| *** heyongli has quit IRC | 00:00 | |
| *** heyongli has joined #openstack-nova | 00:01 | |
| *** Savemech has quit IRC | 00:01 | |
| *** zhangjn has joined #openstack-nova | 00:02 | |
| *** Savemech has joined #openstack-nova | 00:02 | |
| *** ctrath has quit IRC | 00:04 | |
| *** sfinucan has quit IRC | 00:04 | |
| *** cfarquhar has quit IRC | 00:05 | |
| *** cfarquhar has joined #openstack-nova | 00:06 | |
| *** cfarquhar has quit IRC | 00:06 | |
| *** cfarquhar has joined #openstack-nova | 00:06 | |
| *** Kennan has joined #openstack-nova | 00:07 | |
| spandhe | Hi folks, can I get one more +2 on https://review.openstack.org/#/c/240347/ ? | 00:07 |
|---|---|---|
| *** heyongli has quit IRC | 00:10 | |
| *** heyongli has joined #openstack-nova | 00:11 | |
| *** zhangjn has quit IRC | 00:11 | |
| *** markvoelker has quit IRC | 00:11 | |
| *** josecastroleon has quit IRC | 00:15 | |
| *** pixelb has joined #openstack-nova | 00:16 | |
| *** pratikma_ has quit IRC | 00:17 | |
| *** stackdump has joined #openstack-nova | 00:18 | |
| *** rfolco has joined #openstack-nova | 00:18 | |
| *** slagle has quit IRC | 00:19 | |
| *** Jeffrey4l has joined #openstack-nova | 00:20 | |
| *** heyongli has quit IRC | 00:20 | |
| *** heyongli has joined #openstack-nova | 00:21 | |
| spandhe | Can someone from Nova team please review this change to Ironic virt driver? https://review.openstack.org/#/c/240357/ | 00:26 |
| *** vishwanathj has joined #openstack-nova | 00:27 | |
| *** dims has joined #openstack-nova | 00:28 | |
| *** browne has quit IRC | 00:29 | |
| *** sneti has joined #openstack-nova | 00:29 | |
| *** heyongli has quit IRC | 00:31 | |
| *** heyongli has joined #openstack-nova | 00:31 | |
| *** IanGovett has quit IRC | 00:32 | |
| *** edmondsw has quit IRC | 00:32 | |
| *** IanGovett has joined #openstack-nova | 00:33 | |
| *** ndipanov has joined #openstack-nova | 00:37 | |
| *** heyongli has quit IRC | 00:41 | |
| *** diogogmt has quit IRC | 00:42 | |
| *** heyongli has joined #openstack-nova | 00:42 | |
| *** jerrygb has joined #openstack-nova | 00:44 | |
| *** IanGovett has quit IRC | 00:46 | |
| *** bapalm has quit IRC | 00:46 | |
| *** slagle has joined #openstack-nova | 00:47 | |
| *** salv-orlando has joined #openstack-nova | 00:48 | |
| *** bapalm has joined #openstack-nova | 00:48 | |
| *** jerrygb has quit IRC | 00:48 | |
| *** heyongli has quit IRC | 00:51 | |
| *** heyongli has joined #openstack-nova | 00:52 | |
| *** slagle has quit IRC | 00:52 | |
| *** slagle has joined #openstack-nova | 00:54 | |
| *** zhangjn has joined #openstack-nova | 00:58 | |
| *** zhenguo has joined #openstack-nova | 01:00 | |
| *** heyongli has quit IRC | 01:01 | |
| *** ssurana has quit IRC | 01:02 | |
| *** achanda has quit IRC | 01:02 | |
| *** heyongli has joined #openstack-nova | 01:03 | |
| *** mrkz has quit IRC | 01:03 | |
| *** ndipanov has quit IRC | 01:07 | |
| *** salv-orlando has quit IRC | 01:07 | |
| *** RA_ has quit IRC | 01:07 | |
| *** RA_ has joined #openstack-nova | 01:09 | |
| *** heyongli has quit IRC | 01:12 | |
| *** markvoelker has joined #openstack-nova | 01:12 | |
| *** heyongli has joined #openstack-nova | 01:13 | |
| *** stackdump has quit IRC | 01:13 | |
| *** electrocucaracha has quit IRC | 01:15 | |
| *** RuiChen has joined #openstack-nova | 01:15 | |
| *** markvoelker has quit IRC | 01:17 | |
| *** thorst has joined #openstack-nova | 01:22 | |
| *** heyongli has quit IRC | 01:22 | |
| *** heyongli has joined #openstack-nova | 01:23 | |
| openstackgerrit | Tracy Jones proposed openstack/nova: Cleanup of compute api reboot method https://review.openstack.org/127427 | 01:24 |
| *** hemna is now known as hemnafk | 01:24 | |
| *** angdraug has quit IRC | 01:25 | |
| *** signed8bit has quit IRC | 01:28 | |
| *** gyee has quit IRC | 01:28 | |
| *** tjones has joined #openstack-nova | 01:30 | |
| tjones | @melwitt - thanks for your comments on https://review.openstack.org/#/c/127427. i'll address them soon | 01:31 |
| *** heyongli has quit IRC | 01:32 | |
| *** heyongli has joined #openstack-nova | 01:33 | |
| *** su_zhang has quit IRC | 01:37 | |
| *** Jeffrey4l has quit IRC | 01:41 | |
| *** chenzeng has joined #openstack-nova | 01:41 | |
| *** Jeffrey4l has joined #openstack-nova | 01:41 | |
| *** armax has joined #openstack-nova | 01:42 | |
| *** heyongli has quit IRC | 01:42 | |
| *** jwcroppe has quit IRC | 01:43 | |
| *** heyongli has joined #openstack-nova | 01:44 | |
| *** pixelbeat_ has joined #openstack-nova | 01:45 | |
| *** pixelb has quit IRC | 01:46 | |
| openstackgerrit | lvdongbing proposed openstack/nova: Add oslo_config PortOpt support https://review.openstack.org/239569 | 01:48 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova-specs: VMware: placement policies within cluster https://review.openstack.org/237805 | 01:48 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: add flavor extra_spec for DRS vm group https://review.openstack.org/241436 | 01:50 |
| *** smatzek has joined #openstack-nova | 01:51 | |
| *** ssurana has joined #openstack-nova | 01:52 | |
| *** heyongli has quit IRC | 01:53 | |
| *** heyongli has joined #openstack-nova | 01:54 | |
| *** jerrygb has joined #openstack-nova | 02:00 | |
| *** browne has joined #openstack-nova | 02:01 | |
| *** tjones has quit IRC | 02:02 | |
| *** heyongli has quit IRC | 02:03 | |
| *** oomichi has joined #openstack-nova | 02:03 | |
| *** oomichi has quit IRC | 02:03 | |
| *** heyongli has joined #openstack-nova | 02:04 | |
| *** rex_lee_ has joined #openstack-nova | 02:06 | |
| *** electrocucaracha has joined #openstack-nova | 02:09 | |
| *** dims has quit IRC | 02:13 | |
| *** heyongli has quit IRC | 02:13 | |
| *** heyongli has joined #openstack-nova | 02:14 | |
| *** ndipanov has joined #openstack-nova | 02:14 | |
| *** salv-orlando has joined #openstack-nova | 02:16 | |
| *** pratikmallya has joined #openstack-nova | 02:17 | |
| *** huanxie has quit IRC | 02:18 | |
| *** klkumar has joined #openstack-nova | 02:19 | |
| *** gonzalo2kx has joined #openstack-nova | 02:19 | |
| *** heyongli has quit IRC | 02:23 | |
| *** heyongli has joined #openstack-nova | 02:25 | |
| *** xyang1 has quit IRC | 02:28 | |
| *** ndipanov has quit IRC | 02:29 | |
| *** thorst has quit IRC | 02:31 | |
| *** heyongli has quit IRC | 02:34 | |
| *** heyongli has joined #openstack-nova | 02:35 | |
| *** seanw1 has quit IRC | 02:35 | |
| *** seanw1 has joined #openstack-nova | 02:35 | |
| *** salv-orlando has quit IRC | 02:36 | |
| *** unicell has quit IRC | 02:37 | |
| *** flyingtt has quit IRC | 02:38 | |
| *** inteq has joined #openstack-nova | 02:43 | |
| *** markvoelker has joined #openstack-nova | 02:43 | |
| *** heyongli has quit IRC | 02:44 | |
| *** thorst has joined #openstack-nova | 02:45 | |
| *** heyongli has joined #openstack-nova | 02:45 | |
| *** ccarmack has joined #openstack-nova | 02:48 | |
| *** markvoelker has quit IRC | 02:48 | |
| *** thorst has quit IRC | 02:49 | |
| *** dvlwrk_ has joined #openstack-nova | 02:49 | |
| *** thorst has joined #openstack-nova | 02:49 | |
| *** sneti has quit IRC | 02:52 | |
| *** thorst has quit IRC | 02:54 | |
| *** heyongli has quit IRC | 02:54 | |
| *** haomaiwang has joined #openstack-nova | 02:54 | |
| *** heyongli has joined #openstack-nova | 02:55 | |
| *** vilobhmm111 has quit IRC | 02:55 | |
| *** chenzeng has quit IRC | 02:59 | |
| *** dave-mccowan has joined #openstack-nova | 03:00 | |
| *** haomaiwang has quit IRC | 03:01 | |
| *** haomaiwang has joined #openstack-nova | 03:01 | |
| *** smatzek has quit IRC | 03:02 | |
| *** hemna has joined #openstack-nova | 03:03 | |
| *** chenzeng has joined #openstack-nova | 03:04 | |
| *** heyongli has quit IRC | 03:04 | |
| *** hwoarang has quit IRC | 03:05 | |
| *** heyongli has joined #openstack-nova | 03:06 | |
| *** hwoarang has joined #openstack-nova | 03:06 | |
| *** hwoarang has quit IRC | 03:06 | |
| *** hwoarang has joined #openstack-nova | 03:06 | |
| *** haomaiwang has quit IRC | 03:08 | |
| *** haomaiwang has joined #openstack-nova | 03:09 | |
| *** unicell has joined #openstack-nova | 03:11 | |
| *** jdurgin has quit IRC | 03:12 | |
| *** tpeoples has joined #openstack-nova | 03:13 | |
| *** heyongli has quit IRC | 03:15 | |
| *** heyongli has joined #openstack-nova | 03:16 | |
| *** sileht has joined #openstack-nova | 03:19 | |
| *** jmccrory has quit IRC | 03:23 | |
| *** hemna has quit IRC | 03:23 | |
| *** heyongli has quit IRC | 03:25 | |
| *** heyongli has joined #openstack-nova | 03:26 | |
| *** haomaiwang has quit IRC | 03:30 | |
| *** dims has joined #openstack-nova | 03:33 | |
| *** otter768 has joined #openstack-nova | 03:34 | |
| *** RuiChen has quit IRC | 03:34 | |
| *** _RuiChen has joined #openstack-nova | 03:34 | |
| *** heyongli has quit IRC | 03:35 | |
| *** spandhe has quit IRC | 03:36 | |
| *** heyongli has joined #openstack-nova | 03:36 | |
| *** jamielennox is now known as jamielennox|away | 03:38 | |
| *** vilobhmm11 has joined #openstack-nova | 03:38 | |
| *** rfolco has quit IRC | 03:39 | |
| *** jamielennox|away is now known as jamielennox | 03:40 | |
| *** salv-orlando has joined #openstack-nova | 03:43 | |
| *** achanda has joined #openstack-nova | 03:44 | |
| *** heyongli has quit IRC | 03:45 | |
| *** heyongli has joined #openstack-nova | 03:46 | |
| *** pratikmallya has quit IRC | 03:51 | |
| openstackgerrit | Chaoyi Huang proposed openstack/nova-specs: Expose quiece unquiece API https://review.openstack.org/237881 | 03:54 |
| *** vilobhmm11 has quit IRC | 03:54 | |
| *** heyongli has quit IRC | 03:56 | |
| *** heyongli has joined #openstack-nova | 03:57 | |
| *** Jeffrey4l has quit IRC | 03:57 | |
| *** su_zhang has joined #openstack-nova | 03:58 | |
| *** dims has quit IRC | 04:03 | |
| *** salv-orlando has quit IRC | 04:03 | |
| *** ssurana1 has joined #openstack-nova | 04:04 | |
| *** achanda has quit IRC | 04:04 | |
| *** ssurana has quit IRC | 04:06 | |
| *** mgoddard_ has joined #openstack-nova | 04:06 | |
| *** heyongli has quit IRC | 04:06 | |
| *** vishwana_ has joined #openstack-nova | 04:07 | |
| *** heyongli has joined #openstack-nova | 04:07 | |
| *** sballe_ has joined #openstack-nova | 04:07 | |
| *** paul-carlton1 has joined #openstack-nova | 04:09 | |
| *** bapalm_ has joined #openstack-nova | 04:10 | |
| *** Jeffrey4l has joined #openstack-nova | 04:10 | |
| *** kragniz_ has joined #openstack-nova | 04:10 | |
| *** gus_ has joined #openstack-nova | 04:10 | |
| *** hamzy_ has joined #openstack-nova | 04:11 | |
| *** mgoddard has quit IRC | 04:11 | |
| *** RA_ has quit IRC | 04:11 | |
| *** ameade has quit IRC | 04:11 | |
| *** paul-carlton has quit IRC | 04:11 | |
| *** browne has quit IRC | 04:11 | |
| *** adam_g has quit IRC | 04:11 | |
| *** sballe has quit IRC | 04:11 | |
| *** bapalm has quit IRC | 04:11 | |
| *** vishwanathj has quit IRC | 04:11 | |
| *** jhesketh has quit IRC | 04:11 | |
| *** anthonyper has quit IRC | 04:11 | |
| *** rook has quit IRC | 04:11 | |
| *** gus has quit IRC | 04:11 | |
| *** wverdugo500 has quit IRC | 04:11 | |
| *** hamzy has quit IRC | 04:11 | |
| *** kragniz has quit IRC | 04:11 | |
| *** jwcroppe has joined #openstack-nova | 04:11 | |
| *** wverdugo500 has joined #openstack-nova | 04:12 | |
| *** sballe_ is now known as sballe | 04:12 | |
| *** jhesketh has joined #openstack-nova | 04:12 | |
| *** _RuiChen_ has joined #openstack-nova | 04:12 | |
| *** gongysh has quit IRC | 04:13 | |
| *** anthonyper has joined #openstack-nova | 04:14 | |
| *** _RuiChen has quit IRC | 04:15 | |
| *** pixelbeat_ has quit IRC | 04:15 | |
| *** gcb has quit IRC | 04:15 | |
| *** browne has joined #openstack-nova | 04:15 | |
| *** gcb has joined #openstack-nova | 04:16 | |
| *** adam_g has joined #openstack-nova | 04:16 | |
| *** heyongli has quit IRC | 04:16 | |
| *** adam_g has quit IRC | 04:16 | |
| *** adam_g has joined #openstack-nova | 04:16 | |
| *** heyongli has joined #openstack-nova | 04:17 | |
| *** gus_ is now known as gus | 04:18 | |
| *** gongysh has joined #openstack-nova | 04:20 | |
| *** ameade has joined #openstack-nova | 04:21 | |
| *** gongysh has quit IRC | 04:24 | |
| *** chenzeng has quit IRC | 04:25 | |
| *** gongysh has joined #openstack-nova | 04:25 | |
| *** heyongli has quit IRC | 04:26 | |
| *** heyongli has joined #openstack-nova | 04:27 | |
| *** klkumar has quit IRC | 04:28 | |
| *** boris-42 has quit IRC | 04:28 | |
| *** zhangjn has quit IRC | 04:30 | |
| *** zhangjn has joined #openstack-nova | 04:31 | |
| *** zhangjn has quit IRC | 04:31 | |
| *** zhangjn has joined #openstack-nova | 04:32 | |
| *** klkumar has joined #openstack-nova | 04:33 | |
| *** heyongli has quit IRC | 04:37 | |
| *** heyongli has joined #openstack-nova | 04:38 | |
| *** tpeoples has quit IRC | 04:38 | |
| *** dave-mccowan has quit IRC | 04:38 | |
| *** otter768 has quit IRC | 04:41 | |
| *** klkumar has quit IRC | 04:42 | |
| *** greghaynes2 has joined #openstack-nova | 04:43 | |
| *** jmccrory has joined #openstack-nova | 04:44 | |
| *** tpeoples has joined #openstack-nova | 04:44 | |
| *** markvoelker has joined #openstack-nova | 04:44 | |
| *** klkumar has joined #openstack-nova | 04:44 | |
| *** Sree has joined #openstack-nova | 04:46 | |
| *** Sree has quit IRC | 04:46 | |
| *** heyongli has quit IRC | 04:47 | |
| *** Sree has joined #openstack-nova | 04:47 | |
| *** heyongli has joined #openstack-nova | 04:48 | |
| *** markvoelker has quit IRC | 04:49 | |
| *** deepakcs has joined #openstack-nova | 04:51 | |
| *** setuid has quit IRC | 04:52 | |
| *** su_zhang has quit IRC | 04:52 | |
| *** daemontool has quit IRC | 04:53 | |
| *** hparekh has quit IRC | 04:53 | |
| *** hparekh2 has joined #openstack-nova | 04:53 | |
| *** achanda has joined #openstack-nova | 04:54 | |
| *** daemontool has joined #openstack-nova | 04:54 | |
| *** setuid has joined #openstack-nova | 04:57 | |
| *** heyongli has quit IRC | 04:57 | |
| *** heyongli has joined #openstack-nova | 04:58 | |
| *** Sree_ has joined #openstack-nova | 05:01 | |
| *** Sree_ is now known as Guest42639 | 05:01 | |
| *** links has joined #openstack-nova | 05:02 | |
| *** Sree has quit IRC | 05:04 | |
| *** salv-orlando has joined #openstack-nova | 05:04 | |
| *** artom_ has joined #openstack-nova | 05:04 | |
| *** artom has quit IRC | 05:04 | |
| *** mdrabe has joined #openstack-nova | 05:05 | |
| *** Jeffrey4l has quit IRC | 05:06 | |
| *** heyongli has quit IRC | 05:07 | |
| *** heyongli has joined #openstack-nova | 05:08 | |
| *** mdrabe has quit IRC | 05:11 | |
| *** subscope has joined #openstack-nova | 05:13 | |
| *** thorst has joined #openstack-nova | 05:14 | |
| *** dvlwrk_ has quit IRC | 05:15 | |
| *** subscope has quit IRC | 05:16 | |
| *** subscope has joined #openstack-nova | 05:16 | |
| *** Guest42639 has quit IRC | 05:17 | |
| *** Sree has joined #openstack-nova | 05:18 | |
| *** heyongli has quit IRC | 05:18 | |
| *** kragniz_ is now known as kragniz | 05:18 | |
| *** thorst has quit IRC | 05:18 | |
| *** heyongli has joined #openstack-nova | 05:19 | |
| *** Jeffrey4l has joined #openstack-nova | 05:19 | |
| *** ssurana1 has quit IRC | 05:21 | |
| *** changbl has joined #openstack-nova | 05:22 | |
| *** davideagnello has joined #openstack-nova | 05:22 | |
| *** ssurana has joined #openstack-nova | 05:25 | |
| *** jwcroppe has quit IRC | 05:28 | |
| *** heyongli has quit IRC | 05:28 | |
| *** heyongli has joined #openstack-nova | 05:29 | |
| *** salv-orlando has quit IRC | 05:30 | |
| *** greghaynes is now known as greghaynes_ | 05:30 | |
| *** greghaynes2 is now known as greghaynes | 05:30 | |
| *** haomaiwang has joined #openstack-nova | 05:31 | |
| *** dvlwrk_ has joined #openstack-nova | 05:32 | |
| *** ssurana has quit IRC | 05:32 | |
| openstackgerrit | Tang Chen proposed openstack/nova: WIP: Implement status machine for live-migration. https://review.openstack.org/241476 | 05:34 |
| openstackgerrit | Tang Chen proposed openstack/nova: WIP: Implement status machine for resize. https://review.openstack.org/241477 | 05:34 |
| openstackgerrit | Tang Chen proposed openstack/nova: WIP: Implement status machine for evacuation. https://review.openstack.org/241478 | 05:34 |
| openstackgerrit | Tang Chen proposed openstack/nova: WIP: Implement status machine for migration. https://review.openstack.org/241479 | 05:34 |
| openstackgerrit | Tang Chen proposed openstack/nova: WIP: Use status machines to manage the migraiton status. https://review.openstack.org/241480 | 05:34 |
| *** jraju has joined #openstack-nova | 05:36 | |
| *** links has quit IRC | 05:37 | |
| *** heyongli has quit IRC | 05:38 | |
| *** heyongli has joined #openstack-nova | 05:39 | |
| *** shuquan has joined #openstack-nova | 05:45 | |
| *** cheikh_abc has joined #openstack-nova | 05:46 | |
| *** ramishra has quit IRC | 05:46 | |
| cheikh_abc | hello? | 05:46 |
| *** ramishra has joined #openstack-nova | 05:46 | |
| cheikh_abc | nobody here? | 05:47 |
| *** heyongli has quit IRC | 05:48 | |
| *** davideagnello has quit IRC | 05:49 | |
| *** tpeoples has quit IRC | 05:49 | |
| *** heyongli has joined #openstack-nova | 05:49 | |
| *** eliqiao1 has joined #openstack-nova | 05:51 | |
| *** ZZelle_ has joined #openstack-nova | 05:51 | |
| *** eliqiao1 has quit IRC | 05:55 | |
| *** cheikh_abc has left #openstack-nova | 05:56 | |
| *** heyongli has quit IRC | 05:59 | |
| *** vilobhmm11 has joined #openstack-nova | 05:59 | |
| *** heyongli has joined #openstack-nova | 06:00 | |
| *** haomaiwang has quit IRC | 06:01 | |
| *** haomaiwa_ has joined #openstack-nova | 06:01 | |
| *** jerrygb has quit IRC | 06:01 | |
| openstackgerrit | ChangBo Guo(gcb) proposed openstack/nova: Use version convert methods from oslo_utils.versionutils https://review.openstack.org/203322 | 06:02 |
| *** irenab has joined #openstack-nova | 06:03 | |
| *** ZZelle_ has quit IRC | 06:06 | |
| *** heyongli has quit IRC | 06:09 | |
| *** heyongli has joined #openstack-nova | 06:10 | |
| *** eliqiao has joined #openstack-nova | 06:11 | |
| *** cfriesen_ has quit IRC | 06:11 | |
| *** jraju has quit IRC | 06:15 | |
| openstackgerrit | lvdongbing proposed openstack/nova: Add oslo_config PortOpt support https://review.openstack.org/239569 | 06:16 |
| *** heyongli has quit IRC | 06:19 | |
| *** heyongli has joined #openstack-nova | 06:20 | |
| *** electrocucaracha has quit IRC | 06:22 | |
| openstackgerrit | Chaoyi Huang proposed openstack/nova-specs: Expose quiece unquiece API https://review.openstack.org/237881 | 06:23 |
| *** Kevin_Zheng has joined #openstack-nova | 06:24 | |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: vm placement on group of hosts within cluster https://review.openstack.org/210811 | 06:26 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: add flavor extra_spec for vm group https://review.openstack.org/241436 | 06:26 |
| *** heyongli has quit IRC | 06:29 | |
| *** heyongli has joined #openstack-nova | 06:30 | |
| *** shuquan has quit IRC | 06:31 | |
| *** salv-orlando has joined #openstack-nova | 06:31 | |
| *** shuquan has joined #openstack-nova | 06:32 | |
| *** jlvillal has quit IRC | 06:32 | |
| *** jlvillal has joined #openstack-nova | 06:33 | |
| *** gongysh has quit IRC | 06:34 | |
| *** gongysh has joined #openstack-nova | 06:35 | |
| tangchen | cheikh_abc: Hi, I'm here. But I'm not a core developer or reviewer. | 06:36 |
| tangchen | cheikh_abc: If you are looking for someone who can +2 your patch, please wait two more hours. | 06:37 |
| *** unicell1 has joined #openstack-nova | 06:40 | |
| *** heyongli has quit IRC | 06:40 | |
| *** eliqiao has left #openstack-nova | 06:40 | |
| *** shuquan has quit IRC | 06:40 | |
| *** abhishekk has joined #openstack-nova | 06:40 | |
| *** e0ne has joined #openstack-nova | 06:40 | |
| *** heyongli has joined #openstack-nova | 06:41 | |
| *** unicell has quit IRC | 06:41 | |
| *** otter768 has joined #openstack-nova | 06:41 | |
| *** links has joined #openstack-nova | 06:42 | |
| *** subscope has quit IRC | 06:43 | |
| *** chenzeng has joined #openstack-nova | 06:43 | |
| *** baoli has joined #openstack-nova | 06:45 | |
| *** markvoelker has joined #openstack-nova | 06:45 | |
| *** otter768 has quit IRC | 06:46 | |
| *** baoli has quit IRC | 06:47 | |
| *** shuquan has joined #openstack-nova | 06:48 | |
| *** markvoelker has quit IRC | 06:49 | |
| *** e0ne has quit IRC | 06:50 | |
| *** heyongli has quit IRC | 06:50 | |
| *** heyongli has joined #openstack-nova | 06:51 | |
| *** veena has joined #openstack-nova | 06:52 | |
| *** markvoelker has joined #openstack-nova | 07:00 | |
| *** heyongli has quit IRC | 07:00 | |
| *** haomaiwa_ has quit IRC | 07:01 | |
| *** heyongli has joined #openstack-nova | 07:01 | |
| *** haomaiwa_ has joined #openstack-nova | 07:01 | |
| *** jamielennox is now known as jamielennox|away | 07:02 | |
| *** dstepanenko has quit IRC | 07:02 | |
| *** salv-orlando has quit IRC | 07:02 | |
| *** dstepanenko has joined #openstack-nova | 07:02 | |
| *** josecastroleon has joined #openstack-nova | 07:02 | |
| *** lsmola has joined #openstack-nova | 07:03 | |
| *** cristicalin has joined #openstack-nova | 07:04 | |
| *** rajesht has quit IRC | 07:04 | |
| *** markvoelker has quit IRC | 07:05 | |
| *** vilobhmm11 has quit IRC | 07:05 | |
| *** heyongli has quit IRC | 07:10 | |
| *** heyongli has joined #openstack-nova | 07:11 | |
| *** shuquan has quit IRC | 07:13 | |
| *** dstepanenko has quit IRC | 07:14 | |
| *** dstepanenko has joined #openstack-nova | 07:14 | |
| *** nkrinner has joined #openstack-nova | 07:15 | |
| *** thorst has joined #openstack-nova | 07:15 | |
| *** garyk1 has joined #openstack-nova | 07:16 | |
| *** shuquan has joined #openstack-nova | 07:17 | |
| *** thorst has quit IRC | 07:20 | |
| *** heyongli has quit IRC | 07:21 | |
| gibi | andrearosa: sorry, but you just missed me yesterday | 07:22 |
| *** heyongli has joined #openstack-nova | 07:22 | |
| *** gongysh has quit IRC | 07:24 | |
| *** inara` has joined #openstack-nova | 07:24 | |
| *** gongysh has joined #openstack-nova | 07:25 | |
| *** nithyag__ has quit IRC | 07:27 | |
| *** inara has quit IRC | 07:27 | |
| *** sc68cal has quit IRC | 07:27 | |
| *** _cerberus_ has quit IRC | 07:27 | |
| *** jamiec has quit IRC | 07:27 | |
| *** nithyag__ has joined #openstack-nova | 07:27 | |
| *** shz has quit IRC | 07:28 | |
| *** _cerberus_ has joined #openstack-nova | 07:29 | |
| *** heyongli has quit IRC | 07:31 | |
| *** heyongli has joined #openstack-nova | 07:32 | |
| *** chenzeng has quit IRC | 07:32 | |
| *** chenzeng has joined #openstack-nova | 07:32 | |
| *** sc68cal has joined #openstack-nova | 07:34 | |
| *** alexschm has joined #openstack-nova | 07:35 | |
| *** romainh has joined #openstack-nova | 07:36 | |
| *** heyongli has quit IRC | 07:41 | |
| *** heyongli has joined #openstack-nova | 07:42 | |
| *** moshele has joined #openstack-nova | 07:42 | |
| *** jerrygb has joined #openstack-nova | 07:47 | |
| *** mpavone has joined #openstack-nova | 07:49 | |
| *** veena has quit IRC | 07:49 | |
| *** heyongli has quit IRC | 07:51 | |
| *** jerrygb has quit IRC | 07:52 | |
| *** mpavone has quit IRC | 07:52 | |
| *** heyongli has joined #openstack-nova | 07:52 | |
| *** gongysh has quit IRC | 07:58 | |
| *** gongysh has joined #openstack-nova | 07:59 | |
| *** haomaiwa_ has quit IRC | 08:01 | |
| *** Sree_ has joined #openstack-nova | 08:01 | |
| *** Sree_ is now known as Guest86298 | 08:01 | |
| *** haomaiwa_ has joined #openstack-nova | 08:01 | |
| *** salv-orlando has joined #openstack-nova | 08:01 | |
| *** heyongli has quit IRC | 08:02 | |
| *** chenzeng has quit IRC | 08:02 | |
| *** heyongli has joined #openstack-nova | 08:03 | |
| *** Sree has quit IRC | 08:04 | |
| *** shuquan has quit IRC | 08:05 | |
| *** hemnafk has quit IRC | 08:06 | |
| *** hemnafk has joined #openstack-nova | 08:06 | |
| *** achanda has quit IRC | 08:07 | |
| *** hemnafk has quit IRC | 08:07 | |
| *** hemnafk has joined #openstack-nova | 08:07 | |
| *** gonzalo2kx has quit IRC | 08:11 | |
| *** gongysh has quit IRC | 08:11 | |
| *** heyongli has quit IRC | 08:12 | |
| *** heyongli has joined #openstack-nova | 08:13 | |
| bauzas | good morning Nova | 08:13 |
| kashyap | Morning | 08:14 |
| *** salv-orlando has quit IRC | 08:14 | |
| *** salv-orlando has joined #openstack-nova | 08:14 | |
| *** gongysh has joined #openstack-nova | 08:14 | |
| tangchen | Hi | 08:14 |
| tangchen | bauzas: I have post some WIP patches of state machine to Gerrit. | 08:15 |
| tangchen | bauzas: https://review.openstack.org/241476 ~ 241480 | 08:16 |
| bauzas | tangchen: ack, will review them | 08:16 |
| tangchen | bauzas: Thx | 08:16 |
| tangchen | BTW, one small question, do you know if there is any interface in cli or GUI that can modify VM's RTC ? | 08:17 |
| *** thorst has joined #openstack-nova | 08:17 | |
| tangchen | e.g, <timer name='rtc' tickpolicy='catchup'/> | 08:18 |
| tangchen | This is the xml file of libvirt. | 08:18 |
| tangchen | If I want to modify it, any interface in Nova ? | 08:18 |
| *** Guest86298 has quit IRC | 08:18 | |
| *** Sree has joined #openstack-nova | 08:19 | |
| *** rdopiera has joined #openstack-nova | 08:21 | |
| *** thorst has quit IRC | 08:22 | |
| *** heyongli has quit IRC | 08:22 | |
| *** heyongli has joined #openstack-nova | 08:23 | |
| *** subscope has joined #openstack-nova | 08:25 | |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: SameHostFilter should fail if host does not have instances https://review.openstack.org/229030 | 08:26 |
| *** pratikmallya has joined #openstack-nova | 08:27 | |
| openstackgerrit | Alvaro Lopez Garcia proposed openstack/nova: libvirt: make live_migration_uri flag dependent on virt_type https://review.openstack.org/175780 | 08:28 |
| bauzas | tangchen: no, it's not possible using the REST API | 08:28 |
| bauzas | tangchen: since it's really libvirt specific | 08:28 |
| *** markmc has quit IRC | 08:30 | |
| tangchen | bauzas: Then if I want to change the RTC config of a VM, is it doable in openstack ? | 08:30 |
| *** gongysh has quit IRC | 08:31 | |
| *** chenzeng has joined #openstack-nova | 08:31 | |
| *** heyongli has quit IRC | 08:32 | |
| *** heyongli has joined #openstack-nova | 08:33 | |
| *** josecastroleon has quit IRC | 08:34 | |
| *** rotbeard has joined #openstack-nova | 08:36 | |
| bauzas | tangchen: I'd defer that to a libvirt expert like kashyap but I don't feel it's possible using a REST PUT like I said | 08:36 |
| bauzas | of course, you still have virsh for running instances | 08:36 |
| andrearosa | gibi: morning, just a question about the sub-team for notification, as I told you in Tokyo I am interested but at the moment I can't tell you how much time I can invest on it, do you want me to put my name on the etherpad? | 08:37 |
| *** gongysh has joined #openstack-nova | 08:37 | |
| *** ZZelle has quit IRC | 08:38 | |
| *** veena has joined #openstack-nova | 08:38 | |
| *** ZZelle has joined #openstack-nova | 08:38 | |
| *** rk4n has joined #openstack-nova | 08:39 | |
| *** sahid has joined #openstack-nova | 08:42 | |
| *** otter768 has joined #openstack-nova | 08:42 | |
| *** heyongli has quit IRC | 08:43 | |
| *** rotbeard has quit IRC | 08:43 | |
| tangchen | bauzas: Thx. I'll try to ask him. | 08:43 |
| *** rotbeard has joined #openstack-nova | 08:43 | |
| *** heyongli has joined #openstack-nova | 08:44 | |
| *** rk4n has quit IRC | 08:45 | |
| *** otter768 has quit IRC | 08:47 | |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: vm placement on group of hosts within cluster https://review.openstack.org/210811 | 08:48 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: affinity and anti-affinity within cluster https://review.openstack.org/210813 | 08:48 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: anti-affinity placement scheduling for cluster https://review.openstack.org/210812 | 08:48 |
| openstackgerrit | Giridhar Jayavelu proposed openstack/nova: VMware: add flavor extra_spec for vm group https://review.openstack.org/241436 | 08:48 |
| *** josecastroleon has joined #openstack-nova | 08:49 | |
| *** sahid has quit IRC | 08:50 | |
| gibi | andrearosa: I think it is worth putting your name in the etherpad. It is no problem if you will just follow what is happening until you can commit to more | 08:51 |
| andrearosa | gibi: ack | 08:53 |
| *** heyongli has quit IRC | 08:53 | |
| gibi | andrearosa: thanks :) | 08:53 |
| *** gongysh has quit IRC | 08:54 | |
| *** heyongli has joined #openstack-nova | 08:54 | |
| tangchen | kashyap: Hi kashyap, one of my VM's clock is far behind the real clock. And the qemu RTC catchup policy will catch up to the real clock. But the clock runs too fast, and user can feel that. For example, a double click will become a single click. | 08:58 |
| tangchen | kashyap: If I disable the catchup policy in libvirt, it becomes OK. | 08:58 |
| kashyap | tangchen: Hmm, time-keeping is a notorious problem. | 08:59 |
| tangchen | kashyap: So, is there any interface in openstack which is able to control this config ? | 08:59 |
| kashyap | tangchen: Still parsing your question, and looking at libvirt formatdomain page | 08:59 |
| *** e0ne has joined #openstack-nova | 09:00 | |
| *** gongysh has joined #openstack-nova | 09:00 | |
| *** haomaiwa_ has quit IRC | 09:01 | |
| *** markvoelker has joined #openstack-nova | 09:01 | |
| *** haomaiwa_ has joined #openstack-nova | 09:01 | |
| tangchen | kashyap: Thanks. in libvirt xml, it is <timer name='rtc' tickpolicy='catchup'/> | 09:01 |
| kashyap | tangchen: Yeah, found that. Looking through Nova source to see if there's any Nova config | 09:02 |
| kashyap | When danpb is here, I'm sure he'd know for sure -- he fixed a couple of bugs in timers/clocks involving Windows guests, in Nova. | 09:02 |
| kashyap | Hmm, Nova hardcodes the policy to 'catchup': | 09:03 |
| kashyap | nova/virt/libvirt/driver.py: tmrtc.name = "rtc" | 09:03 |
| kashyap | nova/virt/libvirt/driver.py: tmrtc.tickpolicy = "catchup" | 09:03 |
| *** heyongli has quit IRC | 09:03 | |
| *** haomaiwa_ has quit IRC | 09:03 | |
| *** heyongli has joined #openstack-nova | 09:04 | |
| kashyap | tangchen: So, when you say "disable" | 09:04 |
| kashyap | What option did you use? 'discard'? | 09:05 |
| bauzas | kashyap: like I said to tangchen, I feel that as very libvirt specific | 09:05 |
| *** markus_z has joined #openstack-nova | 09:06 | |
| *** markvoelker has quit IRC | 09:06 | |
| kashyap | bauzas: Yeah, timer policy is generated by the libvirt driver. | 09:07 |
| tangchen | kashyap, bauzas: By disable, I mean use 'delay', instead of 'catchup'. | 09:07 |
| tangchen | kashyap, bauzas: There are only two policies here, 'catchup' and 'delay' | 09:08 |
| tangchen | kashyap, bauzas: If we don't specify anything, it is delay by default. | 09:08 |
| *** jistr has joined #openstack-nova | 09:08 | |
| *** ihrachys has joined #openstack-nova | 09:08 | |
| kashyap | tangchen: Yeah the _set_kvm_timers() function | 09:09 |
| kashyap | in driver.py | 09:09 |
| *** haomaiwang has joined #openstack-nova | 09:10 | |
| kashyap | tangchen: The underlying libvirt driver allows more policies though (http://libvirt.org/formatdomain.html#elementsTime), but they're not exposed in Nova. | 09:11 |
| *** haomaiwang has quit IRC | 09:11 | |
| *** sahid has joined #openstack-nova | 09:13 | |
| *** heyongli has quit IRC | 09:13 | |
| *** heyongli has joined #openstack-nova | 09:14 | |
| *** Sree_ has joined #openstack-nova | 09:15 | |
| tangchen | kashyap: I found the comment says: | 09:16 |
| *** Sree_ is now known as Guest42311 | 09:16 | |
| tangchen | # TODO(berrange) One day this should be per-guest | 09:16 |
| tangchen | # OS type configurable | 09:16 |
| kashyap | tangchen: Yep, was about to point that | 09:16 |
| tangchen | kashyap: Do we have any plan for this ? If not, I'd like to post a BP. Of course, before I do, I need to do more investigation. | 09:17 |
| kashyap | tangchen: Go for it, I don't know of any existing plans. | 09:17 |
| kashyap | tangchen: Maybe write a small note to the upstream -dev list? | 09:18 |
| tangchen | kashyap: Sure. Thx for the help. :) | 09:18 |
| openstackgerrit | Zhenyu Zheng proposed openstack/nova: Add project-id and user-id when list server-groups https://review.openstack.org/209018 | 09:18 |
| kashyap | Heh, I did nothing. I just know time-keeping is a hard problem others smarter than me are solving it :-) | 09:18 |
| *** Sree has quit IRC | 09:18 | |
| *** pixelbeat_ has joined #openstack-nova | 09:19 | |
| *** klkumar has quit IRC | 09:20 | |
| kashyap | Writing to the list would be best, as that'll let others who're not on IRC know that there's such a thing in the works. | 09:20 |
| *** danpb has joined #openstack-nova | 09:20 | |
| *** aloga has quit IRC | 09:20 | |
| *** flyingtt_ has joined #openstack-nova | 09:21 | |
| *** yassine__ has joined #openstack-nova | 09:21 | |
| *** aloga has joined #openstack-nova | 09:22 | |
| *** klkumar has joined #openstack-nova | 09:23 | |
| *** heyongli has quit IRC | 09:24 | |
| *** heyongli has joined #openstack-nova | 09:25 | |
| *** pratikmallya has quit IRC | 09:28 | |
| openstackgerrit | garyk proposed openstack/nova-specs: VMware: expand opaque network support https://review.openstack.org/230425 | 09:28 |
| *** alex_klimov has joined #openstack-nova | 09:29 | |
| *** jwcroppe has joined #openstack-nova | 09:31 | |
| *** LukeHinds has joined #openstack-nova | 09:34 | |
| *** heyongli has quit IRC | 09:34 | |
| tangchen | kashyap: Sorry, where is the dev list ? | 09:34 |
| kashyap | tangchen: You posted to it previously -- openstack-dev mailing list | 09:35 |
| *** heyongli has joined #openstack-nova | 09:35 | |
| kashyap | http://lists.openstack.org/pipermail/openstack-dev/ | 09:35 |
| tangchen | kashyap: Oh, you mean the email list. OK. :) | 09:35 |
| kashyap | Yeah | 09:35 |
| tangchen | One more thing that has been discussed before, nova wants to desperate host maintenance mode. | 09:36 |
| tangchen | I was going to post a BP to do that. But now, there is only one user of host maintenance mode, that is Xen. | 09:37 |
| kashyap | danpb: When you get a moment later, a question on time-keeping in Nova (from tangchen): "one of my VM's clock is far behind the real clock. And the qemu RTC catchup policy will catch up to the real clock. But the clock runs too fast, and user can feel that. For example, a double click will become a single click. If I disable the catchup policy in libvirt, it becomes OK. By disable, I mean use 'del | 09:37 |
| kashyap | ay', instead of 'catchup'". | 09:37 |
| kashyap | Some back and forth on that here: http://eavesdrop.openstack.org/irclogs/%23openstack-nova/%23openstack-nova.2015-11-04.log.html#t2015-11-04T08:58:17 | 09:37 |
| *** salv-orlando has quit IRC | 09:37 | |
| *** browne has quit IRC | 09:37 | |
| tangchen | If we desperate host maintenance mode, what should we do about the Xen code ? | 09:37 |
| bauzas | deprecating it | 09:38 |
| tangchen | Reading the Xen code, I guess they want to implement their own migration using xen server's pool-migration functionality. | 09:38 |
| danpb | kashyap: really need to know more about the guest os in question | 09:39 |
| kashyap | tangchen: ^ | 09:39 |
| *** salv-orlando has joined #openstack-nova | 09:39 | |
| danpb | i'm not convinced we need to change the clock timer modes for any mainstream guest OS | 09:39 |
| kashyap | He wanted to work on the TODO item you noted in driver.py _set_kvm_timers() function: | 09:39 |
| kashyap | "One day this should be per-guest OS type configurable" | 09:39 |
| danpb | the settings we have by default should work for any mainstream OS in general. we support custom settings for windows now assuming the image is tagged correctly | 09:40 |
| kashyap | Yep, I'm faimilar your work on hyper-v timer enlightenments for windows guests. | 09:41 |
| danpb | so the only real time I'd expect we'd need to change is for dealing with old legacy OSs or special purpose embedded OS | 09:41 |
| danpb | which might have very specific timing keep needs due to limited or outdated hardware support they might have | 09:41 |
| danpb | the intent is that we'll make use of libosinfo for this - ie have a section in libosinfo database for recording timer settings for OS | 09:42 |
| danpb | so we're really waiting for the libosinfo work to land before persuing it further | 09:42 |
| *** heyongli has quit IRC | 09:44 | |
| *** pratikmallya has joined #openstack-nova | 09:45 | |
| *** heyongli has joined #openstack-nova | 09:45 | |
| kashyap | Ah, I see. Thanks for the details. | 09:46 |
| kashyap | tangchen: Please take a look at the above details from Dan. | 09:47 |
| *** aix has joined #openstack-nova | 09:48 | |
| danpb | i'd still be interested to know what guest OS is seeing trouble with current settings | 09:48 |
| *** markmc has joined #openstack-nova | 09:48 | |
| *** salv-orlando has quit IRC | 09:50 | |
| *** jgriffith has quit IRC | 09:50 | |
| *** salv-orlando has joined #openstack-nova | 09:51 | |
| *** jgriffith has joined #openstack-nova | 09:51 | |
| *** jgriffith is now known as Guest90242 | 09:51 | |
| *** dvlwrk_ has quit IRC | 09:54 | |
| *** alex_xu has quit IRC | 09:54 | |
| *** eglynn has joined #openstack-nova | 09:54 | |
| *** heyongli has quit IRC | 09:54 | |
| *** jamielennox|away is now known as jamielennox | 09:55 | |
| *** derekh has joined #openstack-nova | 09:55 | |
| *** heyongli has joined #openstack-nova | 09:55 | |
| tangchen | kashyap: yes, reading. | 09:56 |
| *** veena has quit IRC | 09:56 | |
| tangchen | danpb: Well, I'm not sure if I have found the real cause of my problem. | 09:56 |
| *** MarcF has joined #openstack-nova | 09:56 | |
| tangchen | danpb: One of my VM (windows7 64bit), the clock is far behind the real clock. | 09:57 |
| *** alex_xu has joined #openstack-nova | 09:57 | |
| tangchen | Then 'catchup' policy will make the clock run faster, which made the double click became single click. | 09:57 |
| tangchen | I'd like to disable 'catchup' | 09:58 |
| danpb | tangchen: was the disk image for the windows VM tagged with os_type=windows ? | 09:59 |
| danpb | tangchen: as doing that enables the hyperv clock & other hyperv extensions which provides a far more reliable paravirtualized clock source | 10:00 |
| danpb | at which point, IIUC, the other clock settings become pretty much irrelevant | 10:00 |
| tangchen | danpb: The tag is in xml file ? | 10:01 |
| *** zhenguo has quit IRC | 10:03 | |
| tangchen | danpb: I don't have such a tag in my xmlfile. os_type=windows | 10:03 |
| tangchen | danpb: I'll try it. Thx. | 10:04 |
| kashyap | I think it's a Glance image metadata property | 10:04 |
| *** heyongli has quit IRC | 10:05 | |
| kashyap | tangchen: Look up "os_type" here: http://docs.openstack.org/cli-reference/content/chapter_cli-glance-property.html | 10:05 |
| *** pratikmallya has quit IRC | 10:05 | |
| tangchen | kashyap: Ah, OK. I'll take a look. Thx. | 10:06 |
| *** heyongli has joined #openstack-nova | 10:06 | |
| tangchen | kashyap, danpb, bauzas: Thank you for your nice help. It is a little later here, and I have to go. See you tomorrow. Thx. :) | 10:07 |
| openstackgerrit | Daniel Berrange proposed openstack/nova-specs: VIF port config versioned objects and driver plugin library https://review.openstack.org/193668 | 10:10 |
| danpb | tangchen: no, in the glance image | 10:10 |
| *** heyongli has quit IRC | 10:15 | |
| *** MarcF has quit IRC | 10:15 | |
| *** heyongli has joined #openstack-nova | 10:16 | |
| *** rk4n has joined #openstack-nova | 10:22 | |
| *** alex_xu has quit IRC | 10:23 | |
| *** alex_xu has joined #openstack-nova | 10:24 | |
| *** jichen has joined #openstack-nova | 10:24 | |
| *** heyongli has quit IRC | 10:25 | |
| *** heyongli has joined #openstack-nova | 10:26 | |
| *** moshele has quit IRC | 10:32 | |
| *** heyongli has quit IRC | 10:35 | |
| *** heyongli has joined #openstack-nova | 10:36 | |
| *** salv-orlando has quit IRC | 10:37 | |
| openstackgerrit | Balazs Gibizer proposed openstack/nova-specs: Versioned notification API https://review.openstack.org/224755 | 10:40 |
| openstackgerrit | Tomi Juvonen proposed openstack/nova: Add host_status attribute for servers/detail and servers/{server_id} https://review.openstack.org/241562 | 10:42 |
| *** otter768 has joined #openstack-nova | 10:43 | |
| *** dvlwrk_ has joined #openstack-nova | 10:44 | |
| *** heyongli has quit IRC | 10:46 | |
| *** heyongli has joined #openstack-nova | 10:47 | |
| *** otter768 has quit IRC | 10:48 | |
| *** jistr_ has joined #openstack-nova | 10:48 | |
| *** ihrachys_ has joined #openstack-nova | 10:48 | |
| *** jistr has quit IRC | 10:48 | |
| *** ihrachys has quit IRC | 10:48 | |
| *** ihrachys has joined #openstack-nova | 10:49 | |
| *** subscope has quit IRC | 10:50 | |
| *** daemontool has quit IRC | 10:51 | |
| *** jistr_ has quit IRC | 10:53 | |
| *** ihrachys_ has quit IRC | 10:53 | |
| *** Kennan has quit IRC | 10:53 | |
| *** khomkrit has joined #openstack-nova | 10:53 | |
| openstackgerrit | Tomi Juvonen proposed openstack/nova: Add host_status attribute for servers/detail and servers/{server_id} https://review.openstack.org/241562 | 10:53 |
| *** heyongli has quit IRC | 10:56 | |
| *** Kennan has joined #openstack-nova | 10:56 | |
| *** heyongli has joined #openstack-nova | 10:57 | |
| *** veena has joined #openstack-nova | 10:57 | |
| *** rotbeard has quit IRC | 10:58 | |
| *** markvoelker has joined #openstack-nova | 11:02 | |
| * alex_xu always learned knowledge from irc | 11:04 | |
| *** jistr_ has joined #openstack-nova | 11:05 | |
| *** markvoelker has quit IRC | 11:06 | |
| *** heyongli has quit IRC | 11:06 | |
| *** heyongli has joined #openstack-nova | 11:07 | |
| *** Guest42311 has quit IRC | 11:08 | |
| *** mmdurrant has quit IRC | 11:09 | |
| *** romainh has left #openstack-nova | 11:14 | |
| *** heyongli has quit IRC | 11:16 | |
| *** heyongli has joined #openstack-nova | 11:17 | |
| mdbooth | alaski: If you're awake yet, do you have a complete command line for this bug: https://bugs.launchpad.net/nova/+bug/1501851 . I get "Mapping image to local is not supported" when trying to do this. | 11:18 |
| openstack | Launchpad bug 1501851 in OpenStack Compute (nova) "Nova can incorrectly think an instance is volume backed" [Low,In progress] - Assigned to Nimish Joshi (jnimish77) | 11:18 |
| *** rotbeard has joined #openstack-nova | 11:18 | |
| *** gszasz has joined #openstack-nova | 11:21 | |
| *** drankis has joined #openstack-nova | 11:23 | |
| johnthetubaguy | mdbooth: not sure, but I think he is back from paternity tomorrow | 11:25 |
| mdbooth | Fun times :) | 11:25 |
| mdbooth | If anybody else happens to know what that bug might be about that would be useful | 11:25 |
| *** heyongli has quit IRC | 11:27 | |
| *** heyongli has joined #openstack-nova | 11:27 | |
| *** pratikmallya has joined #openstack-nova | 11:30 | |
| *** deepakcs has quit IRC | 11:30 | |
| *** josecastroleon has quit IRC | 11:31 | |
| *** moshele has joined #openstack-nova | 11:34 | |
| openstackgerrit | Radomir Dopieralski proposed openstack/nova: Make the create_instance call accept security group ids as well as names https://review.openstack.org/217292 | 11:35 |
| johnthetubaguy | mdbooth: hmm, interesting, I see what you mean, unsure how that bug is hit given the check you are hitting | 11:35 |
| mdbooth | johnthetubaguy: It's not desperately important, but if I'd been able to reproduce easily it looked like a good example bug | 11:36 |
| *** heyongli has quit IRC | 11:37 | |
| *** heyongli has joined #openstack-nova | 11:38 | |
| *** mwagner has quit IRC | 11:40 | |
| johnthetubaguy | mdbooth: yeah, totally | 11:40 |
| *** salv-orlando has joined #openstack-nova | 11:41 | |
| *** aix has quit IRC | 11:41 | |
| mdbooth | johnthetubaguy: While I'm at it, this bug: https://bugs.launchpad.net/nova/+bug/1412993 | 11:42 |
| openstack | Launchpad bug 1412993 in OpenStack Compute (nova) "Nova resize for boot-from-volume instance does not resize volume" [Undecided,Confirmed] - Assigned to Claudiu Belu (cbelu) | 11:42 |
| mdbooth | I read this and had the same thoughts as both you and pmurray | 11:42 |
| mdbooth | What's the procedure for closing this without a resolution? | 11:43 |
| mdbooth | In bugzilla this would be CLOSED NOTABUG | 11:43 |
| johnthetubaguy | mdbooth: good call out, yeah, lets mark this as invalid | 11:43 |
| mdbooth | johnthetubaguy: Spotted, and done. | 11:44 |
| mdbooth | Thanks | 11:44 |
| johnthetubaguy | mdbooth: no problems, thanks for digging | 11:44 |
| *** aysyd has joined #openstack-nova | 11:45 | |
| *** salv-orlando has quit IRC | 11:46 | |
| *** heyongli has quit IRC | 11:47 | |
| *** subscope has joined #openstack-nova | 11:48 | |
| *** heyongli has joined #openstack-nova | 11:48 | |
| *** markmc has quit IRC | 11:49 | |
| *** markmc has joined #openstack-nova | 11:49 | |
| *** pratikmallya has quit IRC | 11:50 | |
| *** sahid has quit IRC | 11:51 | |
| *** Kennan has quit IRC | 11:51 | |
| *** dvlwrk_ has quit IRC | 11:53 | |
| *** Kennan has joined #openstack-nova | 11:54 | |
| *** jaypipes has joined #openstack-nova | 11:55 | |
| sdague | morning folks | 11:57 |
| *** heyongli has quit IRC | 11:57 | |
| *** heyongli has joined #openstack-nova | 11:58 | |
| *** markus_z has quit IRC | 11:59 | |
| *** thorst has joined #openstack-nova | 12:00 | |
| *** jwcroppe has quit IRC | 12:01 | |
| *** rfolco has joined #openstack-nova | 12:01 | |
| *** jwcroppe has joined #openstack-nova | 12:02 | |
| *** zhangjn has quit IRC | 12:02 | |
| *** ferest_ has joined #openstack-nova | 12:03 | |
| *** daemontool has joined #openstack-nova | 12:03 | |
| alex_xu | sdague: morning | 12:04 |
| *** thorst has quit IRC | 12:04 | |
| *** jwcroppe has quit IRC | 12:06 | |
| *** salv-orlando has joined #openstack-nova | 12:07 | |
| *** ferest_ has quit IRC | 12:07 | |
| *** heyongli has quit IRC | 12:08 | |
| *** ericksonsantos has quit IRC | 12:08 | |
| *** heyongli has joined #openstack-nova | 12:08 | |
| *** ericksonsantos has joined #openstack-nova | 12:09 | |
| *** veena has quit IRC | 12:09 | |
| *** veena has joined #openstack-nova | 12:11 | |
| jaypipes | sdague: mornin. | 12:12 |
| jaypipes | alex_xu: mornin :) | 12:13 |
| jaypipes | alex_xu: or evening :) | 12:13 |
| alex_xu | jaypipes: :) | 12:13 |
| *** aix has joined #openstack-nova | 12:13 | |
| *** wasmum has quit IRC | 12:14 | |
| *** markvoelker has joined #openstack-nova | 12:17 | |
| *** dims has joined #openstack-nova | 12:18 | |
| *** heyongli has quit IRC | 12:18 | |
| *** heyongli has joined #openstack-nova | 12:19 | |
| johnthetubaguy | danpb: jaypipes: is there any chance we could chat about this spec after lunch: https://review.openstack.org/#/c/190917/7/specs/mitaka/approved/nova-neutron-binding-negotiation.rst,cm | 12:19 |
| johnthetubaguy | danpb: jaypipes: it feels like the list of "preferred_binding_types" should live is OS VIF Lib, but I don't really know that code well enough | 12:20 |
| danpb | johnthetubaguy: sure, my feeling is that it is really just part of the os-vif work | 12:20 |
| johnthetubaguy | danpb: I guess the question is do we add it now, and port it to os vif lib afterwards, or do we do it in os vif lib to start with? | 12:21 |
| *** Kennan has quit IRC | 12:22 | |
| *** juzuluag has joined #openstack-nova | 12:22 | |
| *** markvoelker has quit IRC | 12:22 | |
| jaypipes | johnthetubaguy: sure thing. so, in about 30 minutes? | 12:22 |
| *** thorst has joined #openstack-nova | 12:23 | |
| johnthetubaguy | jaypipes: that works for me, I should run off and get something to eat, so quick chat at 13.00UTC | 12:23 |
| jaypipes | sure, sounds good. I'll read that spec in the meantime. | 12:23 |
| *** Kennan has joined #openstack-nova | 12:25 | |
| *** heyongli has quit IRC | 12:28 | |
| *** heyongli has joined #openstack-nova | 12:29 | |
| *** EinstCrazy has joined #openstack-nova | 12:30 | |
| *** heyongli has quit IRC | 12:38 | |
| *** A110602877_ has joined #openstack-nova | 12:39 | |
| *** heyongli has joined #openstack-nova | 12:39 | |
| *** chenzeng has quit IRC | 12:39 | |
| *** josecastroleon has joined #openstack-nova | 12:40 | |
| *** achanda has joined #openstack-nova | 12:40 | |
| sdague | http://tinyurl.com/p6l4lvd - the data on the grenade multinode seems pretty solid so far | 12:41 |
| *** haomaiwang has joined #openstack-nova | 12:41 | |
| *** achanda has quit IRC | 12:42 | |
| *** dims has quit IRC | 12:43 | |
| *** jistr_ is now known as jistr | 12:44 | |
| *** otter768 has joined #openstack-nova | 12:44 | |
| *** klkumar has quit IRC | 12:45 | |
| *** moshele has quit IRC | 12:46 | |
| *** dave-mccowan has joined #openstack-nova | 12:47 | |
| *** jamielennox is now known as jamielennox|away | 12:48 | |
| *** heyongli has quit IRC | 12:49 | |
| *** klkumar has joined #openstack-nova | 12:49 | |
| *** otter768 has quit IRC | 12:49 | |
| *** heyongli has joined #openstack-nova | 12:49 | |
| *** wasmum has joined #openstack-nova | 12:50 | |
| *** ferest has joined #openstack-nova | 12:50 | |
| *** baoli has joined #openstack-nova | 12:53 | |
| *** baoli has quit IRC | 12:53 | |
| *** danpb has quit IRC | 12:55 | |
| *** baoli has joined #openstack-nova | 12:56 | |
| *** danpb has joined #openstack-nova | 12:56 | |
| *** daemontool has quit IRC | 12:57 | |
| *** ihrachys has quit IRC | 12:57 | |
| *** daemontool has joined #openstack-nova | 12:57 | |
| *** ihrachys has joined #openstack-nova | 12:58 | |
| alaski | mdbooth: just here for a second, but it appears that the --image flag is needed as well in that case http://git.openstack.org/cgit/openstack/nova/tree/nova/block_device.py#n205 | 12:58 |
| *** heyongli has quit IRC | 12:59 | |
| *** jlanoux has joined #openstack-nova | 13:00 | |
| *** inara` has quit IRC | 13:00 | |
| *** heyongli has joined #openstack-nova | 13:00 | |
| *** ihrachys has quit IRC | 13:00 | |
| johnthetubaguy | sdague: those grenade multi-node things are looking very cool :) | 13:00 |
| *** haomaiwang has quit IRC | 13:01 | |
| johnthetubaguy | danpb: jaypipes: no is good for a quick chat, if you are free | 13:01 |
| johnthetubaguy | doh, now | 13:01 |
| *** haomaiwang has joined #openstack-nova | 13:01 | |
| danpb | ok | 13:03 |
| *** dvlwrk_ has joined #openstack-nova | 13:04 | |
| johnthetubaguy | just wondering what we want to do with that vif negotiation spec | 13:05 |
| johnthetubaguy | do we just let that happen in paralell with os vif lib to start with? | 13:05 |
| *** pratikmallya has joined #openstack-nova | 13:06 | |
| danpb | so the os vif lib will need negotiation support | 13:06 |
| danpb | i guess my concern with doing it in parallel is whether we'd implement negotiation once, then find it doesn't do what we need and then have to do it again | 13:07 |
| danpb | and end up having to support 2 negotiation mechanisms :-( | 13:07 |
| johnthetubaguy | yeah, that seems like a valid concern to me | 13:07 |
| danpb | that's not neccessarily to say we shouldn't approve the spec | 13:08 |
| danpb | it could just be something we resolve during impl time | 13:08 |
| johnthetubaguy | well, yes, but I think they have code they want to merge now | 13:09 |
| *** dims has joined #openstack-nova | 13:09 | |
| *** heyongli has quit IRC | 13:09 | |
| johnthetubaguy | I think we are saying, lets add the API using the OS VIF lib object-ey approach, to start with, so we don't have to do it twice, and that sounds good to me | 13:09 |
| *** heyongli has joined #openstack-nova | 13:10 | |
| jaypipes | johnthetubaguy, danpb: I have some concerns on this spec. Still adding feedback to it... | 13:10 |
| *** klkumar has quit IRC | 13:11 | |
| danpb | johnthetubaguy: yeah, that's pretty much exactly what i'm saying | 13:11 |
| johnthetubaguy | jaypipes: so I was wondering if we wanted a more a set of "hits" like interface, rather than a list of supported things | 13:11 |
| johnthetubaguy | danpb: cool, just wanted to make sure I am understanding you correctly | 13:11 |
| danpb | i guess ian wells' concern is that if os-vif doesn't get merged in M, then his spec misses a cycle again | 13:11 |
| danpb | which is a valid concern, but I hope we've mitigated against failure by making os-vif a priority | 13:12 |
| johnthetubaguy | danpb: +1 | 13:12 |
| *** mwagner has joined #openstack-nova | 13:16 | |
| *** abhishekk has quit IRC | 13:17 | |
| *** ferest has quit IRC | 13:17 | |
| *** salv-orlando has quit IRC | 13:19 | |
| *** heyongli has quit IRC | 13:19 | |
| jaypipes | johnthetubaguy, danpb: ok, reviewed. please take a look at my comments and let me know if I'm crazy. | 13:20 |
| *** ferest has joined #openstack-nova | 13:20 | |
| *** dims has quit IRC | 13:20 | |
| *** heyongli has joined #openstack-nova | 13:20 | |
| *** jwcroppe has joined #openstack-nova | 13:22 | |
| *** inteq has quit IRC | 13:23 | |
| *** markvoelker has joined #openstack-nova | 13:23 | |
| *** baoli_ has joined #openstack-nova | 13:23 | |
| johnthetubaguy | jaypipes: your comment on ln 135, I has assumed neutron needed to do something different for each case, so tells nova which one it has chosen, but you make a very good point, we probably need some kind of retry I guess? | 13:24 |
| *** baoli has quit IRC | 13:26 | |
| *** pratikmallya has quit IRC | 13:26 | |
| *** heyongli has quit IRC | 13:30 | |
| *** heyongli has joined #openstack-nova | 13:31 | |
| *** su_zhang has joined #openstack-nova | 13:33 | |
| *** edmondsw has joined #openstack-nova | 13:34 | |
| *** markus_z has joined #openstack-nova | 13:35 | |
| *** alex_klimov has quit IRC | 13:37 | |
| *** heyongli has quit IRC | 13:40 | |
| *** links has quit IRC | 13:41 | |
| *** heyongli has joined #openstack-nova | 13:41 | |
| *** dvlwrk_ has quit IRC | 13:43 | |
| *** salv-orlando has joined #openstack-nova | 13:43 | |
| *** heyongli has quit IRC | 13:50 | |
| *** heyongli has joined #openstack-nova | 13:51 | |
| *** rlrossit has joined #openstack-nova | 13:52 | |
| *** Sree has joined #openstack-nova | 13:55 | |
| *** jwcroppe has quit IRC | 13:55 | |
| *** dave-mccowan has quit IRC | 13:56 | |
| *** salv-orlando has quit IRC | 13:58 | |
| *** juzuluag has quit IRC | 13:59 | |
| *** ctrath has joined #openstack-nova | 14:00 | |
| *** mdrabe has joined #openstack-nova | 14:00 | |
| *** heyongli has quit IRC | 14:00 | |
| *** haomaiwang has quit IRC | 14:01 | |
| *** haomaiwa_ has joined #openstack-nova | 14:01 | |
| *** heyongli has joined #openstack-nova | 14:01 | |
| *** eglynn has quit IRC | 14:02 | |
| *** ijuwang has joined #openstack-nova | 14:02 | |
| *** mriedem has joined #openstack-nova | 14:02 | |
| *** Piet has quit IRC | 14:02 | |
| *** xyang1 has joined #openstack-nova | 14:02 | |
| *** burgerk has joined #openstack-nova | 14:03 | |
| *** dustins has joined #openstack-nova | 14:06 | |
| *** otter768 has joined #openstack-nova | 14:06 | |
| *** otter768 has quit IRC | 14:07 | |
| *** otter768 has joined #openstack-nova | 14:08 | |
| garyk1 | mriedem: johnthetubaguy: can you please see https://review.openstack.org/#/c/230425/ | 14:08 |
| *** heyongli has quit IRC | 14:11 | |
| *** mriedem has quit IRC | 14:11 | |
| *** heyongli has joined #openstack-nova | 14:11 | |
| johnthetubaguy | garyk1: thanks for getting a good group of reviewers on that, will take a look | 14:12 |
| *** otter768 has quit IRC | 14:12 | |
| garyk1 | np | 14:12 |
| *** ferest has quit IRC | 14:13 | |
| johnthetubaguy | garyk1: while I am thinking about it, once libvirt has adopted OS VIF Lib, that should help you folks with adding these kinds of things? I guess? (BTW, explicitly not blocking folks adding VIF things while we get that OS VIF lib stuff in place) | 14:13 |
| *** mriedem has joined #openstack-nova | 14:14 | |
| *** moshele has joined #openstack-nova | 14:14 | |
| *** dave-mccowan has joined #openstack-nova | 14:14 | |
| *** mc_nair has joined #openstack-nova | 14:15 | |
| garyk1 | johnthetubaguy: i am not sure that that will help. this only uses the vif type. so it is orthogonal. | 14:15 |
| garyk1 | i have a ton of concerns with the vif type lib. the main one is that if it is a external project to nova. getting the version updates into nova will be hard and also vendor specific | 14:16 |
| garyk1 | that will also require all kinds of tweaks to libvirt which at the moment is very tricky | 14:16 |
| *** veena_ has joined #openstack-nova | 14:17 | |
| *** links has joined #openstack-nova | 14:17 | |
| garyk1 | for example the request earlier today for th eopencontrail driver for libvirt (i think that was the name) | 14:17 |
| *** veena has quit IRC | 14:17 | |
| garyk1 | i am also not sure if and when the neutron community will make these changes in the plugins | 14:17 |
| garyk1 | so we will need at least a cycle or two to ensure support of existing and working neutron plugins | 14:18 |
| garyk1 | sorry for being the party pooper | 14:18 |
| johnthetubaguy | there are a few bits to the os vif lib thing, its totally going to take a cycle or two, agreed there | 14:18 |
| garyk1 | i would take the approach of doing it in nova and when it is working in nova spin it out to a seperate repo | 14:19 |
| johnthetubaguy | I think there is a whole class of things that go only into os vif lib, rather than in the core of Nova, so it should be able to move faster | 14:19 |
| openstackgerrit | Gábor Antal proposed openstack/nova: Replaced deprecated timeutils methods (isotime() and strtime()) https://review.openstack.org/241179 | 14:19 |
| *** jichen has quit IRC | 14:19 | |
| *** burgerk has quit IRC | 14:19 | |
| garyk1 | having it in a seperate repo will have the additional overhead of requirements project etc. | 14:19 |
| garyk1 | once it is in nova and it 'compiles' we can start to see how neutron can consume and pass the data to nova | 14:20 |
| johnthetubaguy | yes, I think the idea is to release every week or so, like oslo, to try make it quick to get updates | 14:20 |
| *** READ10 has joined #openstack-nova | 14:20 | |
| *** krtaylor has quit IRC | 14:20 | |
| johnthetubaguy | so, it totally depends where the line is drawn, but gut says it looks good, but I will be more certain one I see the patches up I guess | 14:21 |
| *** heyongli has quit IRC | 14:21 | |
| *** romainh has joined #openstack-nova | 14:21 | |
| sdague | mriedem: you respinning this today - https://review.openstack.org/#/c/225199/ ? | 14:22 |
| *** ctrath has quit IRC | 14:22 | |
| mriedem | sdague: i can | 14:22 |
| *** heyongli has joined #openstack-nova | 14:22 | |
| mriedem | sdague: are you around today? | 14:22 |
| sdague | mriedem: I am | 14:22 |
| mriedem | can we get this done? https://review.openstack.org/#/c/240020/ | 14:22 |
| sdague | mostly un jetlagged | 14:22 |
| * mriedem throws sdague a scooby snack | 14:22 | |
| mriedem | i want to get that novaclient release out before something else is merged | 14:23 |
| sdague | mriedem: yes, though we should probably get dhellmann or dims to do the thing | 14:23 |
| sdague | because the actual release mechanics are still a thing that I've not done in the new model yet | 14:23 |
| *** klkumar has joined #openstack-nova | 14:23 | |
| *** Sree has quit IRC | 14:23 | |
| mriedem | i'm just poking you b/c i poked in the release channel yesterday | 14:23 |
| mriedem | but you have +2 power | 14:24 |
| mriedem | which helps with the pokey | 14:24 |
| mriedem | garyk1: https://review.openstack.org/#/c/230425/ is ready to go | 14:24 |
| *** mrkz has joined #openstack-nova | 14:25 | |
| mriedem | johnthetubaguy: the bp is approved, want to drop your -2? https://review.openstack.org/#/c/165750/ | 14:25 |
| garyk1 | mriedem: johnthetubaguy: thanks! | 14:25 |
| *** markus_z has quit IRC | 14:27 | |
| *** ccarmack has quit IRC | 14:28 | |
| sdague | mriedem: right, except, the +2 doesn't really do a thing. Basically it's a 1 +2 repo, that after you land the change you do the manual thing to do the release | 14:28 |
| *** jerrygb has joined #openstack-nova | 14:28 | |
| mriedem | i know i know, but it's symbolic | 14:29 |
| mriedem | sometimes they want ack from nova peeps | 14:29 |
| mriedem | on a nova release | 14:29 |
| mriedem | even though i'm a nova release guy | 14:29 |
| mriedem | garyk1: is the opaque network thing considered a high priority for the vmware subteam in mitaka? i'm assuming yes | 14:31 |
| *** heyongli has quit IRC | 14:31 | |
| mriedem | i'm wondering if subteam high priority + overall project low priority = medium priority in the bp in lp | 14:31 |
| garyk1 | mriedem: yes, it is high priority. it is a huge pain point for us as the plugin was released in liberty in neutron. | 14:31 |
| *** heyongli has joined #openstack-nova | 14:32 | |
| *** su_zhang has quit IRC | 14:33 | |
| garyk1 | mriedem: i also need to add a few patches following the comments in the spec. | 14:33 |
| *** krtaylor has joined #openstack-nova | 14:33 | |
| mriedem | yeah, just noted some of that in the code review | 14:33 |
| garyk1 | mriedem: yes, just saw that. thanks | 14:34 |
| *** ihrachys has joined #openstack-nova | 14:34 | |
| *** lbelivea has joined #openstack-nova | 14:34 | |
| mriedem | sdague: this is a logging only change to help figure out a gate issue for the cells job, has a +2 on it https://review.openstack.org/#/c/241366/ | 14:35 |
| sdague | mriedem: +A | 14:35 |
| *** ccarmack has joined #openstack-nova | 14:35 | |
| *** Jeffrey4l has quit IRC | 14:35 | |
| mriedem | zzzeek: does a sqlalchemy result object from a statement exec (insert or update) provide some level of detail to tell us if we actually did an insert or update? | 14:36 |
| mriedem | like here https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L3935 | 14:36 |
| *** pece has joined #openstack-nova | 14:37 | |
| sdague | mriedem: if you turn on debug logging for sqla it will show exactly the sql statement getting executed | 14:37 |
| sdague | we turned it off because it was so verbose | 14:37 |
| *** moshele has quit IRC | 14:38 | |
| mriedem | i have a change at the end of the series which creates a unique constraint on the bdm teable | 14:38 |
| mriedem | *table | 14:38 |
| mriedem | so we'll literally blow up when we try a duplicate insert | 14:38 |
| mriedem | just need to recheck until we hit that | 14:38 |
| *** moshele has joined #openstack-nova | 14:38 | |
| mriedem | i have some other ideas about avoiding the calls to block_device_mapping_update_or_create altogether | 14:39 |
| mriedem | since it seems hokey | 14:39 |
| *** heyongli has quit IRC | 14:41 | |
| mriedem | why is it a request id originating from tempest to nova api is not the same request id that we find in the cells logs? | 14:42 |
| bauzas | mriedem: say what ? it is | 14:42 |
| *** pradk_ has joined #openstack-nova | 14:42 | |
| *** pradk_ is now known as pradk | 14:42 | |
| *** heyongli has joined #openstack-nova | 14:43 | |
| mriedem | this is our 400 to create a server http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/console.html#_2015-11-03_21_50_38_500 | 14:43 |
| mriedem | the request id is req-f4c02d3a-b257-4e97-992d-54f5f56a414f right? | 14:43 |
| bauzas | right req-f4c02d3a-b257-4e97-992d-54f5f56a414f | 14:43 |
| mriedem | that doesn't show up here http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz | 14:43 |
| bauzas | you're looking at the cells/messaging call ? | 14:44 |
| mriedem | yeah | 14:44 |
| mriedem | i at least know the instance uuid now from previous logging we added yesterday | 14:45 |
| mriedem | b4d6d91b-a94c-45d9-80dc-3e40d83afcee | 14:45 |
| sdague | mriedem: it's because it's replayed, right? | 14:45 |
| mriedem | but...we don't log that in the cells log | 14:45 |
| sdague | that request id is created in the API layer by middleware | 14:46 |
| bauzas | I'm quickly digging in the cells messaging code to see if we're providing the same context or another one | 14:46 |
| *** mmdurrant has joined #openstack-nova | 14:46 | |
| sdague | my guess is when you hit the child cell api it's regenerated | 14:46 |
| bauzas | sdague: oh sec, that reminds me something I saw in the cells code | 14:46 |
| *** openstackgerrit has quit IRC | 14:47 | |
| bauzas | yeah, we're serializing the context | 14:47 |
| bauzas | lemme show you | 14:47 |
| *** electrocucaracha has joined #openstack-nova | 14:47 | |
| *** openstackgerrit has joined #openstack-nova | 14:47 | |
| *** cfriesen_ has joined #openstack-nova | 14:49 | |
| *** tpeoples has joined #openstack-nova | 14:49 | |
| mriedem | hold the phone | 14:50 |
| mriedem | bauzas: in the bdm list on the 2nd server create where it fails b/c the boot_index on both is 0, http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-api.txt.gz#_2015-11-03_21_45_18_449 | 14:50 |
| mriedem | the snapshot_id is different between those two | 14:50 |
| bauzas | mriedem: sdague: so yeah the current context is serialized when calling the child cell https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L305-L306 | 14:51 |
| bauzas | and then rehydrated there https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L1344-L1346 | 14:52 |
| *** heyongli has quit IRC | 14:52 | |
| *** pratikmallya has joined #openstack-nova | 14:52 | |
| *** heyongli has joined #openstack-nova | 14:53 | |
| *** inteq has joined #openstack-nova | 14:54 | |
| bauzas | mriedem: yeah, that's what we identified yesterday, 2 distinct snapshot ids, but the same boot index | 14:54 |
| mriedem | i hadn't noticed the snapshot ids were the same | 14:55 |
| mriedem | the plot thickens | 14:55 |
| mriedem | here is some of the new logging in the cells logs http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz#_2015-11-03_21_44_58_273 | 14:56 |
| mriedem | i have a meeting, back in 45min | 14:56 |
| *** ccarmack has quit IRC | 14:56 | |
| *** electrocucaracha has quit IRC | 14:56 | |
| *** mriedem is now known as mriedem_meeting | 14:56 | |
| *** links has quit IRC | 14:56 | |
| *** salv-orlando has joined #openstack-nova | 14:56 | |
| mriedem_meeting | ^ is the first initial server boot from volume | 14:57 |
| mriedem_meeting | btw | 14:57 |
| bauzas | mriedem_meeting: ack, will continue investigating | 14:57 |
| *** sfinucan has joined #openstack-nova | 14:58 | |
| *** mgoddard__ has joined #openstack-nova | 14:59 | |
| *** ccarmack has joined #openstack-nova | 15:00 | |
| *** haomaiwa_ has quit IRC | 15:01 | |
| *** 17SADZFC8 has joined #openstack-nova | 15:01 | |
| *** ccarmack has quit IRC | 15:01 | |
| *** heyongli has quit IRC | 15:02 | |
| *** otter768 has joined #openstack-nova | 15:02 | |
| *** mgoddard_ has quit IRC | 15:02 | |
| dansmith | sdague: so what are you thinking about timing of replacing the partial job with the multinode one? | 15:03 |
| *** moshele has quit IRC | 15:03 | |
| *** heyongli has joined #openstack-nova | 15:03 | |
| sdague | dansmith: I think the last thing we need is a way for grenade to verify both computes are working | 15:03 |
| sdague | I'm going to get another cup of coffee and contemplate that | 15:04 |
| dansmith | sdague: okay | 15:04 |
| *** ccarmack has joined #openstack-nova | 15:05 | |
| *** burgerk has joined #openstack-nova | 15:06 | |
| *** otter768 has quit IRC | 15:12 | |
| *** heyongli has quit IRC | 15:12 | |
| *** pumaranikar has joined #openstack-nova | 15:13 | |
| *** heyongli has joined #openstack-nova | 15:13 | |
| *** smurke has joined #openstack-nova | 15:13 | |
| *** signed8bit has joined #openstack-nova | 15:13 | |
| *** sneti has joined #openstack-nova | 15:14 | |
| *** emagana has joined #openstack-nova | 15:16 | |
| *** garyk1 has quit IRC | 15:19 | |
| *** burt has joined #openstack-nova | 15:21 | |
| *** Piet has joined #openstack-nova | 15:21 | |
| *** heyongli has quit IRC | 15:22 | |
| *** heyongli has joined #openstack-nova | 15:24 | |
| *** veena_ has quit IRC | 15:24 | |
| *** sneti has quit IRC | 15:26 | |
| *** sneti has joined #openstack-nova | 15:26 | |
| dansmith | johnthetubaguy: you ready to land the mitaka priorities review? | 15:28 |
| johnthetubaguy | dansmith: yeah, I think its good, lots of +1s, including two folks that were not at the summit that I wanted to see vote on it | 15:29 |
| *** mriedem_meeting is now known as mriedem | 15:29 | |
| dansmith | yep, agree | 15:29 |
| dansmith | I'll hit the button | 15:29 |
| *** tonytan4ever has joined #openstack-nova | 15:30 | |
| *** ZZelle is now known as ZZelle_away | 15:31 | |
| *** tpeoples has quit IRC | 15:31 | |
| dansmith | bauzas: are you checking into this py34 fail? http://logs.openstack.org/98/217798/10/check/gate-nova-python34/f54e951/console.html.gz | 15:31 |
| johnthetubaguy | dansmith: :D | 15:31 |
| *** tpeoples has joined #openstack-nova | 15:31 | |
| *** inara has joined #openstack-nova | 15:31 | |
| *** jamiec has joined #openstack-nova | 15:31 | |
| *** dave-mccowan has quit IRC | 15:31 | |
| *** inara has quit IRC | 15:31 | |
| bauzas | meh, fun | 15:32 |
| *** zhangjn has joined #openstack-nova | 15:32 | |
| bauzas | dansmith: that remembers me some job failure | 15:32 |
| *** heyongli has quit IRC | 15:33 | |
| bauzas | dansmith: IIRC, the py34 job is whitelisting what he runs vs. what's not, so probably someone added those xenapi stuff | 15:33 |
| dansmith | really? | 15:33 |
| openstackgerrit | Michal Dubiel proposed openstack/nova: libvirt, vhostuser: Add support for OpenContrail vRouter https://review.openstack.org/241652 | 15:33 |
| *** heyongli has joined #openstack-nova | 15:34 | |
| bauzas | dansmith: yup, lemme see you | 15:34 |
| *** cristicalin has quit IRC | 15:34 | |
| *** inara has joined #openstack-nova | 15:34 | |
| *** subscope has quit IRC | 15:34 | |
| johnthetubaguy | mriedem: just approved the blueprint behind this patch, need a -2 removing: https://review.openstack.org/#/c/209018/ | 15:34 |
| mriedem | done | 15:35 |
| bauzas | dansmith: https://github.com/openstack/nova/blob/master/tox.ini#L47 | 15:36 |
| bauzas | which is https://github.com/openstack/nova/blob/master/tests-py3.txt | 15:36 |
| dansmith | ah, blacklist | 15:37 |
| dansmith | you said whitelist | 15:37 |
| bauzas | yup, my bad, checking logstash | 15:37 |
| mriedem | "Policy check for os_compute_api:os-extended-server-attributes failed" shows up a ton in a gate run | 15:38 |
| mriedem | def soft_core_authorizer(api_name, extension_name): | 15:41 |
| mriedem | great name on that method | 15:41 |
| *** ctrath has joined #openstack-nova | 15:41 | |
| mriedem | let's just call it cinemax_authorizer | 15:41 |
| bauzas | mriedem: so, back to the cell failure, I'm puzzled | 15:41 |
| *** ssurana has joined #openstack-nova | 15:42 | |
| bauzas | dansmith: will recheck, I don't understand why my code sucks | 15:42 |
| dansmith | bauzas: I'm trying it locally | 15:42 |
| bauzas | (I need anyway to recheck thanks to the lovely cells job failure) | 15:42 |
| mriedem | bauzas: i'm trying to track the requests that lead to us getting 2 bdms with different snapshot ids | 15:43 |
| mriedem | which would happen from the first instance that has the volume attached | 15:43 |
| bauzas | mriedem: that's what I don't understand | 15:43 |
| *** heyongli has quit IRC | 15:43 | |
| sdague | dansmith: how long before a compute is marked not up if there was an rpc mismatch failure? | 15:44 |
| bauzas | we have distinct snapshot ids, but when looking at the cells parent log, can't find those | 15:44 |
| *** heyongli has joined #openstack-nova | 15:44 | |
| bauzas | mriedem: (speaking of the bdm_update logs you had) | 15:44 |
| *** jwcroppe has joined #openstack-nova | 15:44 | |
| *** ociuhandu has joined #openstack-nova | 15:44 | |
| mriedem | bauzas: yeah i noticed that too | 15:44 |
| dansmith | sdague: you mean if RPCs are failing to check in as a service? or the fail-on-startup-if-we're-too-old thing? | 15:45 |
| dansmith | bauzas: works locally, that's weird | 15:45 |
| sdague | in the grenade case, the compute is up | 15:45 |
| sdague | then we update conductor | 15:45 |
| sdague | and it stops talking correctly | 15:45 |
| *** subscope has joined #openstack-nova | 15:45 | |
| bauzas | dansmith: I already saw that behaviour, but logstash doesn't show me a bad list of problems so I guess I went to a race for the mocks calls | 15:46 |
| dansmith | sdague: that would only be the case if something significant has changed in the service object path, which is probably not a strong enough indication that things are happy or unhappy | 15:46 |
| dansmith | bauzas: okay | 15:46 |
| bauzas | mriedem: so I guess the call to bdm_update_or_create_at_top is made when creating the volume, right? | 15:47 |
| sdague | dansmith: ok, I'm just trying to think through what kind of check I could provide here to ensure that we've got as many working computes as we expect | 15:47 |
| johnthetubaguy | mriedem: got another one here: https://review.openstack.org/#/c/205282/ | 15:47 |
| dansmith | sdague: yeah, I'm thinking | 15:47 |
| dansmith | sdague: I mean, checking that both computes are "up" at the end of the new run is definitely some indication | 15:47 |
| mriedem | bauzas: creating the volume? | 15:47 |
| mriedem | the bdm you mean? | 15:47 |
| sdague | dansmith: right, as in service-list up? | 15:47 |
| dansmith | sdague: but we could have failed every boot to node 1 and rescheduled to node 2 | 15:47 |
| bauzas | mriedem: yup, tired | 15:48 |
| dansmith | sdague: right | 15:48 |
| sdague | dansmith: right | 15:48 |
| mriedem | bauzas: yeah i think we're seeing the cells-region logs for the test only when doing the boot from volume on the first instance | 15:48 |
| mriedem | which is why the bdm has volume_id set | 15:48 |
| dansmith | sdague: is it too heavy to just boot an instance directed at each compute node to make sure it lands there? | 15:48 |
| bauzas | mriedem: that's my understanding too, because also the timings are not there | 15:48 |
| mriedem | johnthetubaguy: done | 15:48 |
| bauzas | I mean | 15:48 |
| sdague | dansmith: probably not | 15:49 |
| sdague | dansmith: we'd need to do that post upgrade, right? | 15:49 |
| *** ociuhandu has quit IRC | 15:49 | |
| dansmith | sdague: yeah | 15:49 |
| dansmith | sdague: http://docs.openstack.org/user-guide-admin/cli_nova_specify_host.html | 15:49 |
| bauzas | mriedem: I mean it gives a 400 at 2015-11-03 21:45:18.449 for n-api | 15:49 |
| sdague | so, conceptually | 15:49 |
| sdague | 1) give grenade an expected node count | 15:50 |
| sdague | 2) check that there are that many hypervisors | 15:50 |
| sdague | 3) check that we can poke something to each hypervisor | 15:50 |
| sdague | 3 only really needs to happen post upgrade | 15:50 |
| bauzas | mriedem: while the call to bdm_update_or_create in the messaging log is 2015-11-03 21:44:58.273 | 15:50 |
| mriedem | bauzas: i think there is something wonky going on when we are creating the snapshot image from the instance with the attached volume | 15:51 |
| dansmith | sdague: yeah | 15:51 |
| sdague | the others could happen pre and post as they are quick sanity checks | 15:51 |
| dansmith | yep | 15:51 |
| mriedem | bauzas: yeah the logs we're seeing in the cells logs are from the first instance, with the attached volume, that we create the snapshot from | 15:51 |
| sdague | ok, this feels like a plan | 15:51 |
| mriedem | bauzas: the 400 is when we try to boot the 2nd instance from the image snapshot from instance 1 | 15:51 |
| sdague | I'll start coding | 15:51 |
| bauzas | mriedem: right, but it doesn't go to cells I guess | 15:51 |
| mriedem | bauzas: which fails b/c of the duplicate bdms | 15:51 |
| bauzas | mriedem: yup yup | 15:52 |
| mriedem | right | 15:52 |
| dansmith | sdague: cool | 15:52 |
| mriedem | yeah, so something is f'ed up in the server create image flow | 15:52 |
| mriedem | so i'm tracking that | 15:52 |
| *** tongli has joined #openstack-nova | 15:52 | |
| bauzas | mriedem: so, we see the BDM created correctly in https://github.com/openstack/tempest/blob/master/tempest/scenario/test_volume_boot_pattern.py#L171-L172 | 15:53 |
| bauzas | but something is trampled in https://github.com/openstack/tempest/blob/master/tempest/scenario/test_volume_boot_pattern.py#L175 | 15:53 |
| bauzas | so I'm tracking that req-id | 15:53 |
| *** heyongli has quit IRC | 15:53 | |
| mriedem | bauzas: yeah, the problem is here i think | 15:54 |
| *** breitz has quit IRC | 15:54 | |
| mriedem | http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/console.html#_2015-11-03_21_50_38_474 | 15:54 |
| mriedem | we have 2 bdms in that snapshot image response | 15:54 |
| *** breitz has joined #openstack-nova | 15:54 | |
| *** heyongli has joined #openstack-nova | 15:54 | |
| bauzas | mriedem: req-e6d08eaf-4dc8-4d11-b68d-f45ee08855db is the snapshot request, right? | 15:56 |
| mriedem | yeah | 15:57 |
| bauzas | mriedem: and the snapshot ids are consistent with the 400 request | 15:58 |
| bauzas | speaking of http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-api.txt.gz#_2015-11-03_21_45_18_449 | 15:59 |
| bauzas | so something gets wrong in req-e6d08eaf-4dc8-4d11-b68d-f45ee08855db, tracing it down | 15:59 |
| mriedem | bauzas: yup | 16:00 |
| *** matrohon has quit IRC | 16:00 | |
| mriedem | tempest is also a bit busted | 16:00 |
| mriedem | if bdm and 'snapshot_id' in bdm[0]: | 16:00 |
| mriedem | tempest should be asserting that there is only 1 bdm | 16:00 |
| *** alexschm has quit IRC | 16:00 | |
| mriedem | for the snapshot image | 16:00 |
| *** 17SADZFC8 has quit IRC | 16:01 | |
| *** haomaiwa_ has joined #openstack-nova | 16:01 | |
| bauzas | melwitt: I won't be available for the cells meeting tonight | 16:01 |
| bauzas | 1700UTC | 16:02 |
| *** daemontool has quit IRC | 16:02 | |
| bauzas | alaski: ^ | 16:02 |
| *** haomaiwa_ has quit IRC | 16:02 | |
| *** heyongli has quit IRC | 16:03 | |
| *** heyongli has joined #openstack-nova | 16:05 | |
| *** klkumar has quit IRC | 16:05 | |
| *** lsmola has quit IRC | 16:05 | |
| bauzas | mriedem: https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/servers.py#L1066-L1067 doesn't sounds correct | 16:06 |
| *** klkumar has joined #openstack-nova | 16:06 | |
| mriedem | why? it uses that to figure out if it's a volume backed instance | 16:08 |
| *** garyk has joined #openstack-nova | 16:08 | |
| bauzas | mriedem: no sec, I think the problem is before that call | 16:09 |
| openstackgerrit | Rodolfo Alonso Hernandez proposed openstack/nova: Added new scheduler filter: AggregateTypeExtraSpecsAffinityFilter https://review.openstack.org/189279 | 16:09 |
| bauzas | mriedem: but rather when creating the BDMs from the first boot from volume | 16:09 |
| bauzas | it should be duplicated | 16:10 |
| bauzas | mriedem: then when creating the image, it got persisted | 16:10 |
| bauzas | mriedem: and then failing when booting from the image | 16:10 |
| mriedem | so you think we're attaching 2 bdms to the first instance (volume bdms) and then creating 2 snapshots for the 2 volume bdms? | 16:10 |
| bauzas | mriedem: I feel so | 16:11 |
| mriedem | yeah maybe | 16:11 |
| mriedem | would be helpful if we had >0 debug logging in here :( | 16:11 |
| bauzas | which explains why the snapshot ids are null when looking at the first bdm creations | 16:11 |
| bauzas | but somehow, we're having duplicates | 16:12 |
| mriedem | yeah, so let's find where it's booting from volume in the compute manager | 16:12 |
| * bauzas looking back at the cells.messaging logs you had | 16:12 | |
| *** zhangjn has quit IRC | 16:12 | |
| bauzas | added | 16:12 |
| goodygum | hey all | 16:13 |
| goodygum | I just want my Nova to inform me about creation and deletion of a VM via notifications | 16:13 |
| mriedem | bauzas: here is where we attach the boot from volume on the first server create http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cpu.txt.gz#_2015-11-03_21_45_00_472 | 16:13 |
| goodygum | BTW I have a Kilo version | 16:14 |
| *** heyongli has quit IRC | 16:14 | |
| goodygum | I want Nova to inform me thru RPC call, RabbitMQ and so on | 16:14 |
| mriedem | goodygum: check the openstack docs | 16:15 |
| *** heyongli has joined #openstack-nova | 16:15 | |
| mriedem | this is not a support channel | 16:15 |
| *** haomaiwa_ has joined #openstack-nova | 16:15 | |
| bauzas | mriedem: I can count 2 update_or_create calls with u'instance_uuid': u'2c9cecc1-c3db-4057-81bd-98e488c45ac2' | 16:15 |
| goodygum | mhm, okay | 16:15 |
| bauzas | mriedem: http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz#_2015-11-03_21_44_58_273 and http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz#_2015-11-03_21_44_58_332 | 16:15 |
| bauzas | so the question is, is the 2nd call to update_or_create doing either create or update ? :D | 16:16 |
| * bauzas looking at the code | 16:16 | |
| mriedem | bauzas: when we delete the first instance (booted from volume), there is only 1 bdm that's terminated http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cpu.txt.gz#_2015-11-03_21_45_16_526 | 16:17 |
| bauzas | mriedem: ah-ha | 16:18 |
| pumaranikar | while snapshoting an instance, if compute service restarts, snapshot continues to stay as queued/saving. Should we move it to error? | 16:20 |
| bauzas | mriedem: honesly, I feel we need more DNM logs | 16:21 |
| *** krtaylor has quit IRC | 16:21 | |
| bauzas | because the DB API checks if the device_name is set for updating the row | 16:21 |
| mriedem | bauzas: https://www.diffchecker.com/gablzhqt | 16:21 |
| mriedem | that's the diff between those bdm_update_or_create_at_top calls | 16:21 |
| *** gonzalo2kx has joined #openstack-nova | 16:22 | |
| mriedem | the latter is probably from when the nova.virt.block_device object does it's save | 16:22 |
| bauzas | mriedem: I'll need to drop off for a bit - family time y'know | 16:23 |
| mriedem | sure | 16:23 |
| bauzas | will continue to dig into that later this evening | 16:23 |
| mriedem | i'll probably add some more derbug logging | 16:23 |
| *** heyongli has quit IRC | 16:24 | |
| bauzas | mriedem: I was just about to say that we need to make sure we only create one BDM object when calling db.bdm_update_or_create | 16:24 |
| bauzas | given the ID is the device_name :// | 16:25 |
| mriedem | yeah, that's what alaski was pointing out yesterday | 16:25 |
| bauzas | that's unfortunate we don't have an uuid... | 16:25 |
| mriedem | but the issue was knowing if we needed to create or update | 16:25 |
| *** heyongli has joined #openstack-nova | 16:25 | |
| bauzas | really need to head off, ttyl | 16:25 |
| mriedem | later | 16:26 |
| edleafe | mriedem: btw, diffchecker looks great! Wish I had known about it earlier | 16:26 |
| *** ZZelle has joined #openstack-nova | 16:26 | |
| *** mtanino has joined #openstack-nova | 16:26 | |
| mriedem | i will take full credit for googling for it | 16:26 |
| *** jdurgin has joined #openstack-nova | 16:27 | |
| *** hemnafk is now known as hemna | 16:27 | |
| *** mdrabe has quit IRC | 16:28 | |
| mriedem | hmm: u'connection_info': u'null' | 16:29 |
| mriedem | connection_info_string = jsonutils.dumps( | 16:29 |
| mriedem | self.get('connection_info')) | 16:29 |
| mriedem | if connection_info_string != self._bdm_obj.connection_info: | 16:29 |
| mriedem | self._bdm_obj.connection_info = connection_info_string | 16:29 |
| *** garyk has quit IRC | 16:29 | |
| mriedem | triggers an update to the bdm object | 16:29 |
| mriedem | b/c we go from None to 'null' due to json | 16:29 |
| *** Kevin_Zheng has quit IRC | 16:32 | |
| *** dustins has quit IRC | 16:32 | |
| *** electrocucaracha has joined #openstack-nova | 16:34 | |
| *** krtaylor has joined #openstack-nova | 16:34 | |
| *** heyongli has quit IRC | 16:34 | |
| *** garyk has joined #openstack-nova | 16:35 | |
| *** vilobhmm11 has joined #openstack-nova | 16:35 | |
| *** heyongli has joined #openstack-nova | 16:35 | |
| *** vilobhmm11 has quit IRC | 16:36 | |
| *** tjones has joined #openstack-nova | 16:36 | |
| *** n0ano has quit IRC | 16:36 | |
| smcginnis | johnthetubaguy: Hey, question for you. The nova spec review etherpad was pointed out to me as something that might be useful for Cinder. | 16:36 |
| smcginnis | johnthetubaguy: https://etherpad.openstack.org/p/mitaka-nova-spec-review-tracking | 16:36 |
| smcginnis | johnthetubaguy: Wondering about the review stats link in there. | 16:37 |
| smcginnis | Is that someone's customer script? | 16:37 |
| smcginnis | s/customer/custom/ | 16:37 |
| *** unicell has joined #openstack-nova | 16:38 | |
| *** unicell1 has quit IRC | 16:40 | |
| dansmith | smcginnis: that's his script | 16:41 |
| smcginnis | dansmith: Ah, OK. Thanks! | 16:41 |
| dansmith | smcginnis: it's based on the reviewstats code, I think | 16:42 |
| dansmith | smcginnis: http://russellbryant.net/openstack-stats/cinder-reviewers-30.txt | 16:42 |
| smcginnis | dansmith: Oh, the one from infra? | 16:42 |
| dansmith | smcginnis: see the bottom | 16:42 |
| smcginnis | dansmith: Got it. Thank you. | 16:43 |
| *** mdrabe has joined #openstack-nova | 16:44 | |
| *** drankis has quit IRC | 16:44 | |
| *** thorst has quit IRC | 16:44 | |
| *** achanda has joined #openstack-nova | 16:44 | |
| *** heyongli has quit IRC | 16:44 | |
| *** thorst has joined #openstack-nova | 16:45 | |
| *** mgoddard__ has quit IRC | 16:45 | |
| *** mgoddard has joined #openstack-nova | 16:45 | |
| *** heyongli has joined #openstack-nova | 16:46 | |
| *** r-daneel has joined #openstack-nova | 16:51 | |
| *** thorst_ has joined #openstack-nova | 16:51 | |
| *** mrkz has quit IRC | 16:52 | |
| *** gszasz has quit IRC | 16:53 | |
| *** mrkz has joined #openstack-nova | 16:54 | |
| *** thorst has quit IRC | 16:54 | |
| *** tongli has quit IRC | 16:55 | |
| *** daemontool has joined #openstack-nova | 16:55 | |
| *** heyongli has quit IRC | 16:55 | |
| *** heyongli has joined #openstack-nova | 16:56 | |
| *** mgoddard has quit IRC | 16:56 | |
| *** su_zhang has joined #openstack-nova | 16:57 | |
| *** tongli has joined #openstack-nova | 16:58 | |
| *** drankis has joined #openstack-nova | 16:59 | |
| *** mgoddard has joined #openstack-nova | 17:00 | |
| *** blackjack1 has joined #openstack-nova | 17:00 | |
| *** haomaiwa_ has quit IRC | 17:01 | |
| *** ihrachys has quit IRC | 17:01 | |
| *** haomaiwang has joined #openstack-nova | 17:01 | |
| blackjack1 | Hi guys. I want to set up nova rabbitmq notifications with no success. I set up nova.conf like in this answer: https://ask.openstack.org/en/question/2224/novaquantum-notifications-in-rabbitmq/ but don't receive any message on the channel. If I change my listener's rabbitmq routing key to "#" (catch-all) I get some service messages but not the one I want | 17:03 |
| *** pumaranikar has quit IRC | 17:03 | |
| blackjack1 | release version is kilo | 17:03 |
| johnthetubaguy | smcginnis: yeah, what dansmith said, its based on that code | 17:03 |
| tjones | blackjack1: i just did this yesterday using this example http://alesnosek.com/blog/2015/05/25/openstack-nova-notifications-subscriber | 17:04 |
| *** pratikma_ has joined #openstack-nova | 17:04 | |
| johnthetubaguy | smcginnis: I just upload the result to a swift container, see: http://reviews.johnthetubaguy.com it links to the code | 17:04 |
| johnthetubaguy | smcginnis: ah, cool, I should read scrollback better, I see you got that now | 17:05 |
| blackjack1 | tjones: thanks, I'll try it now | 17:05 |
| *** heyongli has quit IRC | 17:05 | |
| *** jlanoux has quit IRC | 17:06 | |
| *** heyongli has joined #openstack-nova | 17:06 | |
| *** pratikmallya has quit IRC | 17:07 | |
| *** salv-orlando has quit IRC | 17:08 | |
| *** jistr has quit IRC | 17:08 | |
| smcginnis | johnthetubaguy: Thanks! | 17:08 |
| *** salv-orlando has joined #openstack-nova | 17:09 | |
| johnthetubaguy | smcginnis: that branch is a hacked up mess right now, let me know if thats useful, and I should go back and tidy it up | 17:09 |
| smcginnis | johnthetubaguy: Great. On my list now to look in to. I'll let you know if I have any feedback. | 17:10 |
| *** su_zhang has quit IRC | 17:10 | |
| *** pratikmallya has joined #openstack-nova | 17:11 | |
| *** pratikma_ has quit IRC | 17:11 | |
| *** pratikma_ has joined #openstack-nova | 17:11 | |
| *** otter768 has joined #openstack-nova | 17:13 | |
| *** electrocucaracha has quit IRC | 17:14 | |
| blackjack1 | tjones: thanks, it helped! | 17:14 |
| blackjack1 | So as I see there is no notification generated when new flavor was created/updated, is that right? | 17:14 |
| *** nkrinner has quit IRC | 17:14 | |
| *** heyongli has quit IRC | 17:15 | |
| *** pratikmallya has quit IRC | 17:16 | |
| *** heyongli has joined #openstack-nova | 17:16 | |
| *** rlrossit has quit IRC | 17:17 | |
| *** otter768 has quit IRC | 17:17 | |
| *** derekh has quit IRC | 17:18 | |
| tjones | i had it set to only notify on vm_state change. you can get more by switching it to vm_and_task_state. not sure if you will get anything with a flavor change though | 17:18 |
| *** sayali has quit IRC | 17:19 | |
| johnthetubaguy | blackjack1: I don't think we have any notifications for those, its mostly focused around the instance operations right now | 17:19 |
| *** rlrossit has joined #openstack-nova | 17:20 | |
| blackjack1 | johnthetubaguy: understood, thanks | 17:20 |
| *** sayali has joined #openstack-nova | 17:20 | |
| johnthetubaguy | blackjack1: if its useful, it might be a good add, but we are in the process of adding a new notification versioning system, gibi is setting up a notification subteam, if you are interested | 17:20 |
| *** markmc has quit IRC | 17:22 | |
| sfinucan | johnthetubaguy: Would it be possible to take a look at this spec? I think (think :)) it should be a trivial re-approval https://review.openstack.org/#/c/241355/ | 17:22 |
| sfinucan | the original Liberty spec is here https://review.openstack.org/#/c/194221/ | 17:23 |
| sfinucan | I'd ask Daniel Berrangealso, but he doesn't seem to be available | 17:23 |
| danpb | sfinucan: i'm here | 17:25 |
| danpb | i agree, we could just trivial re-approve that thread's pinning one IMHO | 17:25 |
| sfinucan | danpb: I didn't know your IRC handle, apparently | 17:25 |
| sfinucan | whoops | 17:25 |
| *** heyongli has quit IRC | 17:25 | |
| *** r-daneel has quit IRC | 17:26 | |
| *** heyongli has joined #openstack-nova | 17:27 | |
| johnthetubaguy | sfinucan: danpb: seems like a trivial re-approval to me, not checked the pending ones today | 17:27 |
| *** e0ne has quit IRC | 17:27 | |
| *** r-daneel has joined #openstack-nova | 17:27 | |
| *** y_sawai has joined #openstack-nova | 17:28 | |
| *** davideagnello has joined #openstack-nova | 17:28 | |
| *** vilobhmm11 has joined #openstack-nova | 17:30 | |
| *** garyk has quit IRC | 17:31 | |
| *** dustins has joined #openstack-nova | 17:32 | |
| *** davideagnello has quit IRC | 17:33 | |
| sfinucan | danpb: johnthetubaguy: Thank you. I'll submit together the "clarification" patch later today | 17:33 |
| *** smurke has quit IRC | 17:33 | |
| *** davideagnello has joined #openstack-nova | 17:34 | |
| *** heyongli has quit IRC | 17:36 | |
| *** daemontool has quit IRC | 17:36 | |
| jroll | sfinucan: don't feel bad, in tokyo I asked danpb in person if daniel berrange was going to show up :D | 17:36 |
| dansmith | that was awesome | 17:37 |
| *** heyongli has joined #openstack-nova | 17:37 | |
| mriedem | sdague: are you aware of any bugs for network dealloc failures in the large ops job? like this http://logs.openstack.org/97/241397/1/check/gate-tempest-dsvm-large-ops/be8ca35/logs/screen-n-cpu-1.txt.gz#_2015-11-04_14_46_57_639 | 17:37 |
| *** rk4n has quit IRC | 17:37 | |
| mriedem | jroll: that could go worse | 17:37 |
| jroll | heh | 17:38 |
| mriedem | sdague: nvm, we already have 2 marked as invalid | 17:39 |
| *** pratikma_ has quit IRC | 17:41 | |
| danpb | jroll: next time we can have name badges where the font is actually large enough to read :-) | 17:42 |
| * danpb curses who ever decided to use minute sized font for surnames | 17:42 | |
| jroll | danpb: I didn't even look, so not sure if that's helpful :P | 17:42 |
| *** romainh has quit IRC | 17:43 | |
| *** nelsnelson has quit IRC | 17:43 | |
| sfinucan | quick question - does Nova have any support for load balancing of instances | 17:46 |
| mriedem | no | 17:46 |
| *** heyongli has quit IRC | 17:46 | |
| mriedem | there is a spec | 17:46 |
| sfinucan | OK, and is it on anyone's (public) roadmap yet | 17:46 |
| mriedem | pretty sure it's not going to happen | 17:46 |
| sfinucan | Ah | 17:46 |
| *** heyongli has joined #openstack-nova | 17:47 | |
| sfinucan | Is the spec approved? Any pointers? | 17:48 |
| sfinucan | *pointers to the spec | 17:48 |
| mriedem | trying to find | 17:48 |
| mriedem | might have been ML | 17:48 |
| sfinucan | A college buddy is considering it as a thesis project. Seems ambitious | 17:48 |
| sfinucan | and like something someone might already be working on | 17:48 |
| *** tonytan4ever has quit IRC | 17:48 | |
| *** nelsnelson has joined #openstack-nova | 17:48 | |
| mriedem | there are projects that have already done it outside nova | 17:49 |
| mriedem | i think this new 'watcher' project is doing something like that | 17:49 |
| mriedem | ibm platform resource scheduler (closed source) does some auto-migrate/evac stuff depending on 'policies' | 17:49 |
| mriedem | there are projects that have done this | 17:49 |
| tpeoples | watcher is doing resource optimization, but load balancing is probably not the best term for it | 17:49 |
| mriedem | i want to say some cisco guys had a thing in the ML at one point talking about similar things | 17:50 |
| sfinucan | I thought as much. It seems like a rather obvious feature for some deployments | 17:50 |
| sfinucan | Hmm, OK that's his job to find that mail so | 17:51 |
| mriedem | i can't find anything, i thought i saw a spec at one point though | 17:51 |
| tpeoples | mriedem: we're trying to contribute back what we've done with prs to the community through watcher | 17:51 |
| tpeoples | sfinucan: https://wiki.openstack.org/wiki/Watcher | 17:51 |
| mriedem | tpeoples: i always forget you're one of *those* people | 17:52 |
| tpeoples | :) :| | 17:52 |
| mriedem | but yeah, the answer on stuff like this is generally the same, we want less orchestratoin in nova, not more | 17:52 |
| *** tojuvone has quit IRC | 17:53 | |
| sfinucan | mriedem: tpeoples: That's great info. I'll pass on it on and tell him to carry on searching | 17:54 |
| *** su_zhang has joined #openstack-nova | 17:54 | |
| sfinucan | tpeoples: I remember getting mails about watcher during the summit last week. Now I have reason to go read them :) | 17:54 |
| *** tjones has quit IRC | 17:54 | |
| * mriedem loads up on fun size candy bars and moves rooms | 17:55 | |
| *** heyongli has quit IRC | 17:56 | |
| *** nelsnels_ has joined #openstack-nova | 17:57 | |
| *** heyongli has joined #openstack-nova | 17:57 | |
| *** ctrath has quit IRC | 17:58 | |
| *** nelsnelson has quit IRC | 18:00 | |
| *** mriedem has quit IRC | 18:00 | |
| *** haomaiwang has quit IRC | 18:01 | |
| *** haomaiwa_ has joined #openstack-nova | 18:01 | |
| *** unicell1 has joined #openstack-nova | 18:03 | |
| *** klkumar has quit IRC | 18:03 | |
| *** unicell has quit IRC | 18:03 | |
| *** danpb has quit IRC | 18:04 | |
| *** aix has quit IRC | 18:05 | |
| *** heyongli has quit IRC | 18:07 | |
| *** heyongli has joined #openstack-nova | 18:08 | |
| *** browne has joined #openstack-nova | 18:08 | |
| *** mriedem has joined #openstack-nova | 18:11 | |
| *** dave-mccowan has joined #openstack-nova | 18:16 | |
| *** sahid has joined #openstack-nova | 18:16 | |
| *** heyongli has quit IRC | 18:17 | |
| *** e0ne has joined #openstack-nova | 18:17 | |
| *** heyongli has joined #openstack-nova | 18:18 | |
| *** ctrath has joined #openstack-nova | 18:20 | |
| *** jerrygb has quit IRC | 18:21 | |
| *** pratikmallya has joined #openstack-nova | 18:25 | |
| *** heyongli has quit IRC | 18:27 | |
| *** heyongli has joined #openstack-nova | 18:28 | |
| *** pumaranikar has joined #openstack-nova | 18:29 | |
| *** smurke has joined #openstack-nova | 18:32 | |
| *** rk4n has joined #openstack-nova | 18:33 | |
| *** heyongli has quit IRC | 18:37 | |
| mriedem | confirmed that in this cells job failure we're creating 2 volume snapshots | 18:38 |
| *** heyongli has joined #openstack-nova | 18:38 | |
| *** rlrossit has quit IRC | 18:45 | |
| mriedem | so cells.messaing.bdm_update_or_create_at_top must be doing the bdm creates twice, when it should be doing an update the 2nd time | 18:46 |
| *** ctrath has quit IRC | 18:46 | |
| *** ccarmack has quit IRC | 18:46 | |
| *** dvlwrk_ has joined #openstack-nova | 18:47 | |
| *** heyongli has quit IRC | 18:48 | |
| *** subscope has quit IRC | 18:48 | |
| *** heyongli has joined #openstack-nova | 18:49 | |
| johnthetubaguy | mriedem: ouch thats messy, updates have a habbit of getting out of order up there, so two save calls might be enough | 18:50 |
| mriedem | and the nova.virt.block_device objects that wrap bdms have a save on update mechanism | 18:51 |
| *** ctrath has joined #openstack-nova | 18:51 | |
| *** rlrossit has joined #openstack-nova | 18:51 | |
| mriedem | which will fire off successive updates to cells | 18:51 |
| mriedem | so if we don't have a unique constraint on the bdm table, | 18:51 |
| johnthetubaguy | oh, yeah, so we call save 1 million times, I remember looking at that at some point... curious | 18:52 |
| *** rk4n has quit IRC | 18:52 | |
| mriedem | why can't we enforce some unique constraint type logic in cells.messaging.bdm_update_or_create_at_top? | 18:52 |
| mriedem | like it does when doing an update | 18:52 |
| mriedem | we rely on this hacky create parameter being None/True/False | 18:52 |
| *** Marga_ has joined #openstack-nova | 18:52 | |
| johnthetubaguy | I think we should be able to... but I don't remember the context now | 18:52 |
| johnthetubaguy | like doing compare and swap style update, based on the object changes reported? | 18:53 |
| mriedem | yeah | 18:53 |
| johnthetubaguy | maybe we need to record the original values, not 100% sure if we have those? | 18:53 |
| mriedem | we get the bdms from the db, and compare those to what's passed in | 18:53 |
| mriedem | if there are no bdms in the db yet, it's an insert | 18:53 |
| *** alex_klimov has joined #openstack-nova | 18:54 | |
| mriedem | if there are bdms, we check them against some keys and do an update if there is a match, else insert | 18:54 |
| mriedem | i'd think instance_uuid/volume_id/device_name would do a match ok | 18:54 |
| mriedem | volume_id/snapshot_id/image_id | 18:54 |
| johnthetubaguy | so we did run a custom DB later on the nova-cells works, not sure if we got that removed yet, but alaski would know, we might be holding a patch in there for this | 18:54 |
| *** Marga_ has quit IRC | 18:55 | |
| johnthetubaguy | is device_name and instance_uuid not enough? | 18:55 |
| mriedem | "run a custom DB later on the nova-cells works" | 18:55 |
| mriedem | i'm not sure what that means | 18:55 |
| *** Marga_ has joined #openstack-nova | 18:55 | |
| johnthetubaguy | we patch out the DB api with a special one comstud wrote | 18:55 |
| mriedem | that might be enough yeah, it's checking volume_id today too | 18:55 |
| mriedem | see https://review.openstack.org/#/c/241390/ | 18:56 |
| johnthetubaguy | being trying to drop that for ages | 18:56 |
| mriedem | maybe that's not needed | 18:56 |
| *** ccarmack has joined #openstack-nova | 18:56 | |
| vilobhmm11 | bauzas : ping | 18:56 |
| kashyap | "If you would have asked your question, my response would have been your answer" :-) | 18:57 |
| alaski | johnthetubaguy: we removed it briefly, but had to turn it back on for the db_update_and_get_original method to be fast enough to handle cells updates | 18:57 |
| johnthetubaguy | mriedem: crap sorry, I really need to go an get some food and run out... | 18:57 |
| johnthetubaguy | alaski: ah, interesting, thanks | 18:58 |
| mriedem | alaski: are you guys talking about rax internal stuff? | 18:58 |
| *** heyongli has quit IRC | 18:58 | |
| alaski | mriedem: yeah, the mysql db api | 18:58 |
| mriedem | why do i have to always qualify my questions about cells/lxc with 'are you talking about rax internal patches?" | 18:58 |
| mriedem | ah | 18:58 |
| johnthetubaguy | mriedem: yeah, wondering if we fixed it in that sideways code | 18:58 |
| johnthetubaguy | mriedem: you and me both, getting there slowly | 18:59 |
| *** heyongli has joined #openstack-nova | 18:59 | |
| mriedem | alaski: so in here https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L1167 | 18:59 |
| mriedem | when create is None, | 18:59 |
| mriedem | couldn't we get the list of bdms from the db for the instance and then check for a match on device_name to determine if we do an insert or update? | 18:59 |
| * johnthetubaguy bravely runs away to get some food | 18:59 | |
| mriedem | i assume https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L1186 is here in case device_name is empty, which is apparently a thing | 19:00 |
| alaski | mriedem: that's essentially what the db method does | 19:00 |
| mriedem | it checks device_name, | 19:00 |
| mriedem | not instance_uuid | 19:01 |
| *** haomaiwa_ has quit IRC | 19:01 | |
| mriedem | result = query.filter_by(instance_uuid=values['instance_uuid'], | 19:01 |
| mriedem | device_name=values['device_name']).first() | 19:01 |
| mriedem | f* | 19:01 |
| *** haomaiwang has joined #openstack-nova | 19:01 | |
| mriedem | hmmm | 19:01 |
| mriedem | so if not device_name, | 19:01 |
| alaski | it checks it only if device_name, yeah | 19:01 |
| mriedem | check on volume_id/snapshot_id/image_id? | 19:01 |
| openstackgerrit | Raildo Mascena de Sousa Filho proposed openstack/nova-specs: Moving Nested Quota Driver to Mitaka https://review.openstack.org/209969 | 19:02 |
| raildo | johnthetubaguy: ping, added the API example ^ | 19:02 |
| mriedem | if i can get my unique constraint change to fail then i guess we'd see | 19:02 |
| alaski | mriedem: device_name was the same in the duplicate bdms right? | 19:03 |
| dansmith | mriedem: are you proposing a new constraint on the bdms stuff? | 19:03 |
| *** sahid has quit IRC | 19:03 | |
| *** drankis has quit IRC | 19:03 | |
| mriedem | alaski: it was https://review.openstack.org/#/c/241397/1/nova/db/sqlalchemy/migrate_repo/versions/313_add_bdm_unique_constraint.py | 19:03 |
| mriedem | oh, sec | 19:03 |
| dansmith | ah, I see | 19:04 |
| mriedem | alaski: yeah it's /dev/vda in both bdms | 19:05 |
| mriedem | dansmith: just trying to trigger a failure to see where we're coming in from | 19:05 |
| *** rk4n has joined #openstack-nova | 19:05 | |
| dansmith | okay | 19:05 |
| alaski | so the db method should have correctly done an update since device name matched | 19:06 |
| mriedem | i'm assuming that's after we already have the duplicates though | 19:07 |
| mriedem | we probably create the duplicate bdms when device_name isn't set | 19:07 |
| mriedem | which is a thing that the libvirt driver code does now too | 19:07 |
| mriedem | since liberty | 19:07 |
| mriedem | https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L7232 | 19:08 |
| mriedem | bdm.device_name = None | 19:08 |
| mriedem | block_device_info = driver.get_block_device_info(instance, block_device_mapping) | 19:08 |
| mriedem | ^ does the bdm.save() somewhere | 19:08 |
| *** heyongli has quit IRC | 19:08 | |
| mriedem | cells gets 2 bdms with no device_name and create=False | 19:09 |
| mriedem | *create=None | 19:09 |
| mriedem | and then inserts 2 | 19:09 |
| alaski | we try to prevent that with http://git.openstack.org/cgit/openstack/nova/tree/nova/objects/block_device.py#n141 | 19:09 |
| alaski | though it's possible that it's happening via another path | 19:09 |
| *** heyongli has joined #openstack-nova | 19:09 | |
| dansmith | this would probably be easier to debug if we had object tracing, which is something I've wanted for a while | 19:11 |
| mriedem | http://git.openstack.org/cgit/openstack/nova/tree/nova/objects/block_device.py#n186 ? | 19:11 |
| mriedem | i assume it has to be coming from bdm.save() b/c that's what the nova.virt.block_device bdm wrapper objects do - update on save | 19:11 |
| *** emagana_ has joined #openstack-nova | 19:13 | |
| *** emagana has quit IRC | 19:13 | |
| *** otter768 has joined #openstack-nova | 19:14 | |
| * bauzas raises hand | 19:14 | |
| mriedem | if you search for the original volume id 35909d21-81b8-4fda-82b6-d3d75be61238 in http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz you'll see we call db.block_device_mapping_update_or_create twice, but device_name is set in the bdm values each time | 19:15 |
| bauzas | mriedem: so it seems you agree with me about having 2 inserts ? | 19:15 |
| mriedem | which implies we're doing an update | 19:15 |
| mriedem | bauzas: yeah, i added notes to the bug report, there are definitely 2 volume snapshots being created, i see them in the c-api logs | 19:15 |
| mriedem | for the same volume | 19:15 |
| mriedem | and a single delete request on the vol snapshot, which comes from tempest teardown | 19:16 |
| alaski | mriedem: so I wanted to enhance the check to only send the save() if device_name had been changed from '' to something. But I couldn't find a way to get that information from the object changelist | 19:16 |
| bauzas | oh good point to check the cinder logs | 19:16 |
| *** LukeHinds has quit IRC | 19:16 | |
| mriedem | alaski: in the bdm object or in the nova.virt.block_device code? | 19:16 |
| alaski | dansmith: that's a question for you, is there a way to tell if a field has just been set, vs modified | 19:16 |
| alaski | mriedem: object | 19:16 |
| mriedem | obj_what_changed? thought that was a thing | 19:17 |
| bauzas | +1 | 19:17 |
| dansmith | alaski: you mean if it is newly set vs. changed? | 19:17 |
| alaski | dansmith: yep | 19:17 |
| dansmith | alaski: not through the normal objects api, no. Instance keeps some extra tracking, but bdm does not | 19:18 |
| dansmith | I'm not sure why it matters though | 19:18 |
| *** heyongli has quit IRC | 19:18 | |
| mriedem | obj_get_changes() ? | 19:18 |
| dansmith | no | 19:18 |
| dansmith | that won't distinguish | 19:18 |
| *** otter768 has quit IRC | 19:19 | |
| mriedem | blast | 19:19 |
| dansmith | you'd have to pull the DB one to know. Usually we handle this kind of thing deep in the db api, where we've already pulled the original, or where we can say "update if device_name != table.device_name" or something | 19:19 |
| *** heyongli has joined #openstack-nova | 19:19 | |
| alaski | dansmith: it would allow us to send create=True/False vs the None we use now. If device_name has just been set then create, if it's changed update | 19:20 |
| *** jerrygb has joined #openstack-nova | 19:20 | |
| mriedem | i guess that goes back to the db api call we have | 19:20 |
| dansmith | alaski: it's a pretty obscure case, so I don't think we need to build this in, but we can certainly do this in the BDM object for this field | 19:20 |
| dansmith | alaski: assuming you don't need to know across an RPC call | 19:21 |
| mriedem | http://git.openstack.org/cgit/openstack/nova/tree/nova/db/sqlalchemy/api.py#n3943 | 19:21 |
| *** tonytan4ever has joined #openstack-nova | 19:21 | |
| alaski | dansmith: I don't think we do | 19:21 |
| alaski | the other thing to consider is if device_name can't change after it's set we can just assume an update in that case | 19:21 |
| dansmith | alaski: I can work up a patch to bdm to track this if you want | 19:21 |
| alaski | I just wasn't sure on that | 19:22 |
| dansmith | alaski: ah, you can make it read-only and it will only let you set it once | 19:22 |
| bauzas | just to be clear, we are having like a UK for device_name ? | 19:22 |
| dansmith | if you think that will help | 19:22 |
| mriedem | alaski: i think it can change | 19:22 |
| *** spandhe has joined #openstack-nova | 19:22 | |
| bauzas | I mean an UC like (instance, device_name) right? | 19:23 |
| alaski | dansmith: well, I'm not sure if we support the case where it changes | 19:23 |
| mriedem | i'd have to verify, but if the compute api creates the bdm and then the libvirt driver nulls it out, that's a change | 19:23 |
| mriedem | bauzas: i have a patch that does that but it's DNM for testing | 19:23 |
| mriedem | to find an explosion | 19:23 |
| alaski | dansmith: if you get around to it go for it. I can pick it up as well starting tomorrow | 19:23 |
| mriedem | of course i never know which bdm class i'm dealing with | 19:24 |
| alaski | dansmith: whatever works for you | 19:24 |
| dansmith | alaski: I can do it fairly quick.. mriedem you know what to do with it once I write that? | 19:24 |
| alaski | bauzas: I don't like having a UC on device_name because it's not static. But I'm not sure what else we would use right now. | 19:24 |
| alaski | I would love to look at adding uuids for bdms, but that might be too much | 19:25 |
| dansmith | alaski: we can't solve that with an update where clause? | 19:25 |
| bauzas | alaski: +1 for uuids | 19:25 |
| mriedem | dansmith: yeah | 19:25 |
| bauzas | alaski: that's something I thought this EU afternoon | 19:25 |
| bauzas | but agreed as it's a big change | 19:26 |
| *** bradjones has quit IRC | 19:26 | |
| alaski | dansmith: hmm, not sure. I need to consider that | 19:26 |
| *** vishwana_ has quit IRC | 19:27 | |
| *** electrocucaracha has joined #openstack-nova | 19:28 | |
| *** heyongli has quit IRC | 19:29 | |
| alaski | I'm not sure that will work because we need to insert or update based on the where clause | 19:29 |
| *** heyongli has joined #openstack-nova | 19:30 | |
| dansmith | alaski: well, you update where and if that fails, insert, inside a transaction? | 19:30 |
| dansmith | I guess two transactions can still race without the UC | 19:30 |
| *** rlrossit has quit IRC | 19:31 | |
| alaski | yeah. I really think a UC is necessary here | 19:31 |
| *** rlrossit has joined #openstack-nova | 19:32 | |
| dansmith | so, adding a UC to a live DB is going to lock that table, right? | 19:32 |
| alaski | looks like it | 19:35 |
| openstackgerrit | Dan Smith proposed openstack/nova: DNM: Add bdm_field_is_new() to BlockDeviceMapping https://review.openstack.org/241737 | 19:35 |
| dansmith | that will suck :( | 19:35 |
| dansmith | mriedem: ^ | 19:35 |
| alaski | yeah | 19:35 |
| *** diogogmt has joined #openstack-nova | 19:36 | |
| alaski | we could look at creating the bdm in one place, and then always update after | 19:36 |
| alaski | create it early in the boot process | 19:36 |
| dansmith | I think that probably has some unfortunate side effects | 19:37 |
| *** aysyd has quit IRC | 19:37 | |
| *** gjayavelu has joined #openstack-nova | 19:38 | |
| dansmith | we could create a new table with the constraint and hot migrate things to that table, but we probably need to evaluate the pain vs. reward for that | 19:38 |
| dansmith | I really don't want to regress our live upgrade abilities, but we should be pragmatic | 19:38 |
| bauzas | so, instead of that, could we think having only one way to insert into the table ? | 19:38 |
| dansmith | if people can choose to apply our migrations early, during a maintenance window, without having to touch any other code, maybe it's okay to require that level of impact in a migration, I dunno | 19:39 |
| *** heyongli has quit IRC | 19:39 | |
| dansmith | obviously solving this at a higher layer would be much nicer | 19:39 |
| *** ctrath has quit IRC | 19:39 | |
| bauzas | well, that kind of migration for a cells v1 failure seems really overkill | 19:39 |
| bauzas | given cells v1 | 19:39 |
| *** heyongli has joined #openstack-nova | 19:40 | |
| alaski | well, the race exists outside of cells. it's just very unlikely in that case | 19:40 |
| bauzas | that's why I wonder why we shouldn't try to find a way to only have one way to call to insert - and then do like a transaction | 19:40 |
| bauzas | alaski: I don't get the conditionals for the race when it's not cells v1 | 19:41 |
| alaski | dansmith: a lock would also solve it, which is essentially what the UC gives us, but we would need a distributed lock | 19:41 |
| bauzas | oh man | 19:41 |
| *** ctrath has joined #openstack-nova | 19:42 | |
| bauzas | don't say that too loud | 19:42 |
| *** salv-orlando has quit IRC | 19:42 | |
| alaski | bauzas: I'm not sure it would be possible to trigger it in the current code base, but create_or_update with no UC and no lock makes it possible to end up with two creates | 19:42 |
| *** salv-orlando has joined #openstack-nova | 19:42 | |
| cfriesen_ | anyone know why nova doesn't have quotas for local/ephemeral storage? is it just assumed to be to plentiful to ration? | 19:43 |
| bauzas | right, just considering what could be the 2 different services calling it | 19:43 |
| alaski | bauzas: I thought everyone agreed on their goodness at the summit :) | 19:43 |
| dansmith | alaski: ssshhh | 19:43 |
| bauzas | alaski: well | 19:43 |
| dansmith | alaski: and we're doing this too early to serialize ourselves at the compute node? | 19:43 |
| bauzas | alaski: it's.... complicated | 19:43 |
| anteaya | would someone care to add the nova mid-cycle details: http://lists.openstack.org/pipermail/openstack-dev/2015-October/078054.html to the sprints wikipage: https://wiki.openstack.org/wiki/Sprints please? | 19:44 |
| mriedem | not it | 19:44 |
| mmmpork | i can do it :) | 19:45 |
| alaski | dansmith: I htink we could serialize the sending of the update/create message, but in this case the processing happens on the api cell service | 19:45 |
| patrickeast | hey so i'm seeing a problem with test_create_ebs_image_and_check_boot in gate-tempest-dsvm-full-ceph on a stable/kilo review | 19:45 |
| dansmith | alaski: right, right | 19:45 |
| patrickeast | on https://review.openstack.org/#/c/229152/ recent runs are failing the test http://logs.openstack.org/52/229152/11/check/gate-tempest-dsvm-full-ceph/11bddbf/console.html#_2015-11-04_19_20_07_226 | 19:45 |
| bauzas | heh | 19:46 |
| *** su_zhang has quit IRC | 19:46 | |
| *** jichen has joined #openstack-nova | 19:46 | |
| patrickeast | older ones where it was passing it looks like the test was being skipped http://logs.openstack.org/52/229152/8/check/gate-tempest-dsvm-full-ceph/e74fd3b/console.html#_2015-11-02_21_51_13_928 | 19:46 |
| *** su_zhang has joined #openstack-nova | 19:46 | |
| mriedem | patrickeast: the test was re-enabld in the last 2 days | 19:46 |
| mriedem | it's killing the cells job for reasons that we're discussing above | 19:46 |
| patrickeast | mriedem: :D that would explain the change | 19:46 |
| * bauzas actually tempted to ask to skip again this test until we figure out a clear solution path | 19:46 | |
| mriedem | patrickeast: ceph failure is something else | 19:46 |
| patrickeast | derp | 19:46 |
| mmmpork | anteaya: i don't see a section for mitaka sprints on that page | 19:46 |
| * patrickeast should have read the scrollback first | 19:47 | |
| anteaya | you will have to add it | 19:47 |
| *** markmc has joined #openstack-nova | 19:47 | |
| mriedem | patrickeast: don't bother | 19:47 |
| bauzas | mriedem: patrickeast: the ceph failure seems a bit different | 19:47 |
| anteaya | follow the examples of the prior sprints | 19:47 |
| bauzas | at least the rootcause | 19:47 |
| anteaya | copy paste should work and then use the review button to check you work before you publish | 19:47 |
| mmmpork | anteaya: maybe next time you ask people to add their info, you could make it a bit easier by making it easier for us? :) | 19:47 |
| *** sfinucan has quit IRC | 19:48 | |
| *** smurke has quit IRC | 19:48 | |
| *** achanda has quit IRC | 19:49 | |
| *** heyongli has quit IRC | 19:49 | |
| *** heyongli has joined #openstack-nova | 19:50 | |
| *** achanda has joined #openstack-nova | 19:52 | |
| *** sfinucan has joined #openstack-nova | 19:53 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova: Add logging to snapshot_volume_backed method https://review.openstack.org/241739 | 19:54 |
| mriedem | i can update the wiki if needed | 19:54 |
| mriedem | copy/paste liberty table, replace with nova mitaka info from ML | 19:54 |
| *** smurke has joined #openstack-nova | 19:55 | |
| mriedem | gather karma points from anteaya | 19:55 |
| anteaya | mriedem: you got 'em | 19:55 |
| * anteaya knocks the mickey mouse ears off mriedem's head | 19:55 | |
| mmmpork | i'm working on it, it's just that the request was lacking some information | 19:55 |
| mmmpork | ie, we needed to do more than just add our info to the wiki | 19:56 |
| *** sneti has quit IRC | 19:57 | |
| *** heyongli has quit IRC | 19:59 | |
| *** baoli_ has quit IRC | 20:00 | |
| *** heyongli has joined #openstack-nova | 20:00 | |
| *** sneti has joined #openstack-nova | 20:00 | |
| *** haomaiwang has quit IRC | 20:01 | |
| *** haomaiwang has joined #openstack-nova | 20:01 | |
| sdague | mmmpork: probably worth stubbing out a Sprints/NovaMitakaSprint page while you are at it | 20:01 |
| mmmpork | sdague: thanks I was just looking for it ;) | 20:02 |
| *** su_zhang has quit IRC | 20:04 | |
| *** su_zhang has joined #openstack-nova | 20:06 | |
| openstackgerrit | Matt Riedemann proposed openstack/nova: cells: create bdm at top if device_name is set for first time in save() https://review.openstack.org/241742 | 20:07 |
| mriedem | alaski: ^ is that what you were looking for? based on dansmith's change | 20:07 |
| *** smurke has quit IRC | 20:08 | |
| *** smurke has joined #openstack-nova | 20:08 | |
| *** heyongli has quit IRC | 20:09 | |
| *** su_zhang has quit IRC | 20:09 | |
| *** heyongli has joined #openstack-nova | 20:10 | |
| *** su_zhang has joined #openstack-nova | 20:11 | |
| *** smurke has quit IRC | 20:11 | |
| *** smurke has joined #openstack-nova | 20:12 | |
| mriedem | alright, i need to take a break from this cells thing for awhile - time to review some more specs! | 20:12 |
| *** ZZelle has quit IRC | 20:14 | |
| bauzas | mriedem: shouldn't we consider skipping that test until we figure out a good way to fix that? | 20:15 |
| *** gonzalo2kx has quit IRC | 20:15 | |
| mriedem | umm | 20:15 |
| mriedem | well | 20:15 |
| bauzas | mriedem: http://goo.gl/xGnRW6 | 20:15 |
| *** smurke has quit IRC | 20:16 | |
| mriedem | i sort of feel like pulling an old school leave it broken until people step up to help fix it | 20:16 |
| mriedem | it's definitely not like a top gate blocker | 20:16 |
| *** smurke has joined #openstack-nova | 20:16 | |
| mriedem | but it's painful enough that people should be helping out to debug | 20:16 |
| bauzas | well, it's hard to debug because of $bdm and $cells | 20:17 |
| mriedem | i know, | 20:17 |
| mriedem | but that's where the fun is! | 20:17 |
| bauzas | it requires some knowledge that a very few have | 20:17 |
| * mriedem laughs maniacally | 20:17 | |
| mmmpork | anteaya: updated the wiki per your request | 20:17 |
| mmmpork | sdague: stub page created | 20:17 |
| mriedem | bauzas: we have a few changes up for debug help, we can monitor those and recheck as needed | 20:18 |
| mriedem | i still haven't seen the UC patch fail on cells | 20:18 |
| bauzas | mriedem: sure, it's a good learning path - hence me continuing digging - but the benefits are not huge | 20:18 |
| sdague | mmmpork: thanks | 20:18 |
| mriedem | bauzas: i would recommend sleep | 20:18 |
| bauzas | -ETOOEARLY | 20:18 |
| mriedem | i guess it's still early there | 20:18 |
| mriedem | yeah | 20:18 |
| anteaya | mmmpork: thank you | 20:19 |
| mriedem | i'm getting used to DST | 20:19 |
| anteaya | mriedem: I'll still remove your mickey ears for you | 20:19 |
| *** heyongli has quit IRC | 20:19 | |
| mriedem | you know i'd never wear those right | 20:19 |
| mriedem | there were a lot of people though with them on that didn't have children | 20:19 |
| sdague | mriedem: the best way to not get tripped up by DST | 20:19 |
| mriedem | which concerns me | 20:19 |
| sdague | be adjusting from JST at the same time | 20:19 |
| mmmpork | sdague: i was just gonna say | 20:19 |
| mmmpork | go to tokyo? | 20:20 |
| mmmpork | hahaha | 20:20 |
| *** heyongli has joined #openstack-nova | 20:20 | |
| *** smurke has quit IRC | 20:20 | |
| bauzas | no DST change for US yet, right? | 20:21 |
| dansmith | bauzas: yes | 20:21 |
| *** rfolco has quit IRC | 20:22 | |
| bauzas | ack, because EU shifted one week before | 20:22 |
| bauzas | and I know we have like 3 weeks difference | 20:22 |
| *** smurke has joined #openstack-nova | 20:23 | |
| *** angdraug has joined #openstack-nova | 20:23 | |
| *** su_zhang has quit IRC | 20:25 | |
| *** sfinucan has quit IRC | 20:25 | |
| *** sfinucan has joined #openstack-nova | 20:25 | |
| sdague | bauzas: RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU | 20:25 |
| *** khomkrit has quit IRC | 20:26 | |
| sdague | it's just that the first of nov was the first sunday this year | 20:26 |
| bauzas | I see | 20:26 |
| *** edtubill has joined #openstack-nova | 20:28 | |
| *** heyongli has quit IRC | 20:29 | |
| openstackgerrit | Chris Friesen proposed openstack/nova-specs: Clarify the behaviour for the "require" thread policy https://review.openstack.org/241751 | 20:30 |
| *** heyongli has joined #openstack-nova | 20:30 | |
| *** fawadkhaliq has joined #openstack-nova | 20:31 | |
| *** edtubill has quit IRC | 20:31 | |
| anteaya | mriedem: oh I picture you in mickey ears | 20:36 |
| *** jamielennox|away is now known as jamielennox | 20:36 | |
| *** heyongli has quit IRC | 20:40 | |
| *** heyongli has joined #openstack-nova | 20:40 | |
| *** dave-mccowan has quit IRC | 20:43 | |
| *** edtubill has joined #openstack-nova | 20:44 | |
| cfriesen_ | I'm working on a bug where if we delete an instance while it's resizing we end up never calling self.vif_driver.unplug to unplug the network port. (https://bugs.launchpad.net/nova/+bug/1512907) Who would be the right people to talk to? | 20:46 |
| openstack | Launchpad bug 1512907 in OpenStack Compute (nova) "leak of vswitch port if delete an instance while resizing" [Undecided,New] | 20:46 |
| mriedem | cfriesen_: you're the point person on that one | 20:48 |
| mriedem | just mark it low-hanging-fruit... | 20:48 |
| mriedem | that's about the only way you can get anyone to look at nova bugs | 20:48 |
| cfriesen_ | mriedem: :) | 20:48 |
| *** edtubill has quit IRC | 20:49 | |
| *** edtubill has joined #openstack-nova | 20:49 | |
| cfriesen_ | mriedem: I'm considering just delaying the delete if a resize/migration is in progress until it gets to a vm_state of vm_states.RESIZED, at which point the deletion code can auto-confirm the resize/migration. | 20:49 |
| *** heyongli has quit IRC | 20:50 | |
| *** rotbeard has quit IRC | 20:50 | |
| dansmith | cfriesen_: we generally consider a delete operation to be valid at any point | 20:50 |
| dansmith | otherwise users get fairly mad | 20:50 |
| sgordon | arent you reliant on someone actually confirming the resize to get to that point | 20:50 |
| sgordon | oh nm | 20:50 |
| *** heyongli has joined #openstack-nova | 20:51 | |
| cfriesen_ | dansmith: the problem is that if we do a delete after the new host is set but before we set vm_states.RESIZED, then nobody cleans up the resources on the source host. | 20:51 |
| cfriesen_ | dansmith: or at least nobody cleans up the neutron ports | 20:51 |
| *** aysyd has joined #openstack-nova | 20:52 | |
| *** huats has quit IRC | 20:52 | |
| vilobhmm11 | bauzas : can you please review this https://review.openstack.org/#/c/241016/ | 20:52 |
| dansmith | cfriesen_: seems like a thing to fix, it would just be good to not solve it by blocking delete | 20:52 |
| *** edtubill has quit IRC | 20:53 | |
| cfriesen_ | dansmith: I think git commit 18d6b5cc added an audit to clean up the instance files left by a similar case. | 20:55 |
| mriedem | sdague: i'm interested in your thoughts on this https://review.openstack.org/#/c/240271/ | 20:55 |
| *** raildo is now known as raildo-afk | 20:56 | |
| dansmith | cfriesen_: yep | 20:56 |
| cfriesen_ | dansmith: would it make sense to try to modify that to also clean up neutron ports? Or would you prefer to get rid of that audit and clean everything up in the resize/migration code? (Which I suspect might involve a new RPC call.) | 20:56 |
| bauzas | vilobhmm11: very slowly catching up my review backlog from the summit, but FWIW CC'd your change | 20:57 |
| dansmith | cfriesen_: seems like adding that (carefully) to the cleanup would be reasonable, | 20:58 |
| dansmith | cfriesen_: however these are different cases since the user sees the port residue and doesn't see the image residue | 20:58 |
| dansmith | so cleaning the images up later is not a problem | 20:58 |
| dansmith | having ports remain and then disappear minutes later may be confusing | 20:58 |
| cfriesen_ | dansmith: true enough. I'll poke around a bit more, see if I can come up with anything better. | 21:00 |
| mriedem | can you check something when refreshing the network info cache? | 21:00 |
| *** heyongli has quit IRC | 21:00 | |
| *** haomaiwang has quit IRC | 21:01 | |
| *** edtubill has joined #openstack-nova | 21:01 | |
| mriedem | the port has a host in it | 21:01 |
| *** heyongli has joined #openstack-nova | 21:01 | |
| mriedem | if port.host != CONF.host, unplug? | 21:01 |
| mriedem | well, not that easy | 21:01 |
| *** haomaiwa_ has joined #openstack-nova | 21:01 | |
| cfriesen_ | mriedem: I'm not really familiar with that code. | 21:01 |
| mriedem | i was thinking of how host is used here https://github.com/openstack/nova/blob/master/nova/network/neutronv2/api.py#L1840 | 21:03 |
| mriedem | which is called when migration/resize finishes | 21:03 |
| *** achanda has quit IRC | 21:03 | |
| *** edtubill has quit IRC | 21:04 | |
| *** e0ne has quit IRC | 21:06 | |
| vilobhmm11 | bauzas : np…jfyi , the agrregatespec filter seems broken and would lead to perf issues if this is not fixed…as this change allows to return immediately if extra_specs are empty | 21:06 |
| *** yassine__ has quit IRC | 21:07 | |
| cfriesen_ | mriedem: I think what's missing is a call to virt.libvirt.driver.LibvirtDriver.unplug_vifs() (at least when using libvirt). So I'm not sure we can do anything from just in the neutron API. | 21:07 |
| mriedem | cfriesen_: yeah the neutron api code wouldn't be doing it | 21:07 |
| mriedem | but we refresh the network info cache in lots of places, | 21:08 |
| mriedem | which should have the port host in it | 21:08 |
| mriedem | and i was wondering if that could be used for some kind of logic in delete | 21:08 |
| cfriesen_ | mriedem: normally this would be called from confirm_migration(), which is called by delete once the resize/migration gets far enough along | 21:08 |
| cfriesen_ | I wonder if the delete code could just confirm the resize on the source host even if the migration hasn't reached the vm_states.RESIZED state | 21:10 |
| *** READ10 has quit IRC | 21:10 | |
| *** heyongli has quit IRC | 21:10 | |
| mriedem | so are you looking to delete the port? | 21:11 |
| *** dave-mccowan has joined #openstack-nova | 21:11 | |
| *** heyongli has joined #openstack-nova | 21:11 | |
| dansmith | doesn't it matter if the neutron port has been updated for the new host? | 21:12 |
| dansmith | either way, I'm not sure why we need to involve the source host | 21:12 |
| dansmith | the destination should be able to cleanup everything, right? | 21:12 |
| *** salv-orlando has quit IRC | 21:14 | |
| *** rk4n has quit IRC | 21:14 | |
| cfriesen_ | dansmith: we've got a custom vswitch running on each compute node. the lack of cleanup on the source leaves a vswitch port allocated on the source host. | 21:14 |
| mriedem | i was thinking about this more from a generic orphaned ports standpoint, | 21:14 |
| mriedem | and wondering if neutron could send an event to cleanup | 21:14 |
| dansmith | cfriesen_: so you're talking about unplug, not port delete? | 21:14 |
| cfriesen_ | dansmith: at least I think that's what is happening. the networking details belong to someone else | 21:14 |
| cfriesen_ | dansmith: yes, I believe so. | 21:14 |
| *** salv-orlando has joined #openstack-nova | 21:14 | |
| dansmith | cfriesen_: I officially care zero amount about your custom network stuff, of course :) | 21:15 |
| *** otter768 has joined #openstack-nova | 21:15 | |
| cfriesen_ | dansmith: oh, sure | 21:15 |
| cfriesen_ | :) | 21:15 |
| cfriesen_ | but I think upstream will be susceptible to similar issues | 21:15 |
| *** su_zhang has joined #openstack-nova | 21:15 | |
| mriedem | cfriesen_: you might try to track down arosen to talk about orphaned port cleanup ideas, i know he and i used to talk about that a few times | 21:16 |
| dansmith | it'd be good if you could prove that before continuing down this path | 21:16 |
| dansmith | because if not, knowing why is probably your quickest path to success | 21:16 |
| mriedem | those cleanup ideas usually invovled a periodic task though | 21:16 |
| dansmith | mriedem: those are port cleanup things, not local unplug things I think | 21:16 |
| cfriesen_ | dansmith: fair enough. I figured that the introduction of the cleanup audit for the instance files meant that cleanup wasn't happening. :) | 21:16 |
| dansmith | I think the image problem is sufficiently different | 21:17 |
| dansmith | obviously if we're susceptible to it with standard stuff, then we should fix it | 21:17 |
| *** rowagner has quit IRC | 21:18 | |
| cfriesen_ | is there an easy way to check whether the network cleanup is being done? I should be able to reproduce the race easily enough, just not sure what to check for with the standard vswitch. | 21:18 |
| dansmith | I think you'll have to ask neutron people that know about vswitch for that | 21:19 |
| cfriesen_ | okay | 21:19 |
| mriedem | i was thinking more like neutron sending an network-vif-deleted event which then nova does an unplug, but at that point the instance would be deleted so it's too late | 21:19 |
| dansmith | that you're detecting this problem through some non-standard interface to non-standard ovs makes me really curious | 21:20 |
| *** otter768 has quit IRC | 21:20 | |
| *** pratikmallya has quit IRC | 21:20 | |
| dansmith | mriedem: but we don't do that for the source host currently, only for the new one, AFAIK | 21:20 |
| *** heyongli has quit IRC | 21:21 | |
| mriedem | yeah, idk, seems like this would be hard to handle depending on where things are deleted in the migration operation | 21:21 |
| mriedem | a periodic task is easy for cleanup - thigns are gone, cleanup what yo ucan | 21:21 |
| *** heyongli has joined #openstack-nova | 21:21 | |
| cfriesen_ | dansmith: the output from our tool looks like this: http://paste.openstack.org/show/478015/ The "avp-guest" ones with a "link-state" of "down" are ones that have been "leaked" by my testing. There's only one instance running with only one port. | 21:21 |
| mriedem | also, don't delete things in the middle of a resize and expect everything to be hunky dory i guess | 21:21 |
| dansmith | cfriesen_: I can't click that link or I'll burst into flame | 21:22 |
| mriedem | oh cmon, it loaded in less than 10 seconds | 21:22 |
| dansmith | mriedem: I meant because it contains unclean things | 21:22 |
| cfriesen_ | dansmith: got it. :) joy of legal | 21:22 |
| mriedem | 80085 like on a calculator? | 21:22 |
| mriedem | alternatively, cfriesen_ implements the task work so a delete can cancel an in-flight migration which does it's proper cleanup | 21:23 |
| cfriesen_ | mriedem: not likely. :) | 21:24 |
| *** rfolco has joined #openstack-nova | 21:26 | |
| *** e0ne has joined #openstack-nova | 21:26 | |
| mriedem | so, this happened in my city today http://www.postbulletin.com/news/crime/broadway-avenue-businesses-closed-for-bomb/article_b1470825-9a4b-511a-b2e4-edf0274deb8c.html | 21:27 |
| mriedem | "The incident began about noon, when a woman called 911 to report a man driving erratically, yelling and masturbating near Second Street and Third Avenue Southwest. " | 21:27 |
| *** devananda has left #openstack-nova | 21:27 | |
| *** pece has quit IRC | 21:27 | |
| *** devananda has joined #openstack-nova | 21:27 | |
| rlrossit | I didn't see mriedem around noon................ | 21:27 |
| rlrossit | mriedem: does tend to yell... | 21:28 |
| dansmith | hah | 21:28 |
| *** rfolco has quit IRC | 21:29 | |
| rlrossit | should we switch up the meetup from bristol to rochester again? | 21:29 |
| *** e0ne has quit IRC | 21:29 | |
| *** tjones has joined #openstack-nova | 21:29 | |
| *** jamielennox is now known as jamielennox|away | 21:30 | |
| *** baoli has joined #openstack-nova | 21:30 | |
| *** heyongli has quit IRC | 21:31 | |
| *** gonzalo2kx has joined #openstack-nova | 21:31 | |
| *** heyongli has joined #openstack-nova | 21:32 | |
| *** haomaiwa_ has quit IRC | 21:33 | |
| mriedem | sfinucan: do you need to update the spec for this? https://review.openstack.org/#/c/202647/ | 21:34 |
| dansmith | mriedem: that test patch passed cells | 21:34 |
| dansmith | mriedem: are we going to recheck it a bit? | 21:34 |
| mriedem | dansmith: yeah, lots of things are passing cells | 21:34 |
| mriedem | which is why i don't want to skip the test yet | 21:34 |
| *** mwagner has quit IRC | 21:34 | |
| bauzas | so I looked at the object | 21:34 |
| bauzas | and something is weird | 21:35 |
| bauzas | https://github.com/openstack/nova/blob/master/nova/objects/block_device.py#L132-L148 | 21:35 |
| bauzas | given update_or_create = None | 21:35 |
| sfinucan | mriedem: I don't know, tbh | 21:35 |
| bauzas | or False rather | 21:36 |
| mriedem | sfinucan: well, i guess question is, should the enum be 'require' or 'prefer' | 21:36 |
| mriedem | since they mean different thigns | 21:36 |
| mriedem | we should probably be specific | 21:36 |
| bauzas | then it will call db.bdm_create and then call the messaging.bdm_update_or_create, right? | 21:36 |
| *** jerrygb has quit IRC | 21:36 | |
| sfinucan | Oh, I thought you were asking something else | 21:36 |
| *** haomaiwang has joined #openstack-nova | 21:36 | |
| sfinucan | I think I reworded that in the spec | 21:37 |
| *** jamielennox|away is now known as jamielennox | 21:37 | |
| sfinucan | mriedem: Yes, I did https://review.openstack.org/#/c/241355/1//COMMIT_MSG | 21:37 |
| sfinucan | The purpose remains the same. The 'prefer' name didn't make sense, however | 21:37 |
| *** jerrygb has joined #openstack-nova | 21:38 | |
| sfinucan | If you "prefer" something but then go reject all other options, you're not really "preferring" it :) | 21:38 |
| mriedem | sfinucan: ok i was looking at the liberty spec then, thanks | 21:38 |
| mriedem | i'd like the enum descriptions from the spec in the code where they are defined too, but that could be a separate change so you don't have to rebase the series | 21:39 |
| sfinucan | No problem. Cheers for the feedback | 21:39 |
| *** aix has joined #openstack-nova | 21:39 | |
| sfinucan | Yes, I need to address feedback edleafe to a similar end | 21:39 |
| sfinucan | Actually, check in case a later patch addresses that particular issue | 21:40 |
| *** e0ne has joined #openstack-nova | 21:40 | |
| *** baoli has quit IRC | 21:40 | |
| mriedem | bauzas: yeah, but i don't think we're getting called there with update_or_create=False | 21:40 |
| sfinucan | The comments from edleafe concerned UTs. I'm going to address those in a follow up patch to avoid rebasing/gate issues | 21:40 |
| bauzas | mriedem: so I wonder why we have 2 messaging calls | 21:40 |
| mriedem | bauzas: b/c of this https://github.com/openstack/nova/blob/master/nova/compute/api.py#L1281 | 21:40 |
| mriedem | bauzas: that'd be a question for alaski | 21:41 |
| *** e0ne has quit IRC | 21:41 | |
| *** heyongli has quit IRC | 21:41 | |
| mriedem | different db's i guess? | 21:41 |
| bauzas | mriedem: what I wonder then is we would have a bdm.update_or_create followed by a bdm.savez | 21:41 |
| bauzas | mriedem: no | 21:41 |
| bauzas | mriedem: because it's _at_top() | 21:42 |
| *** baoli has joined #openstack-nova | 21:42 | |
| bauzas | which means it's the API DB | 21:42 |
| *** heyongli has joined #openstack-nova | 21:42 | |
| bauzas | not the child d | 21:42 |
| bauzas | db | 21:42 |
| mriedem | well, | 21:42 |
| mriedem | if cell_type == 'compute' and db_bdm.get('device_name') is not None: | 21:42 |
| mriedem | so we don't do the messaging stuff to the cells api if we're already in the api db right? | 21:43 |
| mriedem | i just noticed this https://github.com/openstack/nova/blob/master/nova/objects/block_device.py#L118 | 21:44 |
| bauzas | by cells api, you mean the compute cells api ? | 21:44 |
| mriedem | why don't we have that in save()? | 21:44 |
| dansmith | hmm | 21:45 |
| dansmith | maybe cells is creating the BDM at the top via a cells message going up from the child? | 21:45 |
| bauzas | zactly | 21:46 |
| bauzas | that's my guess | 21:46 |
| *** baoli has quit IRC | 21:46 | |
| *** baoli has joined #openstack-nova | 21:46 | |
| bauzas | https://github.com/openstack/nova/blob/master/nova/compute/cells_api.py#L204 | 21:46 |
| bauzas | mriedem: ^ to answer your question | 21:46 |
| *** pratikmallya has joined #openstack-nova | 21:46 | |
| dansmith | bauzas: well, | 21:47 |
| dansmith | bauzas: that only answers the question if it's creating it without the object | 21:47 |
| dansmith | but it probably is | 21:47 |
| *** Piet has quit IRC | 21:47 | |
| mriedem | but bdm.save() can send the create up to the parent right? | 21:47 |
| *** baoli has quit IRC | 21:47 | |
| mriedem | b/c of https://github.com/openstack/nova/blob/master/nova/objects/block_device.py#L193 | 21:48 |
| dansmith | in the child | 21:48 |
| bauzas | mriedem: that's called by the child IMHO | 21:48 |
| bauzas | mriedem: which then uses the messaging proxy to ask the parent to save it | 21:48 |
| dansmith | yeah, which is creating it with the raw DB method | 21:49 |
| dansmith | https://github.com/openstack/nova/blob/master/nova/cells/messaging.py#L1156-L1194 | 21:49 |
| *** baoli has joined #openstack-nova | 21:49 | |
| openstackgerrit | Mark Doffman proposed openstack/nova: Cleanup rescue images on unrescue when using RBD. https://review.openstack.org/241777 | 21:50 |
| bauzas | that looks like a way to sync the BDM update between the child cell DB and the API DB | 21:51 |
| *** heyongli has quit IRC | 21:51 | |
| bauzas | because it creates a row for the child DB before calling the messaging stuff to call the parent to also save it | 21:51 |
| dansmith | right | 21:51 |
| dansmith | so we don't create at the api with the object, we punt, | 21:51 |
| dansmith | create with the object in the child, and then call back to the parent to do this special create-at-top function | 21:52 |
| bauzas | yup | 21:52 |
| dansmith | just a *tad circuitous | 21:52 |
| *** heyongli has joined #openstack-nova | 21:52 | |
| *** aysyd has quit IRC | 21:52 | |
| bauzas | which can be racy then | 21:53 |
| *** baoli has quit IRC | 21:53 | |
| dansmith | I wasn't paying attention to the early triage, so I'm not actually sure what is happening | 21:53 |
| *** baoli has joined #openstack-nova | 21:53 | |
| bauzas | given 2 calls for updating a BDM | 21:54 |
| bauzas | since it's a RPC payload | 21:54 |
| bauzas | and then going to the parent api cell, it could then be racy I guess | 21:54 |
| *** edtubill has joined #openstack-nova | 21:54 | |
| dansmith | well, what really should be happening is the device name gets allocated at the child and then the parent create should be deterministic I think | 21:55 |
| dansmith | right? | 21:55 |
| bauzas | per the logs, the device names are set | 21:56 |
| mriedem | well | 21:56 |
| *** rdopiera has quit IRC | 21:56 | |
| bauzas | before calling the messaging proxy | 21:56 |
| mriedem | from the logs we have | 21:56 |
| mriedem | but the values in the cells region logs for the bdm create or update do have the device_name in them yeah | 21:56 |
| dansmith | so what actually happens? | 21:56 |
| bauzas | so the problem is that the parent decides twice to create a new entry | 21:56 |
| cfriesen_ | dansmith: mriedem: just thought of something...what about cleaning up PCI devices that were allocated to instances? would we maybe leak those too if we delete during resize? | 21:57 |
| dansmith | cfriesen_: you can't migrate things with pci devices | 21:57 |
| dansmith | currently | 21:57 |
| cfriesen_ | dansmith: ah....I think we've got someone proposing a fix for that. | 21:58 |
| *** baoli has quit IRC | 21:59 | |
| *** baoli has joined #openstack-nova | 21:59 | |
| *** achanda has joined #openstack-nova | 22:00 | |
| *** haomaiwang has quit IRC | 22:01 | |
| *** su_zhang has quit IRC | 22:01 | |
| mriedem | i cannot get https://review.openstack.org/#/c/241397/ to fail | 22:01 |
| *** haomaiwang has joined #openstack-nova | 22:01 | |
| bauzas | heh | 22:01 |
| *** heyongli has quit IRC | 22:02 | |
| mriedem | all of the changes under it can't pass the cells job though | 22:02 |
| mriedem | wtf | 22:02 |
| bauzas | so, given we have 2 messages for updating the api cell with 2 identical BDMs, it means that 2 calls are sent | 22:02 |
| *** heyongli has joined #openstack-nova | 22:02 | |
| bauzas | from the object, there are actually 2 places where you can fire a message to the parent cell : either .create_or_update() or save() | 22:03 |
| bauzas | so, it's either that we call twice in the code create_or_update() or that we call both the above | 22:04 |
| *** baoli has quit IRC | 22:04 | |
| mriedem | i know we call both | 22:04 |
| *** baoli has joined #openstack-nova | 22:04 | |
| bauzas | where? that's what I'm looking at now | 22:04 |
| mriedem | oh nvm, _create_block_device_mapping shouldn't happen for cells | 22:05 |
| mriedem | from what i can tell, that's the only thing that calls bdm.create_or_update | 22:05 |
| sdague | mriedem: commented on the nova-manage logging review | 22:07 |
| *** e0ne has joined #openstack-nova | 22:07 | |
| *** e0ne has quit IRC | 22:08 | |
| *** mdorman has joined #openstack-nova | 22:10 | |
| *** y_sawai has quit IRC | 22:11 | |
| *** heyongli has quit IRC | 22:12 | |
| *** su_zhang has joined #openstack-nova | 22:12 | |
| *** heyongli has joined #openstack-nova | 22:13 | |
| *** pratikmallya has quit IRC | 22:13 | |
| mdorman | so i understand that this blueprint needs to have a spec written up in order to move forward: https://blueprints.launchpad.net/nova/+spec/cells-v1-interface-events | 22:14 |
| mdorman | (i was the one that posted the original patch which led to this.) i’m happy to work on writing up the spec, would i just do the normal nova-spec process, or is this a different situation since a blueprint already exists? | 22:15 |
| dansmith | mdorman: I thought we nak'd this? | 22:15 |
| mdorman | tbh, unsure what the status is. last i heard there was an ask for a spec | 22:15 |
| *** baoli has quit IRC | 22:16 | |
| mdorman | are you saying this is rejected? | 22:16 |
| mriedem | the code change would require a bp and the bp would require a spec imo b/c it requires a cells + neutron job, which we don't have | 22:16 |
| mriedem | and which i started playing with before summit | 22:16 |
| mdorman | yeah i saw that other patch for that | 22:16 |
| dansmith | mdorman: I thought the string of deps was getting scary for it | 22:16 |
| mriedem | there is concern about it detracting attention from cells v2 work | 22:17 |
| mdorman | i guess what i really want to know is: as an operator, what can i do to help with this? if anything… | 22:17 |
| mriedem | maybe it should be on the nova meeting agenda | 22:17 |
| mriedem | before you write a spec | 22:17 |
| mriedem | or anyone | 22:17 |
| mdorman | yeah, i mean the thinking from the large deployers team is to just try to get some of these v1 patches into the tree so that we can stop carrying them locally. understand that the major efforts are against v2 now, but we’re willing to help shepherd these v1 patches in, since we all need them. | 22:17 |
| mdorman | alright. | 22:17 |
| mdorman | what’s the proper way to get that as an agenda item? | 22:18 |
| dansmith | mdorman: and that's noble, it's just that we have said we're not doing any new features or even large bug fixes for v1, and especially if we don't have test coverage... | 22:18 |
| dansmith | mdorman: put it on the agenda wiki | 22:19 |
| mdorman | dansmith: yeah that’s fair, i understand that perspective. | 22:19 |
| mriedem | mdorman: just fit it in here https://wiki.openstack.org/wiki/Meetings/Nova | 22:19 |
| *** sneti has quit IRC | 22:19 | |
| mriedem | probably under open discussion | 22:19 |
| mdorman | kk will do | 22:20 |
| *** baoli has joined #openstack-nova | 22:20 | |
| mriedem | bauzas: dansmith: so i'm assuming that if we're hitting bdm.save() twice, it's b/c of this as the 2nd call: | 22:20 |
| mriedem | http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cpu.txt.gz#_2015-11-03_21_44_58_252 | 22:20 |
| mriedem | there is code in the compute manager where it sets the bdm.device_name and does bdm.save() for the root bdm, | 22:20 |
| *** jwcroppe has quit IRC | 22:21 | |
| mdorman | mriedem dansmith thanks for guidance! | 22:21 |
| mriedem | and then calls the virt driver to set default device names, which for libvirt means nulling it out and saving | 22:21 |
| mriedem | those would happen right after another | 22:21 |
| *** andrewbogott has joined #openstack-nova | 22:22 | |
| *** heyongli has quit IRC | 22:22 | |
| bauzas | mriedem: speaking of https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L1690 ? | 22:22 |
| mriedem | bauzas: yup | 22:23 |
| mriedem | and then that calls _default_device_names_for_instance which goes down into the virt driver which nulls it out | 22:23 |
| *** heyongli has joined #openstack-nova | 22:23 | |
| mriedem | https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L7236 | 22:23 |
| mriedem | then when that calls blockinfo.default_device_names... | 22:23 |
| bauzas | right | 22:23 |
| mriedem | https://github.com/openstack/nova/blob/master/nova/virt/libvirt/blockinfo.py#L468 | 22:23 |
| mriedem | on that 2nd one from the libvirt driver, when we get to the db | 22:24 |
| mriedem | https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L3945 | 22:24 |
| mriedem | device_name isn't set so that evals to False and we do an insert | 22:24 |
| bauzas | yup | 22:24 |
| bauzas | two saves | 22:25 |
| *** lbelivea has quit IRC | 22:25 | |
| bauzas | and the second one having a nullable device name | 22:25 |
| *** yassine__ has joined #openstack-nova | 22:25 | |
| bauzas | hem, sec, no | 22:26 |
| *** daemontool has joined #openstack-nova | 22:26 | |
| bauzas | http://logs.openstack.org/66/241366/1/check/gate-tempest-dsvm-cells/1d7551e/logs/screen-n-cell-region.txt.gz#_2015-11-03_21_44_58_332 was pointing out that device_name was also set for the 2nd call | 22:26 |
| *** achanda has quit IRC | 22:26 | |
| mriedem | honestly i don't know why device_name is not null on that 2nd call | 22:28 |
| bauzas | I'm tempted to provide a DNM with a print_stack() :) | 22:29 |
| mriedem | blockinfo.get_info_from_bdm() might be re-setting the bdm.device_name before the 2nd bdm.save() | 22:30 |
| *** achanda has joined #openstack-nova | 22:30 | |
| *** pumaranikar has quit IRC | 22:31 | |
| mriedem | the 2nd save is where device_type and disk_bus is set | 22:31 |
| mriedem | b/c of get_info_from_bdm | 22:31 |
| *** pumaranikar has joined #openstack-nova | 22:31 | |
| bauzas | ++ | 22:31 |
| *** baoli has quit IRC | 22:31 | |
| bauzas | okay, I feel we're quite finding the root problem | 22:32 |
| *** lbelivea has joined #openstack-nova | 22:32 | |
| *** baoli has joined #openstack-nova | 22:32 | |
| *** rlrossit has left #openstack-nova | 22:32 | |
| *** heyongli has quit IRC | 22:32 | |
| *** mmedvede has quit IRC | 22:32 | |
| bauzas | now, I wonder if removing the first save() should be fine | 22:32 |
| *** mmedvede has joined #openstack-nova | 22:33 | |
| *** heyongli has joined #openstack-nova | 22:33 | |
| bauzas | because I'd prefer to find a way to avoid changing a scheme | 22:33 |
| mriedem | i'm not really sure where the first save is coming from | 22:34 |
| mriedem | if it's really this https://github.com/openstack/nova/blob/master/nova/compute/manager.py#L1690 | 22:34 |
| mriedem | we could use some debug logging in the conditions in _default_block_device_names | 22:36 |
| mriedem | anyway, i have to head out to pick up my kid | 22:36 |
| *** pumaranikar has quit IRC | 22:36 | |
| *** smurke has quit IRC | 22:36 | |
| *** baoli has quit IRC | 22:37 | |
| *** mriedem has quit IRC | 22:37 | |
| *** baoli has joined #openstack-nova | 22:37 | |
| bauzas | laters | 22:39 |
| andrewbogott | I’m looking at the default nova policy.json and, wow, it sure has a lot of “” policies. | 22:40 |
| andrewbogott | I must not understand how policies work, because | 22:40 |
| andrewbogott | "compute:resize": "", | 22:40 |
| andrewbogott | looks to me like any user can resize any instance in any project | 22:40 |
| andrewbogott | what am I missing? | 22:40 |
| *** heyongli has quit IRC | 22:43 | |
| *** heyongli has joined #openstack-nova | 22:43 | |
| *** baoli has quit IRC | 22:46 | |
| *** mwagner has joined #openstack-nova | 22:48 | |
| *** baoli has joined #openstack-nova | 22:48 | |
| *** pratikmallya has joined #openstack-nova | 22:49 | |
| *** pratikma_ has joined #openstack-nova | 22:50 | |
| *** edtubill has quit IRC | 22:51 | |
| *** tonytan4ever has quit IRC | 22:51 | |
| *** heyongli has quit IRC | 22:53 | |
| *** heyongli has joined #openstack-nova | 22:54 | |
| *** pratikmallya has quit IRC | 22:55 | |
| *** stackdump has joined #openstack-nova | 22:55 | |
| *** ashyoung has joined #openstack-nova | 22:55 | |
| *** jichen has quit IRC | 22:59 | |
| *** baoli has quit IRC | 23:00 | |
| *** baoli has joined #openstack-nova | 23:00 | |
| *** haomaiwang has quit IRC | 23:01 | |
| *** haomaiwa_ has joined #openstack-nova | 23:01 | |
| *** notmorgan has joined #openstack-nova | 23:03 | |
| *** heyongli has quit IRC | 23:03 | |
| *** mdrabe has quit IRC | 23:03 | |
| *** pradk has quit IRC | 23:04 | |
| *** heyongli has joined #openstack-nova | 23:04 | |
| *** stackdump has quit IRC | 23:04 | |
| *** alex_klimov has quit IRC | 23:04 | |
| mgagne | is there a reason for network_data.json being injected within this loop instead of with the others above? https://github.com/openstack/nova/blob/master/nova/api/metadata/base.py#L481-L483 | 23:05 |
| *** yassine__ has quit IRC | 23:06 | |
| mgagne | I tried to loosely backport the patch to kilo and when injection is done from within the loop, network_data.json doesn't appear in openstack/2015-10-15/ as expected. When moved outside the loop, everything is fine. | 23:07 |
| *** baoli has quit IRC | 23:11 | |
| *** stackdump has joined #openstack-nova | 23:11 | |
| *** baoli has joined #openstack-nova | 23:12 | |
| *** su_zhang has quit IRC | 23:12 | |
| *** dvlwrk_ has quit IRC | 23:13 | |
| *** heyongli has quit IRC | 23:13 | |
| *** heyongli has joined #openstack-nova | 23:14 | |
| *** dave-mccowan has quit IRC | 23:14 | |
| *** tjones has quit IRC | 23:15 | |
| *** tjones has joined #openstack-nova | 23:15 | |
| *** otter768 has joined #openstack-nova | 23:16 | |
| *** dustins has quit IRC | 23:18 | |
| *** diogogmt has quit IRC | 23:20 | |
| *** dave-mccowan has joined #openstack-nova | 23:20 | |
| *** otter768 has quit IRC | 23:21 | |
| *** su_zhang has joined #openstack-nova | 23:21 | |
| *** burgerk has quit IRC | 23:22 | |
| *** baoli has quit IRC | 23:22 | |
| *** baoli has joined #openstack-nova | 23:23 | |
| *** heyongli has quit IRC | 23:24 | |
| *** thorst_ has quit IRC | 23:24 | |
| *** ctrath has quit IRC | 23:24 | |
| *** heyongli has joined #openstack-nova | 23:24 | |
| *** edmondsw has quit IRC | 23:25 | |
| *** diogogmt has joined #openstack-nova | 23:26 | |
| *** burt has quit IRC | 23:26 | |
| *** mriedem has joined #openstack-nova | 23:33 | |
| *** heyongli has quit IRC | 23:34 | |
| *** signed8bit is now known as signed8bit_ZZZzz | 23:35 | |
| *** heyongli has joined #openstack-nova | 23:35 | |
| *** baoli has quit IRC | 23:35 | |
| *** zhangjn has joined #openstack-nova | 23:39 | |
| *** signed8bit_ZZZzz is now known as signed8bit | 23:39 | |
| *** otter768 has joined #openstack-nova | 23:41 | |
| *** baoli has joined #openstack-nova | 23:43 | |
| *** heyongli has quit IRC | 23:44 | |
| *** haomaiwa_ has quit IRC | 23:45 | |
| *** jaypipes has quit IRC | 23:45 | |
| *** heyongli has joined #openstack-nova | 23:45 | |
| *** haomaiwa_ has joined #openstack-nova | 23:45 | |
| *** inteq has quit IRC | 23:47 | |
| *** _RuiChen_ has quit IRC | 23:47 | |
| *** mrkz has quit IRC | 23:48 | |
| *** dave-mccowan has quit IRC | 23:49 | |
| *** baoli has quit IRC | 23:52 | |
| *** n0ano has joined #openstack-nova | 23:52 | |
| *** baoli has joined #openstack-nova | 23:53 | |
| *** heyongli has quit IRC | 23:54 | |
| *** heyongli has joined #openstack-nova | 23:55 | |
| *** jerrygb has quit IRC | 23:56 | |
| *** ijuwang has quit IRC | 23:57 | |
| *** zhangjn has quit IRC | 23:57 | |
| *** sfinucan has quit IRC | 23:58 | |
| *** takedakn has joined #openstack-nova | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!