | |
| openstackgerrit | Eric Brown proposed openstack/nova: Remove dead vmvc code https://review.openstack.org/168778 | 04:54 |
| *** pradipta has joined #openstack-nova | 04:54 | |
| *** shakamunyi_ has quit IRC | 05:04 | |
| *** pixelb has quit IRC | 05:05 | |
| *** oro has joined #openstack-nova | 05:08 | |
| *** oro_ has joined #openstack-nova | 05:08 | |
| *** takedakn has quit IRC | 05:13 | |
| *** mrodden has quit IRC | 05:17 | |
| *** oro_ has quit IRC | 05:17 | |
| *** shakamunyi has quit IRC | 05:18 | |
| *** oro has quit IRC | 05:19 | |
| *** nkrinner has joined #openstack-nova | 05:19 | |
| *** Alexandra_ has joined #openstack-nova | 05:19 | |
| *** takedakn has joined #openstack-nova | 05:20 | |
| *** arnaud___ has joined #openstack-nova | 05:20 | |
| *** arnaud__ has quit IRC | 05:21 | |
| *** unicell has quit IRC | 05:22 | |
| *** unicell has joined #openstack-nova | 05:22 | |
| *** achanda has quit IRC | 05:22 | |
| *** achanda_ has joined #openstack-nova | 05:22 | |
| *** ijw_ has quit IRC | 05:22 | |
| *** shakamunyi has joined #openstack-nova | 05:22 | |
| *** takedakn has quit IRC | 05:24 | |
| *** arnaud___ has quit IRC | 05:25 | |
| *** ildikov has quit IRC | 05:27 | |
| *** achanda has joined #openstack-nova | 05:29 | |
| *** vivekd has joined #openstack-nova | 05:30 | |
| *** achanda_ has quit IRC | 05:30 | |
| *** hblixt has joined #openstack-nova | 05:31 | |
| *** mrodden has joined #openstack-nova | 05:32 | |
| *** deepthi has quit IRC | 05:33 | |
| *** pradipta has quit IRC | 05:35 | |
| *** otter768 has joined #openstack-nova | 05:39 | |
| *** hdd has quit IRC | 05:39 | |
| *** yingjun has joined #openstack-nova | 05:39 | |
| *** ijw has joined #openstack-nova | 05:40 | |
| *** shaohe_feng has joined #openstack-nova | 05:40 | |
| *** otter768 has quit IRC | 05:43 | |
| *** deepthi has joined #openstack-nova | 05:49 | |
| *** VW_ has joined #openstack-nova | 05:50 | |
| *** vmtrooper has joined #openstack-nova | 05:50 | |
| openstackgerrit | Eli Qiao proposed openstack/nova: API: Don't reload policy files in policy.d every time https://review.openstack.org/168784 | 05:50 |
| openstackgerrit | yanheven proposed openstack/nova: replace "hpy.host" with "hyp.get('host')" https://review.openstack.org/168785 | 05:51 |
| *** VW_ has quit IRC | 05:55 | |
| openstackgerrit | garyk proposed openstack/nova: DB: add support for volume_type_id by BDMS https://review.openstack.org/139992 | 05:55 |
| openstackgerrit | garyk proposed openstack/nova: Object: add volume_type_id to block_device objects https://review.openstack.org/139994 | 05:55 |
| *** moshele has joined #openstack-nova | 05:57 | |
| *** Marga_ has joined #openstack-nova | 05:58 | |
| *** Marga_ has quit IRC | 06:01 | |
| eliqiao | garyk: hi, u round? | 06:01 |
| *** Marga_ has joined #openstack-nova | 06:01 | |
| garyk | eliqiao: yes, how are you? | 06:02 |
| eliqiao | garyk: cool, thanks for your comments. | 06:02 |
| eliqiao | garyk: it's re https://review.openstack.org/#/c/156112/ , I'v added some comments, can you help to check it again? | 06:03 |
| *** stockpirate has joined #openstack-nova | 06:05 | |
| *** rajesht has joined #openstack-nova | 06:07 | |
| *** vmtrooper has quit IRC | 06:10 | |
| *** vmtrooper has joined #openstack-nova | 06:11 | |
| *** vmtrooper has quit IRC | 06:15 | |
| *** zhangtralon has joined #openstack-nova | 06:17 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Imported Translations from Transifex https://review.openstack.org/162030 | 06:17 |
| stockpirate | where can i find original nova.conf files+ | 06:18 |
| stockpirate | samples | 06:18 |
| garyk | eliqiao: yes, your comments make sense. i just think that maybe add to the docstring that the neutron client should be an admin one | 06:19 |
| *** Nikolay_St has joined #openstack-nova | 06:20 | |
| *** jamielennox is now known as jamielennox|away | 06:20 | |
| *** tralon_ has joined #openstack-nova | 06:20 | |
| *** markvoelker has joined #openstack-nova | 06:22 | |
| *** zhangtralon has quit IRC | 06:23 | |
| *** vilobhmm__ has joined #openstack-nova | 06:24 | |
| *** vilobhmm__ has left #openstack-nova | 06:25 | |
| *** shakamunyi has quit IRC | 06:25 | |
| *** markvoelker has quit IRC | 06:27 | |
| *** dsanders has quit IRC | 06:27 | |
| *** sahid has joined #openstack-nova | 06:35 | |
| *** dsanders has joined #openstack-nova | 06:35 | |
| *** takedakn has joined #openstack-nova | 06:37 | |
| *** erkules has quit IRC | 06:43 | |
| *** erkules has joined #openstack-nova | 06:43 | |
| *** takedakn has quit IRC | 06:47 | |
| *** ildikov has joined #openstack-nova | 06:47 | |
| *** belmoreira has joined #openstack-nova | 06:48 | |
| openstackgerrit | Rui Chen proposed openstack/nova: Fix failing of stopping intance when init host https://review.openstack.org/167926 | 06:49 |
| openstackgerrit | Rui Chen proposed openstack/nova: Fix empty instance nw_info cache https://review.openstack.org/146036 | 06:50 |
| *** atuvenie has joined #openstack-nova | 06:50 | |
| *** salv-orlando has joined #openstack-nova | 06:54 | |
| openstackgerrit | Lan Qi Song proposed openstack/nova: VMware: Attach iSCSI volume support CHAP authentication https://review.openstack.org/168795 | 06:56 |
| *** salv-orl_ has joined #openstack-nova | 06:58 | |
| *** salv-orlando has quit IRC | 06:58 | |
| *** doude has joined #openstack-nova | 06:59 | |
| *** nuritv has joined #openstack-nova | 07:00 | |
| *** afazekas has joined #openstack-nova | 07:02 | |
| *** hblixt has quit IRC | 07:02 | |
| *** ankit_ag has joined #openstack-nova | 07:03 | |
| *** achanda has quit IRC | 07:08 | |
| *** dsanders has quit IRC | 07:10 | |
| *** veena has joined #openstack-nova | 07:12 | |
| *** garyk has quit IRC | 07:12 | |
| *** mrda is now known as mrda-away | 07:12 | |
| veena | Hi zz_johnthetubagu | 07:13 |
| veena | Hi Mikal | 07:14 |
| veena | Hi ttx | 07:15 |
| *** Marga_ has quit IRC | 07:16 | |
| eliqiao | garyk: thanks, any suggestion where I should put that docstring? sorry for response late.. :) | 07:17 |
| *** doude_ has joined #openstack-nova | 07:18 | |
| *** doude has quit IRC | 07:22 | |
| *** markvoelker has joined #openstack-nova | 07:23 | |
| *** TobiasE has joined #openstack-nova | 07:24 | |
| *** markvoelker has quit IRC | 07:27 | |
| *** deepthi has quit IRC | 07:28 | |
| *** vivekd has quit IRC | 07:29 | |
| *** ildikov has quit IRC | 07:30 | |
| *** unicell has quit IRC | 07:30 | |
| *** Alexandra_ has quit IRC | 07:30 | |
| *** unicell has joined #openstack-nova | 07:31 | |
| *** amotoki has joined #openstack-nova | 07:38 | |
| *** markus_z has joined #openstack-nova | 07:39 | |
| *** romainh has joined #openstack-nova | 07:39 | |
| *** stockpirate has quit IRC | 07:39 | |
| *** otter768 has joined #openstack-nova | 07:39 | |
| *** annashen has joined #openstack-nova | 07:42 | |
| *** salv-orl_ has quit IRC | 07:42 | |
| *** jlanoux has joined #openstack-nova | 07:43 | |
| *** liusheng has quit IRC | 07:44 | |
| *** otter768 has quit IRC | 07:44 | |
| *** dulek has joined #openstack-nova | 07:44 | |
| *** vivekd has joined #openstack-nova | 07:45 | |
| *** romainh has quit IRC | 07:45 | |
| *** ildikov has joined #openstack-nova | 07:45 | |
| *** jyuso1 has joined #openstack-nova | 07:46 | |
| *** annashen has quit IRC | 07:46 | |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix config option types https://review.openstack.org/168481 | 07:49 |
| *** veena_ has joined #openstack-nova | 07:50 | |
| *** vivekd has quit IRC | 07:50 | |
| *** veena has quit IRC | 07:51 | |
| *** garyk has joined #openstack-nova | 07:52 | |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix config option types https://review.openstack.org/168481 | 07:52 |
| *** mpaolino has joined #openstack-nova | 07:52 | |
| *** dims has joined #openstack-nova | 07:53 | |
| *** ndipanov has joined #openstack-nova | 07:53 | |
| *** deepthi has joined #openstack-nova | 07:53 | |
| *** lpetrut has quit IRC | 07:53 | |
| *** Maike has joined #openstack-nova | 07:55 | |
| openstackgerrit | Deepak C Shetty proposed openstack/nova: libvirt: Use 'relative' flag for online snapshot's commit/rebase operations https://review.openstack.org/168805 | 07:57 |
| *** vivekd has joined #openstack-nova | 07:58 | |
| *** dims has quit IRC | 07:58 | |
| *** Mike_D_wk has quit IRC | 07:58 | |
| *** vmtrooper has joined #openstack-nova | 07:59 | |
| *** gszasz has joined #openstack-nova | 08:00 | |
| *** eglynn has joined #openstack-nova | 08:01 | |
| *** Mike_D_wk has joined #openstack-nova | 08:02 | |
| *** vmtrooper has quit IRC | 08:04 | |
| *** vigneshvar has joined #openstack-nova | 08:05 | |
| *** vigneshvar_ has joined #openstack-nova | 08:05 | |
| *** romainh has joined #openstack-nova | 08:07 | |
| *** browne has quit IRC | 08:08 | |
| *** subscope has joined #openstack-nova | 08:09 | |
| *** jistr has joined #openstack-nova | 08:09 | |
| *** lucasagomes has joined #openstack-nova | 08:13 | |
| openstackgerrit | Ghanshyam Mann proposed openstack/nova: Merge V2 and V2.1 services functional tests https://review.openstack.org/168808 | 08:15 |
| openstackgerrit | Moshe Levi proposed openstack/nova: Add options to ssh to prevent prompting https://review.openstack.org/168680 | 08:16 |
| *** haigang has quit IRC | 08:18 | |
| *** haigang has joined #openstack-nova | 08:18 | |
| *** haigang has quit IRC | 08:19 | |
| *** ildikov has quit IRC | 08:23 | |
| *** Maike has quit IRC | 08:30 | |
| *** apuimedo has joined #openstack-nova | 08:31 | |
| *** xek has quit IRC | 08:31 | |
| *** xek has joined #openstack-nova | 08:31 | |
| *** pradipta has joined #openstack-nova | 08:33 | |
| *** ildikov has joined #openstack-nova | 08:38 | |
| openstackgerrit | yanheven proposed openstack/nova: replace "obj.attr" with "obj['attr']" https://review.openstack.org/168785 | 08:38 |
| *** oro_ has joined #openstack-nova | 08:39 | |
| *** oro has joined #openstack-nova | 08:39 | |
| *** salv-orlando has joined #openstack-nova | 08:43 | |
| *** salv-orlando has quit IRC | 08:47 | |
| veena_ | Hi dims | 08:48 |
| *** Longgeek_ has quit IRC | 08:53 | |
| *** haigang has joined #openstack-nova | 08:55 | |
| openstackgerrit | vsaienko proposed openstack/nova: add ironic hypervisor type https://review.openstack.org/168820 | 08:55 |
| *** vigneshvar has quit IRC | 08:56 | |
| *** ZZelle has joined #openstack-nova | 08:56 | |
| *** vigneshvar_ has quit IRC | 08:56 | |
| *** gszasz has quit IRC | 09:02 | |
| *** zz_johnthetubagu is now known as johnthetubaguy | 09:07 | |
| *** rdo has quit IRC | 09:11 | |
| *** rdo has joined #openstack-nova | 09:13 | |
| *** yingjun has quit IRC | 09:17 | |
| *** ociuhandu has joined #openstack-nova | 09:17 | |
| *** yingjun has joined #openstack-nova | 09:17 | |
| *** bkopilov has quit IRC | 09:18 | |
| *** bkopilov has joined #openstack-nova | 09:23 | |
| *** markvoelker has joined #openstack-nova | 09:24 | |
| *** mpaolino has quit IRC | 09:25 | |
| openstackgerrit | Takenori Yoshimatsu proposed openstack/nova: Catch AttributeError at instance.info_cache.delete https://review.openstack.org/167905 | 09:26 |
| *** smccully has joined #openstack-nova | 09:26 | |
| *** markvoelker has quit IRC | 09:29 | |
| *** eliqiao has quit IRC | 09:29 | |
| *** yingjun has quit IRC | 09:32 | |
| *** derekh has joined #openstack-nova | 09:36 | |
| *** bkopilov has quit IRC | 09:37 | |
| *** bkopilov has joined #openstack-nova | 09:39 | |
| *** vivekd has quit IRC | 09:39 | |
| *** otter768 has joined #openstack-nova | 09:40 | |
| *** annashen has joined #openstack-nova | 09:43 | |
| *** otter768 has quit IRC | 09:45 | |
| *** annashen has quit IRC | 09:48 | |
| *** vmtrooper has joined #openstack-nova | 09:48 | |
| *** vmtrooper has quit IRC | 09:53 | |
| *** dims has joined #openstack-nova | 09:54 | |
| *** vigneshvar has joined #openstack-nova | 09:58 | |
| *** Longgeek has joined #openstack-nova | 09:58 | |
| openstackgerrit | garyk proposed openstack/nova: DB: add support for volume_type_id by BDMS https://review.openstack.org/139992 | 09:59 |
| *** vigneshvar_ has joined #openstack-nova | 09:59 | |
| openstackgerrit | garyk proposed openstack/nova: Object: add volume_type_id to block_device objects https://review.openstack.org/139994 | 10:00 |
| openstackgerrit | garyk proposed openstack/nova: Neutron: add in API method for updating VNIC index https://review.openstack.org/147126 | 10:04 |
| *** oro has quit IRC | 10:05 | |
| *** oro_ has quit IRC | 10:05 | |
| openstackgerrit | garyk proposed openstack/nova: Metadata: support proxying loadbalancers https://review.openstack.org/147539 | 10:05 |
| *** tralon_ has quit IRC | 10:12 | |
| *** flip214 has quit IRC | 10:17 | |
| *** flip214 has joined #openstack-nova | 10:17 | |
| *** Longgeek has quit IRC | 10:17 | |
| *** salv-orlando has joined #openstack-nova | 10:23 | |
| *** markvoelker has joined #openstack-nova | 10:25 | |
| *** markvoelker has quit IRC | 10:30 | |
| *** IanGovett has joined #openstack-nova | 10:31 | |
| *** romainh has quit IRC | 10:33 | |
| *** tralon_ has joined #openstack-nova | 10:35 | |
| *** bkopilov has quit IRC | 10:37 | |
| *** haigang has quit IRC | 10:40 | |
| *** johnthetubaguy is now known as zz_johnthetubagu | 10:41 | |
| *** iamjarvo has joined #openstack-nova | 10:46 | |
| *** haomaiwang has quit IRC | 10:47 | |
| *** bkopilov has joined #openstack-nova | 10:47 | |
| *** mpavone has joined #openstack-nova | 10:48 | |
| *** popw has quit IRC | 10:49 | |
| *** mwagner_lap has quit IRC | 10:52 | |
| *** vigneshvar_ has quit IRC | 11:04 | |
| *** vigneshvar has quit IRC | 11:04 | |
| *** vigneshvar has joined #openstack-nova | 11:05 | |
| *** vigneshvar_ has joined #openstack-nova | 11:05 | |
| *** yasu_ has quit IRC | 11:09 | |
| *** oro has joined #openstack-nova | 11:11 | |
| *** VW_ has joined #openstack-nova | 11:14 | |
| *** Longgeek has joined #openstack-nova | 11:15 | |
| *** jaypipes has joined #openstack-nova | 11:17 | |
| *** VW_ has quit IRC | 11:19 | |
| *** pixelb has joined #openstack-nova | 11:19 | |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Cleanup conductor before RPC major version bump https://review.openstack.org/168856 | 11:20 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Add version 3.0 of conductor RPC interface https://review.openstack.org/168857 | 11:20 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Remove conductor 2.x RPC API https://review.openstack.org/168858 | 11:20 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Remove lazy-loading property compute_task_api from compute api https://review.openstack.org/168859 | 11:20 |
| openstackgerrit | vsaienko proposed openstack/nova: add ironic hypervisor type https://review.openstack.org/168820 | 11:22 |
| sdague | dims: you triaging bugs this morning as well? | 11:25 |
| dims | sdague: i just did one so far | 11:25 |
| *** markvoelker has joined #openstack-nova | 11:26 | |
| sdague | oh, ok, someone else definitely is doing it as well, as the counts keep jumping down every time I come back to the new list | 11:26 |
| dims | interesting | 11:26 |
| *** markus_z has quit IRC | 11:27 | |
| *** lucasagomes is now known as lucas-hungry | 11:28 | |
| *** markvoelker has quit IRC | 11:31 | |
| *** TobiasE1 has joined #openstack-nova | 11:34 | |
| *** TobiasE has quit IRC | 11:36 | |
| *** vmtrooper has joined #openstack-nova | 11:37 | |
| *** iamjarvo has quit IRC | 11:37 | |
| openstackgerrit | zhangtralon proposed openstack/nova: creating ordinary numa vm fails with large_page_vm https://review.openstack.org/167917 | 11:38 |
| *** Longgeek has quit IRC | 11:40 | |
| *** romainh has joined #openstack-nova | 11:41 | |
| *** otter768 has joined #openstack-nova | 11:41 | |
| *** vmtrooper has quit IRC | 11:42 | |
| openstackgerrit | Deepak C Shetty proposed openstack/nova: libvirt: Use 'relative' flag for online snapshot's commit/rebase operations https://review.openstack.org/168805 | 11:42 |
| *** annashen has joined #openstack-nova | 11:45 | |
| *** otter768 has quit IRC | 11:46 | |
| *** iamjarvo has joined #openstack-nova | 11:47 | |
| *** oro has quit IRC | 11:48 | |
| *** dave-mccowan has joined #openstack-nova | 11:48 | |
| *** vigneshvar has quit IRC | 11:49 | |
| *** annashen has quit IRC | 11:49 | |
| *** vigneshvar_ has quit IRC | 11:50 | |
| *** dave-mcc_ has quit IRC | 11:51 | |
| *** pradipta has quit IRC | 11:54 | |
| *** deepthi has quit IRC | 11:57 | |
| *** iamjarvo has quit IRC | 11:57 | |
| *** nkrinner has quit IRC | 12:01 | |
| *** pixelb has quit IRC | 12:01 | |
| *** ferest has joined #openstack-nova | 12:02 | |
| *** dulek_ has joined #openstack-nova | 12:03 | |
| *** raildo has joined #openstack-nova | 12:04 | |
| *** nkrinner has joined #openstack-nova | 12:05 | |
| *** dulek has quit IRC | 12:06 | |
| *** ferest has quit IRC | 12:06 | |
| *** jistr has quit IRC | 12:07 | |
| *** jistr has joined #openstack-nova | 12:08 | |
| sdague | alex_xu: any chance you could respin this patch with the requested fixes? | 12:08 |
| sdague | https://review.openstack.org/#/c/109201 | 12:08 |
| *** markvoelker has joined #openstack-nova | 12:10 | |
| *** dulek___ has joined #openstack-nova | 12:11 | |
| *** pixelb has joined #openstack-nova | 12:12 | |
| *** dulek_ has quit IRC | 12:14 | |
| *** tralon_ has quit IRC | 12:14 | |
| *** shaohe_feng has quit IRC | 12:14 | |
| *** ildikov has quit IRC | 12:17 | |
| *** pixelb has quit IRC | 12:17 | |
| *** dims has quit IRC | 12:17 | |
| *** dims has joined #openstack-nova | 12:18 | |
| *** krtaylor has quit IRC | 12:19 | |
| afazekas | sdague, ping | 12:21 |
| sdague | afazekas: pong | 12:21 |
| afazekas | I think creating spec kind bugs are ok | 12:22 |
| sdague | afazekas: I don't | 12:22 |
| sdague | here is the workflow for specs like bugs | 12:22 |
| afazekas | I am going to write a spec now for that, but some else who has time for that can do it based on the bug report | 12:22 |
| sdague | they get moved to Opinion / Wishlist, then ignored | 12:22 |
| sdague | and at some point we just invalidate them in bulk | 12:22 |
| afazekas | sdague: you aso made it invalid | 12:22 |
| sdague | yes | 12:23 |
| *** dulek___ is now known as dulek | 12:23 | |
| sdague | because from active contributors, I'm going to shortcut that | 12:23 |
| sdague | because they should know better | 12:23 |
| *** aysyd has joined #openstack-nova | 12:23 | |
| *** Maike has joined #openstack-nova | 12:23 | |
| sdague | the bug tracker should be used for bugs not features | 12:23 |
| sdague | especially when it's as big as it is | 12:24 |
| *** dave-mccowan has quit IRC | 12:24 | |
| afazekas | Long time ago I made mistake and created a blueprint for an issue, and then someone expected me to implement that | 12:24 |
| *** ferest has joined #openstack-nova | 12:24 | |
| afazekas | How to propose design changes when you are not going to implement them your self ? | 12:24 |
| *** pixelb has joined #openstack-nova | 12:25 | |
| openstackgerrit | sahid proposed openstack/nova: servicegroup: remove get_all method never used as public https://review.openstack.org/168876 | 12:25 |
| *** gtt116 has joined #openstack-nova | 12:26 | |
| afazekas | I also have a pending spec like document for effective scheduling, but it is definitely longer than a bug's scope | 12:26 |
| sdague | afazekas: so, honestly, wishlist features with no implementor really don't have much value, either in bugs or specs. | 12:26 |
| afazekas | sdague: There are contributors who are looking for new tasks, a confirmed wish list bug is searchable for them | 12:27 |
| *** salv-orlando has quit IRC | 12:27 | |
| sdague | a confirmed wishlist bug is often going to be rejected on code review for lacking a spec | 12:27 |
| *** dkliban_afk is now known as dkliban | 12:27 | |
| sdague | afazekas: there are already 1000 open bugs, I suggest that's a better place for people looking for something to do | 12:28 |
| afazekas | Do we have bugs for the spec repo ? :) | 12:28 |
| *** ferest has quit IRC | 12:28 | |
| *** salv-orlando has joined #openstack-nova | 12:30 | |
| *** smccully has quit IRC | 12:31 | |
| *** rfolco has joined #openstack-nova | 12:32 | |
| *** lucas-hungry is now known as lucasagomes | 12:32 | |
| *** nuritv has quit IRC | 12:36 | |
| *** Longgeek has joined #openstack-nova | 12:36 | |
| *** sandywalsh has quit IRC | 12:37 | |
| *** vladikr has joined #openstack-nova | 12:39 | |
| *** sandywalsh has joined #openstack-nova | 12:39 | |
| *** dave-mccowan has joined #openstack-nova | 12:41 | |
| *** baoli has joined #openstack-nova | 12:42 | |
| *** ildikov has joined #openstack-nova | 12:43 | |
| openstackgerrit | Moshe Levi proposed openstack/nova: Add option to ssh to prevent prompting https://review.openstack.org/168680 | 12:43 |
| *** pixelb has quit IRC | 12:45 | |
| *** lascii is now known as alaski | 12:52 | |
| *** suresh12 has joined #openstack-nova | 12:52 | |
| *** mwagner_lap has joined #openstack-nova | 12:53 | |
| *** dboik has quit IRC | 12:56 | |
| openstackgerrit | vsaienko proposed openstack/nova: add ironic hypervisor type https://review.openstack.org/168820 | 12:57 |
| *** shakamunyi has joined #openstack-nova | 12:57 | |
| *** eglynn has quit IRC | 12:59 | |
| *** iamjarvo has joined #openstack-nova | 12:59 | |
| afazekas | sdague, Do you mind if I change the status to opinion ? At least it does not means the bug is completely stupid. | 12:59 |
| *** krtaylor has joined #openstack-nova | 13:00 | |
| *** neelashah has joined #openstack-nova | 13:00 | |
| sdague | afazekas: sure, that's fine | 13:01 |
| sdague | but in future, please don't file things like this, because it's not helpful | 13:01 |
| afazekas | sdague: I see the ComputeFilter and AZ filter looks like has the 1 row response query in loop case, can it be considered as bug or it is design issue ? | 13:04 |
| *** veena_ has quit IRC | 13:05 | |
| *** atuvenie has quit IRC | 13:07 | |
| *** annegentle has joined #openstack-nova | 13:08 | |
| *** jecarey has joined #openstack-nova | 13:09 | |
| *** iamjarvo has quit IRC | 13:09 | |
| *** Longgeek has quit IRC | 13:10 | |
| *** haigang has joined #openstack-nova | 13:12 | |
| *** wverdugo500 has quit IRC | 13:12 | |
| *** shakamunyi has quit IRC | 13:13 | |
| *** wverdugo500 has joined #openstack-nova | 13:14 | |
| *** wverdugo500 has left #openstack-nova | 13:14 | |
| *** dboik has joined #openstack-nova | 13:16 | |
| *** whenry has joined #openstack-nova | 13:16 | |
| dims | sdague: anything i can help with? | 13:17 |
| *** dprince has joined #openstack-nova | 13:18 | |
| *** kaufer has joined #openstack-nova | 13:18 | |
| alex_xu | sdague: sure, no problem, does it hurry for today? if not, I will update it tomorrow | 13:20 |
| sdague | afazekas: I'm not sure what you believe the bug is? is it a micro optimization? | 13:20 |
| sdague | alex_xu: no prob | 13:20 |
| *** lpetrut has joined #openstack-nova | 13:20 | |
| garyk | sdague: dims: are patches with log messages and exception forbidden at the moment due to a string freeze? | 13:20 |
| garyk | maybe forbidden is too harsh.. | 13:20 |
| *** zz_johnthetubagu is now known as johnthetubaguy | 13:21 | |
| sdague | garyk: adding new strings means we should have a string freeze exception as part of it | 13:21 |
| *** TobiasE has joined #openstack-nova | 13:21 | |
| sdague | ttx: can maybe state what our approach should be there | 13:22 |
| dims | garyk: you may have seen this already... if not, https://wiki.openstack.org/wiki/StringFreeze | 13:22 |
| *** Marga_ has joined #openstack-nova | 13:22 | |
| *** thomasem has joined #openstack-nova | 13:23 | |
| *** thomasem has quit IRC | 13:24 | |
| *** TobiasE1 has quit IRC | 13:24 | |
| afazekas | sdague: If you have 10 hypervisor it is micro if you have 1000++, it is more | 13:24 |
| dims | boris-42: i see some chatter last night about slowness, did you get to the bottom of it? | 13:25 |
| *** sdake has joined #openstack-nova | 13:25 | |
| sdague | afazekas: yeh, so that's true with the entire scheduler filter architecture as it's a giant set of nested reduce loops | 13:25 |
| sdague | so I think a bug is useless, this really needs to be part of a spec to make the filter methodology more optimized | 13:25 |
| afazekas | sdague: bat at least later I can reference the solved bugs on the etherpad and in a spec | 13:26 |
| *** vmtrooper has joined #openstack-nova | 13:26 | |
| *** eglynn has joined #openstack-nova | 13:26 | |
| garyk | dims: thanks - guess that makes sense | 13:27 |
| sdague | just put that information in the spec | 13:27 |
| afazekas | sdague: A scheduler can do little smart change, without a huge architecture change (spec). For example query more once, and process them in the filter. Some kind of singletone query with a manual join | 13:27 |
| *** sdake_ has joined #openstack-nova | 13:28 | |
| sdague | when there are < 500 open bugs in nova, we can consider using it for tracking items like that. But until that time, bugs like this just reduce the chance that any bugs get fixed, because the bug pile is too high | 13:28 |
| boris-42 | dims: nope not yet | 13:28 |
| boris-42 | dims: I am not able to reproduce it | 13:28 |
| boris-42 | dims: except in our gates | 13:28 |
| *** hdd has joined #openstack-nova | 13:29 | |
| dims | boris-42: url? | 13:31 |
| *** thomasem has joined #openstack-nova | 13:31 | |
| *** vmtrooper has quit IRC | 13:31 | |
| *** sdake has quit IRC | 13:31 | |
| afazekas | sdague, I feel like the performance or scale related issues does not get enough attention, because of the small more simpler things. | 13:31 |
| sdague | afazekas: when the bug list is this large, nothing gets enough attension. What I'm saying is help reduce the existing backlog, and we can talk about new things like that. | 13:33 |
| *** exploreshaifali has joined #openstack-nova | 13:33 | |
| sdague | also, I don't think perf / scale issues are typically addressable in narrow bug scope, they mostly need a larger scale arch shift, like jaypipes work on the lockless db updates | 13:33 |
| *** johnthetubaguy is now known as zz_johnthetubagu | 13:34 | |
| sdague | if you want to bring up target performance things, I think mailing list is a better forum. So few people are actually looking at the bug list it's not really going to drive the conversation you want. | 13:34 |
| kashyap | Yeah, for intricate perf related topics, best discussed on the list. | 13:35 |
| *** VW_ has joined #openstack-nova | 13:36 | |
| bauwser | afazekas: as already explained somewhere in the ML, we know that there is some perf issues with the scheduler, but that's why we need to first reduce the tech debt | 13:37 |
| kashyap | Also, I notice a lot of bugs like "Bug, error, figure it out!" and never respond back when asked for details. | 13:37 |
| kashyap | Eventually getting closed as "Invalid" | 13:37 |
| bauwser | afazekas: there are some design issues that can't be resolved just because of a bug | 13:38 |
| *** hdd has quit IRC | 13:38 | |
| sdague | kashyap: yes, there are not enough folks helping drive the bug tracker into something useful | 13:39 |
| *** Nikolay_St has quit IRC | 13:40 | |
| *** rfolco has quit IRC | 13:40 | |
| *** Nikolay_St has joined #openstack-nova | 13:40 | |
| *** jlanoux has quit IRC | 13:40 | |
| *** eharney has joined #openstack-nova | 13:41 | |
| kashyap | sdague: Agreed, and I should admit - I'm guilty of it as much as anyone. | 13:41 |
| kashyap | sdague: Given that we're in RC-phase, maybe people will start looking at bugs. | 13:42 |
| *** otter768 has joined #openstack-nova | 13:42 | |
| *** rfolco has joined #openstack-nova | 13:42 | |
| *** mriedem has joined #openstack-nova | 13:42 | |
| *** markus_z has joined #openstack-nova | 13:43 | |
| *** ankit_ag has quit IRC | 13:43 | |
| afazekas | bauwser: For the scheduler I actually planning to write 2-3 spec, without assignee | 13:44 |
| sdague | kashyap: maybe, I don't seem to see a huge number of new bug fixes inbound though | 13:44 |
| boris-42 | dims: hm | 13:44 |
| *** bauwser is now known as bauzas | 13:45 | |
| boris-42 | dims: what urls ? | 13:45 |
| sdague | afazekas: so, honestly, before doing that, why don't you make in an ML thread? | 13:45 |
| boris-42 | dims: before/after issue? | 13:45 |
| afazekas | bauzas, : it involves secirtygroup driver change | 13:45 |
| *** jgrimm is now known as zz_jgrimm | 13:45 | |
| bauzas | afazekas: well, that's okay, but if you want to help us, it would be better to join us for our weekly meetings | 13:45 |
| bauzas | afazekas: just to be clear, we're not missing ideas, we're missing contributors :) | 13:46 |
| afazekas | sdague: I have 232 concpt text so far (it will be 500 line), I will create an etherpad and post it to the ML, is it ok ? | 13:46 |
| bauzas | afazekas: but sure, you can share your thoughts | 13:46 |
| kashyap | afazekas: Please, just write to the list so it's archived, etherpads are horribly slow and the resyncing is a world of pain. | 13:47 |
| *** rfolco has quit IRC | 13:47 | |
| *** otter768 has quit IRC | 13:47 | |
| *** annashen has joined #openstack-nova | 13:47 | |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Don't query compute_node through service object in nova-manage https://review.openstack.org/168907 | 13:47 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Mirror db schema changes to compute_node and service objects https://review.openstack.org/168908 | 13:47 |
| dims | boris-42: yes, one before and one afer? | 13:48 |
| afazekas | desing goal 100 sched/sec to >>100k host, without dropping cpython | 13:48 |
| *** rfolco has joined #openstack-nova | 13:48 | |
| afazekas | sched/sec can be significantly greater , it is based on your rules mainly | 13:49 |
| afazekas | bauzas: which meeting and when ? | 13:49 |
| *** sdake has joined #openstack-nova | 13:49 | |
| bauzas | afazekas: Gantt meeting Tuesdays 1600UTC | 13:50 |
| bauzas | I actually hate this being called a "gantt" meeting, because that's the nova sched subteam but that's it :( | 13:51 |
| *** vivekd has joined #openstack-nova | 13:51 | |
| bauzas | sounds like "nova scheduler subteam" is longer to type than "gantt" | 13:51 |
| afazekas | :) | 13:51 |
| boris-42 | dims: ok | 13:52 |
| *** annashen has quit IRC | 13:52 | |
| boris-42 | dims: before http://logs.openstack.org/42/160142/8/check/gate-rally-dsvm-rally/76fecde/rally-plot/results.html.gz#/NovaServers.boot_and_delete_server | 13:52 |
| bauzas | afazekas: https://wiki.openstack.org/wiki/Meetings#Gantt_.28Scheduler.29_team_meeting | 13:52 |
| boris-42 | after http://logs.openstack.org/36/168636/3/check/gate-rally-dsvm-rally/ce18c9f/rally-plot/results.html.gz#/NovaServers.boot_and_delete_server | 13:52 |
| dims | thanks boris-42 | 13:52 |
| mriedem | v2 API is basically feature frozen in liberty right? i.e. no new extensions? we just add new API microversions in v2.1 now. | 13:52 |
| boris-42 | dims: you know where are logs? | 13:52 |
| bauzas | afazekas: and oops, Tues 1500UTC | 13:52 |
| dims | boris-42: no | 13:52 |
| bauzas | afazekas: not 4pm | 13:53 |
| boris-42 | dims: so you need to remove from URL evertyhing starting from rally-plot | 13:53 |
| *** sdake_ has quit IRC | 13:53 | |
| boris-42 | dims: so like http://logs.openstack.org/36/168636/3/check/gate-rally-dsvm-rally/ce18c9f/ | 13:53 |
| *** tonytan4ever has joined #openstack-nova | 13:53 | |
| dims | boris-42: gotcha thanks | 13:53 |
| boris-42 | dims: after that logs from all services | 13:53 |
| *** xyang1 has joined #openstack-nova | 13:53 | |
| *** alexpilotti has joined #openstack-nova | 13:53 | |
| boris-42 | dims: and std dsvm job logs | 13:53 |
| superdan | mriedem: I believe so, yeah | 13:56 |
| *** superdan is now known as dansmith | 13:56 | |
| mriedem | cool, easy spec for a wishlist bug in liberty, i'd like to get my feet wet on a microversion | 13:56 |
| edleafe | dims: Here's a quick patch to review: https://review.openstack.org/#/c/165849 | 13:57 |
| *** atuvenie has joined #openstack-nova | 13:57 | |
| *** baoli has quit IRC | 13:59 | |
| openstackgerrit | Bartosz Fic proposed openstack/nova: Live migration failure in API doesn't revert task_state to None https://review.openstack.org/168916 | 13:59 |
| *** baoli_ has joined #openstack-nova | 14:00 | |
| *** haomaiwang has joined #openstack-nova | 14:01 | |
| *** gokrokve has joined #openstack-nova | 14:02 | |
| *** TobiasE1 has joined #openstack-nova | 14:03 | |
| *** TobiasE has quit IRC | 14:03 | |
| *** dboik_ has joined #openstack-nova | 14:03 | |
| sdague | bauzas: a request, please stop using 'gantt' in communications, as it doesn't mean anything to anyone but a small number of people | 14:03 |
| sdague | nova scheduler subteam is much better | 14:03 |
| bauzas | sdague: 100% agreed, that's confusing for a lot of people | 14:04 |
| afazekas | bauzas: thx | 14:04 |
| bauzas | sdague: I'll do a communication on the ML | 14:04 |
| dims | edleafe: those are database instances right? | 14:04 |
| *** rushil has joined #openstack-nova | 14:04 | |
| *** pixelb has joined #openstack-nova | 14:05 | |
| *** jyuso1 has quit IRC | 14:05 | |
| edleafe | dims: yes | 14:06 |
| *** TobiasE has joined #openstack-nova | 14:06 | |
| *** dboik has quit IRC | 14:06 | |
| edleafe | dims: a small optimization suggested by johnthetubaguy during a previous review | 14:06 |
| mriedem | ah, so when a thing is v2.4 in the API, it's v2.4 everywhere for all extensions, not just the extension that's changing | 14:07 |
| *** matrohon has joined #openstack-nova | 14:07 | |
| mriedem | like racing for db migration numbers | 14:07 |
| sdague | mriedem: https://bugs.launchpad.net/nova/+bug/1357437 | 14:07 |
| openstack | Launchpad bug 1357437 in OpenStack Compute (nova) "nova.tests.virt.libvirt.test_driver should use constants from fakelibvirt" [Low,Confirmed] | 14:07 |
| dims | edleafe: ack. thanks. left a tiny note | 14:07 |
| *** TobiasE1 has quit IRC | 14:08 | |
| mriedem | sdague: i think that's probably fixed by other stuff at this point | 14:08 |
| mriedem | i can dupe it | 14:08 |
| sdague | mriedem: can you look at that bug you filed and figure out where we stand? | 14:08 |
| sdague | mriedem: ok | 14:08 |
| sdague | mriedem: and, yes, API version is global | 14:08 |
| mriedem | k | 14:08 |
| edleafe | dims: I felt it was bad form to add my own patch to the trivial patch list :) | 14:08 |
| *** nelsnelson has joined #openstack-nova | 14:09 | |
| *** doude_ has quit IRC | 14:09 | |
| afazekas | bauzas: BTW do you know about published benchmark about the current scheduler's performance ? How many sched/sec it is able to do `officially` at scale ? | 14:09 |
| *** doude has joined #openstack-nova | 14:09 | |
| dims | edleafe: y, i leave mine out as well | 14:09 |
| sdague | mrda-away: which was intentional, and the code was structured such that competing versions would always merge conflict, like db migrations | 14:09 |
| sdague | mriedem: ^^^ | 14:09 |
| bauzas | afazekas: I heard some figures but not real benchs | 14:09 |
| mriedem | sdague: k, yeah, makes sense | 14:09 |
| mriedem | sdague: i guess for specs you just say the version will be whatever is available when it merges | 14:10 |
| openstackgerrit | Chris Friesen proposed openstack/nova: Fix "down" nova-compute service spuriously marked as "up" https://review.openstack.org/163060 | 14:10 |
| bauzas | afazekas: I was told that RAX ran some performance tests, but that's it | 14:10 |
| edleafe | bauzas: RAX uses the caching scheduler, too, so the numbers aren't representative | 14:11 |
| sdague | mriedem: yes | 14:11 |
| bauzas | afazekas: just to give raw numbers, I've been told that the scheduler is able to scale up to 1000 computrs | 14:11 |
| afazekas | bauzas: After Google I found benchmark around 10/sec at 340..1000 node | 14:11 |
| jogo | sdague: there is a governance patch to retire the gantt repo all together in fact | 14:12 |
| bauzas | afazekas: I know about RAX, but they wrote the caching scheduler because of the figures I mentioned :) | 14:12 |
| bauzas | jogo: could you please give me the change number ? | 14:12 |
| jogo | bauzas: https://review.openstack.org/#/c/168883/2 | 14:13 |
| bauzas | jogo: thanks | 14:13 |
| sdague | jogo: oh, now that you are here - https://bugs.launchpad.net/nova/+bug/1153827 - where does that stand? | 14:13 |
| openstack | Launchpad bug 1153827 in OpenStack Compute (nova) "soft delete values in InstanceSystemMetadata on instance delete " [Low,Triaged] | 14:13 |
| jogo | sdague: actually no idea what the status of that is | 14:14 |
| afazekas | bauzas: Is it still one worker process scheduler or is it able to use multiple workers ? | 14:15 |
| dansmith | sdague: jogo: is that still a bug? | 14:15 |
| jogo | skimming instance_destroy in the db api it looks like its still valid | 14:16 |
| jogo | this was part of the failed effort to stop reading deleted values | 14:16 |
| jogo | to make it easier to hard delete etc. | 14:16 |
| sdague | jogo: can you move it into confirmed then? | 14:17 |
| afazekas | bauzas: My pending document contains things related to multiple worker, and an option to cache flavor related data (I do not need more) | 14:17 |
| bauzas | afazekas: well, n-sch having multiple workers is doable but racy | 14:17 |
| sdague | mriedem: https://bugs.launchpad.net/nova/+bug/1349515 - where does that bug stand? it's one of the year old db2 bugs | 14:17 |
| openstack | Launchpad bug 1349515 in OpenStack Compute (nova) "Commit 0e98f5a breaks db migrations with DB2" [Low,Triaged] - Assigned to Matt Riedemann (mriedem) | 14:18 |
| *** jcoufal has joined #openstack-nova | 14:18 | |
| *** tonytan_brb has joined #openstack-nova | 14:18 | |
| garyk | sdague: the db patches are still in review (as far as i last saw) | 14:18 |
| dansmith | sdague: jogo: I can work on that, but I'm not entirely sure what the point is | 14:18 |
| *** mtanino has joined #openstack-nova | 14:18 | |
| *** whenry has quit IRC | 14:18 | |
| dansmith | sdague: jogo: deleted metadata items are showing up in notifications? | 14:18 |
| mriedem | sdague: i'll close it, it's handled in the db2 change with a backend specific check | 14:19 |
| mriedem | i.e. https://review.openstack.org/#/c/69047/45/nova/db/sqlalchemy/migrate_repo/versions/247_nullable_mismatch.py | 14:19 |
| jogo | dansmith: goal was to make it safe to remove soft deleted data from the db | 14:19 |
| *** suresh12 has quit IRC | 14:19 | |
| jogo | and to not have a huge list of 'undeleted' data even though its really deleted | 14:19 |
| jogo | orphaned InstanceSystemMetadata | 14:20 |
| sdague | dansmith: right, like jogo said. The previous attempt to do that caused stack traces because notification happens after the delete | 14:20 |
| dansmith | sdague: okay, just because we tried to read_deleted=yes a joined query with sysmeta? | 14:21 |
| afazekas | bauzas: I propose optimistic concurrency control, the min max value in the nova boot command will mean something.. | 14:21 |
| jogo | dansmith: see https://bugs.launchpad.net/nova/+bug/1185190 | 14:21 |
| openstack | Launchpad bug 1185190 in OpenStack Compute (nova) grizzly "instance_type data cannot be extracted from sys_meta after instance deleted" [Critical,Fix released] - Assigned to Chris Behrens (cbehrens) | 14:21 |
| *** tonytan4ever has quit IRC | 14:22 | |
| dansmith | jogo: ah, flavor shit | 14:22 |
| jogo | yeah, this stuff is all a rats nest | 14:23 |
| dansmith | jogo: well, sysmeta is mostly decoupled from flavor now, which makes it easier, | 14:23 |
| *** Marga_ has quit IRC | 14:23 | |
| jogo | although I think changes since should look at soft deleted data explicitly no? | 14:23 |
| dansmith | although for this release we kinda need to support both | 14:23 |
| jogo | dansmith: this but is by no means a priority | 14:24 |
| jogo | dansmith: so if punting it to Lemming makes it easier then lets just do that | 14:24 |
| dansmith | jogo: yeah,was just about to say it would be easier if we punt until we can drop some compat code | 14:24 |
| *** haomaiwang has quit IRC | 14:24 | |
| jogo | the impact is a slightly dirtier db | 14:24 |
| jogo | dansmith: cool can you add that comment to the bug | 14:25 |
| *** haomaiw__ has joined #openstack-nova | 14:25 | |
| dansmith | that reminds me that I need to check the release notes for migrating flavor stuff | 14:25 |
| dansmith | yep, I will | 14:25 |
| jogo | thanks | 14:25 |
| *** jlanoux has joined #openstack-nova | 14:26 | |
| *** toan-tran has joined #openstack-nova | 14:27 | |
| *** toan-tran has left #openstack-nova | 14:27 | |
| *** tralon has joined #openstack-nova | 14:28 | |
| *** tralon has quit IRC | 14:29 | |
| *** jlanoux_ has joined #openstack-nova | 14:31 | |
| *** pbrooko has joined #openstack-nova | 14:33 | |
| *** achanda has joined #openstack-nova | 14:33 | |
| *** jlanoux has quit IRC | 14:34 | |
| sdague | bauzas: is there a more appropriate bug for this one to duplicate into - https://bugs.launchpad.net/nova/+bug/1341420 | 14:34 |
| openstack | Launchpad bug 1341420 in OpenStack Compute (nova) "gap between scheduler selection and claim causes spurious failures when the instance is the last one to fit" [High,Triaged] | 14:34 |
| sdague | jogo: this is your bug, what do you want to do about it - https://bugs.launchpad.net/nova/+bug/1257788 ? it doesn't seem particularly actionable in current state | 14:35 |
| openstack | Launchpad bug 1257788 in OpenStack Compute (nova) "nova-network doesn't handle large amount of concurrent booting instances well" [Low,Triaged] | 14:35 |
| *** artom has joined #openstack-nova | 14:35 | |
| bauzas | sdague: erm, I think this one can't be resolved until we fix some other stuff which would probably require a spec | 14:35 |
| bauzas | sdague: we should be working on the resource claims by Liberty, which could address that bug | 14:36 |
| *** tonytan_brb is now known as tonytan4ever | 14:36 | |
| *** sseverson has quit IRC | 14:37 | |
| *** cbader has quit IRC | 14:37 | |
| *** asselin has quit IRC | 14:37 | |
| sdague | bauzas: ok, I'll move it out of triaged and into confirmed then | 14:37 |
| bauzas | sdague: ok | 14:37 |
| *** sseverson has joined #openstack-nova | 14:37 | |
| *** cbader has joined #openstack-nova | 14:37 | |
| *** asselin has joined #openstack-nova | 14:37 | |
| sdague | I wasn't sure if there were other issues being addressed here that would handle it | 14:38 |
| *** ildikov has quit IRC | 14:38 | |
| bauzas | sdague: none I'm thinking abouyt | 14:38 |
| *** cbader has quit IRC | 14:39 | |
| *** cbader has joined #openstack-nova | 14:39 | |
| bauzas | alaski: around ? | 14:40 |
| bauzas | alaski: I'm still struggled with the cells bug | 14:40 |
| *** asselin has quit IRC | 14:40 | |
| alaski | bauzas: yep | 14:40 |
| bauzas | alaski: ie. melwitt's -1 on https://review.openstack.org/#/c/157185/9 | 14:40 |
| *** asselin has joined #openstack-nova | 14:41 | |
| bauzas | alaski: as I'm fed up loosing time not testing it, I'm trying to deploy a cells-ish devstack | 14:41 |
| bauzas | alaski: luckily, there is https://etherpad.openstack.org/p/nova-cells-testing | 14:41 |
| *** asselin has quit IRC | 14:41 | |
| *** asselin_ has joined #openstack-nova | 14:41 | |
| sdague | garyk: is this going anywhere - https://bugs.launchpad.net/nova/+bug/1330873 ? | 14:41 |
| openstack | Launchpad bug 1330873 in OpenStack Compute (nova) "Instance name set in horizon is not set in VMware vCenter" [Medium,Confirmed] - Assigned to Jun Xie (junxiebj) | 14:41 |
| bauzas | alaski: just to be clear, a AIO devstack can still be cell-ish ? | 14:42 |
| alaski | bauzas: yes. it's just one parent and one child cell, but it works fine | 14:42 |
| sdague | bauzas: yes, see the localrc in the cells job | 14:42 |
| bauzas | alaski: yeah, and I guess that the dsvm job is doing that way | 14:42 |
| *** asselin_ has quit IRC | 14:42 | |
| garyk | sdague: i actually have a patch for that - it has been in review for a few days | 14:42 |
| *** patrickeast has joined #openstack-nova | 14:42 | |
| *** cbader has quit IRC | 14:42 | |
| openstackgerrit | Radoslav Gerganov proposed openstack/nova: VMware driver: Increasing speed of downloading image https://review.openstack.org/135625 | 14:42 |
| alaski | bauzas: yeah. I run it that way locally and it's great | 14:43 |
| sdague | garyk: ok, any idea why it didn't mark it in progress? | 14:43 |
| *** cbader has joined #openstack-nova | 14:43 | |
| *** asselin_ has joined #openstack-nova | 14:43 | |
| garyk | sdague: i was not aware of the bug :) | 14:43 |
| sdague | can you put the patch reference in there manually? | 14:43 |
| bauzas | alaski: sdague: okay thanks for the clarification, I'm just struggling with installing heat but that should be fixed | 14:43 |
| garyk | sdague: https://review.openstack.org/#/c/166608/ | 14:44 |
| *** asselin_ has quit IRC | 14:44 | |
| sdague | garyk: it was tagged vmware correctly it seems, who's responsible for keeping on top of vmware bugs? | 14:44 |
| *** achanda has quit IRC | 14:44 | |
| bauzas | sdague: btw. I'm really sorry but I had no time working on https://bugs.launchpad.net/nova/+bug/1431291 yet | 14:44 |
| openstack | Launchpad bug 1431291 in OpenStack Compute (nova) "Scheduler Failures are no longer logged with enough detail for a site admin to do problem determination" [High,Confirmed] - Assigned to Sylvain Bauza (sylvain-bauza) | 14:44 |
| *** asselin_ has joined #openstack-nova | 14:44 | |
| bauzas | sdague: I was really focused on fixing the cells bugs for RC1 | 14:44 |
| garyk | sdague: tracy and i. it was just tagged a very long time ago. | 14:44 |
| *** zz_jgrimm is now known as jgrimm | 14:44 | |
| sdague | bauzas: ok, can you un-assign yourself then? | 14:44 |
| bauzas | sdague: sure | 14:45 |
| bauzas | done | 14:45 |
| sdague | garyk: ok, so no one is doing full review of the tag on a regular basis? | 14:45 |
| *** cbader has quit IRC | 14:45 | |
| garyk | sdague: we are. | 14:45 |
| bauzas | sdague: I'll try to provide a patch by this week but RC1 is close | 14:45 |
| *** asselin_ has quit IRC | 14:45 | |
| *** cbader has joined #openstack-nova | 14:45 | |
| *** asselin_ has joined #openstack-nova | 14:45 | |
| garyk | sdague: high bugs do not get reviews so why even address low and mdium ones? | 14:45 |
| garyk | not to mention critical ones? | 14:46 |
| *** cbader has quit IRC | 14:46 | |
| *** sseverson has quit IRC | 14:46 | |
| *** tsekiyama has joined #openstack-nova | 14:46 | |
| *** asselin_ has quit IRC | 14:46 | |
| alaski | bauzas: it's most likely be turned into a dict there by _TargetedMessage.process() | 14:47 |
| *** asselin has joined #openstack-nova | 14:47 | |
| *** cbader has joined #openstack-nova | 14:47 | |
| *** sseverson has joined #openstack-nova | 14:47 | |
| bauzas | alaski: that's my thoughts | 14:47 |
| bauzas | alaski: but I looked at the code, and it's reconstructed | 14:47 |
| *** haomaiwa_ has joined #openstack-nova | 14:47 | |
| *** asselin has quit IRC | 14:47 | |
| openstackgerrit | garyk proposed openstack/nova: VMware: improve instance names on VC https://review.openstack.org/166608 | 14:47 |
| *** sseverson has quit IRC | 14:47 | |
| *** zul has quit IRC | 14:47 | |
| *** cbader has quit IRC | 14:47 | |
| sdague | garyk: my concern is that one of the reasons we have an out of control bug list is because of situations like this | 14:47 |
| bauzas | alaski: so I was about to introspect it | 14:47 |
| sdague | where there are bugs, and there are patches fixing those bugs, but they don't reference those | 14:48 |
| *** sseverson has joined #openstack-nova | 14:48 | |
| *** cbader has joined #openstack-nova | 14:48 | |
| garyk | sdague: i understand your concern. i will go over the list again and see what can and ha been addressed. | 14:48 |
| *** asselin has joined #openstack-nova | 14:48 | |
| sdague | garyk: thanks | 14:48 |
| *** sseverson has quit IRC | 14:48 | |
| *** cbader has quit IRC | 14:48 | |
| *** asselin has quit IRC | 14:49 | |
| *** cbader has joined #openstack-nova | 14:49 | |
| *** asselin_ has joined #openstack-nova | 14:49 | |
| *** sseverson has joined #openstack-nova | 14:49 | |
| *** armax has joined #openstack-nova | 14:49 | |
| *** haomaiw__ has quit IRC | 14:49 | |
| *** burt has joined #openstack-nova | 14:50 | |
| *** sseverson has quit IRC | 14:50 | |
| *** asselin_ has quit IRC | 14:50 | |
| *** cbader has quit IRC | 14:50 | |
| *** asselin_ has joined #openstack-nova | 14:50 | |
| *** sseverson has joined #openstack-nova | 14:50 | |
| *** cbader has joined #openstack-nova | 14:50 | |
| *** asselin_ has quit IRC | 14:51 | |
| *** sseverson has quit IRC | 14:51 | |
| *** cbader has quit IRC | 14:51 | |
| alaski | bauzas: http://git.openstack.org/cgit/openstack/nova/tree/nova/cells/messaging.py#n448 I don't think anything converts it after that point | 14:51 |
| *** cbader has joined #openstack-nova | 14:51 | |
| *** asselin_ has joined #openstack-nova | 14:51 | |
| *** sseverson has joined #openstack-nova | 14:51 | |
| *** hemnafk is now known as hemna | 14:52 | |
| bauzas | alaski: agreed, that's why I think it's just serialized over RPC and then reconstructed | 14:52 |
| *** sseverson has quit IRC | 14:52 | |
| *** cbader has quit IRC | 14:52 | |
| bauzas | alaski: if that's a NovaObject, that's reversible | 14:52 |
| *** asselin_ has quit IRC | 14:52 | |
| bauzas | alaski: and we know that it works with the cn_get() stuff AFAIK | 14:52 |
| *** sseverson has joined #openstack-nova | 14:52 | |
| *** cbader has joined #openstack-nova | 14:52 | |
| *** asselin_ has joined #openstack-nova | 14:52 | |
| *** thangp has joined #openstack-nova | 14:53 | |
| *** haigang has quit IRC | 14:53 | |
| *** zul has joined #openstack-nova | 14:53 | |
| jogo | sdague: https://bugs.launchpad.net/nova/+bug/1257788 has been fixed | 14:55 |
| openstack | Launchpad bug 1257788 in OpenStack Compute (nova) "nova-network doesn't handle large amount of concurrent booting instances well" [Low,Fix released] - Assigned to Dan Smith (danms) | 14:55 |
| *** Longgeek has joined #openstack-nova | 14:55 | |
| *** rajesht has quit IRC | 14:55 | |
| alaski | bauzas: gotcha. My test setup is running highy modified code atm or I would test it real quick | 14:56 |
| *** thedodd has joined #openstack-nova | 14:56 | |
| jogo | sdague: as for 1341420 not sure | 14:56 |
| *** wverdugo500 has joined #openstack-nova | 14:57 | |
| bauzas | alaski: no worries, I wanted to deploy a cell devstack anyway, so that's a good case | 14:57 |
| *** wverdugo500 has left #openstack-nova | 14:57 | |
| mriedem | hmm, this doesn't seem to work http://docs.openstack.org/developer/nova/devref/api_microversions.html#a-method-with-only-small-changes-between-versions | 14:57 |
| *** _gryf is now known as _gryf|afk | 14:59 | |
| sdague | dansmith: how related is this - https://bugs.launchpad.net/nova/+bug/1391816 to the pci bug you were tracking last week? | 15:00 |
| openstack | Launchpad bug 1391816 in OpenStack Compute (nova) "[pci-passthrough] nova-scheduler throws a 500 when PCI passthrough is used if the PCI passthrough scheduler has not been added" [Low,Triaged] | 15:00 |
| openstackgerrit | jichenjc proposed openstack/nova: Remove 2 class variables not used. https://review.openstack.org/168500 | 15:00 |
| openstackgerrit | jichenjc proposed openstack/nova: No need to create APIVersionRequest every time https://review.openstack.org/168501 | 15:00 |
| *** rwsu has joined #openstack-nova | 15:01 | |
| dansmith | sdague: not related, I don't think | 15:01 |
| sdague | ok | 15:01 |
| dansmith | sdague: related in sadness only.. | 15:01 |
| sdague | in the sadness of pci being a mess? :) | 15:01 |
| *** gtt116 has quit IRC | 15:02 | |
| dansmith | sdague: aye | 15:02 |
| openstackgerrit | jichenjc proposed openstack/nova: Remove unused class variables in extended_volumes https://review.openstack.org/168500 | 15:02 |
| openstackgerrit | jichenjc proposed openstack/nova: No need to create APIVersionRequest every time https://review.openstack.org/168501 | 15:02 |
| *** zul has quit IRC | 15:02 | |
| *** zul has joined #openstack-nova | 15:03 | |
| *** apuimedo has quit IRC | 15:03 | |
| *** iamjarvo has joined #openstack-nova | 15:04 | |
| jogo | jaypipes: https://review.openstack.org/#/c/168883/2 | 15:04 |
| *** dboik_ has quit IRC | 15:04 | |
| *** dboik has joined #openstack-nova | 15:04 | |
| sdague | dansmith: so is the crux of some of these issues that we're assuming pci is default enabled, but our scheduler filters require manual enablement? | 15:06 |
| dansmith | sdague: that may be the crux of that issue, but not the others I've seen | 15:07 |
| *** jlanoux_ has quit IRC | 15:07 | |
| dansmith | sdague: the others have all been "everything works until you enable PCI and then things break in unexpected ways" | 15:07 |
| sdague | it does seem like odd that pci code isn't behind a conditional given that the filter isn't in the default filter list | 15:07 |
| *** zul has quit IRC | 15:08 | |
| *** zul has joined #openstack-nova | 15:08 | |
| *** jlanoux has joined #openstack-nova | 15:08 | |
| *** smccully has joined #openstack-nova | 15:11 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Remove compute/api.py::update() https://review.openstack.org/138522 | 15:11 |
| *** nkrinner has quit IRC | 15:12 | |
| *** Longgeek has quit IRC | 15:13 | |
| *** vmtrooper has joined #openstack-nova | 15:15 | |
| *** tonytan4ever has quit IRC | 15:15 | |
| *** afazekas has quit IRC | 15:16 | |
| *** tonytan4ever has joined #openstack-nova | 15:16 | |
| *** zzzeek has joined #openstack-nova | 15:16 | |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 15:17 |
| *** TobiasE1 has joined #openstack-nova | 15:17 | |
| *** haomaiwa_ has quit IRC | 15:18 | |
| *** TobiasE has quit IRC | 15:18 | |
| *** baoli_ has quit IRC | 15:18 | |
| *** haomaiwa_ has joined #openstack-nova | 15:18 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: WIP: Avoid expensive loop when NBD module is absent https://review.openstack.org/168950 | 15:18 |
| *** annegentle has quit IRC | 15:19 | |
| bauzas | sdague: oh gosh, reading https://bugs.launchpad.net/nova/+bug/1391816 | 15:19 |
| openstack | Launchpad bug 1391816 in OpenStack Compute (nova) "[pci-passthrough] nova-scheduler throws a 500 when PCI passthrough is used if the PCI passthrough scheduler has not been added" [Low,Confirmed] | 15:19 |
| *** annegentle has joined #openstack-nova | 15:19 | |
| bauzas | sdague: IMHO, that's just a config problem, because the filter has to be enabled | 15:19 |
| sdague | bauzas: right, but it's not by default | 15:20 |
| sdague | and all the pci code is unconditional | 15:20 |
| dims | boris-42: can you please trigger a rally run that Depends-On https://review.openstack.org/168950 ? | 15:20 |
| boris-42 | dims: i can try | 15:20 |
| *** vmtrooper has quit IRC | 15:20 | |
| boris-42 | dims: just saw patch | 15:20 |
| boris-42 | dims: thanks=) | 15:20 |
| bauzas | sdague: that should not be | 15:20 |
| boris-42 | dims: I hope to get this some day merged https://review.openstack.org/#/c/147765/ | 15:21 |
| bauzas | sdague: PCI tracking of resources is optional IIRC | 15:21 |
| sdague | so you are opening up an interface, but exploding horribly if a user didn't also change a config file value, which there was no real warning that they needed to do | 15:21 |
| bauzas | sdague: just to be clear, that's a conditional in the RT AFAIK | 15:22 |
| *** baoli_ has joined #openstack-nova | 15:22 | |
| sdague | then why does it explode? | 15:22 |
| bauzas | sdague: https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L416-L430 | 15:22 |
| dims | boris-42: please add it to https://etherpad.openstack.org/p/liberty-nova-summit-ideas - maybe we can reach a consensus at the summit | 15:23 |
| garyk | boris-42: dims: rally causes ulcers! | 15:23 |
| *** dulek has quit IRC | 15:24 | |
| boris-42 | dims: I can add this image http://i.ytimg.com/vi/hHDnD62QWTU/maxresdefault.jpg =) | 15:24 |
| bauzas | sdague: oooooooh, found something really ugly | 15:24 |
| sdague | bauzas: apply_requests is getting called unconditionally | 15:24 |
| *** dsanders has joined #openstack-nova | 15:24 | |
| dims | boris-42: garyk: i remain an optimist | 15:24 |
| dims | :) | 15:24 |
| boris-42 | dims: https://i.imgflip.com/jimy3.jpg =) | 15:25 |
| *** shakamunyi has joined #openstack-nova | 15:25 | |
| bauzas | sdague: yeah, that's just something I saw | 15:25 |
| bauzas | sdague: that's even worse, we're adding anyway something in the resources dict at the virt layer, even if there is nothing to track | 15:26 |
| sdague | yep | 15:26 |
| bauzas | okay, I don't have the bandwitdth yet to address that bug but indeed you're right, that's a confirmed bug | 15:26 |
| *** sdake_ has joined #openstack-nova | 15:26 | |
| *** belmoreira has quit IRC | 15:27 | |
| bauzas | the resolution being to make it *really* optionnal | 15:27 |
| sdague | numa has the same issue IIRC. It actually feels like we should block the creation of flavors or booting of images with pci / numa attrs at the API level if the scheduler in the env isn't configured to support those features | 15:27 |
| *** browne has joined #openstack-nova | 15:27 | |
| dansmith | sdague: can't block creating flavors, IMHO | 15:27 |
| bauzas | sdague: that would require a filters discoverabily which I tend to disagree | 15:27 |
| dansmith | yeah | 15:27 |
| sdague | dansmith: why not? | 15:27 |
| dansmith | sdague: because the numa bits are just random extra_specs on the flavor | 15:28 |
| bauzas | sdague: in general, we assume that the operator is able to provide distributed configuration right | 15:28 |
| dansmith | having api know about that stuff is a major layering violation | 15:28 |
| sdague | even in the admin api? | 15:28 |
| bauzas | sdague: yeah | 15:28 |
| dansmith | the fact that that's how we expose numa knobs is a reasonable point to argue, but that's how it is right now | 15:28 |
| sdague | this seems like basic integrity constraints seems like we should address this somehow | 15:29 |
| bauzas | sdague: in general, we leave the operator provide a consistent configuration for both the API, the scheduler and the n-cpus | 15:29 |
| bauzas | here, the problem is that even the operator doesn't want to track PCI, we still provide PCI stats to the scheduler | 15:29 |
| sdague | that's terrible | 15:29 |
| dansmith | sdague: making the scheduler fail fast and explicitly would be the thing we should do, if that's what we want | 15:30 |
| *** dboik_ has joined #openstack-nova | 15:30 | |
| dansmith | sdague: like, if pci filter is not enabled and pci stuff is specified, fail fast with "PCI is not enabled and your flavor has $foo in it" | 15:30 |
| *** baoli_ has quit IRC | 15:30 | |
| *** sdake has quit IRC | 15:30 | |
| bauzas | sdague: what is terrible ? having a distributed config or something not optional ? | 15:30 |
| dansmith | not sure that the current filter organization makes that easy, but it would be reasonable to do | 15:30 |
| sdague | right, current filter stack doesn't make that easy | 15:30 |
| *** suresh12 has joined #openstack-nova | 15:30 | |
| bauzas | dansmith: yeah, we can bug that | 15:30 |
| *** baoli_ has joined #openstack-nova | 15:31 | |
| sdague | bauzas: because we're setting people up for failure. The filters are all optional, half of them are off by default. All these other attributes can be added with no warning that we're going to ignore them all. And hopefully you are then running your entire cluster a debug so you might ever figure out why it's not doing what you thought you told it to do. | 15:31 |
| bauzas | sdague: I was unclear then | 15:32 |
| bauzas | sdague: I'm not saying that we should prevent the operators to have better understanding of a failure | 15:32 |
| *** dboik has quit IRC | 15:32 | |
| bauzas | sdague: I'm just saying that making sure that n-cpu nova.conf and n-sch nova.conf are consistent is something mandatory for the operator | 15:33 |
| dansmith | see, | 15:33 |
| sdague | you firmly believe it should not be an error to create a pci requiring flavor in an OpenStack which can't support it? | 15:33 |
| dansmith | I don't like t hat | 15:33 |
| openstackgerrit | Alexander Schmidt proposed openstack/nova: Prevent access to image when filesystem resize is disabled https://review.openstack.org/168961 | 15:33 |
| dansmith | not because I think they shouldn't be, but because I think making the scheduler look at compute bits and compute look at scheduler bits is breaking another layer | 15:33 |
| *** tralon_ has joined #openstack-nova | 15:33 | |
| bauzas | dansmith: that's my point | 15:34 |
| dansmith | bauzas: you know I was addressing your "all configs should be the same" point right? | 15:34 |
| sdague | ok, but you agree the end behavior is incorrect | 15:34 |
| *** matrohon has quit IRC | 15:34 | |
| dansmith | sdague: I agree that trying to schedule something that is disabled is probably non-obvious | 15:34 |
| dansmith | sdague: how about this: | 15:34 |
| sdague | just a concern about how one might actually implement it with the existing architecture | 15:34 |
| *** suresh12 has quit IRC | 15:35 | |
| dansmith | sdague: what if we had a "newbie filter" that was enabled by default, that tried to fail fast based on config bits | 15:35 |
| dansmith | people could disable that filter if there was some reason | 15:35 |
| dansmith | but otherwise, it defaults to running first and doesn't actually filter, | 15:35 |
| dansmith | but just checks sanity | 15:35 |
| sdague | dansmith: that would be better, for sure | 15:36 |
| boris-42 | dims: I added to the etherpad | 15:36 |
| boris-42 | dims: plus https://review.openstack.org/#/c/168962/ here is Rally patch that depends on your patch | 15:36 |
| boris-42 | dims: I hope this will fix issue=) | 15:36 |
| sdague | but I still think long term we need to think about the consistency model here. Like if both computes and scheduler reported back capabilties into the service structure | 15:37 |
| sdague | so that at an API layer you could prevent the creation of flavors that can't be satisfied in the cluster | 15:37 |
| dansmith | sdague: the only way I'm on board with that is if these things become first class citizens of flavor | 15:38 |
| *** artom has quit IRC | 15:38 | |
| dansmith | because right now, they're magic metadata | 15:38 |
| dansmith | and, if we actually store and can compute usable capabilities | 15:38 |
| dansmith | doing it based on config is perilous because of upgrade situations as well, where we might have old API that doesn't think we should be able to do things that we actually can, | 15:39 |
| bauzas | sorry, on a call right now, so limited bandwitdth for discussing | 15:39 |
| dansmith | because everything behind us is plenty new | 15:39 |
| *** tralon_ has quit IRC | 15:40 | |
| *** dsanders has quit IRC | 15:40 | |
| dansmith | I should have used more periods above ^. I blame distraction from the same call as bauzas :) | 15:41 |
| openstackgerrit | Eric Brown proposed openstack/nova: Remove dead vmrc code https://review.openstack.org/168778 | 15:41 |
| dansmith | I think the summary is: there are some things we can do short-term to be better. The longer-term thing is complicated and requires architectural changes to be reasonable I think | 15:42 |
| openstackgerrit | Benjamin GRASSART proposed openstack/nova: Add vnc_keymap support for vmware compute https://review.openstack.org/167258 | 15:42 |
| bauzas | dansmith: +1 | 15:42 |
| *** dboik_ has quit IRC | 15:42 | |
| bauzas | dansmith: that comes up to better understanding of a scheduler failure right ? | 15:43 |
| *** dboik has joined #openstack-nova | 15:43 | |
| *** otter768 has joined #openstack-nova | 15:43 | |
| dansmith | bauzas: well, I think this is different than just that | 15:43 |
| dansmith | bauzas: sdague wants things to fail that currently don't fail in the scheduler | 15:43 |
| openstackgerrit | Matt Riedemann proposed openstack/nova: Show 'reserved' status in os-fixed-ips https://review.openstack.org/168966 | 15:43 |
| dansmith | well, he wants them to fail earlier than that, but... :) | 15:43 |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 15:44 |
| sdague | dansmith: right, I want things that *can't* ever work to not be allowed | 15:44 |
| bauzas | sdague: like you said creating a flavor using PCI metadata ? | 15:44 |
| sdague | but I agree that's a longer architectural concern | 15:44 |
| dansmith | sdague: right, and based on the current architecture, "not allowed" should be enforced at the scheduler; early and clearly | 15:44 |
| sdague | dansmith: right, which gets hung up on the fact that nothing really is clear out of the scheduler | 15:45 |
| dansmith | yeah | 15:45 |
| openstackgerrit | Matt Riedemann proposed openstack/nova: Show 'reserved' status in os-fixed-ips https://review.openstack.org/168966 | 15:45 |
| sdague | but we can poke at that one | 15:45 |
| *** haomaiwa_ has quit IRC | 15:46 | |
| *** zz_johnthetubagu is now known as johnthetubaguy | 15:46 | |
| *** otter768 has quit IRC | 15:48 | |
| *** tsekiyama has quit IRC | 15:49 | |
| *** tsekiyama has joined #openstack-nova | 15:49 | |
| mriedem | alex_xu: cyeoh: either of you around? | 15:50 |
| *** jcoufal has quit IRC | 15:51 | |
| mriedem | i guess in general, it looks like we don't have any json schema validation for os-fixed-ips in the v3 tree, is that an oversight? | 15:51 |
| sdague | dansmith: I'm assuming - https://bugs.launchpad.net/nova/+bug/1039065 remains unfixed right? | 15:51 |
| openstack | Launchpad bug 1039065 in OpenStack Compute (nova) "scheduler hints should persist with instance for use in migration, resize, and evacuate" [Wishlist,Confirmed] - Assigned to Yassine (yassine-lamgarchal) | 15:51 |
| sdague | mriedem: yes, missing json schema is just a bug afaik | 15:52 |
| mriedem | sdague: ok, reporting | 15:52 |
| *** shakamunyi has quit IRC | 15:52 | |
| *** iamjarvo has quit IRC | 15:52 | |
| mriedem | v | 15:54 |
| mriedem | https://bugs.launchpad.net/nova/+bug/1438286 | 15:54 |
| openstack | Launchpad bug 1438286 in OpenStack Compute (nova) "os-fixed-ips does not have json schema validation" [Medium,Triaged] - Assigned to Matt Riedemann (mriedem) | 15:54 |
| dansmith | sdague: I don't know | 15:56 |
| *** jcoufal has joined #openstack-nova | 15:56 | |
| *** jcoufal has quit IRC | 15:56 | |
| *** markus_z has quit IRC | 15:57 | |
| jogo | sdague: any nova specific check jobs we can drop? we are starting to get a lot | 15:58 |
| jogo | 14 jobs is a lot | 15:59 |
| sdague | I remain unconvinced that check-tempest-dsvm-ironic-pxe_ssh is useful | 15:59 |
| jogo | devananda: ^ | 16:00 |
| *** tonytan4ever has quit IRC | 16:00 | |
| sdague | I'd be pro dropping the pg job | 16:00 |
| jogo | sdague: I think we can drop neutron-large-ops to a periodic job that I can baby sit | 16:00 |
| jogo | sdague: me too, but that may be a political thing | 16:00 |
| *** gokrokve_ has joined #openstack-nova | 16:00 | |
| jogo | but fun fact http://superuser.openstack.org/articles/openstack-user-survey-insights-november-2014 | 16:00 |
| jogo | more people use devstack that postgres | 16:00 |
| sdague | also, if bauzas addresses his patch issues on cells, we should be able to turn cells tempest job on and drop the devstack one | 16:00 |
| devananda | I'm ++ on dropping the pg job | 16:01 |
| jogo | sdague: cool | 16:01 |
| jogo | devananda: the ironic-pxe_ssh job? | 16:01 |
| devananda | sdague: but i'm curious why you don't see value in the check-tempest-dsvm-ironic-pxe_ssh job | 16:01 |
| devananda | jogo: the postgres variant of it | 16:01 |
| sdague | devananda: because I don't believe that's testing anything useful from the nova side | 16:02 |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Allow specifying cpu model in flavor extra-specs or image metadata https://review.openstack.org/168982 | 16:02 |
| jogo | devananda: right we are talking about two separate things at once | 16:02 |
| *** gokrokv__ has joined #openstack-nova | 16:02 | |
| *** gokrokve_ has quit IRC | 16:02 | |
| *** yjiang5 has joined #openstack-nova | 16:03 | |
| *** yjiang5 has left #openstack-nova | 16:03 | |
| jogo | sdague: devananda FYI https://review.openstack.org/#/c/168486/ | 16:03 |
| jogo | testing the waters with that one | 16:03 |
| jogo | but happy to do that everywhere if there is enough support | 16:03 |
| *** artom has joined #openstack-nova | 16:03 | |
| *** gokrokve has quit IRC | 16:04 | |
| *** TobiasE has joined #openstack-nova | 16:04 | |
| *** TobiasE1 has quit IRC | 16:05 | |
| *** annashen has joined #openstack-nova | 16:05 | |
| *** Marga_ has joined #openstack-nova | 16:06 | |
| *** ndipanov has quit IRC | 16:06 | |
| *** annegentle has quit IRC | 16:07 | |
| *** unicell has quit IRC | 16:07 | |
| *** Maike has quit IRC | 16:08 | |
| mriedem | sdague: how are json schemas handled in tempest with microversion changes? i'm assuming when tempest tests v2.1 it's testing the latest version? | 16:09 |
| mriedem | so anything added in a response after v2.1 is optional in the response for tempest? | 16:09 |
| sdague | mriedem: no, it's testing v2.1 | 16:09 |
| sdague | I don't think there are yet tempest tests for microversion adds | 16:09 |
| mriedem | sdague: ok, so if we wanted to test v2.3 we'd have to change something? | 16:10 |
| mriedem | ok | 16:10 |
| *** shakamunyi has joined #openstack-nova | 16:10 | |
| sdague | yeh, you'd have to add microversion header support to the tempest client, which actually should be pretty easy | 16:10 |
| *** thedodd has quit IRC | 16:10 | |
| devananda | jogo: so I was referring to the tempest-dsvm-ironic-pxe_ssh-postgres jobs which appear to only run on Ironic anyway | 16:10 |
| *** veena has joined #openstack-nova | 16:10 | |
| sdague | devananda: right, we're not talking about that job at all :) | 16:11 |
| devananda | sdague: gotcha. so re the -ironic-pxe_ssh job, it's testing the nova.virt.ironic driver. how is that not useful? | 16:12 |
| *** pixelb has quit IRC | 16:12 | |
| sdague | because that should not require a tempest job to be tested | 16:13 |
| devananda | what should it require? | 16:13 |
| *** garyk has quit IRC | 16:13 | |
| sdague | unit tests should be sufficient | 16:14 |
| dansmith | wait, what? | 16:14 |
| dansmith | we ask all the other drivers to have tempest CI, right? | 16:14 |
| jaypipes | jogo: I'll prolly just defer to jeblair and fungi's knowledge there w.r.t. how to handle this... | 16:16 |
| devananda | sdague: there are a bucketload of interactions between scheduler, conductor, compute, and then the Ironic service itself ... and issues inbetween these have been caught by that test plenty of times. it's definitely a test I want to keep around. | 16:17 |
| sdague | dansmith: I guess in that context, however, this is a *really* weird an synthetic iteration of ironic | 16:17 |
| jogo | jaypipes: well the part where I am interested in your thoughts are what is the plan going forward | 16:17 |
| dansmith | sdague: synthetic is what way? because they're VMs? | 16:17 |
| sdague | dansmith: yes | 16:17 |
| dansmith | sdague: but they're PXEing, right? | 16:18 |
| devananda | sdague: the only part that's synthetic is the power driver inside ironic | 16:18 |
| jogo | jaypipes: when you expect the split (if we are even doing a split). do you want to keep the name gantt etc. | 16:18 |
| sdague | right, but the pxe mechanism is on the ironic side, there is no nova code for that, right? | 16:18 |
| jaypipes | jogo: well, it's tough to say. the end goal is to have a functional gantt project by mid-Liberty I believe. whether or not to delete the original repo or not is more of an infra question, no? | 16:18 |
| *** sahid has quit IRC | 16:18 | |
| dansmith | sdague: still, it seems pretty valuable to me. The power driver aside. | 16:18 |
| jogo | jaypipes: well do you want the same name I mean? | 16:19 |
| jogo | gantt vs ??? | 16:19 |
| devananda | sdague: that test is testing the nova<->ironic integration AND all the ironic plumbing aside from the specific power driver | 16:19 |
| *** armax has quit IRC | 16:19 | |
| devananda | sdague: I'm all for making it better, but it's the most important test we have right now | 16:19 |
| sdague | ok, if someone shows me patches that were blocked by that job in nova for real issues being introduced, I'll change my opinion on it | 16:19 |
| jaypipes | jogo: I honestly don't care :) whatever is easier and makes sense for infra is fine by me. honestly, I have trouble seeing that the name Gantt would *not* be legally problematic. | 16:20 |
| devananda | sdague: if you really want I can go look for scheduler changes that broke us before that job was voting | 16:20 |
| devananda | at the end of the Juno cycle | 16:20 |
| dansmith | yeah | 16:20 |
| devananda | it was exceptionaly frustrating | 16:20 |
| dansmith | that was going to be my first example | 16:20 |
| devananda | to everyone on the ironic team | 16:20 |
| sdague | devananda: we've been through that one before, and every instance you pointed at wasn't valid any more | 16:20 |
| devananda | (probably to the nova folks too) | 16:20 |
| *** Mike_D_laptop has joined #openstack-nova | 16:21 | |
| jogo | jaypipes: well just leave your thoughts on the patch in question and infra can figure out what makes the most sense once they have all the information | 16:21 |
| jaypipes | will do right now. | 16:21 |
| jaypipes | thx jogo | 16:21 |
| jogo | jaypipes: thank you! | 16:21 |
| sdague | we put the contract interface in place, and that addressed it | 16:21 |
| devananda | sdague: so the point of having cross project integration tests is specifically, in my opinion, to guard us against introducing changes that break the way projects rely on unstable APIs between each other | 16:21 |
| devananda | sdague: and nova has repeatedly said the driver API is unstable and not something we can rely on | 16:21 |
| sdague | except the driver isn't in ironic | 16:22 |
| devananda | sdague: but that test, in nova, is testing the integration between nova and that driver | 16:22 |
| sdague | so I'm not sure how that's an issue any more | 16:22 |
| bauzas | alaski: bingo http://paste.openstack.org/show/197591/ | 16:22 |
| jaypipes | jogo: dne | 16:22 |
| jaypipes | done even. | 16:22 |
| devananda | removing that test means nova could delete the whole driver and there'd be no issue. or make any breaking change to it | 16:22 |
| dansmith | sdague: what contract? | 16:22 |
| devananda | sdague: i'm really confused now | 16:23 |
| sdague | so, in early juno this was breaking a lot | 16:23 |
| bauzas | alaski: sounds like the object was primitived but not correctlyu | 16:23 |
| devananda | sdague: we removed that contract code because a) it didn't actually work, b) we added this job | 16:23 |
| devananda | (or I thought we removed it) | 16:23 |
| * devananda checks | 16:23 | |
| bauzas | alaski: because we lost the version and metadata fields on the primitive | 16:23 |
| sdague | devananda: no, you removed the contract code because you moved the driver into the nova tree | 16:24 |
| sdague | so there was no longer a point for it | 16:24 |
| sdague | because now changes can be picked up through in tree testing | 16:24 |
| devananda | sdague: does nova test integration with other services, eg. swift or cinder? | 16:25 |
| dansmith | sdague: I really don't see how this is any different than xenapi or vmware CI | 16:25 |
| dansmith | alaski: 13 failures on the cells test on that updated update (heh) patch. That seems right to me | 16:25 |
| devananda | dansmith: it's different in that openstack-infra runs it. aside from that, it's also different because this is anothe ropenstack service that we're integrating with | 16:26 |
| sdague | dansmith: maybe it's not, I think the current approach is pretty low use in finding actual issues however. | 16:26 |
| devananda | but I need to run | 16:26 |
| *** vivekd has quit IRC | 16:26 | |
| dansmith | devananda: those seem to be arguments to strengthen the case, not weaken it :) | 16:26 |
| veena | Hi, I need to understand the nova code base and would like to trace the code based on different funtionalities. Please suggest how do I start | 16:27 |
| bauzas | veena: start at the API level | 16:28 |
| bauzas | veena: and look at how nova is booting an instance from the API level to the virt level | 16:28 |
| dansmith | sdague: well, I disagree. I probably can't argue based on actual test results, but the things being tested seem worthwhile to me | 16:29 |
| sdague | dansmith: more useful than multinode? | 16:29 |
| dansmith | sdague: are we going to say we can't have multinode if we keep ironic testing? | 16:29 |
| *** baoli_ has quit IRC | 16:29 | |
| sdague | the reason this came up at all is that jogo is concerned about proposing multinode into the check queue because node utilization in the system is too high | 16:30 |
| sdague | so I'm pointing at what I consider the least useful jobs we have | 16:30 |
| dansmith | I have to jump on a cal | 16:30 |
| dansmith | but will be glad to continue in a bit | 16:30 |
| sdague | but, yes, we're going to have to make priority calls in the near term if we want multinode running in the gate, due to resource constraints | 16:31 |
| bauzas | veena: like begin with nova.api.openstack.compute.servers module and look at the create() method corresponding to the REST PUT for booting a new instance | 16:31 |
| alaski | dansmith: looking. I reviewed the cells code as well and didn't see any hidden use of that method | 16:31 |
| veena | bauzas, Thanks. would like to understand the infra of the code first, like there's manager.py as an entry point for every nova process. | 16:32 |
| *** Nic has joined #openstack-nova | 16:32 | |
| veena | bauzas, would like to have more insight on the driver.py files. But i get lost in the middle as the code base is huge :) | 16:32 |
| *** aysyd has quit IRC | 16:32 | |
| bauzas | veena: then docs.openstack.org/developer/nova/devref/ can help but some design pages are outdated | 16:32 |
| *** cfriesen has joined #openstack-nova | 16:33 | |
| cfriesen | looking for some help with nova-specs...I'm writing a new spec for liberty. The README.rst file says to run "tox" to validate that the specification is syntactically correct. When I run "tox" it appears to complete and it generates a bunch of info for the juno/kilo specs, but doesn't seem to pick up my new liberty spec. | 16:33 |
| *** oro has joined #openstack-nova | 16:33 | |
| *** marun has joined #openstack-nova | 16:33 | |
| veena | bauzas, okay. will check the link | 16:34 |
| sdague | cfriesen: can you push your patch ? | 16:34 |
| *** suresh12 has joined #openstack-nova | 16:34 | |
| *** emagana has joined #openstack-nova | 16:34 | |
| *** aysyd has joined #openstack-nova | 16:35 | |
| *** oro has quit IRC | 16:36 | |
| *** oro has joined #openstack-nova | 16:36 | |
| *** rushil has quit IRC | 16:37 | |
| alaski | bauzas: good catch, re http://paste.openstack.org/show/197591/ . Have you found where it's happening yet? | 16:38 |
| *** suresh12 has quit IRC | 16:38 | |
| *** baoli has joined #openstack-nova | 16:38 | |
| *** unicell has joined #openstack-nova | 16:39 | |
| cfriesen | sdague: pushed....https://review.openstack.org/#/c/168982/ | 16:39 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: WIP: Avoid expensive loop when NBD module is absent https://review.openstack.org/168950 | 16:39 |
| bauzas | alaski: still chasing but babysitting time | 16:39 |
| bauzas | :) | 16:39 |
| cfriesen | sdague: problems seems to be related to the fact that there is no symlink under doc/source/specs/liberty/ | 16:40 |
| bauzas | alaski: will further look at it later this evening | 16:40 |
| alaski | bauzas: ok. I'll poke around too when I find a moment | 16:40 |
| bauzas | alaski: no worries, it's pretty easy to follow the down path | 16:40 |
| *** apuimedo has joined #openstack-nova | 16:40 | |
| bauzas | alaski: I'm already set up, that's just matter of finding time :) | 16:41 |
| openstackgerrit | Matt Riedemann proposed openstack/nova-specs: Show 'reserved' status in os-fixed-ips API https://review.openstack.org/168996 | 16:41 |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Consolidate code to get the correct availability zone of an instance https://review.openstack.org/149562 | 16:43 |
| openstackgerrit | Sean Dague proposed openstack/nova-specs: add approved symlink to allow liberty specs to be tested https://review.openstack.org/169000 | 16:43 |
| sdague | dansmith / jogo: fast land on ^^^ is probably handy | 16:44 |
| *** pbrooko has quit IRC | 16:44 | |
| openstackgerrit | Hans Lindgren proposed openstack/nova: Fix the inconsistent info of availability zone https://review.openstack.org/159730 | 16:44 |
| *** moshele has quit IRC | 16:44 | |
| *** yamahata has joined #openstack-nova | 16:44 | |
| *** jistr has quit IRC | 16:44 | |
| *** Nic has quit IRC | 16:46 | |
| *** pbrooko has joined #openstack-nova | 16:46 | |
| *** pixelb has joined #openstack-nova | 16:47 | |
| *** ecosta has joined #openstack-nova | 16:48 | |
| mriedem | sdague: was just testing docs build with specs, let me try that quick | 16:48 |
| *** alanf-mc has joined #openstack-nova | 16:48 | |
| openstackgerrit | Marian Horban proposed openstack/nova: WIP: Graceful shutdown WSGI server https://review.openstack.org/129167 | 16:48 |
| *** jaypipes has quit IRC | 16:49 | |
| *** rushil has joined #openstack-nova | 16:50 | |
| mriedem | sdague: so specs should be put under specs/liberty/approved? | 16:51 |
| sdague | yeh | 16:51 |
| mriedem | ok, wasn't clear in the ML (actually said the opposite i think) | 16:51 |
| sdague | mriedem: reference? | 16:51 |
| mriedem | sdague: http://lists.openstack.org/pipermail/openstack-dev/2015-March/058107.html | 16:52 |
| *** pbrooko has quit IRC | 16:52 | |
| *** rushil has quit IRC | 16:52 | |
| mriedem | 2-3 threads in | 16:52 |
| sdague | this is how it was always done in the past | 16:52 |
| *** ildikov has joined #openstack-nova | 16:52 | |
| sdague | Copy your spec from the specs/<oldrelease>/approved directory to | 16:52 |
| sdague | the specs/liberty/approved directory. Note that if we declared your | 16:52 |
| sdague | spec to be a "partial" implementation in Kilo, it might be in the | 16:52 |
| sdague | implemented directory. This was rare however. | 16:52 |
| mriedem | yeah, i know, hence the confusion | 16:53 |
| sdague | I think there was a typo at the end | 16:53 |
| sdague | For other proposals, the process is the same as Juno and Kilo... | 16:53 |
| sdague | Propose a spec review against the specs/kilo/approved directory and | 16:53 |
| sdague | we'll review it from there. | 16:53 |
| sdague | s/kilo/liberty/ | 16:53 |
| sdague | unless you are reading something else that I'm missing | 16:53 |
| *** Marga_ has quit IRC | 16:54 | |
| *** Marga_ has joined #openstack-nova | 16:54 | |
| *** romainh has left #openstack-nova | 16:56 | |
| *** derekh has quit IRC | 16:57 | |
| cfriesen | sdague: my interpretation was that new specs should be put directly under "liberty". From the README.rst file: " | 16:58 |
| cfriesen | Specifications are proposed for a given release by adding them to the | 16:58 |
| cfriesen | ``specs/<release>`` directory and posting it for review. | 16:58 |
| *** erlon has joined #openstack-nova | 16:59 | |
| *** ecosta has left #openstack-nova | 16:59 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova-specs: Show 'reserved' status in os-fixed-ips API https://review.openstack.org/168996 | 16:59 |
| sdague | cfriesen: that language goes back pre juno | 17:00 |
| sdague | so I'm pretty sure the approved directory is still the right place for things to go | 17:00 |
| *** lucasagomes is now known as lucas-dinner | 17:00 | |
| sdague | it could probably be cleaned up, I'm not a nova-driver so I don't know for sure | 17:00 |
| *** apuimedo has quit IRC | 17:01 | |
| cfriesen | according to the email mriedem linked, only previously-approved specs go in "approved", for new ones the process is the same as before (which I take to mean that they go directly under "liberty") | 17:01 |
| cfriesen | whoops, though it does say in the email to go under approved. Missed that part. | 17:01 |
| cfriesen | okay, I'll move mine | 17:01 |
| *** vmtrooper has joined #openstack-nova | 17:03 | |
| *** adalbas has joined #openstack-nova | 17:04 | |
| *** mpavone has quit IRC | 17:04 | |
| *** suresh12 has joined #openstack-nova | 17:05 | |
| *** Marga_ has quit IRC | 17:05 | |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Fix up instructions for proposing spec for approval https://review.openstack.org/169007 | 17:06 |
| *** hdd has joined #openstack-nova | 17:06 | |
| sdague | dims: in https://bugs.launchpad.net/nova/+bug/1438260 - why is rally using nbd at all? Are it's systems just misconfigured? | 17:08 |
| openstack | Launchpad bug 1438260 in OpenStack Compute (nova) "Expensive Device allocation loop when NBD is absent" [Undecided,In progress] - Assigned to Davanum Srinivas (DIMS) (dims-v) | 17:08 |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Allow specifying cpu model in flavor extra-specs or image metadata https://review.openstack.org/168982 | 17:08 |
| bauzas | alaski: okay, a few findings show me that there is a problem with the messaging service providing a serialized dict while we wait for a serialized object | 17:08 |
| *** vmtrooper has quit IRC | 17:08 | |
| dansmith | this removes dead code: https://review.openstack.org/#/c/138522/ | 17:09 |
| dansmith | and makes tests not lie | 17:09 |
| *** dsanders has joined #openstack-nova | 17:10 | |
| sdague | dansmith: nice | 17:11 |
| *** dsanders has quit IRC | 17:11 | |
| dansmith | sdague: so going back to the other thing, | 17:11 |
| *** dsanders has joined #openstack-nova | 17:11 | |
| dansmith | sdague: if ironic precludes multinode then certainly I choose multinode, but I think that to be fair and covered, we need to have someone running third party against ironic just like any other thing | 17:11 |
| dansmith | sdague: given that ironic is part of openstack and not just any other hypervisor, I'd be really sad to see us lose the ability to actually gate on it | 17:12 |
| *** Nic has joined #openstack-nova | 17:12 | |
| dansmith | given that I don't think it's unusefully synthetic | 17:12 |
| openstackgerrit | Chris Friesen proposed openstack/nova: Conditionally expose "last_seen_up" in service list https://review.openstack.org/168418 | 17:13 |
| mriedem | when did nova core also become nova-specs core? | 17:13 |
| sdague | mriedem: oh, no idea, I assumed it wasn't | 17:14 |
| *** atuvenie has quit IRC | 17:14 | |
| mriedem | i'm going to so abuse nova-specs | 17:14 |
| *** armax has joined #openstack-nova | 17:16 | |
| *** rushil has joined #openstack-nova | 17:16 | |
| alaski | bauzas: interesting. do you know where it's being serialized? | 17:17 |
| *** oro has quit IRC | 17:18 | |
| *** yamahata has quit IRC | 17:18 | |
| sdague | mriedem: look, it even caught an issue in your spec :) | 17:18 |
| sdague | dansmith: you want to approve - https://review.openstack.org/#/c/169000/ ? | 17:19 |
| bauzas | alaski: still chasing | 17:19 |
| bauzas | alaski: on the child cell probably | 17:20 |
| bauzas | alaski: because it's badly serialized | 17:20 |
| *** jgrimm is now known as zz_jgrimm | 17:22 | |
| alaski | bauzas: okay, I'm wondering if it's something explicitly done, or the messaging layer doesn't handle objects real well when passing up | 17:23 |
| *** achanda has joined #openstack-nova | 17:24 | |
| *** melwitt has joined #openstack-nova | 17:25 | |
| *** harlowja_away is now known as harlowja | 17:25 | |
| *** gokrokve has joined #openstack-nova | 17:25 | |
| *** iamjarvo has joined #openstack-nova | 17:26 | |
| openstackgerrit | Andrew Laski proposed openstack/nova: Pass instance object to add_instance_fault_from_exc https://review.openstack.org/169014 | 17:26 |
| alaski | ^^ small fix for error reporting during scheduling failures | 17:27 |
| *** atuvenie has joined #openstack-nova | 17:28 | |
| boris-42 | dims: http://logs.openstack.org/62/168962/1/check/gate-rally-dsvm-rally/2e213c7/rally-plot/results.html.gz#/NovaServers.boot_and_delete_server | 17:28 |
| *** Maike has joined #openstack-nova | 17:28 | |
| *** gokrokv__ has quit IRC | 17:28 | |
| boris-42 | dims: your fix works | 17:29 |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix tenant_network test https://review.openstack.org/169016 | 17:30 |
| *** gokrokve has quit IRC | 17:30 | |
| dims | boris-42: yay | 17:30 |
| boris-42 | dims: you my hero | 17:30 |
| dims | boris-42: don't know why this started showing up though... | 17:31 |
| dims | :) | 17:31 |
| *** igordcard has quit IRC | 17:32 | |
| *** igordcard has joined #openstack-nova | 17:32 | |
| cfriesen | what happens when a bugfix requires an API change...does it turn into a feature that needs a blueprint/spec? | 17:33 |
| mriedem | sdague: yeah, would have caught that locally but docs won't build for me locally | 17:33 |
| mriedem | bbiab | 17:34 |
| *** mriedem has quit IRC | 17:34 | |
| sdague | boris-42: so... why are the rally jobs using nbd at all? | 17:34 |
| sdague | that's the confusing part | 17:34 |
| boris-42 | sdague: I don't know * what ht hell is that | 17:35 |
| boris-42 | sdague: maybe somebody changed something in infra related to rally jobs? | 17:35 |
| *** dboik has quit IRC | 17:35 | |
| dansmith | sdague: I'm not sure that's right, so I'll leave it for jogo | 17:35 |
| sdague | so it's the really old legacy way of injecting content into guest disks | 17:35 |
| *** dboik has joined #openstack-nova | 17:35 | |
| *** pixelb has quit IRC | 17:36 | |
| *** tonytan4ever has joined #openstack-nova | 17:36 | |
| *** amotoki has quit IRC | 17:36 | |
| *** eglynn has quit IRC | 17:37 | |
| boris-42 | sdague: very very strange | 17:37 |
| boris-42 | sdague: sooo https://github.com/openstack-infra/project-config/blob/master/jenkins/jobs/rally.yaml#L45-L65 | 17:37 |
| *** annegentle has joined #openstack-nova | 17:37 | |
| boris-42 | sdague: I don't see anything strange here in dvsm job | 17:38 |
| boris-42 | sdague: you? | 17:38 |
| sdague | nope, but you saw my logstash query right? | 17:38 |
| sdague | this seems to only be in rally and cells jobs | 17:38 |
| *** yamahata has joined #openstack-nova | 17:38 | |
| sdague | and started 3 days ago | 17:39 |
| boris-42 | sdague: yep | 17:39 |
| sdague | so it would be good to know what's different here | 17:39 |
| boris-42 | sdague: and 3 days ago we got this issue with performance | 17:39 |
| boris-42 | sdague: we didn't merge too much https://github.com/stackforge/rally/commits/master for 3 days | 17:39 |
| *** sbfox has joined #openstack-nova | 17:39 | |
| boris-42 | sdague: just few small patches | 17:39 |
| *** promulo has joined #openstack-nova | 17:40 | |
| boris-42 | sdague: maybe something in devstack? | 17:40 |
| *** ZZelle_ has joined #openstack-nova | 17:41 | |
| sdague | yeh, looking there as well | 17:41 |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix config option types https://review.openstack.org/168481 | 17:41 |
| sdague | boris-42: these are run as post jobs? | 17:42 |
| *** jaypipes has joined #openstack-nova | 17:42 | |
| *** annegentle has quit IRC | 17:43 | |
| dansmith | alaski: on that notification patch, | 17:43 |
| dansmith | alaski: I got burned on a thing with notifications triggering lazy loads of the object you pass it | 17:43 |
| dansmith | alaski: I wonder if we should neuter the object during that call to prevent that? | 17:43 |
| dansmith | alaski: ...since we just constructed it without much inside | 17:43 |
| *** otter768 has joined #openstack-nova | 17:44 | |
| *** annegentle has joined #openstack-nova | 17:44 | |
| alaski | dansmith: yeah, that would work | 17:44 |
| alaski | really it only needs the instance uuid, so I could update the api but that has a larger surface area | 17:44 |
| alaski | er, larger amount of required changes | 17:45 |
| boris-42 | sdague: nope regular checks and gates pipelines | 17:45 |
| dansmith | alaski: commented | 17:45 |
| dansmith | alaski: hmm, you know that nothing in the notification path hits more? | 17:46 |
| dansmith | alaski: when I hit it, it was pretty deep in there, inside "add instance info" or some such | 17:46 |
| *** annashen has quit IRC | 17:46 | |
| alaski | dansmith: nothing in add_instance_fault_from_exc needs it | 17:46 |
| dansmith | alaski: oh, I just misread the line | 17:47 |
| dansmith | I thought it was modifying the send_update() contents | 17:47 |
| dansmith | which is already getting the object | 17:47 |
| *** annashen has joined #openstack-nova | 17:47 | |
| alaski | gotcha. I would be wary of that too | 17:47 |
| dansmith | alaski: I still have flavor scars. | 17:48 |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Allow explicit numa pinning in flavor extra specs or image metadata https://review.openstack.org/169026 | 17:48 |
| *** otter768 has quit IRC | 17:48 | |
| *** Sukhdev has joined #openstack-nova | 17:48 | |
| sdague | boris-42: what's the payload for the tests? because this seems like rally is somehow driving things down an nbd path which shouldn't be touched | 17:49 |
| sdague | and that's the actual issue to be figured out | 17:49 |
| alaski | dansmith: heh | 17:49 |
| *** iamjarvo has quit IRC | 17:49 | |
| alaski | dansmith: I'm starting to think we need a make_me_impotent() method for objects we can call in places like this | 17:49 |
| boris-42 | sdague: what do you mean by payload? | 17:50 |
| dansmith | alaski: well, we have the temporary context thing | 17:50 |
| *** pradipta has joined #openstack-nova | 17:50 | |
| sdague | boris-42: the real issue in the rally jobs is this - http://logs.openstack.org/09/169009/1/check/check-devstack-dsvm-cells/f590346/logs/screen-n-cpu.txt.gz#_2015-03-30_17_47_19_300 | 17:50 |
| *** mriedem has joined #openstack-nova | 17:51 | |
| boris-42 | sdague: okay but if we didn't touch rally jobs at all (they are not touched in infra/rally repos) | 17:51 |
| boris-42 | sdague: and we didn't merge anything in rally related to this | 17:51 |
| sdague | boris-42: that's fine, however, figuring out what changed is kind of important | 17:51 |
| jogo | dansmith: the general issue are facing is don't have a lot of extra nodes in nodepool usually | 17:53 |
| boris-42 | sdague: nothing https://github.com/openstack/nova/commits/master/nova/virt/disk changed there | 17:53 |
| jogo | dansmith: we can add multi node testing without dropping anything if we want, and remove it, if it takes up too many resources | 17:53 |
| sdague | yeh, I know what triggered it | 17:53 |
| alaski | dansmith: yeah, that would work here. I was thinking more permanently, but I think passing a neutered object over RPC would fix it anyways | 17:53 |
| dansmith | jogo: yes, and ironic is different here in that it consumes upstream nodes instead of vendor-supplied ones | 17:53 |
| jogo | dansmith: right, same for ceph | 17:53 |
| dansmith | alaski: ah, no, whenever you pass an object over rpc, it magically gets endowed with the context you used in the call | 17:54 |
| boris-42 | sdague: http://logstash.openstack.org/#eyJzZWFyY2giOiJcIlVuYWJsZSB0byBpbXBvcnQgZ3Vlc3Rmc1wiIiwiZmllbGRzIjpbXSwib2Zmc2V0IjowLCJ0aW1lZnJhbWUiOiIxNzI4MDAiLCJncmFwaG1vZGUiOiJjb3VudCIsInRpbWUiOnsidXNlcl9pbnRlcnZhbCI6MH0sInN0YW1wIjoxNDI3NzM4MDQ0NTI0fQ== | 17:54 |
| sdague | though, I kind of don't understand why rally is the only thing tripping over it | 17:54 |
| dansmith | alaski: so if we want to defeat that, we'd need something else like you say | 17:54 |
| jogo | dansmith: long term, if multi node works, we can probably drop the regular job and switch to multi node | 17:54 |
| jogo | dansmith: and hopefully remove postgres as well | 17:54 |
| dansmith | jogo: yeah, that seems much better to me | 17:54 |
| jogo | to balance things out | 17:55 |
| jogo | dansmith: but aioppcu is too new and complex to make gating at the moment | 17:55 |
| sdague | boris-42: so that's the thing which is important to figure out, what is rally doing that modifies the guests before they boot? | 17:55 |
| dansmith | jogo: yep | 17:55 |
| boris-42 | sdague: okay let analyze boot-and-delete-server scenario | 17:55 |
| jogo | dansmith: will post a patch getting the multi node test running non voting on nova | 17:56 |
| boris-42 | sdague: https://github.com/stackforge/rally/blob/master/rally/benchmark/scenarios/nova/servers.py#L102-L104 just 2 methods here | 17:56 |
| jogo | dansmith: and we can take it from there | 17:56 |
| dansmith | jogo: cool.. was nice to get to work with clarkb on that last week | 17:56 |
| dansmith | jogo: he only used your name as a curse word like three or four times | 17:56 |
| boris-42 | sdague: this codes boot vm https://github.com/stackforge/rally/blob/master/rally/benchmark/scenarios/nova/utils.py#L118-L119 | 17:56 |
| *** lucas-dinner has quit IRC | 17:57 | |
| jogo | dansmith: :) | 17:57 |
| boris-42 | sdague: self.clients("nova") - is just plain nova python client | 17:57 |
| sdague | boris-42: what else is in your kwargs? | 17:57 |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix tenant_network test https://review.openstack.org/169016 | 17:58 |
| sdague | do you have the full request json somewhere? | 17:58 |
| boris-42 | sdague: it is passed from input config | 17:58 |
| openstackgerrit | Eric Brown proposed openstack/nova: Fix config option types https://review.openstack.org/168481 | 17:58 |
| boris-42 | sdague: in this case as far as I know nothing | 17:58 |
| sdague | ok, can you point me to where you are logging the request json somewhere? | 17:58 |
| *** browne has quit IRC | 17:59 | |
| boris-42 | sdague: http://logs.openstack.org/62/168962/1/check/gate-rally-dsvm-rally/2e213c7/rally-plot/results.html.gz#/NovaServers.boot_and_delete_server/task | 17:59 |
| sdague | boris-42: that's not the nova request | 17:59 |
| boris-42 | sdague: so we can run in debug rally | 17:59 |
| sdague | I need the actual nova request | 18:00 |
| boris-42 | sdague: and get all logs | 18:00 |
| cfriesen | would it be possible to get someone to review https://review.openstack.org/162746 ? It's got a +1 from garyk, but hasn't really gotten much attention. Should be pretty much a no-brainer. There's also a followup at https://review.openstack.org/164762 that fixes a race that both affinity/anti-affinity server groups are susceptible to. | 18:00 |
| boris-42 | sdague: or I can share logs | 18:00 |
| sdague | boris-42: it's not logged during normal runs? | 18:00 |
| boris-42 | sdague: http://logs.openstack.org/62/168962/1/check/gate-rally-dsvm-rally/2e213c7/logs/ | 18:00 |
| boris-42 | sdague: here is console http://logs.openstack.org/62/168962/1/check/gate-rally-dsvm-rally/2e213c7/console.html | 18:00 |
| boris-42 | sdague: we disabled debug mode because it was generating too much data | 18:01 |
| boris-42 | sdague: but it is easy to turn it on for single run | 18:01 |
| sdague | boris-42: ok, I'm not going to sift all the rally logs here, please provide me a direct link to something with the nova request that was inbound | 18:01 |
| boris-42 | sdague: okay let me make patch in rally it will take some time | 18:02 |
| boris-42 | sdague: to be run in gates | 18:02 |
| *** vladikr has quit IRC | 18:03 | |
| sdague | yeh, so the issue is we accidentally stopped installing python-guestfs in devstack, because guestfs doesn't publish to pypi | 18:05 |
| sdague | and the venv bits for nova broke | 18:05 |
| sdague | however, it exposes that only rally actually uses them for anything | 18:06 |
| dansmith | file injection? | 18:06 |
| sdague | and it would be really good to figure out what is triggering that code | 18:06 |
| kashyap | Can anyone else reproduce this issue? -- https://bugs.launchpad.net/nova/+bug/1438183 | 18:06 |
| openstack | Launchpad bug 1438183 in OpenStack Compute (nova) "Graceful shutdown of nova-compute service fails" [Undecided,New] | 18:06 |
| sdague | dansmith: yeh | 18:06 |
| kashyap | I doubt if the test I noted in comment #2 is helpful which just sends a SIGTERM to nova-compute process immediately after booting an instance | 18:07 |
| sdague | kashyap: there is another bug already for tha | 18:07 |
| dansmith | kashyap: you'll have to time it just right | 18:07 |
| *** greghaynes has quit IRC | 18:07 | |
| dansmith | I should probably take that one | 18:08 |
| dansmith | and make nova kill all event waiters | 18:08 |
| dansmith | kashyap: ^ | 18:08 |
| sdague | oh, the other bug is for the api actually | 18:08 |
| *** jlanoux has quit IRC | 18:09 | |
| kashyap | sdague: Got a URL? | 18:09 |
| kashyap | dansmith: Ah, "time it just right" | 18:09 |
| sdague | kashyap: https://review.openstack.org/#/c/129167/ is the review, the bug is backlinked from it | 18:10 |
| dansmith | kashyap: I'll snagificate this one | 18:10 |
| kashyap | dansmith: Is that word? :-) | 18:10 |
| *** sbfox has quit IRC | 18:11 | |
| kashyap | But, please do. I was just looking at it out of curiosity to see the behavior. | 18:11 |
| dansmith | kashyap: depends on your definition of "is" :) | 18:11 |
| kashyap | sdague: Thanks. | 18:11 |
| *** vladikr has joined #openstack-nova | 18:12 | |
| sdague | dims: you up for updating https://review.openstack.org/#/c/55873 to use DeprecatedOpt so that we aren't trying to deprecate this option for another 2 years? | 18:16 |
| sdague | bauzas: in future, just updating a patch like that is more helpful than the -1 | 18:16 |
| dims | sdague: ack | 18:16 |
| dims | sdague: will take a stab at it in a few mins | 18:17 |
| sdague | dims: cool, it should be quick | 18:17 |
| dims | sdague: so in https://review.openstack.org/#/c/168950/2/nova/virt/disk/mount/nbd.py,cm i was trying to fail fast, as we know that the ndb stuff is not available. is that not right? | 18:17 |
| sdague | dims: well, the only reason we got there is because it was our last resort | 18:18 |
| *** sbfox has joined #openstack-nova | 18:18 | |
| dims | true | 18:18 |
| sdague | realistically in this kind of scenario we should probably fail all the way back to the user and put vms into error state | 18:18 |
| sdague | because nova was told to do a very specific thing, and it couldn't | 18:19 |
| dims | sdague: i'll try again later to submit one that fails all the way (and fail fast) | 18:20 |
| sdague | dims: cool, though it did expose that apparently we aren't using libguestfs at all in normal jobs | 18:21 |
| dims | 1y | 18:21 |
| dims | y | 18:21 |
| sdague | so let's not rush to close it until we figure out how to tickle those bits | 18:21 |
| sdague | or more importantly, let's make sure we're getting libguestfs tested somewhere | 18:22 |
| dims | i had WIP in the title. ack | 18:23 |
| *** johnthetubaguy is now known as zz_johnthetubagu | 18:23 | |
| *** gokrokve has joined #openstack-nova | 18:26 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova-specs: Show 'reserved' status in os-fixed-ips API https://review.openstack.org/168996 | 18:26 |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Allow explicit numa pinning in flavor extra specs or image metadata https://review.openstack.org/169026 | 18:27 |
| *** rushil_ has joined #openstack-nova | 18:28 | |
| *** vilobhmm11 has joined #openstack-nova | 18:29 | |
| vilobhmm11 | sjcazzol : ping | 18:29 |
| *** rushil has quit IRC | 18:31 | |
| vilobhmm11 | does this idea seem useful to the group https://review.openstack.org/#/c/84432/ we are interested in this feature and if the group seems an interest in this kind of a feature would like to work on this…previously this feature was abandoned due to some valid concerns…has anything changed in nova after that ; that will allow to have flavor based quotas ? | 18:32 |
| *** exploreshaifali has quit IRC | 18:33 | |
| *** VW_ has quit IRC | 18:37 | |
| *** VW_ has joined #openstack-nova | 18:37 | |
| *** achanda has quit IRC | 18:38 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Scheduler: deprecate mute_weight_value option on weigher https://review.openstack.org/55873 | 18:38 |
| *** VW_ has quit IRC | 18:39 | |
| *** VW_ has joined #openstack-nova | 18:39 | |
| *** sbfox has quit IRC | 18:46 | |
| *** xyang1 has quit IRC | 18:46 | |
| *** suresh12 has quit IRC | 18:47 | |
| *** pradipta has quit IRC | 18:48 | |
| *** dsanders1 has joined #openstack-nova | 18:50 | |
| *** dsanders has quit IRC | 18:51 | |
| *** Nic has quit IRC | 18:51 | |
| *** iamjarvo has joined #openstack-nova | 18:51 | |
| *** iamjarvo has quit IRC | 18:51 | |
| *** iamjarvo has joined #openstack-nova | 18:52 | |
| *** Maike has quit IRC | 18:52 | |
| *** Maike has joined #openstack-nova | 18:52 | |
| *** vmtrooper has joined #openstack-nova | 18:52 | |
| *** apuimedo has joined #openstack-nova | 18:56 | |
| *** achanda has joined #openstack-nova | 18:57 | |
| *** achanda has quit IRC | 18:57 | |
| *** achanda has joined #openstack-nova | 18:58 | |
| *** vmtrooper has quit IRC | 18:58 | |
| *** baoli has quit IRC | 19:00 | |
| *** xyang1 has joined #openstack-nova | 19:01 | |
| *** matrohon has joined #openstack-nova | 19:01 | |
| *** oro has joined #openstack-nova | 19:02 | |
| *** suresh12 has joined #openstack-nova | 19:04 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Print choices in the config generator https://review.openstack.org/168447 | 19:04 |
| *** claudiub has joined #openstack-nova | 19:07 | |
| *** sbfox has joined #openstack-nova | 19:08 | |
| *** Marga_ has joined #openstack-nova | 19:12 | |
| *** rushil has joined #openstack-nova | 19:13 | |
| *** Nic has joined #openstack-nova | 19:15 | |
| *** rushil_ has quit IRC | 19:16 | |
| *** iamjarvo has quit IRC | 19:18 | |
| *** Maike has quit IRC | 19:19 | |
| openstackgerrit | Joe Gordon proposed openstack/nova-specs: Update Liberty index https://review.openstack.org/169053 | 19:19 |
| jogo | mriedem sdague dansmith: that should finally fix the liberty specs stuff | 19:19 |
| *** Sukhdev has quit IRC | 19:20 | |
| *** nikhil|afk has quit IRC | 19:21 | |
| dansmith | jogo: reviewing that right now would compete with a burrito, and honestly, lose. | 19:21 |
| jogo | dansmith: lol | 19:21 |
| *** annashen has quit IRC | 19:22 | |
| *** penick has joined #openstack-nova | 19:24 | |
| *** Marga_ has quit IRC | 19:25 | |
| vilobhmm11 | dansmith : hi | 19:27 |
| dansmith | vilobhmm11: hello | 19:27 |
| vilobhmm11 | any inputs on "does this idea seem useful to the group https://review.openstack.org/#/c/84432/ we are interested in this feature and if the group seems an interest in this kind of a feature would like to work on this…previously this feature was abandoned due to some valid concerns…has anything changed in nova after that ; that will allow to have flavor based quotas ?" | 19:27 |
| dansmith | vilobhmm11: didn't we have this conversation already? | 19:28 |
| vilobhmm11 | asked this question in the nova channel couple of min ago | 19:28 |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Allow explicit numa pinning in flavor extra specs or image metadata https://review.openstack.org/169026 | 19:28 |
| vilobhmm11 | not really | 19:28 |
| vilobhmm11 | dansmith : ^^ | 19:28 |
| dansmith | vilobhmm11: yeah, I saw, but everyone is kinda busy with this release and thus looking at stuff for L is kinda off the radar at the moment | 19:28 |
| dansmith | vilobhmm11: I had a conversation about flavor quotas with someone.. that wasn't you? | 19:29 |
| dansmith | a couple weeks ago | 19:29 |
| *** exploreshaifali has joined #openstack-nova | 19:29 | |
| vilobhmm11 | no dansmith…I asked about the object model for quotas and not falvor based quotas | 19:29 |
| vilobhmm11 | will check when you are free to consider about L | 19:30 |
| *** lpetrut has left #openstack-nova | 19:30 | |
| *** adalbas has quit IRC | 19:31 | |
| dansmith | vilobhmm11: that would be better | 19:32 |
| openstackgerrit | Dan Smith proposed openstack/nova: Cancel all waiting events during compute node shutdown https://review.openstack.org/169056 | 19:32 |
| openstackgerrit | Dan Smith proposed openstack/nova: Prevent scheduling new external events when compute is shutdown https://review.openstack.org/169057 | 19:32 |
| *** atuvenie has quit IRC | 19:34 | |
| bauzas | sdague: re: https://review.openstack.org/#/c/55873, I'm okay with your PoV, but just wanted to say that I'm actually doing this only for commit msgs atm | 19:34 |
| bauzas | sdague: but of course, will do it in the future | 19:34 |
| *** annashen has joined #openstack-nova | 19:34 | |
| sdague | jogo: it fails tests | 19:35 |
| *** veena has quit IRC | 19:37 | |
| *** VW__ has joined #openstack-nova | 19:40 | |
| *** VW_ has quit IRC | 19:40 | |
| *** exploreshaifali has quit IRC | 19:42 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Fix API links and labels https://review.openstack.org/166985 | 19:42 |
| jogo | sdague: thanks, | 19:44 |
| sdague | mriedem: you still around? | 19:44 |
| sdague | I want to fix that doc bug of yours | 19:44 |
| mriedem | sdague: i am | 19:44 |
| mriedem | i'm also playing with that | 19:44 |
| *** burt has quit IRC | 19:44 | |
| sdague | can you try reordering the requirements.txt ? | 19:44 |
| openstackgerrit | Joe Gordon proposed openstack/nova-specs: Update Liberty index https://review.openstack.org/169053 | 19:44 |
| jogo | take 2 | 19:44 |
| sdague | pbr first | 19:44 |
| jogo | about to board a flight, so if that fails feel free to respin | 19:45 |
| sdague | sphinx second | 19:45 |
| sdague | then the rest | 19:45 |
| mriedem | sdague: ok, will try it | 19:45 |
| *** otter768 has joined #openstack-nova | 19:45 | |
| *** adalbas has joined #openstack-nova | 19:45 | |
| sdague | and rm -rf .tox to make sure we're starting from a clean point | 19:46 |
| *** Mike_D_laptop has quit IRC | 19:47 | |
| *** otter768 has quit IRC | 19:49 | |
| *** mpavone has joined #openstack-nova | 19:50 | |
| *** apuimedo has quit IRC | 19:51 | |
| bauzas | alaski: still there ? | 19:52 |
| bauzas | alaski: I think I found the root proble | 19:52 |
| bauzas | pbb | 19:52 |
| bauzas | problem even | 19:52 |
| alaski | bauzas: yep | 19:52 |
| bauzas | alaski: https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L1854-L1857 | 19:53 |
| *** iamjarvo has joined #openstack-nova | 19:54 | |
| bauzas | alaski: jsonutils.dumps(obj) creates a dict but jsonutils.loads(dict) doesn't rehydrate an object | 19:55 |
| bauzas | we should rather call obj_to_primitive IIUC | 19:55 |
| alaski | bauzas: yeah. it will need to conditionally primitize the object and rehydrate it | 19:57 |
| *** sdgdfgfdsg has joined #openstack-nova | 19:57 | |
| sdgdfgfdsg | samp.bmhp.net:7777 | 19:57 |
| mriedem | sdague: re-ordering the reqs didn't help | 19:57 |
| * dansmith clicks on the url to check it out | 19:57 | |
| bauzas | alaski: so, that means that all the objects methods are serialized into dicts... | 19:58 |
| bauzas | s/serialized/primitived | 19:58 |
| sdgdfgfdsg | samp.bmhp.net: 7777 | 19:58 |
| *** oro has quit IRC | 19:58 | |
| dansmith | sdgdfgfdsg: I'm putting in my personal info, but it says my login is incorrect | 19:58 |
| bauzas | alaski: btw. I don't see why we need to do all of this using oslo.messaging, do you know ? | 19:58 |
| *** oro has joined #openstack-nova | 19:59 | |
| *** dprince has quit IRC | 19:59 | |
| bauzas | alaski: to do all of this *not* using oslo.messaging of course... (late here) | 19:59 |
| alaski | bauzas: the serializaition is only an issue on the responses, most methods don't need a response so not everything is broken | 20:00 |
| alaski | bauzas: I don't know if there's a reason it's not using oslo.messaging | 20:00 |
| bauzas | alaski: agreed, but that's just something broken for any call, right? | 20:00 |
| bauzas | RPC call I mean | 20:00 |
| alaski | any that are expecting an object to be returned | 20:01 |
| bauzas | sure | 20:02 |
| *** pixelb has joined #openstack-nova | 20:02 | |
| sdgdfgfdsg | you hack my server ok samp.bmhp.net:7777 hack me hehhe | 20:02 |
| bauzas | okay, let's fix this once that we know why... | 20:02 |
| alaski | bauzas: compute_node_get_all is doing an obj_to_primitive before it hits that point | 20:03 |
| sdgdfgfdsg | done ahahahha | 20:03 |
| bauzas | alaski: not in my patch | 20:03 |
| alaski | bauzas: ahh | 20:03 |
| NobodyCam | sdague: question, I was just reading the scroll back and saw your chat with devananda this morning about removing the pxe_ssh job. | 20:03 |
| alaski | bauzas: need to get my head on this again | 20:04 |
| bauzas | alaski: np | 20:04 |
| sdgdfgfdsg | you hack my server ok samp.bmhp.net:7777 hack me hehhe | 20:04 |
| NobodyCam | is there a place (review) I can follow this? | 20:04 |
| sdgdfgfdsg | .i. | 20:04 |
| *** annegentle has quit IRC | 20:04 | |
| bauzas | alaski: anyway, I'll fix the above method | 20:05 |
| sdague | NobodyCam: not really, jogo just asked my opinion on low value jobs. And I said that was a low value one as I don't know of any regressions that it's caught since the ironic driver merged | 20:05 |
| sdague | I would be happy to be pointed at instances where it did catch a regression that was uncaught by other mechanisms | 20:06 |
| sdague | *since* the ironic driver merged | 20:06 |
| *** iamjarvo has quit IRC | 20:06 | |
| sdgdfgfdsg | you hack my server ok samp.bmhp.net:7777 hack me hehhe | 20:07 |
| dansmith | sdague: I think there are circumstances where, despite not having caught anything yet, we can make an educated call that testing is valuable, right? | 20:07 |
| dansmith | anyone mind if I boot sdgdfgfdsg out? | 20:07 |
| *** ChanServ sets mode: +o dansmith | 20:08 | |
| sdague | dansmith: go for it | 20:08 |
| *** sdgdfgfdsg was kicked by dansmith (Your behavior is not conducive to the desired environment.) | 20:08 | |
| *** sdgdfgfdsg has joined #openstack-nova | 20:08 | |
| sdgdfgfdsg | .i. | 20:08 |
| *** dansmith sets mode: +b *!~ircap@178-223-226-167.dynamic.isp.telekom.rs | 20:08 | |
| *** sdgdfgfdsg was kicked by dansmith (Your behavior is not conducive to the desired environment.) | 20:08 | |
| *** ChanServ sets mode: -o dansmith | 20:08 | |
| *** suresh12 has quit IRC | 20:10 | |
| alaski | bauzas: ok. you might be able to take some inspiration from https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L302-L312 | 20:10 |
| bauzas | alaski: agreed, I was looking at the pattern | 20:11 |
| sdague | dansmith: does that educated call involve looking at the actual test payload of that job? :) | 20:11 |
| sdague | because it's not what you think | 20:11 |
| sdague | for instance, it doesn't actually run any compute tests | 20:11 |
| dansmith | sdague: how do you know what I think it is? | 20:12 |
| sdague | I don't | 20:12 |
| sdague | well, based on your statements earlier about it being like vmware and hyperv | 20:12 |
| sdague | I assumed you thought it was running compute tests | 20:12 |
| dansmith | sdague: indeed I thought it was running more than it is, because what seems to be running isn't even enough for us to have put it into the tree in the first place | 20:14 |
| dansmith | so presumably a job got dropped somewhere? | 20:14 |
| sdague | not as far as I know | 20:14 |
| *** BadCub has joined #openstack-nova | 20:14 | |
| dansmith | I recall looking at fairly complete tempest runs just before merge | 20:15 |
| *** lpetrut has joined #openstack-nova | 20:15 | |
| sdague | well, it was a regex, I don't know how complete it was | 20:15 |
| sdague | again, this is the job as evolved by the ironic team | 20:15 |
| sdague | and part of the problem with cross linked jobs like this, because someone goes and changes out the job definition | 20:16 |
| dansmith | are they running a full tempest run on their commits? | 20:16 |
| sdague | I don't know | 20:17 |
| sdague | looks like not | 20:17 |
| dansmith | don't they have to in order to be a whatever-we-call-it-these-days project? | 20:17 |
| mtreinish | dansmith: IIRC they are not | 20:17 |
| jroll | sdague: dansmith: we aren't running *full* tempest suite because we don't support everything nova does, like rescue | 20:18 |
| dansmith | how is that possible? | 20:18 |
| sdague | and, hence, why I don't think there is value here | 20:18 |
| jroll | and private networks or whatever | 20:18 |
| jroll | (afaik) | 20:18 |
| dansmith | jroll: yeah, that excuse doesn't work :) | 20:18 |
| dansmith | sdague: doesn't the TC care about this? | 20:18 |
| jroll | dansmith: so you don't think ensuring the ironic driver can boot and delete an instance is valuable? | 20:18 |
| dansmith | I mean, I swear this was a thing | 20:18 |
| sdague | dansmith: right, and that was the thing where no one could ever merge code | 20:19 |
| jroll | dansmith: which includes all the resource interaction etc | 20:19 |
| jroll | resource tracker* | 20:19 |
| dansmith | jroll: what test tests that? | 20:19 |
| sdague | their one scenario test does | 20:19 |
| jroll | dansmith: dsvm-pxe_ssh or whatever | 20:19 |
| sdague | jroll: actually you only have 1 test in there that does that | 20:19 |
| dansmith | jroll: no I mean within that *job* which of the very few tests does that | 20:19 |
| dansmith | jroll: because what I'm seeing looks really skimpy | 20:20 |
| jroll | dansmith: I'd have to look for specifics, adam_g knows more off the top of his head | 20:20 |
| sdague | dansmith: so I've actually been pushing on the idea of not testing everything against everything, per the testing specification up there now. | 20:20 |
| dansmith | sdague: I know you can't speak for the TC thing, but can you comment? | 20:20 |
| dansmith | sdague: not even tempest against ironic on *their* patches? | 20:20 |
| sdague | they are running the baremetal tag of tempest on their patches | 20:21 |
| dansmith | sdague: AFAIK, our requirement (for nova) has been "a reasonably complete tempest run, minus things you don't support", which this doesn't look anywhere close | 20:21 |
| sdague | dansmith: sure, from a nova perspective that's true | 20:21 |
| sdague | this job is not that | 20:21 |
| sdague | hence, it has really low value in my mind | 20:22 |
| dansmith | it also took >5 minutes to boot a single guest? | 20:22 |
| dansmith | sdague: it would have helped a lot if you had led with this :) | 20:22 |
| *** suresh12 has joined #openstack-nova | 20:22 | |
| jroll | dansmith: nested virt is pretty slow :| | 20:22 |
| dims | mriedem: i can't recreate your problem with docs either | 20:22 |
| sdague | honestly, I knew it was incomplete, I did not fully understand it was this incomplete until recently | 20:22 |
| cfriesen | jroll: if you do nested kvm it's not bad | 20:23 |
| adam_g | hi | 20:23 |
| NobodyCam | sdague: is this something nova is looking to remove before the summit. would we be able to sit down and plan it at the summit | 20:23 |
| NobodyCam | Hi adam_g :) | 20:23 |
| adam_g | the tempest job was skimped back last week https://review.openstack.org/#/c/161420/ | 20:23 |
| jroll | cfriesen: I can't control what hypervisor rackspace or HP use | 20:23 |
| sdague | the fact that it had never caught a regression to my understanding was part of not thinking it was useful | 20:23 |
| adam_g | prior to that we had two jobs: the pxe_ssh which still ran a subset, tho not a slim, and another job (running outsie of nova) that ran the whole test suite | 20:23 |
| dansmith | adam_g: can you point me at one of the unslimmed runs? | 20:24 |
| sdague | anyway, I have errands to go run | 20:24 |
| adam_g | we decided just target a smaller subset tests via the baremetal tag after discussion with devananda and mtreinish in -qa | 20:24 |
| adam_g | dansmith, let me find some | 20:24 |
| *** jecarey has quit IRC | 20:24 | |
| dansmith | adam_g: thanks | 20:24 |
| dansmith | 40 minutes of test time to boot a single guest is pretty hard to justify :/ | 20:25 |
| *** annegentle has joined #openstack-nova | 20:25 | |
| adam_g | dansmith, pre-slimming we were still only booting one guest via the scenari | 20:25 |
| dansmith | okay, well, that's what I wanted to see | 20:25 |
| mtreinish | adam_g: that discussion was about the job for running on baremetal commits, I don't think we ever discussed what was required for nova | 20:26 |
| adam_g | dansmith, compute tests that booted instances were explictily skipped IIRC. the other tjob that was running th *entire* suite (with compute tests that spawned instances) took 60min+ | 20:26 |
| dansmith | adam_g: okay | 20:26 |
| adam_g | mtreinish, fair enough | 20:26 |
| adam_g | one sec, let me logstash | 20:26 |
| dansmith | so, I think we need to have a conversation about this, because this is a far lower bar than we hold for anyone else, where ironic's should be at least as high, running in infra and all | 20:27 |
| dansmith | and lower than our stated minimum for being in tree | 20:27 |
| adam_g | dansmith, tempest can run in its entirety now (it didnt before, which is why we were using that regex).. the questin is, is nova okay with having a voting job that takes 1hr+ to run? | 20:27 |
| *** afazekas has joined #openstack-nova | 20:27 | |
| dansmith | adam_g: well, there are two concerns: | 20:28 |
| mtreinish | adam_g: it takes >1hr running in parallel? | 20:28 |
| dansmith | adam_g: first is us wanting to see full runs and second is infra trying to disable even what we have because of resources | 20:28 |
| adam_g | mtreinish, yes. each instance takes minutes to boot | 20:28 |
| dansmith | if getting what we really want is too heavy on resources, then we have a real problem | 20:28 |
| adam_g | dansmith, im not aware of it ever being an issue of resources, the test still runs below the infra global job timeout | 20:29 |
| mtreinish | adam_g: what about trying to bump the concurrency to try and block less on slow server boots? | 20:29 |
| NobodyCam | dansmith: is this some thing wa can discuss at the summit, while leaving whats there inplace until we can draft some kind of action plan? | 20:29 |
| dansmith | adam_g: that's what brought this up today | 20:29 |
| adam_g | mtreinish, we cant: bumping the concurrency requires bigger slaves to spawn more beefy VM nodes in parallel | 20:30 |
| dansmith | NobodyCam: well, the problem is, we are making a release under the assumption that this stuff has seen some testing | 20:30 |
| dansmith | NobodyCam: and that turns out to be only very slightly true :( | 20:30 |
| NobodyCam | but is not slightly true better then not at all true? | 20:31 |
| dansmith | NobodyCam: I think the schedule is working in your favor as I can't imagine making a knee-jerk change here is going to be on anyone's list of pleasant actions, but... | 20:31 |
| dansmith | I'm just really (really) disappointed | 20:31 |
| jroll | dansmith: I mean, there isn't a ton of functionality in the ironic driver. I think it's all tested afaict | 20:31 |
| NobodyCam | I am all of makeing better | 20:31 |
| NobodyCam | just not removing with a bulk swip whats already there | 20:31 |
| *** ijw has quit IRC | 20:31 | |
| dansmith | so the thing that came up here, | 20:32 |
| cfriesen | maybe ironic needs dedicated actual hardware for testing? or at least VMs that could do nested kvm (avoiding the need for qemu)? | 20:32 |
| dansmith | was potentially looking to remove this job altogether because it's not useful enough to justify the "cost" | 20:32 |
| adam_g | dansmith, sorry, looks like there aren't jobs archived going back to before the 20th | 20:32 |
| dansmith | and tbh, I can't really argue that what we're covering is worth the ~40 minutes of nodepool time | 20:32 |
| dansmith | adam_g: okay, thanks | 20:32 |
| mtreinish | adam_g: logstash only has 10 days of history | 20:32 |
| jroll | cfriesen: nested kvm won't work because rackspace runs xen. if we can find actual hardware, that would be ideal. tis kind of hard. | 20:33 |
| *** melwitt has quit IRC | 20:33 | |
| dansmith | jroll: FWIW, every other hypervisor driver has committed significant hardware resources to their CI | 20:33 |
| *** melwitt_ has joined #openstack-nova | 20:33 | |
| dansmith | I know the owner isn't as clear cut in this case | 20:33 |
| jroll | dansmith: I promise you I've tried. I can't even get hardware for a downstream CI environment. | 20:34 |
| *** afazekas has quit IRC | 20:34 | |
| cfriesen | what about piggybacking on the intel CI that's doing the NUMA/PCI testing? | 20:34 |
| *** ociuhandu has quit IRC | 20:34 | |
| adam_g | dansmith, so the idea behind cutting out the compute API tests was that the single scenario test that does run covers stressing all of the integration points between nova+ironic, or at least is supposed to | 20:36 |
| dansmith | adam_g: it doesn't cover all of the driver though.. it can't | 20:36 |
| dansmith | adam_g: I could merge a rebuild() implementation that raises Exception('haha, you're broken') and ironic CI wouldn't catch it, AFAICT | 20:37 |
| adam_g | dansmith, that scenario tests rebuild | 20:37 |
| *** afazekas has joined #openstack-nova | 20:37 | |
| dansmith | adam_g: fine, pick some other operation ironic doesn't support :) | 20:37 |
| adam_g | dansmith, tests that stress opeerations not supported by the driver would just be skipped anyway? | 20:38 |
| mriedem | pause! | 20:38 |
| dansmith | adam_g: you understand my point, right? or do you want to keep arguing like this? :) | 20:39 |
| mriedem | does the scenario test cover interface attach/detach? | 20:39 |
| adam_g | since theres no way for tempest to determine driver capabilities, it uses feature flags to avoid the tests known not to be supported. so enabling the entire suite gets a really long test time and a bunch of skips. the API tests that do end up running burn cycles spinning up ironic nodes to test, say, nova+glance integratino | 20:39 |
| dansmith | adam_g: that scenario test doesn't test reboot, right? | 20:39 |
| dansmith | adam_g: or stop/start | 20:40 |
| dansmith | adam_g: or the code that gets run when security groups are modified on a running instance | 20:40 |
| dims | dansmith: the LOG.debug in https://review.openstack.org/#/c/169056/1/nova/compute/manager.py,cm does not seem right (compare with LOG.debug in say the external_instance_event method) | 20:40 |
| jroll | this entire conversation reads to me as "slow operations shouldn't be tested" | 20:40 |
| dansmith | dims: did it for string freeze.. wanted it to be a warning :/ | 20:40 |
| *** mwagner_lap has quit IRC | 20:40 | |
| jroll | which is... an interesting viewpoint. | 20:40 |
| adam_g | dansmith, fair enough | 20:41 |
| dansmith | jroll: not even a little | 20:41 |
| *** annegentle has quit IRC | 20:41 | |
| dansmith | jroll: what I want is you to run the tests. presumably infra wants their resources back since you're not :) | 20:41 |
| jroll | dansmith: the reason to drop the ironic testing is that it takes too long, AIUI | 20:41 |
| adam_g | dansmith, where was the discussion about resources happening? | 20:41 |
| *** vmtrooper has joined #openstack-nova | 20:41 | |
| jroll | dansmith: we're clearly running tests. just not all the tests you expect. | 20:41 |
| dansmith | jroll: no, the reason to drop what we have is that the value is so much less than the cost.. I'd definitely rather keep what we have than lose it, but it's also hard to defend that viewpoint | 20:42 |
| dansmith | adam_g: here, this morning | 20:42 |
| dansmith | adam_g: I don't know what the impetus was | 20:42 |
| adam_g | ok | 20:42 |
| jroll | dansmith: right, it's too slow to bother testing. | 20:42 |
| dansmith | adam_g: I just jumped in and tried to defend ya'lls resources as absolutely critical | 20:42 |
| dansmith | adam_g: jroll: but then it was pointed out that you're not even meeting nova's minimum | 20:42 |
| *** vladikr has quit IRC | 20:43 | |
| *** mrda-away is now known as mrda | 20:43 | |
| mriedem | couldn't stop/start/reboot etc just be added to test_baremetal_basic_ops? once you have the server up, that's most of the cost isn't it? | 20:43 |
| jroll | dansmith: got a list of the minimum? | 20:43 |
| dansmith | jroll: tempest - anything you don't specifically support | 20:43 |
| jroll | ok | 20:43 |
| dansmith | mriedem: it wouldn't cover a lot of the edge cases the other tests probe for | 20:43 |
| dansmith | mriedem: but would certainly be better | 20:43 |
| adam_g | dansmith, we can easily revert the change and go back to testing the whole thing if it is okay resource-wise. need to read irc logs | 20:43 |
| *** afazekas has quit IRC | 20:44 | |
| adam_g | mriedem, yeah, but then we'd just be duplicating a lot of test code | 20:44 |
| mordred | infra doesn't need any resources back, fwiw | 20:44 |
| dansmith | adam_g: you said you weren't actually building any more than one guest before the slimming right? | 20:44 |
| mordred | we have clouds - we shoudl use them | 20:44 |
| dansmith | mordred: any idea where the thing came from this morning then? | 20:44 |
| mordred | and so far we have avoided solving problems in infrastructure by removing testing | 20:44 |
| mordred | no, I do not | 20:44 |
| mordred | first I've heard of it | 20:44 |
| mordred | one of our clouds is limpy a little bit right now | 20:45 |
| dansmith | mordred: cool, well, that's what I was defending this morning: not dropping these tests because of perceived low value | 20:45 |
| adam_g | dansmith, sorry, i meant we can revert that change and add the full run to nova's pipeline | 20:45 |
| mordred | but that's life - removing a test won't help much :) | 20:45 |
| dansmith | mordred: sounds like we just need to ramp that up to actually cover what we want | 20:45 |
| mordred | dansmith: cool. I mean, it's possible I'm missing some context of course ... | 20:45 |
| mordred | jeblair: ^^ you know anything about this ? | 20:45 |
| mtreinish | mordred, dansmith: just blame jogo, that sounds like his windmill | 20:45 |
| mordred | kk | 20:45 |
| dansmith | adam_g: I can't make that call myself, especially since it has an effect on our total turnaround time, right? | 20:46 |
| dansmith | heh | 20:46 |
| mordred | well, I'm fine with removing or keeping whatever people want to run | 20:46 |
| mordred | but infra is not running out of resources | 20:46 |
| mordred | we're here to run what the projects want to run | 20:46 |
| mordred | and if we need to solve issues around that, we'll do so | 20:46 |
| dansmith | cool, then mtreinish had some ideas about changing the parallelism to make that turnaround time quicker too | 20:46 |
| *** Sukhdev has joined #openstack-nova | 20:46 | |
| mordred | cool! | 20:46 |
| *** vmtrooper has quit IRC | 20:47 | |
| dansmith | adam_g: so when we first had this discussion before we merged ironic, | 20:47 |
| dansmith | adam_g: the thing we wanted was for ironic to be run like 3rd party, which was reporting separately from jenkins, | 20:47 |
| dansmith | adam_g: which means it could run longer and we wouldn't care as much | 20:47 |
| adam_g | the parallelism of the ironic job is limited by the size of the test slaves | 20:47 |
| dansmith | adam_g: but I think that was hard for $reasons | 20:47 |
| mtreinish | dansmith: well maybe, adam_g said the nodes aren't big enough to bring up enough virtual baremetal guests | 20:48 |
| mordred | yah. _that_ is a thing that infra does not support at the moment | 20:48 |
| dansmith | yeah, that would just make it much easier since we wouldn't block jenkins on one very serial test | 20:48 |
| mtreinish | adam_g: but what is the resource constraint ram, cpus? | 20:48 |
| jeblair | mordred, dansmith: i do not know what 'the thing from this morning is'. i agree with what mordred has said. | 20:48 |
| dansmith | and I'd hate to see us not get full runs because that is painful for people | 20:48 |
| dansmith | jeblair: cool, I have no idea where the infra part came from | 20:49 |
| adam_g | mtreinish, RAM. | 20:49 |
| jeblair | mordred, dansmith, mtreinish: and yeah, jogo is removing unecessary jobs, which is good, because that's hard to figure out what jobs are really unecessary. but i don't think we've actually crossed the line to removing jobs that actually test things. :) | 20:49 |
| dansmith | jeblair: excellent | 20:49 |
| jeblair | mordred, dansmith, mtreinish: (ie, jobs that don't actually test neutron running on neutron changes, etc) | 20:49 |
| *** jecarey has joined #openstack-nova | 20:50 | |
| adam_g | wasn't there some work last cycle going on to enable jobs running on upstream infra to report similar to third party jobs? | 20:50 |
| dansmith | so one thing we could do, | 20:50 |
| mriedem | hmm, seems our hypervisor support matrix is not accurate for ironic, e.g. says suspend is partially supported http://docs.openstack.org/developer/nova/support-matrix.html | 20:50 |
| dansmith | which would not be completely ideal, | 20:50 |
| dansmith | but would be to have a full job that only runs on changes to drivers/ironic/* | 20:50 |
| dansmith | which would provide about as much coverage as other hypervisors provide | 20:50 |
| dansmith | I don't think the others vote on all patches, but at least prioritize their driver directory | 20:51 |
| mtreinish | adam_g: why do you need so much ram in the 2nd level guests? Isn't the only difference the ipmi thing running on those images. How much can that take :) | 20:51 |
| *** annegentle has joined #openstack-nova | 20:52 | |
| *** rfolco has quit IRC | 20:54 | |
| jroll | mtreinish: the deploy ramdisk takes quite a bit of RAM | 20:54 |
| lifeless | jroll: are you running the iscsi one now? | 20:55 |
| lifeless | jroll: that should let you thin it out | 20:55 |
| *** ijw has joined #openstack-nova | 20:55 | |
| *** suresh12 has quit IRC | 20:55 | |
| jroll | lifeless: what, we're running the IPA ramdisk now | 20:55 |
| lifeless | jroll: yes but there is an IPA+iscsi option too now | 20:55 |
| lifeless | jroll: I thought that landed | 20:55 |
| jroll | lifeless: yes. that doesn't reduce the size of that ramdisk | 20:56 |
| *** suresh12 has joined #openstack-nova | 20:56 | |
| *** adalbas has quit IRC | 20:56 | |
| lifeless | jroll: I know, but it will reduce the working footprint by the size of the compressed image | 20:56 |
| lifeless | deploy image, not ramdisk image | 20:56 |
| jroll | lifeless: if anyone made an effort to reduce the size of the DIB + IPA ramdisk, it might help, it's currently requiring 3gb | 20:56 |
| jroll | lifeless: oh, the instance image? right | 20:57 |
| jroll | cirros is pretty small, I don't think 20MB or whatever will make much of a difference | 20:57 |
| openstackgerrit | Matt Riedemann proposed openstack/nova: Update hypervisor support matrix for ironic wrt pause/suspend https://review.openstack.org/169086 | 20:57 |
| lifeless | jroll: yeah, because no qcow2 into tmpfs overhead | 20:57 |
| lifeless | ok | 20:57 |
| jroll | right | 20:57 |
| lifeless | so DIB+IPA is 3G. Thats surprisingly big. | 20:57 |
| jroll | lifeless: we only run iscsi driver on nova changes, fwiw | 20:57 |
| jroll | right. | 20:57 |
| jroll | that's why we don't use it | 20:57 |
| jroll | nobody has optimized it yet, by removing build-time deps etc | 20:58 |
| *** ijw_ has joined #openstack-nova | 20:58 | |
| *** signed8bit has joined #openstack-nova | 20:58 | |
| *** sbfox has quit IRC | 20:59 | |
| *** amotoki has joined #openstack-nova | 20:59 | |
| *** ijw has quit IRC | 20:59 | |
| *** vilobhmm11 has quit IRC | 21:01 | |
| *** melwitt_ has quit IRC | 21:01 | |
| *** melwitt has joined #openstack-nova | 21:01 | |
| *** ijw_ has quit IRC | 21:01 | |
| *** vilobhmm1 has joined #openstack-nova | 21:01 | |
| *** dkliban is now known as dkliban_afk | 21:01 | |
| *** melwitt has quit IRC | 21:02 | |
| *** matrohon has quit IRC | 21:02 | |
| *** vilobhmm1 has quit IRC | 21:02 | |
| *** ociuhandu has joined #openstack-nova | 21:02 | |
| *** vilobhmm1 has joined #openstack-nova | 21:02 | |
| *** ijw has joined #openstack-nova | 21:03 | |
| *** ijw_ has joined #openstack-nova | 21:04 | |
| *** raildo is now known as raildo|away | 21:04 | |
| *** Marga_ has joined #openstack-nova | 21:05 | |
| *** afazekas has joined #openstack-nova | 21:06 | |
| *** ijw has quit IRC | 21:07 | |
| *** melwitt has joined #openstack-nova | 21:09 | |
| *** mpavone has quit IRC | 21:11 | |
| *** sbfox has joined #openstack-nova | 21:12 | |
| *** casusbelli has joined #openstack-nova | 21:12 | |
| *** mpavone has joined #openstack-nova | 21:13 | |
| *** burt has joined #openstack-nova | 21:15 | |
| openstackgerrit | Moshe Levi proposed openstack/nova: Add option to ssh to prevent prompting https://review.openstack.org/168680 | 21:15 |
| *** thangp has quit IRC | 21:16 | |
| *** ak22 has joined #openstack-nova | 21:17 | |
| *** sdake has joined #openstack-nova | 21:18 | |
| *** boris-42 has quit IRC | 21:18 | |
| *** afazekas has quit IRC | 21:18 | |
| *** mpavone has quit IRC | 21:20 | |
| *** annegentle has quit IRC | 21:21 | |
| *** sdake_ has quit IRC | 21:21 | |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Expose "last_seen_up" when listing services https://review.openstack.org/169093 | 21:21 |
| *** oro has quit IRC | 21:25 | |
| *** boris-42 has joined #openstack-nova | 21:30 | |
| *** cbader has quit IRC | 21:30 | |
| *** yamahata has quit IRC | 21:34 | |
| openstackgerrit | jichenjc proposed openstack/nova: Do not delete volumes during reschedule https://review.openstack.org/169097 | 21:34 |
| *** jamielennox|away is now known as jamielennox | 21:36 | |
| *** annegentle has joined #openstack-nova | 21:36 | |
| *** burt has quit IRC | 21:40 | |
| *** sbfox1 has joined #openstack-nova | 21:40 | |
| *** Marga_ has quit IRC | 21:40 | |
| *** sbfox has quit IRC | 21:40 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova-specs: Show 'reserved' status in os-fixed-ips API https://review.openstack.org/168996 | 21:41 |
| *** Marga_ has joined #openstack-nova | 21:43 | |
| *** mriedem is now known as mriedem_away | 21:45 | |
| *** otter768 has joined #openstack-nova | 21:46 | |
| *** jamielennox is now known as jamielennox|away | 21:46 | |
| *** lpetrut has quit IRC | 21:48 | |
| *** dave-mccowan has quit IRC | 21:50 | |
| *** otter768 has quit IRC | 21:51 | |
| *** jamielennox|away is now known as jamielennox | 21:53 | |
| *** Marga_ has quit IRC | 21:53 | |
| *** kaufer has quit IRC | 21:54 | |
| *** Marga_ has joined #openstack-nova | 21:54 | |
| *** Marga_ has quit IRC | 21:54 | |
| *** aysyd has quit IRC | 21:55 | |
| *** annashen has quit IRC | 21:57 | |
| *** ijw_ has quit IRC | 21:59 | |
| *** oro has joined #openstack-nova | 21:59 | |
| *** suresh12 has quit IRC | 22:00 | |
| *** unicell has quit IRC | 22:02 | |
| *** unicell has joined #openstack-nova | 22:02 | |
| *** nelsnelson has quit IRC | 22:02 | |
| *** vilobhmm1 has quit IRC | 22:04 | |
| openstackgerrit | Eric Brown proposed openstack/nova: Use oslo_config choices support https://review.openstack.org/158573 | 22:04 |
| *** vilobhmm1 has joined #openstack-nova | 22:05 | |
| *** vilobhmm11 has joined #openstack-nova | 22:08 | |
| *** vilobhmm12 has joined #openstack-nova | 22:09 | |
| *** vilobhmm1 has quit IRC | 22:09 | |
| *** mhorban has quit IRC | 22:10 | |
| *** amotoki has quit IRC | 22:10 | |
| *** paul-- has quit IRC | 22:11 | |
| *** paul-- has joined #openstack-nova | 22:12 | |
| *** vilobhmm11 has quit IRC | 22:13 | |
| *** dboik_ has joined #openstack-nova | 22:15 | |
| *** dboik has quit IRC | 22:18 | |
| *** oro has quit IRC | 22:19 | |
| *** annegentle has quit IRC | 22:19 | |
| *** dboik_ has quit IRC | 22:19 | |
| *** gokrokve has quit IRC | 22:21 | |
| *** mhorban has joined #openstack-nova | 22:22 | |
| *** VW__ has quit IRC | 22:24 | |
| *** vmtrooper has joined #openstack-nova | 22:30 | |
| *** claudiub has quit IRC | 22:35 | |
| *** vmtrooper has quit IRC | 22:35 | |
| *** ijw has joined #openstack-nova | 22:36 | |
| *** Marga_ has joined #openstack-nova | 22:37 | |
| *** Marga_ has quit IRC | 22:37 | |
| *** ajayaa has joined #openstack-nova | 22:37 | |
| *** Marga_ has joined #openstack-nova | 22:38 | |
| *** iamjarvo has joined #openstack-nova | 22:38 | |
| *** iamjarvo has joined #openstack-nova | 22:39 | |
| *** annegentle has joined #openstack-nova | 22:41 | |
| *** annashen has joined #openstack-nova | 22:42 | |
| *** smccully has quit IRC | 22:43 | |
| *** dboik_ has joined #openstack-nova | 22:43 | |
| *** eharney has quit IRC | 22:43 | |
| *** oro has joined #openstack-nova | 22:44 | |
| *** harlowja has quit IRC | 22:45 | |
| *** harlowja_ has joined #openstack-nova | 22:45 | |
| *** Sukhdev has quit IRC | 22:46 | |
| *** neelashah has quit IRC | 22:50 | |
| *** annegentle has quit IRC | 22:55 | |
| openstackgerrit | Dan Smith proposed openstack/nova: Prevent scheduling new external events when compute is shutdown https://review.openstack.org/169057 | 22:57 |
| openstackgerrit | Dan Smith proposed openstack/nova: Cancel all waiting events during compute node shutdown https://review.openstack.org/169056 | 22:57 |
| *** markvoelker has quit IRC | 22:57 | |
| *** annashen has quit IRC | 23:00 | |
| mikal | . | 23:01 |
| *** suresh12 has joined #openstack-nova | 23:02 | |
| *** unicell has quit IRC | 23:02 | |
| *** unicell has joined #openstack-nova | 23:03 | |
| *** oro has quit IRC | 23:06 | |
| *** ijw has quit IRC | 23:06 | |
| *** patrickeast_ has joined #openstack-nova | 23:08 | |
| *** patrickeast has quit IRC | 23:08 | |
| *** patrickeast_ is now known as patrickeast | 23:08 | |
| *** ijw has joined #openstack-nova | 23:09 | |
| *** igordcard_ has joined #openstack-nova | 23:11 | |
| *** zzzeek has quit IRC | 23:13 | |
| *** yasu_ has joined #openstack-nova | 23:16 | |
| *** hblixt has joined #openstack-nova | 23:16 | |
| *** penick has quit IRC | 23:16 | |
| *** ajayaa has quit IRC | 23:17 | |
| *** melwitt has quit IRC | 23:17 | |
| *** baoli has joined #openstack-nova | 23:20 | |
| *** hemna is now known as hemnafk | 23:20 | |
| *** alexpilotti has quit IRC | 23:24 | |
| *** baoli_ has joined #openstack-nova | 23:25 | |
| *** alexpilotti has joined #openstack-nova | 23:26 | |
| *** alexpilotti has quit IRC | 23:26 | |
| *** baoli has quit IRC | 23:28 | |
| *** sbfox1 has quit IRC | 23:34 | |
| *** asettle has quit IRC | 23:37 | |
| *** dims has quit IRC | 23:38 | |
| *** iamjarvo has quit IRC | 23:39 | |
| *** Marga_ has quit IRC | 23:43 | |
| *** Marga_ has joined #openstack-nova | 23:44 | |
| *** otter768 has joined #openstack-nova | 23:46 | |
| *** yasu_ has quit IRC | 23:47 | |
| *** igordcard_ has quit IRC | 23:50 | |
| *** ZZelle_ has quit IRC | 23:50 | |
| *** otter768 has quit IRC | 23:51 | |
| *** markvoelker has joined #openstack-nova | 23:53 | |
| *** zzzeek has joined #openstack-nova | 23:54 | |
| *** Marga_ has quit IRC | 23:55 | |
| *** shakamunyi has quit IRC | 23:55 | |
| *** annegentle has joined #openstack-nova | 23:56 | |
| *** iamjarvo has joined #openstack-nova | 23:56 | |
| *** iamjarvo has quit IRC | 23:57 | |
| *** takedakn has joined #openstack-nova | 23:57 | |
| *** iamjarvo has joined #openstack-nova | 23:57 | |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Fix disconnect_volume issue when find_multipath_device returns None https://review.openstack.org/159626 | 23:58 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: instance termination with update_dns_entries set fails https://review.openstack.org/160858 | 23:58 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Add the last of the oslo libraries to hacking check https://review.openstack.org/166188 | 23:58 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Remove orphaned tables - iscsi_targets, volumes https://review.openstack.org/166214 | 23:58 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Add documentation for nova-cells and nova-idmapshift commands https://review.openstack.org/166553 | 23:58 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Remove stale test + opensssl information from docs https://review.openstack.org/166555 | 23:59 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Stale rc.local file - vestige from cloudpipe.rst https://review.openstack.org/166566 | 23:59 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Fix API links and labels https://review.openstack.org/166985 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
