*** dsanders has quit IRC | 00:04 | |
*** dimsum__ has joined #openstack-nova | 00:06 | |
*** sdake_ has joined #openstack-nova | 00:08 | |
*** sdake_ has quit IRC | 00:08 | |
*** sdake_ has joined #openstack-nova | 00:09 | |
*** sdake has quit IRC | 00:12 | |
*** dimsum__ has quit IRC | 00:14 | |
*** otter768 has joined #openstack-nova | 00:23 | |
*** jogo has quit IRC | 00:25 | |
openstackgerrit | Eric Brown proposed openstack/nova: Minor edits to support-matrix doc https://review.openstack.org/181432 | 00:27 |
---|---|---|
openstackgerrit | sajeesh cs proposed openstack/nova: Create column allocated in Quota table https://review.openstack.org/151327 | 00:29 |
*** Alexandra_ has quit IRC | 00:32 | |
*** yingjun has joined #openstack-nova | 00:36 | |
openstackgerrit | Sam Morrison proposed openstack/nova: Create neutron ports with correct instance AZ value https://review.openstack.org/181776 | 00:36 |
*** dimsum__ has joined #openstack-nova | 00:39 | |
openstackgerrit | Eric Brown proposed openstack/nova: Deprecate nova ironic driver's admin_auth_token https://review.openstack.org/179960 | 00:40 |
*** yingjun has quit IRC | 00:50 | |
*** yamahata has quit IRC | 00:55 | |
*** lsmola_ has joined #openstack-nova | 01:01 | |
*** mriedem has joined #openstack-nova | 01:02 | |
*** lsmola has quit IRC | 01:04 | |
*** mriedem1 has quit IRC | 01:04 | |
*** lsmola_ has quit IRC | 01:11 | |
*** ericksonsantos has quit IRC | 01:12 | |
*** otter768 has quit IRC | 01:12 | |
*** ericksonsantos has joined #openstack-nova | 01:16 | |
*** takedakn has joined #openstack-nova | 01:20 | |
*** yingjun has joined #openstack-nova | 01:20 | |
*** lsmola_ has joined #openstack-nova | 01:25 | |
*** zhangtralon has joined #openstack-nova | 01:27 | |
*** liusheng has quit IRC | 01:27 | |
*** liusheng has joined #openstack-nova | 01:31 | |
*** vilobhmm11 has quit IRC | 01:31 | |
*** takedakn has quit IRC | 01:31 | |
*** coreywright has quit IRC | 01:34 | |
*** markvoelker has quit IRC | 01:35 | |
*** markvoelker has joined #openstack-nova | 01:36 | |
openstackgerrit | sajeesh cs proposed openstack/nova: Create column allocated in Quota table https://review.openstack.org/151327 | 01:37 |
*** markvoelker has quit IRC | 01:37 | |
*** coreywright has joined #openstack-nova | 01:39 | |
*** baoli has joined #openstack-nova | 01:39 | |
*** vilobhmm1 has joined #openstack-nova | 01:41 | |
*** baoli has quit IRC | 01:44 | |
openstackgerrit | Eli Qiao proposed openstack/nova: Set correct power state when evaucate fails due to recreate not supported https://review.openstack.org/173150 | 01:45 |
*** vilobhmm1 has quit IRC | 01:46 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: libvirt: handle code=38 + sigkill (ebusy) in destroy() https://review.openstack.org/181781 | 01:49 |
*** yingjun has quit IRC | 01:52 | |
*** mriedem has quit IRC | 01:53 | |
*** yingjun has joined #openstack-nova | 01:54 | |
*** achanda has joined #openstack-nova | 01:55 | |
*** oomichi has joined #openstack-nova | 02:01 | |
*** oomichi has quit IRC | 02:01 | |
*** yingjun has quit IRC | 02:09 | |
*** dsanders has joined #openstack-nova | 02:10 | |
*** dimsum__ has quit IRC | 02:12 | |
*** otter768 has joined #openstack-nova | 02:14 | |
*** ericksonsantos has quit IRC | 02:19 | |
*** mmedvede_ has quit IRC | 02:20 | |
*** jogo has joined #openstack-nova | 02:21 | |
*** yasu_ has joined #openstack-nova | 02:22 | |
*** park_heijlong has joined #openstack-nova | 02:22 | |
*** vilobhmm1 has joined #openstack-nova | 02:25 | |
openstackgerrit | heijlong proposed openstack/nova: Merge user_data functional tests of v2 and v2.1 https://review.openstack.org/168265 | 02:31 |
*** yingjun has joined #openstack-nova | 02:33 | |
openstackgerrit | heijlong proposed openstack/nova: Merge extended-volume functional tests of v2 and v2.1 https://review.openstack.org/173216 | 02:37 |
*** arnaud____ has joined #openstack-nova | 02:37 | |
*** yingjun has quit IRC | 02:39 | |
*** arnaud____ has quit IRC | 02:39 | |
*** Alexandra_ has joined #openstack-nova | 02:40 | |
*** patrickeast has quit IRC | 02:52 | |
*** vilobhmm1 has quit IRC | 02:54 | |
*** boris-42 has quit IRC | 02:58 | |
*** vilobhmm1 has joined #openstack-nova | 03:00 | |
*** dave-mccowan has quit IRC | 03:00 | |
*** achanda has quit IRC | 03:01 | |
*** achanda_ has joined #openstack-nova | 03:04 | |
*** sdake has joined #openstack-nova | 03:05 | |
*** sdake has quit IRC | 03:05 | |
*** sdake has joined #openstack-nova | 03:05 | |
*** sdake_ has quit IRC | 03:06 | |
*** sdake has quit IRC | 03:08 | |
*** sdake has joined #openstack-nova | 03:09 | |
*** iamjarvo has joined #openstack-nova | 03:13 | |
*** achanda_ has quit IRC | 03:14 | |
*** tobe has joined #openstack-nova | 03:20 | |
*** alex_xu has quit IRC | 03:20 | |
*** baoli has joined #openstack-nova | 03:21 | |
*** alex_xu has joined #openstack-nova | 03:21 | |
*** yingjun has joined #openstack-nova | 03:22 | |
openstackgerrit | heijlong proposed openstack/nova: Merge user_data functional tests of v2 and v2.1 https://review.openstack.org/168265 | 03:23 |
*** vilobhmm1 has quit IRC | 03:24 | |
*** VW_ has joined #openstack-nova | 03:26 | |
*** baoli has quit IRC | 03:26 | |
*** Zhengguang has joined #openstack-nova | 03:27 | |
*** yingjun has quit IRC | 03:27 | |
*** yingjun has joined #openstack-nova | 03:29 | |
*** yingjun_ has joined #openstack-nova | 03:33 | |
*** yingjun has quit IRC | 03:33 | |
*** yingjun has joined #openstack-nova | 03:34 | |
*** yingjun_ has quit IRC | 03:38 | |
*** subscope has joined #openstack-nova | 03:38 | |
*** deepthi has joined #openstack-nova | 03:48 | |
*** yingjun has quit IRC | 03:52 | |
*** sdake_ has joined #openstack-nova | 03:56 | |
sorrison | anyone around and can help me with a failing tempest gate job https://review.openstack.org/#/c/181776/ ? | 03:58 |
*** sdake has quit IRC | 04:00 | |
*** VW_ has quit IRC | 04:06 | |
*** arnaud____ has joined #openstack-nova | 04:06 | |
*** rushiagr_away is now known as rushiagr | 04:06 | |
*** coolsvap|afk is now known as coolsvap | 04:23 | |
*** iamjarvo has quit IRC | 04:25 | |
*** vilobhmm1 has joined #openstack-nova | 04:26 | |
*** ildikov has quit IRC | 04:28 | |
*** ildikov has joined #openstack-nova | 04:31 | |
*** sdake has joined #openstack-nova | 04:34 | |
*** otter768 has quit IRC | 04:35 | |
*** markvoelker has joined #openstack-nova | 04:37 | |
*** sdake_ has quit IRC | 04:38 | |
*** rushiagr is now known as rushiagr_away | 04:46 | |
*** irenab has quit IRC | 04:46 | |
*** pixelbeat_ has quit IRC | 04:49 | |
*** yingjun has joined #openstack-nova | 04:53 | |
*** afazekas has joined #openstack-nova | 04:55 | |
*** rushiagr_away is now known as rushiagr | 04:55 | |
openstackgerrit | heijlong proposed openstack/nova: Merge extension-info functional tests of v2 and v2.1 https://review.openstack.org/175701 | 04:56 |
*** lpetrut has joined #openstack-nova | 04:59 | |
*** tobe has quit IRC | 05:03 | |
*** nihilifer has joined #openstack-nova | 05:04 | |
*** rushiagr is now known as rushiagr_away | 05:08 | |
*** nkrinner has joined #openstack-nova | 05:10 | |
*** sdake_ has joined #openstack-nova | 05:13 | |
*** emagana has joined #openstack-nova | 05:15 | |
*** ericksonsantos has joined #openstack-nova | 05:15 | |
*** sdake has quit IRC | 05:16 | |
*** unicell1 has joined #openstack-nova | 05:20 | |
*** _gryf|afk is now known as _gryf | 05:20 | |
*** ericksonsantos has quit IRC | 05:20 | |
*** unicell has quit IRC | 05:21 | |
*** Alexandra_ has quit IRC | 05:24 | |
*** oro has joined #openstack-nova | 05:28 | |
*** heyongli has quit IRC | 05:29 | |
*** irenab has joined #openstack-nova | 05:30 | |
*** heyongli has joined #openstack-nova | 05:31 | |
*** ildikov has quit IRC | 05:32 | |
*** n0ano has joined #openstack-nova | 05:32 | |
*** Nikolay_St has quit IRC | 05:32 | |
*** yingjun has quit IRC | 05:35 | |
*** Zhengguang has quit IRC | 05:36 | |
*** rushiagr_away is now known as rushiagr | 05:36 | |
*** rushiagr is now known as rushiagr_away | 05:37 | |
*** achanda has joined #openstack-nova | 05:37 | |
*** jogo has quit IRC | 05:39 | |
*** jogo has joined #openstack-nova | 05:39 | |
*** Jeffrey4l has joined #openstack-nova | 05:50 | |
*** vilobhmm1 has quit IRC | 05:54 | |
*** sdake has joined #openstack-nova | 05:57 | |
*** sudipto has joined #openstack-nova | 05:58 | |
*** Longgeek_ has joined #openstack-nova | 05:59 | |
*** sdake_ has quit IRC | 06:01 | |
*** arnaud____ has quit IRC | 06:02 | |
*** yingjun has joined #openstack-nova | 06:03 | |
*** Longgeek has quit IRC | 06:03 | |
*** doron_afk has joined #openstack-nova | 06:04 | |
*** cfriesen has quit IRC | 06:05 | |
*** lpetrut has quit IRC | 06:06 | |
openstackgerrit | Qin Zhao proposed openstack/nova: Update sqlalchemy-migrate to 0.9.6 https://review.openstack.org/170050 | 06:10 |
*** sdake_ has joined #openstack-nova | 06:15 | |
*** sdake__ has joined #openstack-nova | 06:18 | |
*** sdake has quit IRC | 06:18 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Imported Translations from Transifex https://review.openstack.org/179638 | 06:19 |
*** ankit_ag has joined #openstack-nova | 06:19 | |
openstackgerrit | Rajesh Tailor proposed openstack/nova: Fix quota-update of instances stuck in deleting when nova-compute startup finish https://review.openstack.org/122347 | 06:20 |
*** sdake_ has quit IRC | 06:21 | |
*** rushiagr_away is now known as rushiagr | 06:26 | |
*** oro has quit IRC | 06:27 | |
*** jcoufal has joined #openstack-nova | 06:31 | |
*** otter768 has joined #openstack-nova | 06:36 | |
*** lpetrut has joined #openstack-nova | 06:37 | |
*** Murali has joined #openstack-nova | 06:38 | |
*** sahid has joined #openstack-nova | 06:41 | |
*** otter768 has quit IRC | 06:41 | |
*** lpetrut has quit IRC | 06:42 | |
*** park_heijlong has quit IRC | 06:48 | |
*** mpaolino has joined #openstack-nova | 06:52 | |
openstackgerrit | Lan Qi Song proposed openstack/nova: Remove useless volume when boot from volume failed https://review.openstack.org/177084 | 06:53 |
*** yasu_ has quit IRC | 06:56 | |
*** yasu_ has joined #openstack-nova | 06:56 | |
*** destiny has joined #openstack-nova | 06:56 | |
destiny | hi | 06:56 |
destiny | i'm looking for informations about the docker driver, is it dead ? | 06:56 |
*** ildikov has joined #openstack-nova | 06:57 | |
*** park_heijlong has joined #openstack-nova | 07:01 | |
*** Marga_ has joined #openstack-nova | 07:01 | |
*** Ritesh has joined #openstack-nova | 07:01 | |
Ritesh | Hi | 07:02 |
Ritesh | Can we migrate mulitpie instances at a timei ? | 07:02 |
Ritesh | nova live-migration vm2 vm1 | 07:02 |
*** markus_z has joined #openstack-nova | 07:02 | |
*** haypo has joined #openstack-nova | 07:03 | |
*** Marga_ has quit IRC | 07:04 | |
*** subscope has quit IRC | 07:05 | |
openstackgerrit | Qin Zhao proposed openstack/nova: Libvirt: Use tpool to invoke guestfs api https://review.openstack.org/181808 | 07:07 |
*** RuiChen has joined #openstack-nova | 07:12 | |
*** achanda has quit IRC | 07:15 | |
*** moshele has joined #openstack-nova | 07:16 | |
*** jichen has joined #openstack-nova | 07:17 | |
*** Marga_ has joined #openstack-nova | 07:19 | |
*** Marga_ has quit IRC | 07:19 | |
*** romainh has joined #openstack-nova | 07:19 | |
*** yingjun has quit IRC | 07:20 | |
*** yingjun has joined #openstack-nova | 07:22 | |
*** matrohon has joined #openstack-nova | 07:22 | |
*** sahid has quit IRC | 07:24 | |
*** nihilifer has quit IRC | 07:24 | |
*** sahid has joined #openstack-nova | 07:24 | |
*** dsanders has quit IRC | 07:25 | |
*** sdake__ has quit IRC | 07:28 | |
*** mpavone has joined #openstack-nova | 07:30 | |
*** nihilifer has joined #openstack-nova | 07:30 | |
*** ildikov has quit IRC | 07:31 | |
*** Jeffrey4l has quit IRC | 07:31 | |
*** browne has quit IRC | 07:37 | |
*** Jeffrey4l has joined #openstack-nova | 07:37 | |
*** tobe has joined #openstack-nova | 07:42 | |
*** apuimedo has joined #openstack-nova | 07:43 | |
*** Nikolay_St has joined #openstack-nova | 07:44 | |
*** achanda has joined #openstack-nova | 07:45 | |
*** jlanoux has joined #openstack-nova | 07:46 | |
*** ildikov has joined #openstack-nova | 07:47 | |
openstackgerrit | Vincent Hou proposed openstack/nova: Add a new parameter connector into migrate_volume_completion https://review.openstack.org/181818 | 07:54 |
*** romainh has quit IRC | 07:55 | |
*** Nikolay_St has quit IRC | 07:56 | |
*** fifieldt has joined #openstack-nova | 07:57 | |
*** fifieldt has quit IRC | 07:58 | |
*** jistr has joined #openstack-nova | 07:58 | |
*** park_heijlong has quit IRC | 07:59 | |
*** achanda has quit IRC | 08:01 | |
*** lucasagomes has joined #openstack-nova | 08:04 | |
*** Nikolay_St has joined #openstack-nova | 08:09 | |
*** romainh has joined #openstack-nova | 08:10 | |
*** belmoreira has joined #openstack-nova | 08:11 | |
*** nithyag_ has quit IRC | 08:12 | |
*** yingjun has quit IRC | 08:20 | |
*** derekh has joined #openstack-nova | 08:22 | |
*** redbeard has joined #openstack-nova | 08:22 | |
*** yingjun has joined #openstack-nova | 08:23 | |
*** lpetrut has joined #openstack-nova | 08:25 | |
johnthetubaguy | destiny: its not in tree any more, its in stackforge, but they are trying to merge it back | 08:29 |
johnthetubaguy | Ritesh: you can live-migrate multiple VMs, but its two separate API calls | 08:29 |
openstackgerrit | Moshe Levi proposed openstack/nova: libvirt: mlnx_direct vif type removal https://review.openstack.org/181671 | 08:31 |
*** dguitarbite has joined #openstack-nova | 08:32 | |
*** marekd has left #openstack-nova | 08:33 | |
*** sudipto has quit IRC | 08:34 | |
*** Jeffrey4l has quit IRC | 08:35 | |
*** otter768 has joined #openstack-nova | 08:37 | |
*** gilliard_afk is now known as gilliard | 08:37 | |
*** sudipto has joined #openstack-nova | 08:39 | |
gilliard | Hello Nova! | 08:39 |
bauzas | gilliard: aloha | 08:39 |
*** otter768 has quit IRC | 08:42 | |
*** ociuhandu has joined #openstack-nova | 08:44 | |
*** achanda has joined #openstack-nova | 08:45 | |
alex_xu | gilliard: welcome back! | 08:45 |
moshele | bauzas: hi | 08:45 |
bauzas | moshele: hi | 08:46 |
moshele | bauzas: just for my understanding compute.filter is for nova-compute utilities and network.filter is for nova-network utilities, right? | 08:47 |
bauzas | moshele: I don't get your point | 08:47 |
bauzas | moshele: you mean nova.filters ? | 08:47 |
moshele | bauzas: yes | 08:47 |
bauzas | moshele: the module ? | 08:47 |
bauzas | moshele: that's for the scheduler filters yes | 08:48 |
bauzas | moshele: there is an handler used to manage how the filters are called | 08:48 |
moshele | moshele: no rootwrap files in etc | 08:48 |
*** Jeffrey4l has joined #openstack-nova | 08:48 | |
moshele | bauzas: no rootwrap files in etc | 08:48 |
bauzas | moshele: aaaah | 08:49 |
openstackgerrit | Lingxian Kong proposed openstack/nova: Correct az property of instance when specifying force host/node https://review.openstack.org/163842 | 08:49 |
bauzas | moshele: could you please point me out the module full path ? | 08:50 |
moshele | bauzas: can you look on this patch https://review.openstack.org/#/c/181670/ | 08:50 |
bauzas | moshele: tbh I'm pretty busy for preparing the Summit, but I'll try | 08:51 |
bauzas | moshele: my review rate is pretty low these days | 08:51 |
destiny | okay johnthetubaguy, any idea of a date which will be happen ? or something to follow it ? | 08:52 |
*** fawadkhaliq has joined #openstack-nova | 08:52 | |
moshele | bauzas: ok thanks | 08:52 |
johnthetubaguy | destiny: totally unsure if it will ever go back into tree at this point, but you can see the spec review here: https://review.openstack.org/#/c/128753/ | 08:52 |
gilliard | HI alex_xu. Thanks :) | 08:58 |
*** doron_afk is now known as doron | 08:59 | |
alex_xu | gilliard: how was your test | 08:59 |
*** Ritesh has quit IRC | 09:00 | |
alex_xu | gilliard: thanks for the comment on https://review.openstack.org/181784 , if we use field length, it pretty similar to the format of request. Then it totally sounds like different case | 09:01 |
*** RuiChen has quit IRC | 09:01 | |
*** RuiChen has joined #openstack-nova | 09:02 | |
*** tojuvone has joined #openstack-nova | 09:02 | |
*** ociuhandu has quit IRC | 09:03 | |
*** park_heijlong has joined #openstack-nova | 09:04 | |
openstackgerrit | Victor Stinner proposed openstack/nova: Rename _CellProxy.iteritems method to items on py3 https://review.openstack.org/179783 | 09:06 |
openstackgerrit | Victor Stinner proposed openstack/nova: Replace dict.iteritems() with six.iteritems(dict) https://review.openstack.org/179784 | 09:06 |
*** park_heijlong has quit IRC | 09:09 | |
*** oro has joined #openstack-nova | 09:10 | |
destiny | ok thanks johnthetubaguy | 09:10 |
*** doron is now known as doron_afk | 09:14 | |
openstackgerrit | Victor Stinner proposed openstack/nova: Drop L from literal integer numbers for Python 3 https://review.openstack.org/179828 | 09:16 |
openstackgerrit | John Garbutt proposed openstack/nova: devref: add information to clarify nova scope https://review.openstack.org/178623 | 09:18 |
*** aix has joined #openstack-nova | 09:20 | |
*** derekh has quit IRC | 09:27 | |
*** ildikov has quit IRC | 09:27 | |
*** derekh has joined #openstack-nova | 09:28 | |
sahid | zhangtralon: hello | 09:31 |
*** haomaiwa_ has quit IRC | 09:31 | |
zhangtralon | sahid:hi | 09:32 |
*** aswadr has joined #openstack-nova | 09:33 | |
sahid | zhangtralon: so about the review 167917 (sorry i have pb to copy/past link) | 09:34 |
sahid | do you still plan to work on it? | 09:34 |
zhangtralon | let me see | 09:34 |
zhangtralon | I still work on it | 09:35 |
sahid | any blockers? | 09:36 |
*** yingjun has quit IRC | 09:36 | |
*** IanGovett has joined #openstack-nova | 09:36 | |
sahid | i was thinking about to push some changes to update the way we compute free resources from drivers (and so include hugepages) | 09:37 |
*** danpb has joined #openstack-nova | 09:37 | |
sahid | perhaps you can then put your changes on top of it | 09:37 |
zhangtralon | before, I want to discuss the https://bugs.launchpad.net/nova/+bug/1444232 with you and nikola | 09:37 |
openstack | Launchpad bug 1444232 in OpenStack Compute (nova) "Only using huge page may filter the usable host" [Low,Incomplete] - Assigned to zhangtralon (zhangchunlong1) | 09:37 |
zhangtralon | I think the bug 1444232 is related to the review 167917 | 09:38 |
openstack | bug 1444232 in OpenStack Compute (nova) "Only using huge page may filter the usable host" [Low,Incomplete] https://launchpad.net/bugs/1444232 - Assigned to zhangtralon (zhangchunlong1) | 09:38 |
sahid | zhangtralon: so you want to separate numa topology and hugepages, right? | 09:39 |
*** achanda has quit IRC | 09:40 | |
zhangtralon | I am not sure, look the comments from nikola | 09:40 |
*** Nikolay_St has quit IRC | 09:41 | |
*** Nikolay_St has joined #openstack-nova | 09:41 | |
*** ZZelle has joined #openstack-nova | 09:42 | |
*** ildikov has joined #openstack-nova | 09:43 | |
*** subscope has joined #openstack-nova | 09:43 | |
zhangtralon | sahid: If separating numa topology and hugepages, maybe we can collaborate on it | 09:43 |
sudipto | johnthetubaguy, curious about an RT related observation. I see the following log message in the compute.log file: Total usable vcpus: 20, total allocated vcpus: 156 -- is this theoretically possible? | 09:46 |
johnthetubaguy | sudipto: yes | 09:46 |
johnthetubaguy | sudipto: CPU overcommit is quite a common hypervisor feature | 09:47 |
sudipto | johnthetubaguy, yeah I mean the overcommit ratio is probably 1.5x times by default? | 09:47 |
sudipto | This seems like a huge number to me? | 09:47 |
sahid | johnthetubaguy: we have two backports here for kilo and juno, they seem to be reasonable in my point of view I would like to get your feeling 178072 please | 09:47 |
sudipto | even if it's 2x times - this seems like a way off number doesn't it? | 09:48 |
sahid | zhangtralon: i do not believe separate hugepages nad numa topology is the way we want to take | 09:48 |
johnthetubaguy | sudipto: it depends what your users are doing, if most folks don't use all the CPU, thats maybe a great way to go | 09:48 |
johnthetubaguy | sudipto: depends how you have decided to configure your system, I don't think we enforce CPU overcommit by default, its more a scheduler hint by default, I think | 09:48 |
*** pixelbeat_ has joined #openstack-nova | 09:49 | |
sudipto | johnthetubaguy, ok, sorry for asking a dumb question. Is it just controller by the quota from a user irrespective than actually being verified on the system? | 09:49 |
sudipto | *controller = controlled. | 09:49 |
*** dulek has joined #openstack-nova | 09:51 | |
zhangtralon | sahid:oh, separating the numa and hugepages will take the more flexibility, users can using the numa or hugepages based on their cases. | 09:54 |
johnthetubaguy | sudipto: there are user quotas, but thats separate to what the resource tracker enforces, it all depends how you configure it, I don't remember the details of that right now | 09:55 |
sudipto | johnthetubaguy, ok... | 09:55 |
johnthetubaguy | sudipto: as a Xen users, I generally only care about RAM, as I don't want ballooning, and thats the only finite resource, assuming i size flavors correctly to limit overcommit of everything else, like disk/CPU | 09:56 |
*** exploreshaifali has joined #openstack-nova | 09:56 | |
*** fawadkhaliq has quit IRC | 09:56 | |
johnthetubaguy | sahid: that looks like you are backporting a feature? | 09:56 |
johnthetubaguy | sahid: if the original patch has a bug or blueprint attached, it would be much easier to understand the impact | 09:58 |
sudipto | johnthetubaguy, hmm ok. Maybe one of the reasons why we only have a RamFilter and not the one for CPU? | 09:59 |
sahid | yes but unforntunatly it was not the case. ok so rule is rule i'm going to abandon them. | 09:59 |
johnthetubaguy | sudipto: I assumed we had a CPU one, but I haven't looked, feel free to add one if you want one | 10:00 |
*** exploreshaifali has quit IRC | 10:01 | |
johnthetubaguy | sahid: you could make the case, by a bug describing the impact, etc, and how that commit fixes it, and reference that in the cherry-pick | 10:01 |
sudipto | johnthetubaguy, my bad, there's a CoreFilter. | 10:01 |
*** eglynn has joined #openstack-nova | 10:02 | |
*** dimsum__ has joined #openstack-nova | 10:02 | |
*** doron_afk is now known as doron | 10:05 | |
johnthetubaguy | sudipto: ah, no worries | 10:11 |
*** NithyaG has joined #openstack-nova | 10:20 | |
*** dguitarbite has quit IRC | 10:24 | |
*** doron is now known as doron_afk | 10:25 | |
*** dguitarbite has joined #openstack-nova | 10:27 | |
*** romainh has quit IRC | 10:29 | |
*** fawadkhaliq has joined #openstack-nova | 10:31 | |
*** ajayaa has joined #openstack-nova | 10:35 | |
*** baoli has joined #openstack-nova | 10:37 | |
*** otter768 has joined #openstack-nova | 10:38 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Drop explicit suds dependency https://review.openstack.org/181554 | 10:39 |
*** VW_ has joined #openstack-nova | 10:41 | |
*** baoli has quit IRC | 10:41 | |
*** VW_ has quit IRC | 10:42 | |
*** VW_ has joined #openstack-nova | 10:42 | |
*** otter768 has quit IRC | 10:43 | |
*** coolsvap is now known as coolsvap|afk | 10:48 | |
*** gszasz has joined #openstack-nova | 10:48 | |
*** belmoreira has quit IRC | 10:48 | |
*** ildikov has quit IRC | 10:59 | |
openstackgerrit | Tomi Juvonen proposed openstack/nova-specs: New nova API call to mark nova-compute down https://review.openstack.org/169836 | 11:07 |
*** exploreshaifali has joined #openstack-nova | 11:07 | |
*** zhenguo has joined #openstack-nova | 11:10 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Add rootwrap daemon mode support https://review.openstack.org/180695 | 11:12 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP] Run using rootwrap daemon as True https://review.openstack.org/180809 | 11:13 |
*** tojuvone has quit IRC | 11:15 | |
*** ildikov has joined #openstack-nova | 11:15 | |
*** mpaolino has quit IRC | 11:15 | |
*** exploreshaifali has quit IRC | 11:17 | |
openstackgerrit | Michal Rostecki proposed openstack/nova-specs: Dedicate aggregates for specific tenants https://review.openstack.org/181054 | 11:23 |
openstackgerrit | Michal Rostecki proposed openstack/nova-specs: Dedicate aggregates for specific tenants https://review.openstack.org/181054 | 11:26 |
*** EmilienM|afk is now known as EmilienM | 11:28 | |
*** romainh has joined #openstack-nova | 11:29 | |
*** mmedvede_ has joined #openstack-nova | 11:30 | |
*** jaypipes has joined #openstack-nova | 11:32 | |
*** Nikolay_St has quit IRC | 11:33 | |
*** Nikolay_St has joined #openstack-nova | 11:34 | |
*** ociuhandu has joined #openstack-nova | 11:35 | |
openstackgerrit | Michal Rostecki proposed openstack/nova-specs: Dedicate aggregates for specific tenants https://review.openstack.org/181054 | 11:37 |
*** zul has quit IRC | 11:37 | |
*** zul has joined #openstack-nova | 11:38 | |
*** lucasagomes is now known as lucas-hungry | 11:41 | |
*** subscope has quit IRC | 11:44 | |
*** subscope has joined #openstack-nova | 11:47 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Nova serial console access support https://review.openstack.org/145004 | 11:51 |
*** masayukig has quit IRC | 11:53 | |
*** tobe has quit IRC | 11:53 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Hyper-V: Nova serial console access support https://review.openstack.org/145004 | 11:54 |
*** tojuvone has joined #openstack-nova | 11:55 | |
*** deepthi has quit IRC | 11:56 | |
*** baoli has joined #openstack-nova | 11:56 | |
*** claudiub has joined #openstack-nova | 11:58 | |
*** masayukig has joined #openstack-nova | 11:59 | |
moshele | johnthetubaguy: hi | 12:01 |
*** dprince has joined #openstack-nova | 12:02 | |
*** Nikolay_St has quit IRC | 12:03 | |
*** dimsum__ has quit IRC | 12:03 | |
*** boris-42 has joined #openstack-nova | 12:05 | |
*** Nikolay_St has joined #openstack-nova | 12:05 | |
*** kodoku has joined #openstack-nova | 12:05 | |
kodoku | I question about Server personality. This feature use injection file from libvirt ? | 12:06 |
*** mwagner_lap has quit IRC | 12:07 | |
*** lpetrut1 has joined #openstack-nova | 12:09 | |
*** lpetrut has quit IRC | 12:10 | |
*** tzumainn_ has joined #openstack-nova | 12:10 | |
kashyap | Uh, I'm always thrown off by Gerrit's formatting of comments in a way I didn't intend (no long line wrap) - https://review.openstack.org/#/c/181781/ | 12:11 |
openstackgerrit | Radoslav Gerganov proposed openstack/nova: VMware: verify vCenter server certificate https://review.openstack.org/179728 | 12:12 |
*** dave-mccowan has joined #openstack-nova | 12:12 | |
*** Marga_ has joined #openstack-nova | 12:12 | |
johnthetubaguy | moshele: hi | 12:13 |
*** claudiub has quit IRC | 12:13 | |
johnthetubaguy | kodoku: depends on your configuration | 12:13 |
johnthetubaguy | kodoku: we are trying to remove file injection, instead make cloud-init do the heavy lifting based on config drive / metadata service | 12:13 |
kodoku | johnthetubaguy ok because trove use this feature for configure the guest. | 12:14 |
*** Nikolay_St has quit IRC | 12:15 | |
*** atuvenie has joined #openstack-nova | 12:15 | |
kodoku | johnthetubaguy and with vmware backend, you don't have this feature | 12:15 |
johnthetubaguy | kodoku: vmware supports config drive, that should allow something similar using cloud-init in your guest | 12:15 |
lxsli | kashyap: you put a space at the front so you got 'code block' mode | 12:15 |
*** claudiub has joined #openstack-nova | 12:16 | |
*** rfolco has joined #openstack-nova | 12:16 | |
*** nihilifer has quit IRC | 12:18 | |
kashyap | lxsli: Yeah, I normally provide 4 spaces, as that'd wrap lines. And, forgot this time. | 12:18 |
*** raildo has joined #openstack-nova | 12:19 | |
*** htruta has joined #openstack-nova | 12:19 | |
openstackgerrit | Petrut Lucian proposed openstack/nova-specs: Hyper-V: Add storage QoS support https://review.openstack.org/175422 | 12:19 |
lxsli | :) | 12:19 |
*** tzumainn_ has left #openstack-nova | 12:21 | |
*** Marga_ has quit IRC | 12:21 | |
*** Marga_ has joined #openstack-nova | 12:22 | |
*** nihilifer has joined #openstack-nova | 12:22 | |
*** fawadkhaliq has quit IRC | 12:26 | |
*** nihilifer has quit IRC | 12:28 | |
*** sudipto has quit IRC | 12:30 | |
*** Jeffrey4l has quit IRC | 12:30 | |
*** Nikolay_St has joined #openstack-nova | 12:31 | |
openstackgerrit | Brent Eagles proposed openstack/nova: WIP: Refactor Nova VIF model https://review.openstack.org/167287 | 12:32 |
*** VW_ has quit IRC | 12:32 | |
*** VW_ has joined #openstack-nova | 12:33 | |
*** aysyd has joined #openstack-nova | 12:34 | |
*** VW_ has quit IRC | 12:35 | |
*** VW_ has joined #openstack-nova | 12:36 | |
*** artom has joined #openstack-nova | 12:36 | |
*** lucas-hungry is now known as lucasagomes | 12:37 | |
*** oro_ has joined #openstack-nova | 12:37 | |
*** geguileo has joined #openstack-nova | 12:38 | |
*** sdake has joined #openstack-nova | 12:38 | |
*** otter768 has joined #openstack-nova | 12:39 | |
*** sdake_ has joined #openstack-nova | 12:40 | |
*** Jeffrey4l has joined #openstack-nova | 12:42 | |
*** haomaiwa_ has joined #openstack-nova | 12:43 | |
*** otter768 has quit IRC | 12:43 | |
*** fawadkhaliq has joined #openstack-nova | 12:44 | |
*** sdake has quit IRC | 12:44 | |
*** kodoku has quit IRC | 12:46 | |
*** Longgeek_ has quit IRC | 12:48 | |
*** Longgeek has joined #openstack-nova | 12:49 | |
openstackgerrit | jichenjc proposed openstack/nova-specs: Allow adjust ephemeral disks during resize https://review.openstack.org/145736 | 12:49 |
openstackgerrit | Sailaja proposed openstack/nova: Project quota update should not be less than user quota. https://review.openstack.org/181349 | 12:49 |
*** fawadk has joined #openstack-nova | 12:50 | |
*** mwagner_lap has joined #openstack-nova | 12:50 | |
*** fawadkhaliq has quit IRC | 12:50 | |
*** VW_ has quit IRC | 12:53 | |
*** VW_ has joined #openstack-nova | 12:54 | |
*** RuiChen has quit IRC | 12:55 | |
openstackgerrit | Petrut Lucian proposed openstack/nova: Add Hyper-V storage QoS support https://review.openstack.org/175479 | 12:55 |
*** xyang1 has joined #openstack-nova | 12:55 | |
*** RuiChen has joined #openstack-nova | 12:55 | |
*** sdake_ is now known as sdake | 12:55 | |
*** inara has quit IRC | 12:56 | |
*** inara has joined #openstack-nova | 12:56 | |
openstackgerrit | Joel Coffman proposed openstack/nova: libvirt: Disconnect dm-crypt on instance suspend/stop https://review.openstack.org/141485 | 12:59 |
*** VW_ has quit IRC | 12:59 | |
*** VW_ has joined #openstack-nova | 13:00 | |
*** VW_ has quit IRC | 13:00 | |
haypo | johnthetubaguy, hi. for my python 3 spec, would this change be enough to explicit that Tempest tests must pass on Python 3? http://paste.openstack.org/show/219846/ | 13:00 |
*** VW_ has joined #openstack-nova | 13:00 | |
johnthetubaguy | haypo: I would remove the functional tests from that list for now, tempest and unit tests are the big ones for now | 13:02 |
*** READ10 has joined #openstack-nova | 13:02 | |
*** VW__ has joined #openstack-nova | 13:03 | |
*** VW_ has quit IRC | 13:03 | |
*** VW__ has quit IRC | 13:03 | |
*** VW_ has joined #openstack-nova | 13:04 | |
*** VW_ has quit IRC | 13:04 | |
*** haomaiwa_ has quit IRC | 13:04 | |
*** VW_ has joined #openstack-nova | 13:05 | |
openstackgerrit | Victor Stinner proposed openstack/nova-specs: Adding Python 3.4 support to Nova https://review.openstack.org/176868 | 13:06 |
haypo | johnthetubaguy, ok. here is the updated spec, don't hesitate to comment it inline if you want to suggest other changes ^^ | 13:06 |
*** rushiagr is now known as rushiagr_away | 13:07 | |
*** fawadkhaliq has joined #openstack-nova | 13:08 | |
*** fawadk has quit IRC | 13:11 | |
*** yingjun has joined #openstack-nova | 13:11 | |
*** burt has joined #openstack-nova | 13:12 | |
*** VW_ has quit IRC | 13:15 | |
*** VW_ has joined #openstack-nova | 13:15 | |
*** Longgeek has quit IRC | 13:15 | |
*** Longgeek has joined #openstack-nova | 13:16 | |
*** fifieldt has joined #openstack-nova | 13:17 | |
*** fifieldt has quit IRC | 13:17 | |
*** eharney has joined #openstack-nova | 13:17 | |
openstackgerrit | Rajesh Tailor proposed openstack/nova: Allow force-delete irrespective of VM task_state https://review.openstack.org/163798 | 13:18 |
*** VW_ has quit IRC | 13:18 | |
*** VW_ has joined #openstack-nova | 13:19 | |
*** VW_ has quit IRC | 13:19 | |
*** VW_ has joined #openstack-nova | 13:20 | |
*** josecastroleon has joined #openstack-nova | 13:21 | |
*** BobBall_AWOL is now known as BobBall | 13:21 | |
alex_xu | johnthetubaguy: hi, just add two more questions for policy and add some detail for version bump which hope help for discussion https://etherpad.openstack.org/p/YVR-nova-api-2.1-in-liberty | 13:21 |
*** cinerama has quit IRC | 13:22 | |
alex_xu | two etherpads are LGTM | 13:22 |
*** VW_ has quit IRC | 13:22 | |
*** VW_ has joined #openstack-nova | 13:23 | |
johnthetubaguy | alex_xu: whats "user-owned permission"? | 13:24 |
alex_xu | johnthetubaguy: currently all the policy enforce without target. So tenant-owner permission is done by the db. So thinking if we adding target, then we can enable user-owned permission | 13:25 |
alex_xu | johnthetubaguy: it means the rule can be "user_id:%(user_id)s", the user only can see the vm belong to him | 13:25 |
alex_xu | johnthetubaguy: user can't see the vm in the same tenant belong to other user. | 13:25 |
*** doron_afk is now known as doron | 13:26 | |
johnthetubaguy | alex_xu: hierarchical tenant stuff seemed like the way to support that | 13:26 |
*** cinerama has joined #openstack-nova | 13:26 | |
alex_xu | johnthetubaguy: hierarchical tenant is still enforce on tenant-owner? I guess, but let me check the detail hierarchical tenant | 13:27 |
johnthetubaguy | alex_xu: it does, but users can be in multiple tenants in interesting groups, and create their own sub tenants, and add users into those | 13:28 |
*** fawadkhaliq has quit IRC | 13:29 | |
johnthetubaguy | alex_xu: ssh keys are the exception here, I guess | 13:29 |
alex_xu | johnthetubaguy: emm... sub-tenants have resource isolation at here? | 13:29 |
*** baoli has quit IRC | 13:30 | |
alex_xu | johnthetubaguy: the case what I said is pretty similar, just need change authorize(context, action="detail") to authorize(context, action="detail", target=instance) | 13:31 |
alex_xu | s/similar/simple.... | 13:31 |
*** baoli_ has joined #openstack-nova | 13:31 | |
*** jgrimm is now known as zz_jgrimm | 13:32 | |
johnthetubaguy | alex_xu: but if a user gets a shared tenant, and an individual tenant each, you get much of what you need, without fundamentally changing things. | 13:33 |
johnthetubaguy | alex_xu: hmm, bit worried that debate would go on and on, probably best to move it to the end of the session, assuming its not high priority | 13:34 |
johnthetubaguy | alex_xu: I guess I just don't understand the use case | 13:34 |
alex_xu | johnthetubaguy: yes, it pretty low priority, I put at here just in case there really have some one want that. | 13:34 |
*** _gryf is now known as _gryf|afk | 13:36 | |
*** b3nt_pin has quit IRC | 13:36 | |
*** beagles has quit IRC | 13:38 | |
johnthetubaguy | alex_xu: lets ignore it then please, we should only add features where there is a strong need, everything owned by a tenant is a very old assumption to suddenly change | 13:38 |
*** b3nt_pin has joined #openstack-nova | 13:38 | |
*** b3nt_pin is now known as beagles | 13:38 | |
johnthetubaguy | at the end is cool | 13:38 |
*** VW_ has quit IRC | 13:38 | |
*** mpaolino has joined #openstack-nova | 13:38 | |
openstackgerrit | James Page proposed openstack/nova: Handle return code 2 from blkid calls https://review.openstack.org/159797 | 13:39 |
alex_xu | johnthetubaguy: ok, no problem | 13:39 |
*** dimsum__ has joined #openstack-nova | 13:39 | |
*** Jeffrey4l has quit IRC | 13:40 | |
*** jamespage has joined #openstack-nova | 13:40 | |
johnthetubaguy | alex_xu: thanks for reviewing all of those etherpads though, hopefully that helps with the sessions | 13:40 |
alex_xu | johnthetubaguy: np, hope we have wonderful api session | 13:41 |
*** lpetrut1 has quit IRC | 13:41 | |
*** exploreshaifali has joined #openstack-nova | 13:42 | |
*** mriedem has joined #openstack-nova | 13:42 | |
*** nihilifer has joined #openstack-nova | 13:43 | |
*** emagana has quit IRC | 13:43 | |
mriedem | oh hello | 13:43 |
jamespage | any chance anyone could take a look at https://review.openstack.org/#/c/159797/ ? | 13:44 |
jamespage | we're holding that as a patch in ubuntu for kilo release as launching instances without libguestfs without it is pretty broken | 13:44 |
mriedem | jamespage: looking | 13:46 |
*** thangp has joined #openstack-nova | 13:46 | |
gilliard | Hi mriedem. Thanks for fixing the live-migration tempest thing last week. | 13:46 |
jamespage | mriedem, thanks | 13:46 |
*** BobH has joined #openstack-nova | 13:48 | |
*** ajayaa has quit IRC | 13:48 | |
bauzas | johnthetubaguy: if you agree, wanted to volunteer on https://etherpad.openstack.org/p/YVR-nova-scalling-out-scheduler-for-cells | 13:48 |
bauzas | johnthetubaguy: I already have an etherpad for that | 13:49 |
bauzas | https://etherpad.openstack.org/p/liberty-nova-scalable-scheduler | 13:49 |
ankit_ag | mriedem: hi, Could you please review https://review.openstack.org/#/c/145738/. I have addressed all your review comments to make code blocks more readable. | 13:49 |
rwmjones | mriedem: yeah I was going to send you that patch, because I think it's related to the problem you were talking about on Friday | 13:51 |
mriedem | rwmjones: this one? https://review.openstack.org/#/c/159797/ - i'm using guestfs though | 13:51 |
rwmjones | sorry, no, NOT that one | 13:51 |
* rwmjones is confused | 13:52 | |
rwmjones | mriedem: https://review.openstack.org/#/c/181808/ is the one I was going to send you | 13:52 |
johnthetubaguy | bauzas: can you combine your ideas into the official etherpad for me please? | 13:52 |
bauzas | is it me or etherpad is just cracking ? | 13:52 |
bauzas | johnthetubaguy: yup, was about to do that until I saw that etherpad severely suffers | 13:52 |
mriedem | rwmjones: oh yeah, didn't know about that but qin zhao works with me | 13:52 |
mriedem | rwmjones: he was looking at the same issue | 13:53 |
*** dimsum__ has quit IRC | 13:53 | |
johnthetubaguy | bauzas: I had issues at the weekend if I had too many etherpads open | 13:53 |
johnthetubaguy | bauzas: seemed OK otherwise | 13:53 |
rwmjones | mriedem: ah, that explains it .. danpb looked at the patch also and is OK with it | 13:53 |
*** apuimedo has quit IRC | 13:53 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Drop explicit suds dependency https://review.openstack.org/181554 | 13:53 |
bauzas | johnthetubaguy: going to wipe the cache to see | 13:53 |
mriedem | rwmjones: since i have a recreate of the issue i'll try to test this change out today | 13:54 |
mriedem | thanks for pointing it out | 13:54 |
*** exploreshaifali has quit IRC | 13:54 | |
danpb | yeah, that bug will cause the entire of nova-compute to hang during the 30 seconds the guestfs takes to build its appliance the first time | 13:55 |
danpb | so that could certainly cause general nova timeout problems | 13:55 |
dansmith | ouch | 13:55 |
danpb | did i mention eventlet sucks | 13:55 |
mriedem | danpb: yeah, that's the thing i was talking about last friday | 13:55 |
*** exploreshaifali has joined #openstack-nova | 13:55 | |
*** artom has quit IRC | 13:55 | |
danpb | eventlet and native libraries are a just awful mix that should be a reason to never use eventlet | 13:55 |
mriedem | well, there is a session in vancouver on moving off eventlet | 13:56 |
*** markvoelker has quit IRC | 13:56 | |
rwmjones | even in the ordinary case, g.launch() is going to hang green threads for 2 or 3 seconds | 13:57 |
*** sahid has quit IRC | 13:57 | |
*** joshw42 has joined #openstack-nova | 13:57 | |
danpb | rwmjones: yeah, we already have that bit running via a background native thread | 13:59 |
danpb | this function that's just been fixed was something that was put in to probe if guestfs was working properly (since ubuntu installs have it broken by default) | 13:59 |
rwmjones | yup | 13:59 |
*** vladikr has joined #openstack-nova | 14:00 | |
openstackgerrit | javeme proposed openstack/nova: VMware: Update to return the correct ESX iqn https://review.openstack.org/162700 | 14:00 |
*** sahid has joined #openstack-nova | 14:00 | |
*** Longgeek has quit IRC | 14:00 | |
*** Longgeek has joined #openstack-nova | 14:01 | |
*** Jeffrey4l has joined #openstack-nova | 14:01 | |
mriedem | jamespage: +2 on https://review.openstack.org/#/c/159797/ | 14:01 |
jamespage | mriedem, thankyou - much appreciated | 14:01 |
jamespage | really hate holding a delta :-) | 14:01 |
*** smccully has quit IRC | 14:02 | |
*** Longgeek_ has joined #openstack-nova | 14:03 | |
*** Longgeek has quit IRC | 14:04 | |
*** bkopilov has quit IRC | 14:05 | |
*** VW_ has joined #openstack-nova | 14:05 | |
* mriedem is getting used to the 'new' gerrit today, a bit lost | 14:05 | |
*** joshw42 has quit IRC | 14:05 | |
*** openstackgerrit has quit IRC | 14:06 | |
*** openstackgerrit has joined #openstack-nova | 14:07 | |
*** VW_ has quit IRC | 14:07 | |
dansmith | there are some nice things about it | 14:07 |
dansmith | did they change the "old" interface or just the new one? | 14:07 |
kashyap | What is the newness? I only see "Can Merge" | 14:07 |
*** VW_ has joined #openstack-nova | 14:08 | |
*** mpaolino has quit IRC | 14:09 | |
mriedem | 'new' as in whatever was 'new' for a long time now | 14:09 |
*** rook has quit IRC | 14:09 | |
mriedem | the page is just very busy to me compared to the old one | 14:09 |
dansmith | yeah, they call it the "new" interface in the prefs I think | 14:10 |
mriedem | right, that's what i changed to | 14:10 |
mriedem | and now my world is topsy turvy | 14:10 |
dansmith | yeah, not sure why they feel the need to change it so drastically each time | 14:10 |
*** rook has joined #openstack-nova | 14:11 | |
*** dboik has joined #openstack-nova | 14:11 | |
openstackgerrit | Dan Smith proposed openstack/nova: Drop compute RPC 3.x support https://review.openstack.org/179951 | 14:11 |
mriedem | it's like, who would feed big changes to their users every 6 months or so, that's just crazy | 14:11 |
kashyap | mriedem: C'mon, OpenStack _itself_ feeds its operators/users with a new release every 6 months :-) | 14:12 |
mriedem | hence the joke | 14:12 |
kashyap | Ah, didn't read it in the right 'tone' | 14:12 |
*** dave-mccowan has quit IRC | 14:14 | |
sdague | dansmith: yeh, honestly, I'm pretty close to writing client side json to swap the commit message & review columns back | 14:15 |
sdague | client side javascript | 14:15 |
*** sdake_ has joined #openstack-nova | 14:16 | |
dansmith | I don't like that it wraps forcibly now | 14:16 |
*** dims has joined #openstack-nova | 14:16 | |
*** ferest has joined #openstack-nova | 14:16 | |
dansmith | or maybe I just wish people didn't write long lines in their commit message | 14:16 |
dansmith | otherwise I don't really care | 14:17 |
*** zzzeek has joined #openstack-nova | 14:18 | |
*** openstack has joined #openstack-nova | 14:19 | |
*** ankit_ag has quit IRC | 14:20 | |
*** yasu_ has quit IRC | 14:20 | |
*** ferest has quit IRC | 14:21 | |
*** yingjun has quit IRC | 14:21 | |
*** irenab has quit IRC | 14:23 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Drop explicit suds dependency https://review.openstack.org/181554 | 14:24 |
*** lpetrut has joined #openstack-nova | 14:24 | |
*** mpaolino has joined #openstack-nova | 14:25 | |
*** bnemec has joined #openstack-nova | 14:25 | |
*** kmartin has quit IRC | 14:26 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/nova: Updated from global requirements https://review.openstack.org/181913 | 14:28 |
*** dave-mccowan has joined #openstack-nova | 14:29 | |
*** doron is now known as doron_afk | 14:31 | |
johnthetubaguy | dansmith: sdague: is it just me, or is adding comments a bit unreliable after the upgrade, stuff keep getting reopened and added in odd places? | 14:31 |
johnthetubaguy | maybe its just me using chrome? | 14:31 |
dansmith | not that I've noticed | 14:31 |
dansmith | although I haven't done much yet | 14:31 |
dansmith | just a few comments | 14:31 |
*** sdake has joined #openstack-nova | 14:32 | |
openstackgerrit | javeme proposed openstack/nova: Make vm_state consistent with power_state https://review.openstack.org/181661 | 14:33 |
openstackgerrit | Claudiu Belu proposed openstack/nova: Hyper-V: Fixes cold migration / resize issue https://review.openstack.org/181916 | 14:33 |
*** yingjun has joined #openstack-nova | 14:33 | |
*** mtanino_away has quit IRC | 14:35 | |
*** yingjun has quit IRC | 14:35 | |
*** yingjun has joined #openstack-nova | 14:35 | |
*** mtanino has joined #openstack-nova | 14:35 | |
*** yingjun has quit IRC | 14:35 | |
*** sdake_ has quit IRC | 14:35 | |
*** yingjun has joined #openstack-nova | 14:35 | |
*** zz_jgrimm is now known as jgrimm | 14:37 | |
*** mestery has quit IRC | 14:38 | |
sdague | johnthetubaguy: seems to work fine for me | 14:39 |
*** nkrinner has quit IRC | 14:40 | |
*** otter768 has joined #openstack-nova | 14:40 | |
sdague | johnthetubaguy: I'm using chrome. | 14:40 |
*** annegentle has joined #openstack-nova | 14:40 | |
johnthetubaguy | sdague: hmm, OK, maybe its just me | 14:40 |
dansmith | me too | 14:40 |
*** Murali has quit IRC | 14:40 | |
openstackgerrit | Andrea Rosa proposed openstack/nova: WIP: Passing body for all requests but for TRACE https://review.openstack.org/181918 | 14:41 |
mikal | . | 14:41 |
sdague | johnthetubaguy: a shift+reload is always warrented after a gerrit upgrade, the javascript has a weird caching model some times | 14:41 |
mikal | johnthetubaguy: oh, you're online at the same time as me! | 14:41 |
mikal | I find this very confusing | 14:42 |
johnthetubaguy | sdague: yeah, good point | 14:42 |
johnthetubaguy | mikal: :) | 14:42 |
*** emagana has joined #openstack-nova | 14:42 | |
johnthetubaguy | mikal: you in SAT? | 14:42 |
mikal | johnthetubaguy: I just emailed you cause I thought you'd be asleep because I suck at maths | 14:42 |
johnthetubaguy | mikal: heh | 14:43 |
johnthetubaguy | mikal: timezones are just hard, duno why, its not even 4pm yet | 14:44 |
*** otter768 has quit IRC | 14:44 | |
*** zsais has joined #openstack-nova | 14:44 | |
*** amotoki_ has joined #openstack-nova | 14:45 | |
*** anteaya has joined #openstack-nova | 14:46 | |
johnthetubaguy | sdague: oh, so it adds the comment where I last clicked first, I think that is what was confusing me | 14:46 |
dims | johnthetubaguy: i touched up the rootwrap as daemon spec https://review.openstack.org/#/c/105404/ (BP - https://blueprints.launchpad.net/nova/+spec/nova-rootwrap-daemon-mode). Do i add it to the nova meeting agenda? (code is almost ready as well) | 14:46 |
sdague | johnthetubaguy: ah, yes | 14:46 |
johnthetubaguy | dims: only if the review is stuck at this point really | 14:47 |
*** whenry has joined #openstack-nova | 14:47 | |
johnthetubaguy | dims: its stuff without a spec that gets approved in the meeting routinely | 14:47 |
dims | johnthetubaguy: ok thanks. i'll ride it out :) | 14:47 |
johnthetubaguy | dims: I was just reviewing it, it reads like neutron hasn't done the switch yet? | 14:48 |
dims | johnthetubaguy: they are using it. i just picked a run at random to check - http://logs.openstack.org/67/180367/9/check/check-tempest-dsvm-neutron-full/fb27ec5/logs/screen-q-dhcp.txt.gz#_2015-05-09_04_22_15_691 | 14:49 |
dims | cinder has not moved yet | 14:50 |
johnthetubaguy | dims: yeah, the blueprint was marked as implemented, was just looking at some of the patches | 14:50 |
johnthetubaguy | dims: its just the spec is a bit old in places, adding comments | 14:50 |
dims | johnthetubaguy: ack thanks. | 14:50 |
*** hemnafk is now known as hemna | 14:51 | |
openstackgerrit | javeme proposed openstack/nova: VMware: Prefer ShutdownGuest to PowerOffVM_Task https://review.openstack.org/181634 | 14:52 |
*** cfriesen has joined #openstack-nova | 14:52 | |
*** artom has joined #openstack-nova | 14:53 | |
johnthetubaguy | dims: did we see the expected performance boost in neutron? | 14:54 |
*** BobH1 has joined #openstack-nova | 14:54 | |
anteaya | johnthetubaguy: for the sprint at ibm make sure they know you need to have wifi | 14:54 |
anteaya | johnthetubaguy: the sprint for cinder at ibm in austin had now wifi for all 3 days, we all had to tether off of our phones | 14:55 |
*** lpetrut has quit IRC | 14:55 | |
dims | digging it up | 14:55 |
johnthetubaguy | anteaya: mikal and mriedem are looking after all that for me, thats handy feedback about sorting out wifi though | 14:55 |
johnthetubaguy | anteaya: oh, I remember that being mentioned now, that would suck… | 14:55 |
anteaya | it did | 14:55 |
mriedem | way ahead of you | 14:55 |
anteaya | mriedem: awesome | 14:56 |
mriedem | we also have...running water! | 14:56 |
mikal | Heh | 14:56 |
mriedem | and lighting | 14:56 |
mikal | Do you have tornadoes? | 14:56 |
dansmith | mriedem: ibm is notorious for effing that up, FWIW :) | 14:56 |
mriedem | sometimes | 14:56 |
mikal | texas has tornados apparently | 14:56 |
*** BobH has quit IRC | 14:56 | |
mriedem | i suspect there are many semi-pro wrestlers named the texas tornado | 14:57 |
mriedem | we have more thunderstorms than anything | 14:57 |
openstackgerrit | Claudiu Belu proposed openstack/nova: Hyper-V: Fixes cold migration / resize issue https://review.openstack.org/181916 | 14:57 |
dims | johnthetubaguy: long thread with at least one set of numbers - http://markmail.org/message/v7f2adthal3rrgi6 | 14:57 |
mikal | mriedem: is there a recommended hotel in Rochester? | 14:58 |
mikal | mriedem: also can I get a street address? | 14:58 |
mriedem | mikal: hotels? hmm, there are a ton of them | 14:58 |
mikal | i.e. please make https://wiki.openstack.org/wiki/Sprints/NovaLibertySprint less blank | 14:59 |
mikal | mriedem: any have an IBM visitor discount? | 14:59 |
mriedem | not sure about that, will have to ask around | 14:59 |
mikal | Intel had some sort of visitor rate in Oregon | 14:59 |
mikal | Its not a big deal if it doesn't, but people will ask | 14:59 |
johnthetubaguy | dims: i was meaning, sure it helps in isolation, did it actually have an overall impact? I guess yes, but I was curious I supose | 15:00 |
mriedem | i feel like anytime people are visiting they are staying downtown somewhere so there might be a discount there | 15:00 |
edleafe | charge/. | 15:00 |
edleafe | doh - wrong window | 15:01 |
dims | mikal: there's a cluster of hotels around this one http://www.marriott.com/hotels/travel/rstts-towneplace-suites-rochester/ - pretty decent. you can even stay in downtown and drive to the ibm office, just takes 12-15 mins | 15:02 |
dims | mikal: y, just ask for IBM rate, should have one | 15:03 |
openstackgerrit | dave-mcnally proposed openstack/nova: Add policy to cover snapshotting of volume backed instances https://review.openstack.org/176805 | 15:03 |
mriedem | dims: mikal: let me confirm, | 15:03 |
mriedem | i know ibm has a thing with marriotts but wasn't sure if that was both ways | 15:03 |
*** markvoelker has joined #openstack-nova | 15:03 | |
*** wwriverrat has joined #openstack-nova | 15:04 | |
*** jecarey has joined #openstack-nova | 15:05 | |
*** dboik has quit IRC | 15:05 | |
*** rook has quit IRC | 15:07 | |
*** lpetrut has joined #openstack-nova | 15:07 | |
mikal | This new gerrit UI is going to take some getting used to | 15:08 |
dims | mriedem: spent a lot of time putting racks together there. good memories :) | 15:09 |
mriedem | dims: are you going to come for the midcycle? | 15:09 |
dims | mriedem: y | 15:09 |
*** nelsnelson has joined #openstack-nova | 15:11 | |
*** dims has quit IRC | 15:12 | |
*** aix has quit IRC | 15:12 | |
*** aswadr has quit IRC | 15:14 | |
*** yamahata has joined #openstack-nova | 15:14 | |
*** dsanders has joined #openstack-nova | 15:14 | |
*** wwriverrat has left #openstack-nova | 15:16 | |
*** baoli_ has quit IRC | 15:16 | |
*** unicell has joined #openstack-nova | 15:18 | |
*** VW_ has quit IRC | 15:18 | |
*** mtanino has quit IRC | 15:20 | |
*** unicell1 has quit IRC | 15:20 | |
*** rook has joined #openstack-nova | 15:20 | |
*** fifieldt has joined #openstack-nova | 15:22 | |
*** fifieldt has quit IRC | 15:23 | |
*** VW_ has joined #openstack-nova | 15:23 | |
johnthetubaguy | russellb: I think the gerrit upgrade made have broken your stats tools, looking at this: https://groups.google.com/forum/#!topic/repo-discuss/yQgRR5hlS3E | 15:24 |
mtreinish | johnthetubaguy: I was looking at the nova summit sched and noticed the nova functional testing session overlaps with a qa session | 15:24 |
mtreinish | do you think we could swap it with the session before after? | 15:24 |
russellb | johnthetubaguy: thanks for the heads up. indeed i used resume_sortkey ... | 15:24 |
russellb | johnthetubaguy: and based on that thread, it seems there's no workaround | 15:25 |
johnthetubaguy | russellb: yeah, that was my reading of that too :( | 15:26 |
russellb | johnthetubaguy: guess i'll replace my stats files with a passive aggressive comment about gerrit | 15:26 |
johnthetubaguy | russellb: heh | 15:26 |
*** lpetrut has quit IRC | 15:27 | |
dansmith | :( | 15:27 |
*** bkopilov has joined #openstack-nova | 15:27 | |
johnthetubaguy | russellb: I wonder if our version is new enough for this: https://gerrit-review.googlesource.com/#/c/59125/ | 15:28 |
*** deepakcs has joined #openstack-nova | 15:28 | |
*** atuvenie has quit IRC | 15:29 | |
johnthetubaguy | russellb: looks like we have 2.10 with has this: https://gerrit-review.googlesource.com/#/c/59170/ | 15:29 |
deepakcs | dansmith: morning, when u get some time, pls bless https://review.openstack.org/#/c/168805/ - TIA | 15:29 |
*** rushiagr_away is now known as rushiagr | 15:29 | |
openstackgerrit | Doug Hellmann proposed openstack/python-novaclient: Drop use of 'oslo' namespace package https://review.openstack.org/180701 | 15:30 |
russellb | johnthetubaguy: will check | 15:30 |
*** mtanino has joined #openstack-nova | 15:31 | |
*** armax has joined #openstack-nova | 15:33 | |
*** dboik has joined #openstack-nova | 15:35 | |
russellb | johnthetubaguy: seems to work, thanks | 15:36 |
*** annegentle has quit IRC | 15:37 | |
*** markus_z has quit IRC | 15:37 | |
johnthetubaguy | russellb: ah, sweet, np | 15:37 |
*** dsanders has quit IRC | 15:37 | |
*** annegentle has joined #openstack-nova | 15:38 | |
russellb | johnthetubaguy: https://review.openstack.org/181938 | 15:38 |
openstackgerrit | xing-yang proposed openstack/nova-specs: EMC ScaleIO Data Client Libvirt Volume Driver https://review.openstack.org/181941 | 15:38 |
*** atuvenie has joined #openstack-nova | 15:38 | |
openstackgerrit | Michael Still proposed openstack/nova-specs: Add the description about v2.1 API validation https://review.openstack.org/172771 | 15:38 |
sdague | russellb: apparently I have +2 on that | 15:39 |
sdague | that's going to impact a bunch of our tools iirc | 15:39 |
russellb | sdague: heh, yeah, you're stats-core | 15:39 |
russellb | i usually just merge my own stuff in that repo, figure nobody really cares | 15:40 |
*** whenry has quit IRC | 15:40 | |
sdague | yep, no worries | 15:40 |
openstackgerrit | Matt Riedemann proposed openstack/nova: Add missing runtime requirements https://review.openstack.org/181944 | 15:40 |
johnthetubaguy | russellb: does the len(changes) work correctly with the limit 5 bit? | 15:41 |
johnthetubaguy | russellb: oh wait, I see now | 15:41 |
russellb | k :) | 15:42 |
russellb | limit:5 only on the first request | 15:42 |
russellb | just in case our local cache is actually pretty close to up to date | 15:42 |
*** dboik_ has joined #openstack-nova | 15:42 | |
johnthetubaguy | russellb: yeah, I had stuff upside down in my head, looks good | 15:43 |
russellb | the cache db is so sophistacated | 15:43 |
russellb | a pickled dict to local disk! | 15:43 |
russellb | hacky scripts are fun. | 15:43 |
*** atuvenie has quit IRC | 15:43 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: Add missing runtime requirements https://review.openstack.org/181944 | 15:44 |
luv | umm, why do i get both oslo_messaging and oslo.messaging again? | 15:44 |
*** dboik has quit IRC | 15:44 | |
mriedem | luv: ? | 15:44 |
*** mpaolino has quit IRC | 15:44 | |
dhellmann | luv: http://specs.openstack.org/openstack/oslo-specs/specs/kilo/drop-namespace-packages.html | 15:45 |
russellb | RTFS | 15:45 |
russellb | heh | 15:45 |
luv | thanks | 15:45 |
mriedem | i'm still wondering about this: http://logs.openstack.org/05/168805/8/check/gate-nova-tox-functional/e562ffe/console.html.gz#_2015-05-05_04_58_59_882 | 15:45 |
mriedem | /home/jenkins/workspace/gate-nova-tox-functional/.tox/functional/local/lib/python2.7/site-packages/pkg_resources/__init__.py:2351: DeprecationWarning: The oslo namespace package is deprecated. Please use oslo_middleware instead. | 15:45 |
*** markvoelker has quit IRC | 15:45 | |
*** baoli has joined #openstack-nova | 15:46 | |
mriedem | we use oslo_middleware everywhere from what i can tell | 15:46 |
luv | interesting - so it's because of setuptools shortcomings ... unfortunate ... it looks and feels neater with the dot | 15:47 |
dhellmann | mriedem: https://review.openstack.org/178440 should fix that | 15:47 |
*** rwsu has joined #openstack-nova | 15:48 | |
mriedem | dhellmann: heh, ahh, i did see that last week when i was looking at this | 15:48 |
mriedem | since i synced up this morning i noticed that was changed | 15:48 |
mriedem | now i know why :) | 15:48 |
*** Nikolay_St has quit IRC | 15:48 | |
*** arnaud____ has joined #openstack-nova | 15:48 | |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: Proposal for cells v2 scheduling interaction https://review.openstack.org/141486 | 15:49 |
*** tojuvone has quit IRC | 15:51 | |
*** eharney_ has joined #openstack-nova | 15:54 | |
*** tcammann has joined #openstack-nova | 15:55 | |
*** eharney has quit IRC | 15:56 | |
*** ildikov has quit IRC | 15:56 | |
*** dulek has quit IRC | 15:56 | |
*** ildikov has joined #openstack-nova | 15:56 | |
mriedem | and she's down | 15:56 |
*** mpaolino has joined #openstack-nova | 15:58 | |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: Fix pep8 violations in tests https://review.openstack.org/179230 | 15:59 |
*** baoli has quit IRC | 15:59 | |
*** annegentle has quit IRC | 15:59 | |
*** annegentle has joined #openstack-nova | 16:00 | |
*** unicell has quit IRC | 16:01 | |
*** jlanoux has quit IRC | 16:05 | |
mikal | Does anyone else get 404s when they clikc on links in review dashboards? Is that a known issue? | 16:05 |
*** annegentle has quit IRC | 16:06 | |
mriedem | mikal: me too, was watching in -infra | 16:07 |
*** yingjun has quit IRC | 16:07 | |
mikal | Ok, Kevin ust reported something similar on email too | 16:07 |
mriedem | mikal: i'm only seeing it on WIP changes | 16:08 |
*** baoli has joined #openstack-nova | 16:08 | |
*** deepthi has joined #openstack-nova | 16:08 | |
*** rdo has quit IRC | 16:08 | |
mriedem | oh also, if your gerrit is setup for the 'old' screen then it doesn't work getting to changes | 16:08 |
*** matrohon has quit IRC | 16:08 | |
mikal | I thought I was using the new review screen | 16:09 |
mikal | Reviewing was working earlier this morning, but not now | 16:09 |
mriedem | if it's -W I get a 404 | 16:09 |
*** mpavone has quit IRC | 16:09 | |
*** rdo has joined #openstack-nova | 16:10 | |
mriedem | and now my dashboard just changed and all of my -Workflow changes are gone :( | 16:11 |
*** subscope has quit IRC | 16:12 | |
*** browne has joined #openstack-nova | 16:13 | |
*** jecarey has quit IRC | 16:14 | |
mriedem | mikal: working now | 16:15 |
mriedem | ? | 16:15 |
*** haypo has quit IRC | 16:16 | |
*** deepthi has quit IRC | 16:16 | |
*** jistr has quit IRC | 16:17 | |
mikal | mriedem: yep | 16:17 |
mikal | mriedem: the restart seems to have fixed me | 16:17 |
bauzas | again, trying to reload etherpad but fails | 16:17 |
mriedem | old style still doesn't work | 16:17 |
mriedem | so i'm forced to 'new' | 16:17 |
*** jcoufal has quit IRC | 16:18 | |
bauzas | new is always better (c) Barney | 16:18 |
*** jcoufal has joined #openstack-nova | 16:18 | |
claudiub | bauzas: +1 | 16:18 |
edleafe | bauzas: because it's innovative! | 16:19 |
*** jogo has quit IRC | 16:19 | |
*** rdo has quit IRC | 16:19 | |
*** moshele has quit IRC | 16:21 | |
*** rdo has joined #openstack-nova | 16:21 | |
openstackgerrit | Stuart McLaren proposed openstack/nova: Add explicit dependency on python-keystoneclient https://review.openstack.org/181966 | 16:22 |
bauzas | so, the 1M dollar question is : why etherpad is not providing a cache buster ? | 16:22 |
*** baoli has quit IRC | 16:22 | |
*** kmartin has joined #openstack-nova | 16:22 | |
*** claudiub|2 has joined #openstack-nova | 16:22 | |
*** jcoufal has quit IRC | 16:23 | |
claudiub|2 | hello. If you guys have a little bit of time to take a look at a small, crucial bugfix, I would appreciate it. :) https://review.openstack.org/#/c/181916/ | 16:23 |
*** jcoufal has joined #openstack-nova | 16:23 | |
openstackgerrit | David Bingham proposed openstack/nova: Allow cleanup completion when serial ports already released https://review.openstack.org/179978 | 16:24 |
openstackgerrit | sahid proposed openstack/nova: libvirt: introduces a Guest to wrap arround virConnect https://review.openstack.org/174313 | 16:24 |
openstackgerrit | sahid proposed openstack/nova: libvirt: update Host to return a Guest object https://review.openstack.org/181969 | 16:24 |
*** claudiub has quit IRC | 16:25 | |
*** sdake_ has joined #openstack-nova | 16:25 | |
mriedem | claudiub|2: i have an axe to grind, http://lists.openstack.org/pipermail/openstack-dev/2015-May/063183.html | 16:27 |
mriedem | claudiub|2: i couldn't get a hold of alexpilotti in here to address those concerns | 16:27 |
johnthetubaguy | mtreinish: can you email about that session clash please, and I can try shuffle things a little | 16:27 |
johnthetubaguy | mtreinish: just not sure I will get chance right away | 16:28 |
johnthetubaguy | mtreinish: actually, did you say the session just before would work? | 16:28 |
*** baoli has joined #openstack-nova | 16:28 | |
*** lpetrut has joined #openstack-nova | 16:29 | |
claudiub|2 | mriedem: hello. Well, as it was proposed on the ML, I've resubmited the bp spec, with all the details regarding to it. https://review.openstack.org/#/c/103945/ | 16:29 |
*** bkopilov has quit IRC | 16:29 | |
*** alaski has quit IRC | 16:29 | |
*** leifz has quit IRC | 16:29 | |
*** marun has joined #openstack-nova | 16:29 | |
*** sdake has quit IRC | 16:29 | |
*** mmedvede_ has quit IRC | 16:29 | |
claudiub|2 | mriedem: sorry for the inconvenience. :) | 16:30 |
mriedem | claudiub|2: thanks, did the kilo release notes get updated? | 16:30 |
*** matrohon has joined #openstack-nova | 16:30 | |
claudiub|2 | mriedem: checking | 16:31 |
*** sahid has quit IRC | 16:31 | |
mriedem | i don't see anything specific about gen2 images | 16:31 |
mtreinish | johnthetubaguy: yeah the session just before would work. | 16:31 |
mtreinish | johnthetubaguy: sure I can send you an email | 16:31 |
claudiub|2 | mriedem: I see here ( https://wiki.openstack.org/wiki/ReleaseNotes/Kilo#Hyper-V ) that "Support for generation 2 VMs" has been added. | 16:32 |
claudiub|2 | mriedem: what do you mean? | 16:32 |
mriedem | claudiub|2: yeah, but the spec said there was a docs impact | 16:32 |
mriedem | claudiub|2: "The Nova driver documentation should include an entry about this topic including when to use and when not to use generation 2 VMs. A note on the relevant Glance image property should be added as well." | 16:32 |
claudiub|2 | mriedem: ok, I'll take care of that, although that topic was pretty much covered in the spec | 16:34 |
*** unicell has joined #openstack-nova | 16:34 | |
*** alaski has joined #openstack-nova | 16:35 | |
*** Sukhdev has joined #openstack-nova | 16:35 | |
*** leifz has joined #openstack-nova | 16:35 | |
johnthetubaguy | mtreinish: I just updated that now, in the hope the other way around is fine | 16:35 |
*** sdake has joined #openstack-nova | 16:36 | |
mtreinish | johnthetubaguy: perfect that works fine. Thanks | 16:36 |
*** ajayaa has joined #openstack-nova | 16:36 | |
*** pixelbeat has joined #openstack-nova | 16:37 | |
*** Nic has joined #openstack-nova | 16:38 | |
*** pixelbeat_ has quit IRC | 16:38 | |
johnthetubaguy | mtreinish: thanks for spotting it for me, I really appreciate that :) | 16:39 |
*** browne has quit IRC | 16:39 | |
*** deepakcs has quit IRC | 16:40 | |
*** sdake_ has quit IRC | 16:40 | |
*** otter768 has joined #openstack-nova | 16:40 | |
*** arnaud____ has quit IRC | 16:42 | |
*** mmedvede_ has joined #openstack-nova | 16:42 | |
*** Marga_ has quit IRC | 16:42 | |
*** bkopilov has joined #openstack-nova | 16:43 | |
mgagne | the bug #1328546 is considered resolved in Nova but without any link to commits or changes. Can someone point me to the commit? | 16:44 |
openstack | bug 1328546 in neutron "Race condition when hard rebooting instance" [Medium,Invalid] https://launchpad.net/bugs/1328546 - Assigned to Li Ma (nick-ma-z) | 16:44 |
*** derekh_ has joined #openstack-nova | 16:44 | |
mgagne | after a brief glance at the code, I don't see anything that could have fixed the problem | 16:44 |
*** otter768 has quit IRC | 16:45 | |
*** mpaolino has quit IRC | 16:46 | |
*** ZZelle is now known as ZZelle_ | 16:46 | |
*** derekh has quit IRC | 16:47 | |
*** exploreshaifali has quit IRC | 16:48 | |
*** derekh_ has quit IRC | 16:51 | |
*** pixelbeat has quit IRC | 16:51 | |
*** dsanders has joined #openstack-nova | 16:52 | |
*** doron_afk has quit IRC | 16:52 | |
*** dguitarbite has quit IRC | 16:52 | |
*** atuvenie has joined #openstack-nova | 16:52 | |
*** rushiagr is now known as rushiagr_away | 16:53 | |
dansmith | I'm getting some 500s from gerrit loading related changes | 16:54 |
dansmith | and it also seems to be improperly marking my stuff as merge conflicts | 16:55 |
*** afazekas has quit IRC | 16:55 | |
bauzas | johnthetubaguy: I went through the Summit etherpads and you're providing a big TODO - we need less text | 16:56 |
johnthetubaguy | bauzas: yeah, quite a few say, hmm, this needs more focus | 16:57 |
mriedem | dansmith: yeah it's all fubar atm | 16:57 |
mriedem | dansmith: git review isn't working for me either | 16:57 |
bauzas | johnthetubaguy: mmm, just wonder how we can be both short and precise | 16:57 |
johnthetubaguy | bauzas: mostly to be agreed between the session leaders, or at least ensure there are concrete proposals to review for each | 16:57 |
mriedem | dansmith: the infra guys are working on it | 16:57 |
dansmith | mriedem: okay, wasn't sure if it was being discussed in infra or not, based on the convo | 16:57 |
dansmith | okay cool | 16:57 |
mriedem | sounds like it's one of many issues | 16:57 |
johnthetubaguy | bauzas: now if you are happy with the list, feel free to delete the TODO :) | 16:58 |
*** davideagnello has joined #openstack-nova | 16:58 | |
johnthetubaguy | bauzas: just a few look a bit like more than 40 mins of stuff to discuss, but I could be proved wrong, maybe the key thing is putting the most important bits first | 16:58 |
*** yamahata has quit IRC | 16:58 | |
bauzas | johnthetubaguy: sure thing | 16:58 |
*** yamahata has joined #openstack-nova | 16:58 | |
bauzas | johnthetubaguy: jaypipes had nice ideas for the first session at least, so I don't really want to cut the wood | 16:59 |
*** eharney_ is now known as eharney | 16:59 | |
bauzas | johnthetubaguy: but we can prio the bullets for sure | 16:59 |
*** amotoki_ has quit IRC | 16:59 | |
*** pixelbeat has joined #openstack-nova | 17:00 | |
*** davideagnello has quit IRC | 17:00 | |
*** davideagnello has joined #openstack-nova | 17:00 | |
*** matrohon has quit IRC | 17:01 | |
*** vilobhmm1 has joined #openstack-nova | 17:03 | |
openstackgerrit | James Chapman proposed openstack/nova-specs: SRIOV scheduling with NIC capabilities https://review.openstack.org/135331 | 17:05 |
*** annegentle has joined #openstack-nova | 17:05 | |
*** promulo_ has joined #openstack-nova | 17:05 | |
*** promulo_ has quit IRC | 17:05 | |
*** pixelbeat has quit IRC | 17:05 | |
*** promulo_ has joined #openstack-nova | 17:05 | |
*** Marga_ has joined #openstack-nova | 17:06 | |
*** romainh has left #openstack-nova | 17:07 | |
openstackgerrit | James Chapman proposed openstack/nova-specs: SRIOV scheduling with NIC capabilities https://review.openstack.org/135331 | 17:08 |
openstackgerrit | xing-yang proposed openstack/nova-specs: EMC ScaleIO Data Client Libvirt Volume Driver https://review.openstack.org/181941 | 17:09 |
*** baoli has quit IRC | 17:10 | |
openstackgerrit | xing-yang proposed openstack/nova-specs: EMC ScaleIO Data Client Libvirt Volume Driver https://review.openstack.org/181941 | 17:11 |
*** baoli has joined #openstack-nova | 17:12 | |
*** claudiub has joined #openstack-nova | 17:12 | |
*** danpb has quit IRC | 17:14 | |
*** lucasagomes is now known as lucas-dinner | 17:14 | |
*** browne has joined #openstack-nova | 17:14 | |
*** jcoufal has quit IRC | 17:15 | |
*** claudiub|2 has quit IRC | 17:15 | |
*** dguitarbite has joined #openstack-nova | 17:15 | |
*** Marga_ has quit IRC | 17:16 | |
openstackgerrit | James Chapman proposed openstack/nova-specs: SRIOV scheduling with NIC capabilities https://review.openstack.org/135331 | 17:16 |
*** jogo has joined #openstack-nova | 17:19 | |
*** jogo has quit IRC | 17:19 | |
*** jogo has joined #openstack-nova | 17:19 | |
*** rushiagr_away is now known as rushiagr | 17:21 | |
*** haomaiwa_ has joined #openstack-nova | 17:22 | |
*** patrickeast has joined #openstack-nova | 17:24 | |
*** emagana has quit IRC | 17:24 | |
*** krtaylor has quit IRC | 17:24 | |
*** gszasz has quit IRC | 17:25 | |
*** Marga_ has joined #openstack-nova | 17:25 | |
*** emagana has joined #openstack-nova | 17:26 | |
*** baoli has quit IRC | 17:29 | |
*** jodah has quit IRC | 17:30 | |
*** moshele has joined #openstack-nova | 17:30 | |
*** vipul has quit IRC | 17:31 | |
*** baoli has joined #openstack-nova | 17:34 | |
*** jodah- has quit IRC | 17:35 | |
*** rushiagr is now known as rushiagr_away | 17:36 | |
*** krtaylor has joined #openstack-nova | 17:36 | |
-openstackstatus- NOTICE: We have discovered post-upgrade issues with Gerrit affecting nova (and potentially other projects). Some changes will not appear and some actions, such as queries, may return an error. We are continuing to investigate. | 17:39 | |
*** ChanServ changes topic to "We have discovered post-upgrade issues with Gerrit affecting nova (and potentially other projects). Some changes will not appear and some actions, such as queries, may return an error. We are continuing to investigate." | 17:39 | |
*** moshele has quit IRC | 17:40 | |
*** mestery has joined #openstack-nova | 17:41 | |
*** arnaud____ has joined #openstack-nova | 17:42 | |
*** annegentle has quit IRC | 17:43 | |
*** baoli has quit IRC | 17:43 | |
*** mestery has quit IRC | 17:44 | |
*** moshele has joined #openstack-nova | 17:45 | |
*** mestery has joined #openstack-nova | 17:46 | |
*** arnaud____ has quit IRC | 17:46 | |
mriedem | johnthetubaguy: alaski: hey.... http://git.openstack.org/cgit/openstack/nova/tree/nova/virt/xenapi/image/bittorrent.py#n75 | 17:47 |
mriedem | does rax drop some custom entry points in? | 17:47 |
mriedem | i don't see any entry points in nova for a torrent_url | 17:47 |
mriedem | found that by looking at what's using pkg_resources in nova, it's this and versionutils.py from oslo-incubator | 17:47 |
mriedem | this entry point loader thing should be using stevedore now, but this looks like it's not even for anything in tree | 17:48 |
alaski | mriedem: let me check, though I don't believe we're actually using that class at all right now | 17:50 |
dansmith | yeah, | 17:50 |
dansmith | I thought it ended up not being very useful or something | 17:50 |
dansmith | forget who told me that | 17:50 |
mriedem | leif? | 17:50 |
dansmith | no | 17:50 |
mriedem | can we just kill that bittorrent module then? | 17:50 |
dansmith | dunno | 17:51 |
alaski | I wouldn't kill the module, as it could be useful for someone. but requiring the config, and removing the entrypoints stuff would be good | 17:52 |
alaski | if someone needs it they can add it back with stevedor | 17:52 |
mriedem | alaski: ok, i'm on it | 17:52 |
mriedem | after coffee | 17:52 |
mriedem | i didn't see the config option stuff right above it | 17:53 |
*** achanda has joined #openstack-nova | 17:54 | |
alaski | dansmith: I think there was some difficulty with torrent libs during testing, like downloads would stop for some reason and a root cause wasn't found | 17:55 |
dansmith | alaski: okay, I thought it was something more like it ended up saturating the wrong links, or being too unpredictable in the impacts on other resources | 17:56 |
dansmith | *seems* like it should be a damn awesome solution though | 17:56 |
*** mestery has quit IRC | 17:56 | |
alaski | that's possible too, I only heard bits and pieces about it and think there were a number of issues. | 17:57 |
alaski | yeah, we still discuss it occassionally | 17:57 |
*** baoli has joined #openstack-nova | 17:58 | |
alaski | given the ever elusive "time" we'd love to try it out again | 17:58 |
*** VW__ has joined #openstack-nova | 17:58 | |
*** BobH has joined #openstack-nova | 17:58 | |
*** VW_ has quit IRC | 17:59 | |
*** BobH1 has quit IRC | 18:00 | |
*** dguitarbite has quit IRC | 18:00 | |
*** rushiagr_away is now known as rushiagr | 18:02 | |
*** dprince has quit IRC | 18:03 | |
*** redbeard has quit IRC | 18:04 | |
*** jodah has joined #openstack-nova | 18:05 | |
*** ijw has joined #openstack-nova | 18:05 | |
*** ijw has quit IRC | 18:05 | |
*** ijw has joined #openstack-nova | 18:05 | |
mriedem | dansmith: holy cow, there is an ibmvisitor open network already down where the meetup will be. my work is done. | 18:09 |
dansmith | mriedem: does it work? | 18:09 |
dansmith | mriedem: in BVR, the visitor network required a sponsor to pre-generate an access code to activate it | 18:10 |
dansmith | else you couldn't go anywhere | 18:10 |
dansmith | mriedem: you might want to try with a linux laptop too | 18:10 |
dansmith | because if we get there and only you (with your heretical windows box) can get online, you'll be in big trouble | 18:10 |
mriedem | i'm going to take my laptop down there later and try vpn | 18:11 |
*** dsanders has quit IRC | 18:11 | |
*** annegentle has joined #openstack-nova | 18:12 | |
sdague | right, I'm pretty sure IBM visitor required preregistration codes | 18:12 |
sdague | at least it did all the times that I tried using it | 18:12 |
dansmith | yeah | 18:13 |
mriedem | well way to rain on my parade | 18:13 |
lbragstad | mriedem: you could probably bribe mrodden into trying on one of his linux laptops | 18:13 |
mriedem | i'm not trying nothing nohow | 18:14 |
*** emagana has quit IRC | 18:15 | |
mrodden | mriedem, test ur crap plzkthx | 18:16 |
mrodden | wifi networks included | 18:16 |
*** emagana has joined #openstack-nova | 18:17 | |
*** achanda has quit IRC | 18:18 | |
*** tsekiyama has joined #openstack-nova | 18:19 | |
*** tonytan4ever has joined #openstack-nova | 18:20 | |
*** baoli has quit IRC | 18:21 | |
*** baoli has joined #openstack-nova | 18:21 | |
*** artom_ has joined #openstack-nova | 18:22 | |
*** tralon_ has joined #openstack-nova | 18:23 | |
*** obondarev_ has joined #openstack-nova | 18:23 | |
*** lennyb has joined #openstack-nova | 18:23 | |
*** hemna_ has joined #openstack-nova | 18:24 | |
*** unicell has quit IRC | 18:25 | |
*** alaski has quit IRC | 18:25 | |
*** dboik_ has quit IRC | 18:25 | |
*** bnemec has quit IRC | 18:25 | |
*** vladikr has quit IRC | 18:25 | |
*** lsmola_ has quit IRC | 18:25 | |
*** hanzhang has quit IRC | 18:25 | |
*** kaisers has quit IRC | 18:25 | |
*** Guest64787 has quit IRC | 18:25 | |
*** morganfainberg has quit IRC | 18:25 | |
*** russellb has quit IRC | 18:25 | |
*** ir2ivps3 has quit IRC | 18:25 | |
*** sorrison has quit IRC | 18:25 | |
*** raph_ has quit IRC | 18:25 | |
*** mrda has quit IRC | 18:25 | |
*** strictlyb has quit IRC | 18:25 | |
*** larsks|alt has joined #openstack-nova | 18:26 | |
*** Vek has quit IRC | 18:26 | |
*** jodah- has joined #openstack-nova | 18:26 | |
*** mjg has joined #openstack-nova | 18:26 | |
*** emagana has quit IRC | 18:26 | |
*** leifz has quit IRC | 18:26 | |
*** gilliard has quit IRC | 18:26 | |
*** larsks has quit IRC | 18:26 | |
*** tlbr has quit IRC | 18:26 | |
*** hemna has quit IRC | 18:26 | |
*** slagle has quit IRC | 18:26 | |
* mriedem is sad that git review still doesn't work | 18:26 | |
*** claudiub|2 has joined #openstack-nova | 18:26 | |
*** claudiub has quit IRC | 18:26 | |
*** jodah has quit IRC | 18:26 | |
*** artom has quit IRC | 18:26 | |
*** htruta has quit IRC | 18:26 | |
*** zhangtralon has quit IRC | 18:26 | |
*** rwmjones has quit IRC | 18:26 | |
*** obondarev has quit IRC | 18:26 | |
*** lan has quit IRC | 18:26 | |
*** _gryf|afk has quit IRC | 18:26 | |
*** mishkin_ has quit IRC | 18:26 | |
*** lparth has quit IRC | 18:26 | |
sdague | mriedem: ? | 18:26 |
*** _gryf|afk has joined #openstack-nova | 18:26 | |
mriedem | ! [remote rejected] HEAD -> refs/publish/master/bug/1453857 (n/a (unpacker error)) | 18:27 |
openstack | mriedem: Error: "remote" is not a valid command. | 18:27 |
*** lparth has joined #openstack-nova | 18:27 | |
*** mishkin__ has joined #openstack-nova | 18:27 | |
*** leifz has joined #openstack-nova | 18:27 | |
*** slagle has joined #openstack-nova | 18:27 | |
*** alaski has joined #openstack-nova | 18:27 | |
*** unicell has joined #openstack-nova | 18:27 | |
*** dboik_ has joined #openstack-nova | 18:27 | |
*** bnemec has joined #openstack-nova | 18:27 | |
*** vladikr has joined #openstack-nova | 18:27 | |
*** lsmola_ has joined #openstack-nova | 18:27 | |
*** hanzhang has joined #openstack-nova | 18:27 | |
*** kaisers has joined #openstack-nova | 18:27 | |
*** morganfainberg has joined #openstack-nova | 18:27 | |
*** russellb has joined #openstack-nova | 18:27 | |
*** ir2ivps3 has joined #openstack-nova | 18:27 | |
*** sorrison has joined #openstack-nova | 18:27 | |
*** raph_ has joined #openstack-nova | 18:27 | |
*** mrda has joined #openstack-nova | 18:27 | |
*** strictlyb has joined #openstack-nova | 18:27 | |
mriedem | sdague: the thing infra is working on i believe | 18:27 |
*** htruta has joined #openstack-nova | 18:27 | |
*** tlbr has joined #openstack-nova | 18:27 | |
*** slagle has quit IRC | 18:27 | |
*** slagle has joined #openstack-nova | 18:27 | |
*** rwmjones has joined #openstack-nova | 18:28 | |
*** emagana has joined #openstack-nova | 18:28 | |
*** lan has joined #openstack-nova | 18:28 | |
dansmith | I still can't get to this review of mine: | 18:28 |
dansmith | https://review.openstack.org//#/c/173995/ | 18:28 |
*** larsks|alt is now known as larsks | 18:28 | |
jodah- | neither can i | 18:28 |
*** dsanders has joined #openstack-nova | 18:28 | |
*** annegentle has quit IRC | 18:29 | |
*** dan| has joined #openstack-nova | 18:29 | |
dansmith | the level of discussion in #infra has me pretty concerned | 18:29 |
sdague | oh, the corruption issue | 18:29 |
dansmith | like, db corruption causing things like that ^ | 18:29 |
*** annegentle has joined #openstack-nova | 18:29 | |
clarkb | its not db corruption | 18:32 |
clarkb | its jgit inable to read packfiles | 18:32 |
clarkb | (C git works fine) | 18:32 |
*** atuvenie has quit IRC | 18:32 | |
dansmith | clarkb: that's causing my 404 too? | 18:32 |
*** jodah- has quit IRC | 18:32 | |
clarkb | yes | 18:32 |
dansmith | okay cool | 18:32 |
dansmith | the "schema migration XX does YY" looked concerning :) | 18:33 |
*** VW__ has quit IRC | 18:33 | |
clarkb | well looking at what a downgrade will involve | 18:33 |
dansmith | ah | 18:33 |
*** jodah has joined #openstack-nova | 18:33 | |
clarkb | also if you just need refs you can fetch from git.o.o | 18:33 |
dansmith | clarkb: has jgit reached "bane of our existence" phase yet? | 18:33 |
clarkb | I think for today at least :) | 18:33 |
dansmith | :) | 18:34 |
*** ildikov has quit IRC | 18:34 | |
sdague | dansmith: so... https://review.openstack.org/#/c/181420 is now 100% fail on bringing up the second cinder guest ... without an obvious reason for that failure | 18:34 |
*** VW_ has joined #openstack-nova | 18:35 | |
*** Sukhdev has quit IRC | 18:35 | |
sdague | what else do we need in the dumps to help sort that out | 18:35 |
dansmith | nice, let me catch up | 18:35 |
dansmith | so this is 10 normal nova guestsm, | 18:35 |
dansmith | followed by 10 BFV guests? | 18:35 |
sdague | well, 10 and 4 | 18:36 |
dansmith | and #2 BFV fails? | 18:36 |
dansmith | okay | 18:36 |
sdague | because I didn't want to consume the entire volume space | 18:36 |
dansmith | yeah | 18:36 |
sdague | but yes, #2 bvf fails pretty consistently from what I see in those runs | 18:36 |
*** ZZelle has joined #openstack-nova | 18:36 | |
dansmith | cool, this is definitely cloud-scale | 18:36 |
dansmith | 12 guests.. should be plenty | 18:36 |
sdague | well, also, we don't verify the first 10 | 18:36 |
sdague | so this is the second host we've tried to ping | 18:37 |
sdague | and ssh into | 18:37 |
dansmith | oh | 18:37 |
sdague | we *would* verify these later | 18:37 |
*** zhenguo has quit IRC | 18:37 | |
sdague | but never get that far because of early fail for the 2nd bfv to be sshable | 18:37 |
dansmith | okay, wait, | 18:38 |
dansmith | so we boot 10 nova guests and don't check them, | 18:38 |
sdague | yep | 18:38 |
dansmith | but boot two cinder guests and check those and the second one fails? | 18:38 |
sdague | yes | 18:38 |
dansmith | or rather, check each cinder one when we create | 18:38 |
dansmith | was that important to get to reliable fail? not checking the nova ones | 18:39 |
sdague | I don't know | 18:39 |
dansmith | okay | 18:39 |
sdague | the reason the cinder ones are checked as they are created is we ssh in and put content in the volume | 18:39 |
sdague | to ensure that content is persistent | 18:39 |
dansmith | and nothing is shut down right now either, so this is just upgrade-unrelated right? | 18:39 |
sdague | so you have to do a ping loop, then an ssh loop | 18:39 |
sdague | correct | 18:39 |
dansmith | eff. | 18:39 |
sdague | we're still in the create resources stage | 18:40 |
dansmith | right | 18:40 |
dansmith | so, I kinda wonder about not doing the cinder ones to see if we don't hit the upgrade issue without them | 18:40 |
*** annegentle has quit IRC | 18:40 | |
*** yamahata has quit IRC | 18:40 | |
dansmith | as in, maybe the BFV process is causing us to be more racy with secgroups in some way that we're always borking networking somewhere in the BFV phase and noticing later? | 18:40 |
sdague | so, I can put up a patch that does that if you like | 18:41 |
*** annegentle has joined #openstack-nova | 18:41 | |
dansmith | is that a crazy thought? | 18:41 |
sdague | http://logs.openstack.org/20/181420/4/check/check-grenade-dsvm/57f64bd/logs/worlddump-2015-05-11-140416.txt.gz is the fail | 18:41 |
*** eglynn has quit IRC | 18:41 | |
dansmith | we do a lot more waiting for things in the BFV case I think | 18:41 |
sdague | includes the iptables dump | 18:41 |
sdague | dansmith: define "we"? | 18:41 |
*** otter768 has joined #openstack-nova | 18:41 | |
dansmith | nova | 18:41 |
dansmith | nova-compute | 18:42 |
sdague | I had to put a lot of client side waiting in there to be able to progress | 18:42 |
dansmith | no, | 18:42 |
*** jecarey_ has joined #openstack-nova | 18:42 | |
dansmith | I meant in the BFV case, nova does different things, which might cause us to be more likely to race with some networking-related thing or something | 18:42 |
openstackgerrit | sean mooney proposed openstack/nova: fix AggregateTypeAffinityFilter multi values support https://review.openstack.org/141883 | 18:43 |
dansmith | I would think worlddump should have: "brctl show" and "ip link" and "ip addr" and a route listing | 18:43 |
dansmith | also "arp -n" | 18:43 |
sdague | ok, I can add those | 18:43 |
*** ajayaa has quit IRC | 18:45 | |
*** otter768 has quit IRC | 18:46 | |
openstackgerrit | Eric Brown proposed openstack/nova: VMware: update supported vsphere 6.0 os types https://review.openstack.org/181020 | 18:47 |
*** dan| has quit IRC | 18:49 | |
*** arnaud____ has joined #openstack-nova | 18:49 | |
openstackgerrit | Eric Brown proposed openstack/nova: Deprecate nova ironic driver's admin_auth_token https://review.openstack.org/179960 | 18:49 |
openstackgerrit | Matt Riedemann proposed openstack/nova: Fix pip-missing-reqs https://review.openstack.org/181944 | 18:50 |
mriedem | dansmith: welcome back cotter https://review.openstack.org//#/c/173995/ | 18:51 |
mriedem | s/cotter/kotter/ | 18:51 |
dansmith | heh | 18:51 |
dansmith | nice | 18:51 |
*** baoli has quit IRC | 18:51 | |
openstackgerrit | Eric Brown proposed openstack/nova: VMware: Don't raise exception on resize of 0 disk https://review.openstack.org/173170 | 18:51 |
*** harlowja has quit IRC | 18:52 | |
*** harlowja has joined #openstack-nova | 18:52 | |
vilobhmm1 | jaypipes : ping | 18:52 |
jaypipes | vilobhmm1: pong | 18:53 |
vilobhmm1 | wanted to get more insights into the way the rpc layer and service things are tied together | 18:53 |
vilobhmm1 | https://github.com/openstack/nova/blob/master/nova/rpc.py#L136 -> oslo.meesaging | 18:54 |
vilobhmm1 | layer | 18:54 |
vilobhmm1 | to get the target details | 18:54 |
*** arnaud____ has quit IRC | 18:54 | |
vilobhmm1 | but where is the target places I mean does the server refer to one of the services currently stored in the nova.services ? | 18:54 |
*** dan_ has joined #openstack-nova | 18:54 | |
vilobhmm1 | https://github.com/vilobhmm/oslo.messaging/blob/master/oslo_messaging/rpc/server.py#L124 -> https://github.com/vilobhmm/oslo.messaging/blob/master/oslo_messaging/rpc/dispatcher.py#L79 -> https://github.com/openstack/oslo.messaging/blob/master/oslo_messaging/target.py#L74 | 18:55 |
*** dan_ is now known as Guest25580 | 18:55 | |
vilobhmm1 | just to be more specific of what I am talking about | 18:55 |
vilobhmm1 | about the oslo.messaging layer | 18:55 |
vilobhmm1 | https://github.com/openstack/nova/blob/master/nova/rpc.py#L136 -> to get the rpc server details | 18:56 |
dansmith | vilobhmm1: see nova/compute/rpcapi for example | 18:57 |
openstackgerrit | Matt Riedemann proposed openstack/nova: Fix cinder v1 warning with cinder_catalog_info option reference https://review.openstack.org/182015 | 18:57 |
dansmith | vilobhmm1: the topic "compute" in there is what points it at a compute node | 18:57 |
dansmith | vilobhmm1: same for network/rpcapi, scheduler, conductor, etc | 18:57 |
mriedem | vilobhmm1: also https://www.youtube.com/watch?v=wVXiw09XP04 | 18:57 |
dansmith | oh yeah | 18:57 |
dansmith | ffs :) | 18:58 |
mriedem | educational recordings ftw! | 18:58 |
dansmith | too bad we don't do those anymore | 18:58 |
mriedem | i still have to watch more than the first 10 minutes of that | 18:58 |
*** ijw has quit IRC | 18:58 | |
mriedem | dansmith: i think sdague just needs topics and volunteers | 18:58 |
dansmith | is 10m where I start talking? | 18:58 |
dansmith | mriedem: I think sdague needs fewer children | 18:59 |
mriedem | ouch | 18:59 |
*** ijw has joined #openstack-nova | 18:59 | |
dansmith | well, everyone does, IMHO | 18:59 |
mriedem | i think the scheduler would be a decent topic | 18:59 |
mriedem | and a rundown of the bazillion filters | 18:59 |
mriedem | mentioned that to bauzas but TZ is difficult since it's late friday night his time | 19:00 |
dansmith | yeah, I could use a nap | 19:00 |
sdague | mriedem: yeh, honestly, I kind of put it down for a while due to running out of time. | 19:00 |
vilobhmm1 | mriedem : cool thanks…but still dansmith, mridem : https://github.com/openstack/nova/blob/master/nova/service.py#L190 sets the rpcserver but again coming back to my orignal question | 19:00 |
vilobhmm1 | the target to which the rpcserver connects to/talks to is it related to services entry in nova.services ? | 19:01 |
dansmith | vilobhmm1: are you talking about the "target" being the rabbit server? | 19:01 |
dansmith | otherwise rpc servers don't connect to anything | 19:01 |
*** _gryf|afk is now known as _gryf | 19:01 | |
*** rushiagr is now known as rushiagr_away | 19:02 | |
vilobhmm1 | target : here i suppose i the end service like nova-compute or something of that sort | 19:02 |
dansmith | nova compute *is* an rpc server | 19:02 |
vilobhmm1 | TRANSPORT : https://github.com/openstack/nova/blob/master/nova/rpc.py#L49 | 19:02 |
dansmith | vilobhmm1: seriously, you should watch that video | 19:02 |
vilobhmm1 | https://github.com/openstack/nova/blob/master/nova/rpc.py#L136 | 19:03 |
vilobhmm1 | ok | 19:03 |
dansmith | we recorded it so we didn't have to explain it every time someone asks in here | 19:03 |
*** ijw has quit IRC | 19:03 | |
*** yamahata has joined #openstack-nova | 19:03 | |
openstackgerrit | Eric Brown proposed openstack/nova: Use oslo_config choices support https://review.openstack.org/158573 | 19:04 |
*** VW__ has joined #openstack-nova | 19:05 | |
*** emagana has quit IRC | 19:05 | |
*** emagana has joined #openstack-nova | 19:05 | |
*** dsanders has quit IRC | 19:05 | |
*** VW_ has quit IRC | 19:07 | |
openstackgerrit | Dan Smith proposed openstack/nova: Move our ObjectListBase to subclass from the Oslo one https://review.openstack.org/173995 | 19:08 |
*** VW_ has joined #openstack-nova | 19:08 | |
*** VW__ has quit IRC | 19:09 | |
*** VW_ has quit IRC | 19:10 | |
jaypipes | vilobhmm1: sorry, stepped away. so, I think you are asking whether the services table in the Nova database is involved in the setup of the RPC endpoints (either consumer or producer). Is that correct? | 19:10 |
*** VW_ has joined #openstack-nova | 19:10 | |
*** Marga_ has quit IRC | 19:11 | |
*** eharney has quit IRC | 19:12 | |
*** eglynn has joined #openstack-nova | 19:13 | |
openstackgerrit | jichenjc proposed openstack/nova: Only set access_ip_* when instance goes ACTIVE https://review.openstack.org/129333 | 19:14 |
*** dave-mccowan has quit IRC | 19:17 | |
vilobhmm1 | jaypipes : thats exactly i want to know | 19:17 |
*** dsanders has joined #openstack-nova | 19:17 | |
jaypipes | vilobhmm1: ok, I thought so :) | 19:17 |
dansmith | jaypipes: are you answering or just clarifying? | 19:19 |
jaypipes | dansmith: I am in process of answering :) | 19:20 |
dansmith | okay | 19:20 |
*** Jeffrey4l has quit IRC | 19:21 | |
bauzas | that's actually a good topic for discussing at the Summit ;) | 19:21 |
dansmith | is it? I think the answer is "no" :) | 19:21 |
*** iamjarvo has joined #openstack-nova | 19:22 | |
bauzas | dansmith: eh, we're updating a DB table each time a new service is run, but the operator can actually use an other driver but the DB one for SG :) | 19:22 |
vilobhmm1 | bauzas : +1 | 19:22 |
dansmith | right, but it doesn't have anything to do with the rpc bit | 19:22 |
bauzas | dansmith: that's my point, this is confusing | 19:23 |
openstackgerrit | Michael Still proposed openstack/nova: Move more qemu calls into virt.images I'd like to move calls to qemu into a central place because it will make it easier to work around the vagaries of qemu in a consistent manner. https://review.openstack.org/182030 | 19:23 |
bauzas | dansmith: I'm okay if we say that the table is just for knowing the RPC servers | 19:23 |
dansmith | it's not though | 19:23 |
bauzas | dansmith: I'm not okay if this is for knowing if a service is up | 19:23 |
dansmith | it really has nothing at all to do with it | 19:23 |
jaypipes | vilobhmm1: so, the services database table really isn't involved in much of that. The different RPC services (nova-compute, nova-conductor, etc) when they start up, they pass in two variables to their Service.create() constructor: topic and binary. The topic comes from a configuration variable, like here: https://github.com/openstack/nova/blob/master/nova/cmd/compute.py#L71. The binary is a hard-coded string like 'nova-compute'. In the Service.cr | 19:24 |
jaypipes | eate() method, the host is picked up by default from CONF.host configuration variable, which by default comes from a gethostname() system call from the netconf module: https://github.com/openstack/nova/blob/master/nova/service.py#L242 | 19:24 |
bauzas | dansmith: actually, I'm even unclear myself | 19:24 |
bauzas | dansmith: I mean, we need to draw a line for the services table saying why it's needed | 19:24 |
jaypipes | vilobhmm1: only when the Service.start() method is called does a services database table record get created if there is none matching the host and binary: https://github.com/openstack/nova/blob/master/nova/service.py#L165 | 19:25 |
jaypipes | vilobhmm1: after this point, the database services table records have ZERO to do with RPC communication between a consumer and a producer. | 19:25 |
openstackgerrit | Chris St. Pierre proposed openstack/nova: Allow blacklisting os-api extensions https://review.openstack.org/160986 | 19:25 |
dansmith | yep, that ^ | 19:25 |
bauzas | agreed | 19:25 |
jaypipes | vilobhmm1: the services database table is ONLY used for the crappy-ass-trying-to-be-zookeeper DB servicegroup API driver. | 19:25 |
jaypipes | vilobhmm1: which of course you already are aware of because you submitted https://review.openstack.org/#/c/138607/ | 19:26 |
jaypipes | vilobhmm1: and yes, I know I owe you a review on that :) | 19:26 |
jaypipes | bauzas, dansmith: anything to add or correct to what I wrote above? | 19:27 |
bauzas | jaypipes: MHO is that the services table should just be for referencing the services | 19:28 |
jaypipes | say what? | 19:28 |
bauzas | that's it | 19:28 |
jaypipes | what do you mean? | 19:28 |
dansmith | jaypipes: nope, I thought I was saying that already in my explanation, but not clearly I guess :) | 19:28 |
bauzas | jaypipes: lemme rephrase the best I can | 19:28 |
jaypipes | bauzas: that's a bit of a tautology isn't it? :) | 19:28 |
bauzas | jaypipes: lol | 19:29 |
vilobhmm1 | jaypipes : agree..and thanks for the reply…so in case the service/process is deleted lets say nova-compute is deleted on compute node (the rpc-server is deleted) and the entry in nova.services is marked as deleted….as you mentioned the RPC server and nova.services are not related makes sense now… | 19:29 |
bauzas | jaypipes: not exactly, since currently the services table is more than just a reference :) | 19:29 |
*** subscope has joined #openstack-nova | 19:29 | |
vilobhmm1 | *nova-compute rpc-server* | 19:29 |
jaypipes | vilobhmm1: yep, and if the nova-compute is restarted and the DB record has been deleted, then the Server.start() method will re-create the service database table record for that host. | 19:29 |
bauzas | jaypipes: so, the best I can say is that's just mentioning that a service exists or existed | 19:30 |
bauzas | like the Schrodinger's cat, you don't know if the service is alive | 19:30 |
*** angdraug has joined #openstack-nova | 19:31 | |
*** dave-mccowan has joined #openstack-nova | 19:31 | |
jaypipes | bauzas: if you are saying that relying on the services database table for whether the service is *actually* up or down is not really possible, then yes I agree with you. :) | 19:31 |
bauzas | jaypipes: that's MHO yes | 19:32 |
jaypipes | bauzas: it's MHO as well :) | 19:32 |
bauzas | hence the word "reference" | 19:32 |
bauzas | a pointer if you prefer | 19:32 |
vilobhmm1 | jaypipes : seems logical…thanks | 19:32 |
jaypipes | vilobhmm1: well, it's far from logical, but hey, welcome to the servicegroup API :) | 19:33 |
vilobhmm1 | jaypipes : haha…sure…and I agree to your point of moving the hearbeat thing or liveliness thing that the current nova.services thing is doing to something like zookeeper because of reasons mentioned in the spec https://review.openstack.org/#/c/138607/ | 19:35 |
jaypipes | yup. | 19:35 |
vilobhmm1 | from scheduler point of view, scheduler simply relies on nova.services to make sure if the service is up or not and not whether the service is actually running right | 19:35 |
bauzas | vilobhmm1: that's not exactly that :) | 19:36 |
bauzas | vilobhmm1: the scheduler just makes sure that the compute is referenced to a service, dead or notr | 19:36 |
bauzas | not | 19:36 |
sdague | so, wasn't the point of this that it's an eventually consistent model on the db side, less than ideal, but sometimes good enough. The zookeeper implementation was brought in to be a stronger guaruntee, at the cost of another service yuo had to run | 19:36 |
dansmith | it's often good enough, yes | 19:37 |
bauzas | vilobhmm1: to know if the service is up is the job of the ComputeFilter, which is optional | 19:37 |
dansmith | which is why I don't want to condemn it any time soon | 19:37 |
dansmith | i.e. we can't _replace_ it with zk any time soon because it's really not needed | 19:37 |
sdague | the db implementation was there first, it was only carved into a plug point when yun and ibm research people wanted to bring in zookeeper | 19:37 |
dansmith | but we do need to abstract it out so we _can_ reasonably replace it with things | 19:37 |
sdague | sure | 19:38 |
dansmith | right, and it's not really done, is the problem | 19:38 |
sdague | well, it was half assed done :) | 19:38 |
sdague | then rotted | 19:38 |
vilobhmm1 | dansmith : https://blueprints.launchpad.net/nova/+spec/servicegroup-api-control-plane | 19:38 |
dansmith | I dunno, | 19:38 |
dansmith | maybe half-assed done is a reasonable description | 19:38 |
vilobhmm1 | yep | 19:39 |
sdague | also, it's probably worth figuring out if anyone is actually using the memcache driver, or if that can just be deprecated, because as far as I can tell it was just a half copy paste | 19:39 |
dansmith | sdague: the spec calls that out pretty well I think | 19:40 |
dansmith | dunno if you've read that or not | 19:40 |
vilobhmm1 | sdague : https://review.openstack.org/#/c/138607/ | 19:40 |
jaypipes | vilobhmm1: actually, the scheduler looks at the disabled state as well as the up state: https://github.com/openstack/nova/blob/master/nova/scheduler/driver.py#L54-L57 | 19:40 |
vilobhmm1 | has clearly depicted the problem | 19:41 |
dansmith | my point about half-assed-ness was that it seems more like quarter-ass complete :) | 19:41 |
sdague | heh | 19:41 |
vilobhmm1 | sdague : http://lists.openstack.org/pipermail/openstack-dev/2015-May/063602.html | 19:41 |
bauzas | jaypipes: oh oh, missed that *facepalm* | 19:42 |
*** Marga_ has joined #openstack-nova | 19:42 | |
bauzas | jaypipes: should we then just say bye bye to the hellish ComputeFilter ? | 19:42 |
jaypipes | vilobhmm1: and the "disabled thing" is hidden in the DB API :( https://github.com/openstack/nova/blob/master/nova/db/sqlalchemy/api.py#L478 | 19:42 |
jaypipes | bauzas: what is hellish about the compute filter? | 19:43 |
bauzas | jaypipes: lemme give you a thread | 19:43 |
jaypipes | bauzas: never mind, I remember. | 19:43 |
jaypipes | bauzas: it's not hellish. it's retarded. | 19:43 |
vilobhmm1 | jaypipes : and we have numerous interfaces to change the state of the service :P | 19:43 |
bauzas | jaypipes: correct, it's confusing | 19:43 |
jaypipes | vilobhmm1: well, yeah. :) | 19:43 |
*** tralon__ has joined #openstack-nova | 19:44 | |
sdague | so... I'm not sure why ComputeFilter is so terrible in it's current state | 19:45 |
openstackgerrit | Tomoki Sekiyama proposed openstack/nova: libvirt: Make behavior of os_require_quiesce consistent https://review.openstack.org/167819 | 19:45 |
bauzas | sdague: because it promises something it can stand | 19:46 |
bauzas | can't even | 19:46 |
sdague | bauzas: that's why we have built in retry logic | 19:46 |
jaypipes | sdague: because it's never executed :) | 19:46 |
*** Marga_ has quit IRC | 19:46 | |
sdague | jaypipes: it's never executed? | 19:46 |
bauzas | sdague: the retry design is not a reason for preventing good things to happen :) | 19:47 |
*** tralon_ has quit IRC | 19:47 | |
sdague | bauzas: no, that's fair, but I also know of other systems that started from the perfect scheduling design point | 19:47 |
sdague | and the result is a system that had a pretty small max number of hosts | 19:47 |
sdague | and it took 20 minutes to schedule 1 vm | 19:48 |
dansmith | sdague: and a boot request made in 2007 is still pending | 19:48 |
openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova: Add and use raise_http_not_implemented_error() https://review.openstack.org/163697 | 19:48 |
sdague | right, exactly | 19:48 |
bauzas | \o/ | 19:48 |
jaypipes | sdague: this line is the only return possible from the filter: https://github.com/openstack/nova/blob/master/nova/scheduler/filters/compute_filter.py#L50 | 19:48 |
sdague | so... while I get that their is slop here, the slop was slightly intentional to prevent this from being a key scaling bottleneck | 19:49 |
sdague | jaypipes: because the data model is broken? | 19:49 |
dansmith | jaypipes: how is that? | 19:49 |
jaypipes | sdague: because only non-disabled, non-down hosts are ever passed to the HostStateManager. | 19:49 |
sdague | jaypipes: ok, so just that the contract was satisfied further up? | 19:49 |
*** oomichi has joined #openstack-nova | 19:49 | |
jaypipes | sdague, dansmith: because of this: https://github.com/openstack/nova/blob/master/nova/scheduler/driver.py#L51-L57 | 19:49 |
*** matrohon has joined #openstack-nova | 19:50 | |
dansmith | yeah, that changed though, yes? | 19:50 |
jaypipes | which runs prior to the HostStateManager getting its cache of HostState objects, each of which gets its Service object from there. | 19:50 |
openstackgerrit | Matt Riedemann proposed openstack/nova: xenapi: remove bittorrent entry point lookup code https://review.openstack.org/182048 | 19:50 |
jaypipes | dansmith: nope. | 19:50 |
dansmith | jaypipes: in 2012 with the servicegroup stuff | 19:50 |
dansmith | https://github.com/openstack/nova/commit/64e167eb62bd3483b2947ec8de218453c116bd93 | 19:50 |
*** oomichi has quit IRC | 19:50 | |
jaypipes | dansmith: no, not really. it still does the select all non-disabled hosts from the database call. | 19:51 |
dansmith | it did at that time? | 19:52 |
jaypipes | dansmith: it does it via the objects API now, instead of a direct DB call, but the strategy of doing that hasn't changed. | 19:52 |
dansmith | so it's code that never ever made sense? | 19:52 |
jaypipes | yup. | 19:52 |
sdague | ok, so then lets clean it up, that's cool | 19:52 |
* harlowja lurks | 19:52 | |
jaypipes | sdague: there's 3 blueprints proposed already for that :) | 19:52 |
sdague | jaypipes: ok | 19:52 |
*** jichen has quit IRC | 19:53 | |
sdague | so this was just a vent session because everyone needed one? :) | 19:53 |
jaypipes | sdague: sorry man, I'm not yelling at ya... :) just pointing out that a number of BPs have been created already... | 19:53 |
jaypipes | sdague: well, no, it started as vilobhmm1 wanting answers to some questions about the services database table and its relation to the RPC communication (minimal) | 19:54 |
sdague | sure, that's fine, what can I do to help move this ball forward? Seems like deprecating ComputeFilter should be a simple patch, given that it's functionality is duplicated at the moment | 19:54 |
vilobhmm1 | jaypipes :exactly now uses object layer but does same db call | 19:55 |
jaypipes | yup. | 19:55 |
jaypipes | sdague: a review on this spec first please: https://review.openstack.org/#/c/138607/ | 19:55 |
bauzas | sdague: http://lists.openstack.org/pipermail/openstack-dev/2015-March/058438.html | 19:56 |
openstackgerrit | Claudiu Belu proposed openstack/nova: Hyper-V: Fixes cold migration / resize issue https://review.openstack.org/181916 | 19:56 |
bauzas | sdague: but the scope changed since jaypipes discovered it was doubled | 19:56 |
jaypipes | sdague: these are the other two blueprints involved: https://blueprints.launchpad.net/nova/+spec/make-availability-zone-real-data-model and https://blueprints.launchpad.net/nova/+spec/servicegroup-api-control-plane | 19:56 |
jaypipes | sdague: we're working on spec submissions for both now. | 19:56 |
jaypipes | sdague: I'll ping you them later. | 19:57 |
bauzas | sdague: so I can just respawn https://review.openstack.org/#/c/162180/ | 19:57 |
*** lennyb has left #openstack-nova | 20:00 | |
openstackgerrit | Michael Still proposed openstack/nova: Move more qemu calls into virt.images https://review.openstack.org/182030 | 20:03 |
*** dims has joined #openstack-nova | 20:03 | |
vilobhmm1 | out for lunch..please ping me if needed | 20:05 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Eventlet supports python logging directly https://review.openstack.org/179801 | 20:05 |
sdague | bauzas: so doesn't that patch say that ComputeFilter is not useless, because it's still filtering out disabled hosts? | 20:05 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Use oslo-config-generator instead of generate_sample.sh https://review.openstack.org/180013 | 20:06 |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Add rootwrap daemon mode support https://review.openstack.org/180695 | 20:06 |
bauzas | sdague: that's what I said above, jaypipes just found tonight that the state check was also done elsewhere | 20:06 |
bauzas | sdague: my point was initially to move the logic back to the scheduler (not in a filter) | 20:06 |
_gryf | dansmith, hey | 20:07 |
*** mwagner_lap has quit IRC | 20:07 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: Drop explicit suds dependency https://review.openstack.org/181554 | 20:08 |
dansmith | _gryf: good morning, and welcome! | 20:08 |
_gryf | dansmith, i would say good evening :) | 20:09 |
_gryf | anyway | 20:09 |
*** pixelbeat has joined #openstack-nova | 20:09 | |
*** sdake_ has joined #openstack-nova | 20:10 | |
jaypipes | bauzas: I didn't find this out tonight :) I've known about it for a while. | 20:10 |
_gryf | dansmith, you wrote in a comment, that we will need a notify in case of forced_down service. | 20:10 |
bauzas | jaypipes: while I totally missed it | 20:10 |
dansmith | jaypipes: well, the rest of us did | 20:10 |
jaypipes | heh :) | 20:10 |
dansmith | jaypipes: I knew that line in the filter had changed, but never realized the db query was stupid | 20:11 |
*** nihilifer has quit IRC | 20:11 | |
*** doude has quit IRC | 20:11 | |
_gryf | dansmith, why do we need this and who will be the target of such notification? | 20:11 |
dansmith | _gryf: no, I said I think we shuld | 20:11 |
*** doude has joined #openstack-nova | 20:11 | |
_gryf | dansmith, than way should we have it? ;) | 20:11 |
dansmith | _gryf: the target will be the target of all the notifications.. monitoring tools | 20:11 |
mriedem | _gryf: this might help https://wiki.openstack.org/wiki/SystemUsageData | 20:12 |
*** sdake has quit IRC | 20:13 | |
_gryf | mriedem, thanks | 20:16 |
*** b3nt_pin has joined #openstack-nova | 20:16 | |
_gryf | dansmith, so i should to dig into it aswell | 20:17 |
*** _gryf is now known as _gryf|afk | 20:18 | |
*** ildikov has joined #openstack-nova | 20:19 | |
*** Vek has joined #openstack-nova | 20:21 | |
*** b3nt_pin has quit IRC | 20:25 | |
*** b3nt_pin has joined #openstack-nova | 20:26 | |
*** Marga_ has joined #openstack-nova | 20:29 | |
*** tsekiyam_ has joined #openstack-nova | 20:30 | |
openstackgerrit | Jay Pipes proposed openstack/nova: Fix scheduler issue when multiple-create failed https://review.openstack.org/147048 | 20:32 |
*** tsekiyam_ has quit IRC | 20:32 | |
*** tsekiyam_ has joined #openstack-nova | 20:32 | |
*** tsekiyama has quit IRC | 20:33 | |
*** mtanino has quit IRC | 20:34 | |
*** iamjarvo has quit IRC | 20:37 | |
*** jichen has joined #openstack-nova | 20:38 | |
*** arnaud____ has joined #openstack-nova | 20:38 | |
*** MVenesio has joined #openstack-nova | 20:38 | |
*** lpetrut has quit IRC | 20:39 | |
jogo | mikal: any reason you didn't _W this? https://review.openstack.org/#/c/173995/ | 20:39 |
jogo | +W | 20:39 |
jogo | dansmith: ^^ any reason not too? | 20:40 |
*** baoli has joined #openstack-nova | 20:40 | |
dansmith | figured he was waiting for jenkins | 20:40 |
mikal | jogo: just cause it was too new IIRC | 20:40 |
dansmith | too new? | 20:40 |
jogo | mikal: it was a trivial rebase | 20:40 |
dansmith | almost a month old :) | 20:40 |
*** IanGovett1 has joined #openstack-nova | 20:40 | |
jogo | mikal: thanks, just checking | 20:40 |
*** tsekiyam_ has quit IRC | 20:41 | |
mikal | jogo: yeah, looking at it again I think you're right | 20:41 |
jogo | dansmith: as a role of thumb if the patch isn't old enough to crawl I don't think we should merge it :) | 20:41 |
mikal | I see you fixed it though | 20:41 |
*** tsekiyama has joined #openstack-nova | 20:41 | |
jogo | mikal: yup | 20:41 |
dansmith | jogo: heh | 20:41 |
mikal | You need to remember I am very jet lagged, and not very competent to start with | 20:42 |
dansmith | jogo: I have no ideas when babies start crawling, so I'd need some instruction | 20:42 |
dansmith | mikal: yeah, double whammy, eh? | 20:42 |
*** IanGovett has quit IRC | 20:42 | |
*** ildikov has quit IRC | 20:42 | |
jogo | dansmith: google tells me 7-10 months | 20:42 |
*** otter768 has joined #openstack-nova | 20:42 | |
dansmith | huh | 20:42 |
dansmith | I was thinking five years old | 20:43 |
*** arnaud____ has quit IRC | 20:43 | |
jogo | mikal: what country are you in? | 20:43 |
dansmith | but pretty sure they start talking at 2 months | 20:43 |
jogo | haha | 20:43 |
*** iamjarvo has joined #openstack-nova | 20:43 | |
*** baoli has quit IRC | 20:44 | |
*** openstackgerrit_ has joined #openstack-nova | 20:44 | |
*** tonytan4ever has quit IRC | 20:44 | |
*** jichen has quit IRC | 20:44 | |
*** baoli has joined #openstack-nova | 20:44 | |
mriedem | dansmith: so you're mega in love with rebuild/evacuate right? | 20:46 |
dansmith | totes | 20:46 |
mriedem | so in the compute api before the cast to rebuild, we set the task_state to REBUILDING: http://git.openstack.org/cgit/openstack/nova/tree/nova/compute/api.py#n2370 | 20:46 |
mriedem | oh nvm, just answered the question i was going to ask | 20:46 |
*** tsekiyama has quit IRC | 20:47 | |
*** otter768 has quit IRC | 20:47 | |
mriedem | compute manager has the @reverts_task_state decorator | 20:47 |
mriedem | was looking at this: http://git.openstack.org/cgit/openstack/nova/tree/nova/compute/manager.py#n2968 | 20:47 |
mriedem | hmm, that's evacuate anyway, not rebuild | 20:48 |
*** achanda has joined #openstack-nova | 20:48 | |
*** tonytan4ever has joined #openstack-nova | 20:49 | |
*** Sukhdev has joined #openstack-nova | 20:52 | |
*** thangp has quit IRC | 20:52 | |
*** claudiub has joined #openstack-nova | 20:53 | |
*** cbader has joined #openstack-nova | 20:55 | |
*** claudiub|2 has quit IRC | 20:55 | |
*** VW_ has quit IRC | 20:56 | |
*** vilobhmm1 has quit IRC | 20:57 | |
*** vilobhmm1 has joined #openstack-nova | 20:58 | |
*** ildikov has joined #openstack-nova | 20:58 | |
*** Marga_ has quit IRC | 20:58 | |
*** Marga_ has joined #openstack-nova | 20:59 | |
*** oro has quit IRC | 20:59 | |
*** htruta has quit IRC | 21:00 | |
*** oro_ has quit IRC | 21:00 | |
*** setuid has quit IRC | 21:00 | |
*** ociuhandu has quit IRC | 21:00 | |
*** htruta has joined #openstack-nova | 21:01 | |
*** haomai___ has joined #openstack-nova | 21:02 | |
openstackgerrit | Andrew Laski proposed openstack/nova-specs: Add BuildRequest object https://review.openstack.org/169901 | 21:02 |
*** mwagner_lap has joined #openstack-nova | 21:02 | |
*** VW_ has joined #openstack-nova | 21:03 | |
*** eharney has joined #openstack-nova | 21:03 | |
*** setuid has joined #openstack-nova | 21:03 | |
*** sdake has joined #openstack-nova | 21:04 | |
*** haomaiwa_ has quit IRC | 21:05 | |
dansmith | jogo: that force-delete patch, | 21:06 |
dansmith | jogo: previously it allowed multiple deletes as long as you said "I really mean it for sure" in the request | 21:06 |
dansmith | jogo: I was -1 on that because it just means everyone always force-deletes in a tight loop until their thing goes away, which is what task-state is supposed to prevent | 21:07 |
jogo | right | 21:07 |
jogo | dansmith: AFAIK this goes back to a patch by alaski (?) to no-op extra deletes if already in a delete state | 21:07 |
jogo | the question is do we still want that logic or not | 21:08 |
alaski | I vaguely recall reviewing something like that | 21:08 |
alaski | I didn't write something like that | 21:08 |
*** sdake_ has quit IRC | 21:08 | |
jogo | I think it was rax but may of been someone else | 21:08 |
dansmith | I would like us to fubar our states less so that we don't need the client to keep deleting things that are in deleted state | 21:08 |
dansmith | er, deleting | 21:08 |
bauzas | mmm, is there any recent merged change that then provides a conflict for most of the changes ? | 21:09 |
bauzas | or is it Gerrit 2.10 having a problem ? | 21:09 |
jogo | bauzas: AFAIK it may be gerrit | 21:09 |
*** iamjarvo has quit IRC | 21:09 | |
*** Marga_ has quit IRC | 21:10 | |
*** wwriverrat has joined #openstack-nova | 21:10 | |
bauzas | jogo: ok good to know because my dash is just red for most of the reviews I have to do | 21:10 |
*** moshele has quit IRC | 21:10 | |
jogo | alaski: wow I was wayyy off I2f97f93bd714e0ea3b6d4fa3ac457ab43eed00e1 | 21:10 |
*** Marga_ has joined #openstack-nova | 21:10 | |
mriedem | jogo: it was HP | 21:10 |
*** mtanino has joined #openstack-nova | 21:10 | |
mriedem | jogo: i konw it well and can find it | 21:10 |
*** annegent_ has joined #openstack-nova | 21:10 | |
* jogo hides | 21:10 | |
mriedem | you better | 21:10 |
*** tonytan4ever has quit IRC | 21:11 | |
mriedem | the commit even said, 'in a later patch we'll fix this other thing that this patch breaks' | 21:11 |
jogo | mriedem: https://review.openstack.org/#/c/55444/ | 21:11 |
mriedem | that's the one | 21:11 |
mriedem | "Dealing with delete requests that never got executed is not in scope of this change and will be submitted separately." | 21:11 |
mriedem | that never happened and left a bug in the code | 21:11 |
mriedem | jogo: which you fixed later with a @reverts_task_state decorator | 21:11 |
jogo | Dealing with delete requests that never got executed is not in scope of | 21:12 |
mriedem | you couldn't delete an instance in error state | 21:12 |
jogo | this change and will be submitted separately. | 21:12 |
jogo | I remember that one | 21:12 |
jogo | that was silly | 21:12 |
mriedem | the reverts_task_state thing was regressed with a later change, which we fixed in kilo | 21:12 |
mriedem | fixed in kilo and backported to icehouse | 21:12 |
openstackgerrit | Ankit Agrawal proposed openstack/nova: Detach volume after deleting instance with no host https://review.openstack.org/145738 | 21:12 |
mriedem | i still have nightmares about this as you can tell | 21:12 |
*** apuimedo has joined #openstack-nova | 21:13 | |
*** eglynn has quit IRC | 21:13 | |
mriedem | btw, some kids never crawl, they just go straight to walking | 21:14 |
mriedem | i call them freaks! | 21:14 |
jogo | mriedem: based on the info in https://bugs.launchpad.net/nova/+bug/1248563 can we revert the first patch? | 21:14 |
openstack | Launchpad bug 1248563 in OpenStack Compute (nova) "Instance deletion is prevented when another component locks up" [High,Fix released] - Assigned to Stanislaw Pitucha (stanislaw-pitucha) | 21:14 |
jogo | and allow multiple deletes? | 21:15 |
*** annegentle has quit IRC | 21:15 | |
*** baoli has quit IRC | 21:15 | |
mriedem | jogo: i think we're way beyond reverting https://review.openstack.org/#/c/55444/ at this point no? | 21:15 |
jogo | mriedem: based on https://review.openstack.org/70187 landing | 21:15 |
*** baoli has joined #openstack-nova | 21:15 | |
mriedem | i vaguely remember something related to https://review.openstack.org/#/c/70187/ breaking hyper-v CI | 21:16 |
jogo | mriedem: well do_terminate_instance uses the instance.uuid lock now | 21:17 |
mriedem | basically, that was all a big mess for a long time | 21:17 |
jogo | so good guess | 21:17 |
*** Marga_ has quit IRC | 21:17 | |
mriedem | i don't really want to revert those old changes b/c of how messy they were to begin with | 21:18 |
mriedem | do we have a new critical bug around this? | 21:18 |
*** Marga_ has joined #openstack-nova | 21:18 | |
mriedem | there was another attempt at force-delete here which broke the cells api so we reverted that | 21:18 |
jogo | reading http://lists.openstack.org/pipermail/openstack-dev/2013-October/017454.html this was ugly the old way too | 21:18 |
mriedem | jogo: we're here now https://review.openstack.org/#/c/163798/ | 21:19 |
jogo | mriedem: just wondering since every few months there seems to be a patch related to multiple deletes | 21:19 |
jogo | mriedem: right which I just nacked | 21:19 |
*** tonytan4ever has joined #openstack-nova | 21:19 | |
mriedem | oh | 21:19 |
openstackgerrit | Taylor Peoples proposed openstack/nova: libvirt: safe_decode xml for i18n logging https://review.openstack.org/181708 | 21:19 |
mriedem | hard for me to see in the 'new' gerrit that i'm new to | 21:19 |
mriedem | there is just so much gd informatoin in this style | 21:20 |
jogo | yeah it takes some time to adjust to | 21:20 |
jogo | mriedem: I think what this is really showing is we don't test nova enough around failure conditions in general | 21:21 |
bauzas | mriedem: what's funny for people already using the new Gerrit UI is that the merge conflicts are at the same location than previously the series | 21:22 |
bauzas | so I'm always wondering why patch #1 refers to patch #2 - but that just means both are conflicting :/ | 21:22 |
* jogo slowly walks away from the mess | 21:23 | |
mriedem | jogo: well, we test error conditions and we have negative api testing, | 21:23 |
mriedem | jogo: but yeah, we don't really test 'recovery' scenarios | 21:23 |
jogo | right | 21:24 |
jogo | and things like assume virt drivers other services begin failing | 21:24 |
jogo | chaos monkey stuff | 21:24 |
*** tpeoples has joined #openstack-nova | 21:24 | |
mriedem | like libvirt? | 21:25 |
mriedem | e.g. https://review.openstack.org/#/c/181781/ | 21:26 |
openstackgerrit | Matt Riedemann proposed openstack/nova: libvirt: handle code=38 + sigkill (ebusy) in destroy() https://review.openstack.org/181781 | 21:26 |
jogo | haha | 21:27 |
*** matrohon has quit IRC | 21:30 | |
mriedem | you know what i hate? api_samples tests | 21:31 |
openstackgerrit | Felipe Reyes proposed openstack/nova: Detach iSCSI latest path for latest disk https://review.openstack.org/135382 | 21:31 |
cfriesen | does openstack support using qemu's builtin iscsi support rather than creating a block device on the host? if not, is anyone looking at adding it? | 21:33 |
*** emagana has quit IRC | 21:34 | |
*** emagana has joined #openstack-nova | 21:35 | |
*** dsanders has quit IRC | 21:35 | |
*** dsanders has joined #openstack-nova | 21:35 | |
openstackgerrit | Jay Pipes proposed openstack/nova: Remove db layer hard-code permission checks for service_get_by_compute_host https://review.openstack.org/160089 | 21:36 |
dansmith | cfriesen: isn't that likely to suck more? | 21:38 |
*** VW_ has quit IRC | 21:40 | |
*** MVenesio has quit IRC | 21:41 | |
openstackgerrit | Matt Riedemann proposed openstack/nova: WIP: Show 'reserved' status in os-fixed-ips https://review.openstack.org/168966 | 21:42 |
mriedem | alex_xu: i've added the start of some functional api samples tests there for v2.4 ^ | 21:42 |
*** VW_ has joined #openstack-nova | 21:42 | |
openstackgerrit | Jay Pipes proposed openstack/nova: Remove db layer hard-code permission checks for service_get_by_host* https://review.openstack.org/150349 | 21:43 |
mriedem | they don't work, i keep getting "TypeError: expected string or buffer" so i'm sure my samples are messed up | 21:43 |
*** achanda has quit IRC | 21:43 | |
*** baoli has quit IRC | 21:44 | |
*** jdurgin1 has joined #openstack-nova | 21:44 | |
tpeoples | If anyone has time to review https://review.openstack.org/#/c/181708/ (small 1 line libvirt change) that'd be great | 21:45 |
openstackgerrit | Jay Pipes proposed openstack/nova: API: remove instance_get_all_by_host(_and_node) hard-code admin check from db https://review.openstack.org/177673 | 21:45 |
*** jdurgin has quit IRC | 21:46 | |
*** jdurgin1 has quit IRC | 21:46 | |
*** jdurgin has joined #openstack-nova | 21:46 | |
*** VW_ has quit IRC | 21:47 | |
mriedem | alex_xu: nevermind, think i just figured it out | 21:47 |
openstackgerrit | Matt Riedemann proposed openstack/nova: Show 'reserved' status in os-fixed-ips https://review.openstack.org/168966 | 21:47 |
tpeoples | thanks mriedem | 21:47 |
harlowja | vilobhmm1 jaypipes i put some potential work items @ https://blueprints.launchpad.net/nova/+spec/servicegroup-api-control-plane (if u want to modify) | 21:48 |
harlowja | feel free to change it :-P | 21:48 |
jaypipes | tpeoples: approved. thank you! | 21:48 |
jaypipes | harlowja: thx mate | 21:49 |
harlowja | sure | 21:49 |
cfriesen | dansmith: I came across this: https://lists.gnu.org/archive/html/qemu-devel/2011-04/msg01977.html the data is a few years old, but it shows qemu+libiscsi being way faster than cache=none unless you also set aio=native | 21:50 |
*** sdake has quit IRC | 21:50 | |
cfriesen | dansmith: and that was for reading, not even writing | 21:50 |
dansmith | cfriesen: I don't mean suck more == slower | 21:51 |
dansmith | cfriesen: I just mean general correctness | 21:51 |
*** baoli has joined #openstack-nova | 21:52 | |
dansmith | but given how much the linux iscsi initiator stuff is used, it seems odd to me that it would be slower than the qemu one, unless it's doing safer operations, you know | 21:52 |
cfriesen | dansmith: should be better. I just noticed an issue with udev automounting cinder volumes on the compute host. (which I think could potentially mean replaying the journal on both host and guest) | 21:52 |
*** VW_ has joined #openstack-nova | 21:52 | |
*** annegent_ has quit IRC | 21:52 | |
*** sdake has joined #openstack-nova | 21:52 | |
*** annegentle has joined #openstack-nova | 21:53 | |
mriedem | dims: re: your earlier comment about hotels, that was because you were with ibm at the time and had the marriott discount thingy | 21:55 |
mriedem | dims: that's not going to happen :) | 21:55 |
*** mriedem is now known as mriedem_away | 21:56 | |
*** nickchase has joined #openstack-nova | 21:56 | |
bauzas | ah, good point, so I guess the dates for the midcycle are set now ? | 21:57 |
*** artom_ has quit IRC | 21:57 | |
dansmith | cfriesen: did you read the full thread? | 21:58 |
tpeoples | jaypipes: is there anything else i need to do for that? will the nova team decide to backport or not and then cherry pick it? | 21:58 |
tpeoples | (if they decide to backport) | 21:58 |
dansmith | cfriesen: either way, I assume you could implement this alongside what we have | 21:58 |
dansmith | I'd have to look, but it could be interesting to provide this as an option if there really are benefits | 21:58 |
cfriesen | dansmith: yeah, I didn't see Christoph ever addressing the fact that the testcase was read-only while the cache would only affect writing | 21:59 |
cfriesen | dansmith: I think it'd be interesting to redo the test, at any rate. And yeah, I don't see any reason why it couldn't coexist. | 21:59 |
dansmith | I don't really understand why this is likely to be any better than the default implementation, | 22:00 |
dansmith | so any crazy speedup leads me to expect that there is some mismatch between configs that isn't apparent (i.e. caching issues) | 22:00 |
cfriesen | dansmith: the main difference seemed to be the aio options...when using aio=threads it's slower | 22:00 |
*** emagana has quit IRC | 22:00 | |
*** annegentle has quit IRC | 22:00 | |
dansmith | cfriesen: I mean I don't see why it'd ever be faster | 22:01 |
dansmith | cfriesen: if you do all the iscsi logic in the qemu process you get charged for it, and if you do it in the host kernel you don't | 22:02 |
*** Marga_ has quit IRC | 22:04 | |
*** Marga_ has joined #openstack-nova | 22:04 | |
*** burt has quit IRC | 22:04 | |
cfriesen | dansmith: from http://www-01.ibm.com/support/knowledgecenter/linuxonibm/liaat/liaatbpkvmasynchio.htm: "BM® lab tests show that using AIO support can improve guest I/O performance, especially when there are multiple threads performing I/O operations at the same time. You can enable AIO by specifying aio=native in the qemu-kvm command for a KVM guest." | 22:04 |
*** baoli has quit IRC | 22:05 | |
*** emagana has joined #openstack-nova | 22:05 | |
cfriesen | dansmith: I'm no expert, just been looking around for info | 22:05 |
dansmith | that means aio even for host-based devices, yes/ | 22:05 |
dansmith | I'm no expert either, I just can't think of any *fundamental* reason why this should be faster to do it in qemu | 22:07 |
cfriesen | dansmith: but there are other issues reported for aio=native with cache=none and sparse files, so I wasn't sure if it was okay to suggest switching to aio=native generally | 22:07 |
*** aysyd has quit IRC | 22:08 | |
dansmith | this is probably stuff danpb has context on, I'd guess | 22:08 |
cfriesen | seems likely | 22:08 |
*** ericksonsantos has joined #openstack-nova | 22:09 | |
*** subscope has quit IRC | 22:09 | |
*** moshele has joined #openstack-nova | 22:13 | |
*** subscope has joined #openstack-nova | 22:18 | |
*** claudiub has quit IRC | 22:20 | |
*** jgrimm is now known as zz_jgrimm | 22:24 | |
*** rwsu has quit IRC | 22:25 | |
*** pixelbeat has quit IRC | 22:26 | |
*** tonytan4ever has quit IRC | 22:26 | |
*** arnaud____ has joined #openstack-nova | 22:27 | |
openstackgerrit | Zhenguo Niu proposed openstack/nova: Remove unused decorator on attach/detach volume https://review.openstack.org/165839 | 22:28 |
*** lucas-dinner has quit IRC | 22:28 | |
*** tpeoples has quit IRC | 22:28 | |
*** Nic1 has joined #openstack-nova | 22:29 | |
*** Nic has quit IRC | 22:29 | |
*** jecarey_ has quit IRC | 22:31 | |
*** erickson has joined #openstack-nova | 22:31 | |
*** arnaud____ has quit IRC | 22:32 | |
*** ociuhandu has joined #openstack-nova | 22:32 | |
*** ericksonsantos has quit IRC | 22:34 | |
*** otter768 has joined #openstack-nova | 22:43 | |
*** tpeoples has joined #openstack-nova | 22:44 | |
*** achanda has joined #openstack-nova | 22:46 | |
dansmith | zzzeek: ping | 22:46 |
zzzeek | dansmith: hi | 22:46 |
*** nelsnelson has quit IRC | 22:47 | |
dansmith | zzzeek: just want to make sure you're understanding one aspect of how nova is (usually) deployed, re: this database parallelism thing | 22:47 |
zzzeek | dansmith: ok | 22:47 |
*** otter768 has quit IRC | 22:47 | |
dansmith | zzzeek: when nova-api talks to the db, it completely blocks, but we have lots of nova-api processes (real processes), so it's not *too* bad | 22:48 |
zzzeek | dansmith: sure | 22:48 |
dansmith | zzzeek: nova-compute does not talk to the database, but rather talks to nova-conductor, | 22:48 |
dansmith | which also runs many (many) *processes* each of which can fall on its sword for the blocking call | 22:48 |
zzzeek | dansmith: that is all fine, however, eventlet is still pointles within the database process | 22:48 |
dansmith | yes | 22:49 |
zzzeek | dansmith: the discussions are always about, threads are slow! the GIL ! terrible! | 22:49 |
zzzeek | dansmith: which is idiotic to be honest. we have zero concurrency within a process | 22:49 |
dansmith | your last message made it seem like maybe you thought we were even more serial than we are | 22:49 |
dansmith | yep, okay, just making usre | 22:49 |
zzzeek | dansmith: well fear not, robert was talking about GIL | 22:49 |
dansmith | I know | 22:49 |
zzzeek | dansmith: all discussions about GIL are silly. we are in the MySQLdb-Interpreter-Lock right now | 22:49 |
*** diegows has joined #openstack-nova | 22:50 | |
zzzeek | dansmith: i dont know if you recall, someone wanted to replace all concurrency with asyncio, because, *faster* | 22:50 |
dansmith | they are right now, but if we were able to get to a (real) threadpool for those blocking requests, then we're contending for the GIL within a process | 22:50 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: configuration for interface driver options https://review.openstack.org/128826 | 22:50 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: enable virtio-net multiqueue https://review.openstack.org/128829 | 22:50 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: update NetworkRequest object to support queues https://review.openstack.org/182108 | 22:50 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova: tests: _stub_allocate_for_instance to accept port ids list https://review.openstack.org/182109 | 22:50 |
openstackgerrit | Vladik Romanovsky proposed openstack/nova: libvirt: update VIF metadata with queues in the neutron api https://review.openstack.org/182110 | 22:50 |
zzzeek | if my “we’re slow” arguemnts are tempered by “we have lots of processes”, so are all the “we can be *fast!” arguments about asyncio, if they were even true | 22:50 |
zzzeek | dansmith: the contending for the GIL within a process when most of the work is DB work is not bad at all, and I spent a lot of time doing benchmarks to show this | 22:51 |
zzzeek | dansmith: and it is much much less than contending for individual database conversations as we are right now | 22:51 |
*** Sukhdev has quit IRC | 22:52 | |
dansmith | right, I think that's what he was trying to get at, that we're good at IO-bound work, and can be threaded and not worry about th GIL if that's the case | 22:52 |
dansmith | yes | 22:52 |
zzzeek | dansmith: OK, I didnt exactly read it that way but I will try to see that | 22:52 |
dansmith | I don't actually know how much CPU work we're doing processing the result of the database call | 22:52 |
dansmith | so I don't know if we *are* cpu bound or not | 22:53 |
zzzeek | dansmith: we’re not | 22:53 |
dansmith | good :) | 22:53 |
dansmith | your last point is something I'm pretty concerned about, | 22:54 |
dansmith | although maybe in a different way than you meant | 22:54 |
dansmith | I think the number of race conditions we manage to hit just in things like nova-compute because we're almost completely single threaded, | 22:55 |
dansmith | has me concerned that moving straight to real threads for some of our stuff will expose a bunch of things that only work now because we didn't need to lock across a region | 22:55 |
dansmith | which isn't a reason not to do it, but it is something that I worry about | 22:55 |
zzzeek | dansmith: well as far as that i was alittle more handwavy, but i see the most weird bugs in things once an app goes from talking to 2 or 3 db connections at once to 30 or 40 | 22:55 |
dansmith | yep, but I think that even not related to the db specifically, such a move would likely cause us trouble | 22:56 |
*** BadCub is now known as BadCub-Hiding | 22:56 | |
dansmith | more so in nova-compute than other things, and we could easily leave nova-compute async as long as conductor performs better | 22:56 |
zzzeek | dansmith: what, just using threads? i woudl disagree with that, as long as we aren’t using shared mutable datastructures in process | 22:57 |
zzzeek | dansmith: i think all the “Web request http” stuff should remain as async as it wants to be | 22:57 |
zzzeek | dansmith: when you deal with HTTP connections you have to be more ready for arbitrary behavior | 22:57 |
jaypipes | tpeoples: I tagged the bug with both juno and kilo backport potential... that should be enough :) | 22:58 |
dansmith | zzzeek: nova-compute is *full* of mutable data structures :) | 22:58 |
zzzeek | dansmith: that are shared between requests | 22:58 |
zzzeek | dansmith: i haven’t seen that | 22:58 |
*** tsekiyama has joined #openstack-nova | 22:59 | |
dansmith | zzzeek: yeah, tons of it.. hypervisor state, resource tracker stuff, all that kinda stuff is shared common bits | 22:59 |
dansmith | zzzeek: so if we move nova-compute to real threads, we have a ton of locking to add | 22:59 |
zzzeek | dansmith: I’m only looking specifically at: db/sqlalchemy/api.py . that is the part that I’d put behind the thread pool | 22:59 |
zzzeek | dansmith: basically your relational database service layer | 23:00 |
tpeoples | jaypipes: yeah, but do i need to propose the patch to the other branches, is that done automatically, etc.? (sorry i'm fairly new to openstack) | 23:00 |
dansmith | zzzeek: yeah, that's what I meant by keeping it collected there and leaving nova-compute async | 23:00 |
zzzeek | dansmith: Ok then we agree :) | 23:00 |
dansmith | zzzeek: although conductor could be fully threaded if it helped, and that would insulate us | 23:00 |
jaypipes | tpeoples: ah, yes, you would need to do a cherry-pick to those branches. one sec, lemme grab an explanatory wiki page. | 23:00 |
dansmith | zzzeek: but I think doing it fully at the sqalchemy/api layer is fine if it gets us everything we need | 23:01 |
dansmith | zzzeek: so you know what has been preventing us from doing that in the past right? I'm foggy on the actual details | 23:01 |
jaypipes | tpeoples: https://wiki.openstack.org/wiki/StableBranch | 23:01 |
jaypipes | tpeoples: that has everything you should need :) | 23:01 |
zzzeek | dansmith: well thats as far as I’ve been able to understand the situation, I’m not opposed to using threads more extensively but I try to always limit the territory im staking out in order to maximize the chance of success :) | 23:01 |
tpeoples | thanks jaypipes | 23:02 |
zzzeek | dansmith: what prevents us is, everytime it is brought up, the handwringing begins about the GIL | 23:02 |
dansmith | zzzeek: and less code that needs to be threadsafe, yes? that seems like a good idea :) | 23:02 |
dansmith | since we clearly suck at it | 23:02 |
dansmith | zzzeek: no | 23:02 |
dansmith | zzzeek: we had a way to use a threadpool for mysql stuff, but under heavy load it would stall or something | 23:02 |
dansmith | zzzeek: I think it was related to eventlet unthreadsafeness | 23:03 |
zzzeek | dansmith: that’s a different kind of threadpool | 23:03 |
zzzeek | dansmith: that’s the one dug into eventlet itself | 23:03 |
dansmith | ah, okay | 23:03 |
zzzeek | dansmith: which yes, im sure is buggy as all get out | 23:03 |
zzzeek | dansmith: it is too fine grained | 23:03 |
dansmith | okay | 23:03 |
*** emagana has quit IRC | 23:03 | |
zzzeek | dansmith: taht’s totally where you’d see weird deadlocks happen . b.c. your DB calls are totally concurrent but your eventlet calls are forced to wait, i rememebr seeing this early on | 23:04 |
dansmith | zzzeek: okay, makes sense | 23:04 |
zzzeek | dansmith: they don’t mix well at that level | 23:04 |
dansmith | zzzeek: they didn't, yeah :) | 23:04 |
dansmith | zzzeek: so, "zzeek's awesome thread pool" + which driver? c or pymysql? | 23:04 |
dansmith | seems like pymysql in the last email, but in previous ones, it seemed like you favored the c ones | 23:05 |
zzzeek | dansmith: well we should probably use the new mysqliclient thing that is a fork of python-mysql | 23:05 |
dansmith | that's the forked C one? | 23:05 |
zzzeek | dansmith: yes. | 23:05 |
dansmith | okay | 23:05 |
zzzeek | dansmith: but in truth it hardly matters pymysql is very good also. mysql-python etc. will blow it away in a straight drag race style test but in real world you aren’t going to notice much | 23:06 |
dansmith | TBH, I thought the only way we were going to get the parallelism was with the pure python one | 23:06 |
dansmith | so if not, I'm happy and don't really care after that :) | 23:06 |
dansmith | parallel requests, I mean | 23:07 |
zzzeek | dansmith: i re-read my response and I know why it tweaked you, becuse im going off on “openstack is so slow!” so yes, i know things work out OK in any case | 23:08 |
dansmith | well, | 23:08 |
dansmith | it didn't tweak me, I just wanted to make sure you weren't doing testing with one conductor process and getting the wrong idea :) | 23:08 |
dansmith | anyway, thanks for the 'splainin | 23:09 |
*** subscope has quit IRC | 23:10 | |
zzzeek | dansmith: I’m super glad it helps :) | 23:10 |
*** dboik_ has quit IRC | 23:12 | |
melwitt | dansmith: I'm looking at the last of the db.instance_update calls to get rid of update_cells=True logic and I was wondering, what is the object equivalent to doing an update with a join with system_metadata? I was thinking like creating an Instance(**updates) and then saving it, but how to give sys meta as an expected_attr? https://github.com/openstack/nova/blob/master/nova/conductor/manager.py#L127-L129 | 23:13 |
dansmith | melwitt: well, save() will save the metadata if it's set (and dirty) | 23:15 |
dansmith | melwitt: create() should do it if it's set | 23:15 |
dansmith | is it not? | 23:15 |
melwitt | I could pull the instance with expected_attrs first, put the updates to it, and then save it, but I thought that's adding a db lookup | 23:15 |
melwitt | dansmith: sorry, I didn't mean literally Instance.create because it already exists. making an Instance object. I think maybe I'm understanding now, I wouldn't need to do anything special with system_metadata like the current code is, it's only doing it for _from_db_object? | 23:18 |
*** hemna_ is now known as hemnafk | 23:19 | |
dansmith | ah, yes | 23:19 |
*** Longgeek_ has quit IRC | 23:20 | |
openstackgerrit | Joe Cropper proposed openstack/nova-specs: Proposal for Flexible Resize Scheduling https://review.openstack.org/175115 | 23:21 |
*** VW_ has quit IRC | 23:22 | |
melwitt | dansmith: then I see Instance.save takes care of sending a notification update, in conductor it does notifications.send_update(context, old_ref, inst_obj, service) whereas Instance.save does notifications.send_update(context, old_ref, new_ref). if conductor needs to send "service" as part of the update, does that mean I can't replace this with Instance.save? | 23:26 |
*** mdbooth has quit IRC | 23:26 | |
dansmith | um | 23:27 |
dansmith | I don't recognize that service bit | 23:27 |
*** arif-ali has quit IRC | 23:28 | |
dansmith | at.all. | 23:29 |
*** arif-ali has joined #openstack-nova | 23:29 | |
*** vladikr has quit IRC | 23:30 | |
melwitt | looking... it defaults service="compute" but it looks like conductor takes a service parameter in its instance_update so it can notify with that there. I don't know much about the notifications | 23:30 |
*** tjones1 has joined #openstack-nova | 23:30 | |
dims | mriedem_away: i was told that visitors to the facility could ask for it when they check in | 23:30 |
dansmith | oh | 23:31 |
dansmith | melwitt: yeah, just found that | 23:31 |
dansmith | melwitt: that's weird | 23:31 |
dansmith | I must've written that | 23:31 |
melwitt | because it's weird? :P | 23:32 |
dansmith | oh snap! | 23:32 |
dansmith | no, I just think I did all this stuff | 23:32 |
*** mdbooth has joined #openstack-nova | 23:32 | |
dansmith | so it defaults to compute on the send_update call too | 23:32 |
melwitt | haha, okay. the statements right after each other made it sound funny | 23:32 |
dansmith | heh | 23:33 |
dansmith | so, | 23:33 |
dansmith | I think you can go ahead and do this | 23:33 |
dansmith | the only other place I see it used as other than compute is in scheduler_utils | 23:33 |
dansmith | which should be okay I think | 23:33 |
dansmith | I assume this is used to show in the notification what entity caused the notification | 23:34 |
dansmith | but I assume that it's almost always "compute" regardless of where it's sent, | 23:34 |
dansmith | since I only see it passed explicitly in scheduler | 23:34 |
*** dims has quit IRC | 23:34 | |
*** moshele has quit IRC | 23:36 | |
melwitt | okay, yeah it looks like each service has its own notifier object | 23:36 |
*** marun has quit IRC | 23:37 | |
*** dims has joined #openstack-nova | 23:39 | |
melwitt | dansmith: and now I find it's conductor manager that's sending it as 'compute_task' for all those cases | 23:40 |
dansmith | is it? | 23:40 |
dansmith | I don't see that | 23:41 |
melwitt | dansmith: this? https://github.com/openstack/nova/blob/master/nova/conductor/manager.py#L604-L608 | 23:41 |
melwitt | it looks like it's the only thing calling the scheduler utils thing | 23:42 |
*** dims has quit IRC | 23:43 | |
dansmith | oh, in task manager yeah | 23:43 |
dansmith | so, anything that uses instance.save() will send "as compute" I guess, | 23:44 |
dansmith | but that's probably already the case | 23:44 |
dansmith | I don't think you're making it worse, | 23:44 |
dansmith | and if that's a thing we need to do properly then we need some context somewhere to make that correct | 23:44 |
*** tjones1 has quit IRC | 23:45 | |
melwitt | got it, yeah | 23:45 |
*** Raildo_ has joined #openstack-nova | 23:49 | |
*** IanGovett1 has quit IRC | 23:49 | |
*** gmann has joined #openstack-nova | 23:50 | |
*** oomichi has joined #openstack-nova | 23:52 | |
-openstackstatus- NOTICE: Gerrit is going offline while we perform an emergency downgrade to version 2.8. | 23:54 | |
*** ChanServ changes topic to "Gerrit is going offline while we perform an emergency downgrade to version 2.8." | 23:54 | |
*** pixelbeat has joined #openstack-nova | 23:55 | |
*** dboik has joined #openstack-nova | 23:55 | |
*** ZZelle has quit IRC | 23:55 | |
*** dboik_ has joined #openstack-nova | 23:56 | |
*** sdake_ has joined #openstack-nova | 23:57 | |
openstackgerrit | Joe Cropper proposed openstack/nova-specs: Proposal for Flexible Resize Scheduling https://review.openstack.org/175115 | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!