Wednesday, 2016-03-16

*** ndipanov has quit IRC00:00
*** mriedem has joined #openstack-nova00:01
*** haomaiwa_ has quit IRC00:01
*** haomaiwang has joined #openstack-nova00:01
*** vilobhmm11 has quit IRC00:02
*** Swami has quit IRC00:03
*** vilobhmm11 has joined #openstack-nova00:03
*** vilobhmm11 has quit IRC00:05
*** takedakn has joined #openstack-nova00:06
*** vilobhmm11 has joined #openstack-nova00:07
*** dims has joined #openstack-nova00:08
*** mylu has quit IRC00:08
*** mylu has joined #openstack-nova00:10
*** ianw has quit IRC00:10
*** Jeffrey4l has joined #openstack-nova00:12
*** armax has joined #openstack-nova00:14
*** armax has quit IRC00:14
*** mrkz has quit IRC00:16
*** efried has quit IRC00:16
*** mrkz has joined #openstack-nova00:16
*** yuntongjin has joined #openstack-nova00:17
*** efried has joined #openstack-nova00:18
*** sdague has joined #openstack-nova00:22
*** openstack has joined #openstack-nova00:24
*** nic has quit IRC00:25
*** sdague has quit IRC00:26
*** baoli has joined #openstack-nova00:27
*** mylu has quit IRC00:27
*** mylu has joined #openstack-nova00:27
*** mtanino has quit IRC00:28
*** ssurana1 has quit IRC00:28
*** takedakn has quit IRC00:29
*** fawadkhaliq has quit IRC00:30
*** yuntongjin1 has joined #openstack-nova00:32
*** fawadkhaliq has joined #openstack-nova00:33
*** yuntongjin has quit IRC00:34
*** mylu has quit IRC00:35
*** erlon has quit IRC00:37
*** eharney has quit IRC00:37
*** mylu has joined #openstack-nova00:39
*** yuntongjin has joined #openstack-nova00:40
*** mylu has quit IRC00:40
*** yuntongjin1 has quit IRC00:42
*** baoli has quit IRC00:42
*** yuntongjin1 has joined #openstack-nova00:43
*** angdraug has quit IRC00:43
*** yuntongjin has quit IRC00:45
*** mylu has joined #openstack-nova00:45
*** dims has joined #openstack-nova00:45
*** mylu has quit IRC00:46
openstackgerritMark Doffman proposed openstack/nova-specs: Add new probabilistic placement algorithm.
*** mylu has joined #openstack-nova00:48
*** zz_dimtruck is now known as dimtruck00:49
*** mylu has quit IRC00:50
*** oomichi has quit IRC00:50
*** oomichi has joined #openstack-nova00:51
*** hoangcx has joined #openstack-nova00:52
*** phuongnh has joined #openstack-nova00:52
*** afazekas has quit IRC00:53
*** Jeffrey4l has quit IRC00:53
*** oomichi has quit IRC00:55
*** dave-mccowan has joined #openstack-nova00:57
*** Kennan has quit IRC00:58
*** fawadkhaliq has quit IRC00:58
*** afazekas has joined #openstack-nova00:59
*** oomichi has joined #openstack-nova01:00
*** haomaiwang has quit IRC01:01
*** haomaiwang has joined #openstack-nova01:01
*** fawadkhaliq has joined #openstack-nova01:02
*** fawadkhaliq has quit IRC01:02
*** ZZelle has quit IRC01:03
*** david-lyle has joined #openstack-nova01:04
*** fawadkhaliq has joined #openstack-nova01:06
*** fawadkhaliq has quit IRC01:06
*** dims has quit IRC01:07
*** Kennan has joined #openstack-nova01:10
*** dims has joined #openstack-nova01:12
*** mylu has joined #openstack-nova01:14
*** mylu has quit IRC01:15
*** fawadkhaliq has joined #openstack-nova01:18
*** fawadkhaliq has quit IRC01:19
*** EinstCrazy has joined #openstack-nova01:21
*** haomaiwang has quit IRC01:23
*** ccarmack has joined #openstack-nova01:23
*** tiendc has joined #openstack-nova01:23
*** boris-42 has quit IRC01:24
*** ljxiash has joined #openstack-nova01:25
*** oomichi has quit IRC01:25
*** afazekas has quit IRC01:25
*** ccarmack has quit IRC01:27
*** oomichi has joined #openstack-nova01:28
*** afazekas has joined #openstack-nova01:28
*** dims has quit IRC01:29
*** dims has joined #openstack-nova01:29
*** fawadkhaliq has joined #openstack-nova01:30
*** Daisy has joined #openstack-nova01:31
*** fawadkhaliq has quit IRC01:31
*** MVenesio has joined #openstack-nova01:32
*** akshai has quit IRC01:33
openstackgerritBin Zhou proposed openstack/nova: upgrade connection_info when Ceph mon IP changed
*** ljxiash has quit IRC01:35
*** ljxiash has joined #openstack-nova01:36
*** MVenesio has quit IRC01:36
*** vilobhmm11 has quit IRC01:36
*** mrkz has quit IRC01:36
*** Daisy has quit IRC01:37
*** vishwanathj has joined #openstack-nova01:37
*** Daisy has joined #openstack-nova01:37
*** ljxiash has quit IRC01:40
*** fawadkhaliq has joined #openstack-nova01:42
*** Daisy has quit IRC01:42
*** mylu has joined #openstack-nova01:42
*** mriedem has quit IRC01:42
*** mylu has quit IRC01:44
*** kaisers1 has joined #openstack-nova01:44
*** kaisers has quit IRC01:45
*** ljjjustin has joined #openstack-nova01:50
*** fawadkhaliq has quit IRC01:51
*** fawadkhaliq has joined #openstack-nova01:51
*** dims has quit IRC01:52
*** yamahata has joined #openstack-nova01:53
*** mgoddard has quit IRC01:54
*** baoli has joined #openstack-nova01:55
*** dims has joined #openstack-nova01:56
openstackgerritFeng Xi Yan proposed openstack/nova: <VMWare> Allocate free bus for new SCSI controller
*** haomaiwang has joined #openstack-nova02:00
*** haomaiwang has quit IRC02:01
*** haomaiwa_ has joined #openstack-nova02:01
*** dims has quit IRC02:03
*** apoorvad has quit IRC02:03
*** dims has joined #openstack-nova02:05
*** akshai has joined #openstack-nova02:05
*** fawadkhaliq has quit IRC02:06
*** woodster_ has quit IRC02:07
*** akshai_ has joined #openstack-nova02:07
*** mylu has joined #openstack-nova02:08
*** mylu has quit IRC02:09
openstackgerritAugustina Ragwitz proposed openstack/nova: Lowercase aggregate metadata keys due to case insensitivity
openstackgerritAugustina Ragwitz proposed openstack/nova: Lowercase aggregate metadata keys due to case insensitivity
openstackgerritAugustina Ragwitz proposed openstack/nova: add regression test for bug #1538011
openstackbug 1538011 in OpenStack Compute (nova) "Metadata keys are not case sensitive" [High,In progress] - Assigned to Augustina Ragwitz (auggy)02:10
*** akshai has quit IRC02:10
*** baoli_ has joined #openstack-nova02:11
*** ljxiash has joined #openstack-nova02:12
*** baoli has quit IRC02:14
*** liusheng has quit IRC02:16
*** liusheng has joined #openstack-nova02:16
*** vilobhmm11 has joined #openstack-nova02:18
*** thorst_ has quit IRC02:18
*** mylu has joined #openstack-nova02:20
*** vilobhmm111 has joined #openstack-nova02:20
*** yamahata has quit IRC02:21
*** mylu has quit IRC02:21
*** vilobhmm11 has quit IRC02:22
*** baoli_ has quit IRC02:23
*** baoli has joined #openstack-nova02:23
*** mylu has joined #openstack-nova02:25
*** fawadkhaliq has joined #openstack-nova02:25
*** mylu has quit IRC02:27
*** mylu has joined #openstack-nova02:27
*** Daisy has joined #openstack-nova02:28
*** mylu has quit IRC02:29
*** Daisy_ has joined #openstack-nova02:29
*** baoli has quit IRC02:29
*** yamahata has joined #openstack-nova02:29
*** Daisy__ has joined #openstack-nova02:31
*** fawadkhaliq has quit IRC02:32
*** Daisy has quit IRC02:33
*** Daisy has joined #openstack-nova02:33
*** ducttape_ has joined #openstack-nova02:33
*** mylu has joined #openstack-nova02:34
*** Daisy_ has quit IRC02:34
*** baoli has joined #openstack-nova02:35
*** mylu has quit IRC02:35
*** mylu has joined #openstack-nova02:35
*** houming has joined #openstack-nova02:36
*** Daisy__ has quit IRC02:37
*** unicell has quit IRC02:37
*** Daisy has quit IRC02:37
*** Daisy has joined #openstack-nova02:39
*** yamahata has quit IRC02:39
*** mylu has quit IRC02:40
*** READ10 has quit IRC02:47
*** mylu has joined #openstack-nova02:49
*** tbachman_ has joined #openstack-nova02:49
*** tbachman has quit IRC02:51
*** tbachman_ is now known as tbachman02:51
*** gcb has quit IRC02:51
*** Jeffrey4l has joined #openstack-nova02:57
*** krtaylor has quit IRC02:58
*** haukebruno has joined #openstack-nova02:58
*** mylu has quit IRC02:59
*** haomaiwa_ has quit IRC03:01
*** haomaiwang has joined #openstack-nova03:01
*** tbachman_ has joined #openstack-nova03:03
*** mylu has joined #openstack-nova03:03
*** gcb has joined #openstack-nova03:04
*** ctrath has joined #openstack-nova03:04
*** tbachman has quit IRC03:04
*** tbachman_ is now known as tbachman03:04
*** ssurana has joined #openstack-nova03:06
*** krtaylor has joined #openstack-nova03:10
*** gcb has quit IRC03:11
*** mylu has quit IRC03:13
*** ducttape_ has quit IRC03:14
*** houming has quit IRC03:14
*** houming_ has joined #openstack-nova03:14
*** mylu has joined #openstack-nova03:16
*** dave-mccowan has quit IRC03:17
*** mylu has quit IRC03:18
*** thorst has joined #openstack-nova03:19
*** Daisy has quit IRC03:19
*** ctrath has quit IRC03:22
*** dimtruck is now known as zz_dimtruck03:25
*** ducttape_ has joined #openstack-nova03:26
*** gcb has joined #openstack-nova03:26
*** thorst has quit IRC03:27
*** mylu has joined #openstack-nova03:27
*** yuntongjin1 has quit IRC03:28
*** mylu has quit IRC03:29
*** _ducttape_ has joined #openstack-nova03:29
*** ducttape_ has quit IRC03:29
*** MVenesio has joined #openstack-nova03:32
*** gongysh has joined #openstack-nova03:34
*** dims has quit IRC03:37
*** MVenesio has quit IRC03:37
*** ljxiash has quit IRC03:38
*** baoli has quit IRC03:39
*** sridharg has joined #openstack-nova03:41
*** aswadr has joined #openstack-nova03:47
*** aswadr_ has joined #openstack-nova03:47
*** akshai_ has quit IRC03:52
*** Daisy has joined #openstack-nova03:54
*** haomaiwang has quit IRC04:01
*** haomaiwang has joined #openstack-nova04:01
*** gongysh has quit IRC04:04
*** vishwanathj has quit IRC04:05
*** knikolla has quit IRC04:06
*** josecastroleon1 has joined #openstack-nova04:06
*** josecastroleon has quit IRC04:06
*** whenry has quit IRC04:06
*** links has joined #openstack-nova04:10
*** vilobhmm111 has quit IRC04:11
*** vilobhmm11 has joined #openstack-nova04:11
*** fawadkhaliq has joined #openstack-nova04:14
*** flwang1 has quit IRC04:16
*** Daisy has quit IRC04:20
*** vilobhmm11 has quit IRC04:21
*** _ducttape_ has quit IRC04:24
*** Daisy has joined #openstack-nova04:24
*** thorst has joined #openstack-nova04:25
*** Daisy has quit IRC04:29
*** ducttape_ has joined #openstack-nova04:30
*** thorst has quit IRC04:32
*** akshai has joined #openstack-nova04:36
*** vilobhmm11 has joined #openstack-nova04:42
*** ducttape_ has quit IRC04:44
*** mylu has joined #openstack-nova04:44
*** mylu has quit IRC04:45
*** gcb has quit IRC04:46
openstackgerritFeng Xi Yan proposed openstack/nova: <VMWare> Allocate free bus for new SCSI controller
*** haomaiwang has quit IRC05:01
*** haomaiwang has joined #openstack-nova05:01
*** akshai_ has joined #openstack-nova05:04
*** gcb has joined #openstack-nova05:05
*** mylu has joined #openstack-nova05:06
*** akshai has quit IRC05:07
*** yamahata has joined #openstack-nova05:07
*** ssurana has quit IRC05:08
*** ljxiash has joined #openstack-nova05:09
*** ljxiash has quit IRC05:10
*** ljxiash has joined #openstack-nova05:10
*** clduser_ has quit IRC05:12
*** clduser has joined #openstack-nova05:12
*** ssurana has joined #openstack-nova05:13
*** boris-42 has joined #openstack-nova05:14
*** Daisy has joined #openstack-nova05:15
*** _xin1 has quit IRC05:22
*** _xin has joined #openstack-nova05:22
*** ccarmack has joined #openstack-nova05:24
*** _xin has quit IRC05:25
*** alex_xu has quit IRC05:25
openstackgerritAbhishek Talwar proposed openstack/python-novaclient: aggregate-details changed to aggregate-show
*** sc68cal has joined #openstack-nova05:26
*** _xin has joined #openstack-nova05:28
*** unicell has joined #openstack-nova05:28
openstackgerritAbhishek Talwar proposed openstack/python-novaclient: aggregate-details changed to aggregate-show
*** ccarmack has quit IRC05:29
*** thorst has joined #openstack-nova05:30
*** vishwanathj has joined #openstack-nova05:31
*** alex_xu has joined #openstack-nova05:31
*** vishwanathj has quit IRC05:31
*** Jeffrey4l has quit IRC05:32
*** MVenesio has joined #openstack-nova05:33
*** thorst has quit IRC05:37
*** MVenesio has quit IRC05:37
openstackgerritAbhishek Talwar proposed openstack/python-novaclient: flavor-id should be only integer or UUID
*** fawadkhaliq has quit IRC05:42
*** fawadkhaliq has joined #openstack-nova05:43
*** akshai_ has quit IRC05:44
*** david-lyle has quit IRC05:45
*** david-lyle has joined #openstack-nova05:45
*** mylu has quit IRC05:46
*** deepakcs has joined #openstack-nova05:48
*** Daisy has quit IRC05:48
*** Daisy has joined #openstack-nova05:48
*** vishwanathj has joined #openstack-nova05:48
*** vishwanathj has quit IRC05:51
*** mylu has joined #openstack-nova05:52
*** fawadkhaliq has quit IRC05:52
*** fawadkhaliq has joined #openstack-nova05:53
*** moshele has joined #openstack-nova05:56
*** mylu has quit IRC05:57
*** liverpooler has quit IRC05:59
*** haomaiwang has quit IRC06:01
*** flwang1 has joined #openstack-nova06:01
*** haomaiwang has joined #openstack-nova06:01
openstackgerritTomoki Sekiyama proposed openstack/nova: Fix multipath iSCSI encrypted volume attach failure
*** flwang1 has quit IRC06:09
*** Daisy_ has joined #openstack-nova06:10
*** Daisy has quit IRC06:13
*** aswadr has quit IRC06:18
*** gongysh has joined #openstack-nova06:19
*** astupnikov has joined #openstack-nova06:20
openstackgerritOpenStack Proposal Bot proposed openstack/nova: Imported Translations from Zanata
*** Daisy_ has quit IRC06:26
*** Daisy has joined #openstack-nova06:27
*** ekuris_ has joined #openstack-nova06:29
*** Daisy has quit IRC06:31
*** moshele has quit IRC06:32
*** vishwanathj has joined #openstack-nova06:34
openstackgerritWenzhi Yu proposed openstack/nova-specs: Move calculation of 'disk_available_least' to ResourceTracker
*** vishwanathj is now known as vishwanathj_zzz06:35
*** thorst has joined #openstack-nova06:35
*** sc68cal has quit IRC06:36
*** esracelik has joined #openstack-nova06:40
*** harlowja_at_home has quit IRC06:41
*** thorst has quit IRC06:42
*** ssurana has quit IRC06:44
*** ishant has joined #openstack-nova06:50
*** Daisy has joined #openstack-nova06:55
*** ljjjustin has quit IRC06:58
*** ljjjustin has joined #openstack-nova06:58
*** hoangcx has quit IRC06:59
*** esracelik has quit IRC07:00
*** haomaiwang has quit IRC07:01
*** haomaiwang has joined #openstack-nova07:01
openstackgerritWenzhi Yu proposed openstack/nova-specs: Move calculation of 'disk_available_least' to ResourceTracker
*** fawadkhaliq has quit IRC07:03
*** ihrachys has joined #openstack-nova07:03
*** hoangcx has joined #openstack-nova07:04
*** fawadkhaliq has joined #openstack-nova07:04
*** ihrachys has quit IRC07:06
*** mjura has joined #openstack-nova07:07
*** zenoway has joined #openstack-nova07:10
*** zenoway has quit IRC07:12
*** zenoway has joined #openstack-nova07:12
*** flwang1 has joined #openstack-nova07:13
*** lpetrut has joined #openstack-nova07:15
*** tesseract has joined #openstack-nova07:16
*** tesseract is now known as Guest6496707:17
*** nkrinner has joined #openstack-nova07:17
*** jaosorior has joined #openstack-nova07:18
*** Guest64967 has quit IRC07:21
openstackgerritChangBo Guo(gcb) proposed openstack/nova: Enhance value check for option notify_on_state_change
*** Marga_ has quit IRC07:25
*** fawadkhaliq has quit IRC07:25
*** fawadkhaliq has joined #openstack-nova07:26
*** fawadkhaliq has quit IRC07:26
*** fawadkhaliq has joined #openstack-nova07:27
*** browne has quit IRC07:27
*** ljxiash has quit IRC07:30
*** flwang1 has quit IRC07:30
*** ljxiash has joined #openstack-nova07:31
*** ljxiash has quit IRC07:32
*** MVenesio has joined #openstack-nova07:34
*** ljxiash has joined #openstack-nova07:36
*** tesseract- has joined #openstack-nova07:36
*** gcb has quit IRC07:36
*** moshele has joined #openstack-nova07:36
*** Marga_ has joined #openstack-nova07:38
*** MVenesio has quit IRC07:38
*** thorst has joined #openstack-nova07:40
*** Marga_ has quit IRC07:41
*** Marga_ has joined #openstack-nova07:41
*** Marga_ has quit IRC07:42
*** Marga_ has joined #openstack-nova07:43
*** mariusv_ has quit IRC07:44
*** mariusv has quit IRC07:44
*** mariusv has joined #openstack-nova07:45
*** lunarlamp has joined #openstack-nova07:48
*** thorst has quit IRC07:48
*** gcb has joined #openstack-nova07:49
*** liverpooler has joined #openstack-nova07:49
*** klkumar has joined #openstack-nova07:54
*** scheuran has joined #openstack-nova07:56
*** jwcroppe has quit IRC07:57
*** belmoreira has joined #openstack-nova07:58
*** jwcroppe has joined #openstack-nova07:58
*** haomaiwang has quit IRC08:01
*** haomaiwang has joined #openstack-nova08:01
*** yuywz has joined #openstack-nova08:02
*** jwcroppe has quit IRC08:02
*** erno has joined #openstack-nova08:03
*** lpetrut has quit IRC08:04
*** ihrachys has joined #openstack-nova08:04
*** y_sawai has joined #openstack-nova08:10
*** matrohon has joined #openstack-nova08:11
*** liverpooler has quit IRC08:14
*** hoangcx has quit IRC08:14
*** liverpooler has joined #openstack-nova08:14
*** esracelik has joined #openstack-nova08:17
*** jichen has joined #openstack-nova08:18
*** MCoLo has joined #openstack-nova08:19
*** hoangcx has joined #openstack-nova08:19
*** permalac has joined #openstack-nova08:20
*** kbyrne has joined #openstack-nova08:22
*** EinstCrazy has quit IRC08:23
*** cdent has joined #openstack-nova08:26
*** atuvenie has joined #openstack-nova08:26
*** roeyc has joined #openstack-nova08:27
*** EinstCrazy has joined #openstack-nova08:30
openstackgerritxhzhf proposed openstack/nova: Support-matrix of vmware for chap is wrong
*** yamahata has quit IRC08:33
*** roeyc has quit IRC08:36
*** ig0r_ has quit IRC08:39
*** slaweq_ has joined #openstack-nova08:43
slaweq_I'm looking for API calls rate limit in nova08:43
slaweq_I found that there is such middleware (legacy) in v2 api08:43
slaweq_but there is no such think in current v2.1 api08:43
slaweq_can You maybe tell me why (or give me some links where it is explained)08:44
slaweq_is this middleware deprecated?08:44
slaweq_or maybe I should use something else in v21 api?08:44
*** andymaier has joined #openstack-nova08:45
*** thorst has joined #openstack-nova08:46
doffmDoesn't rate limiting sit in-front of the API in a proxy now?08:48
doffmI actually don't know. slaweq_ You may be better off asking in #openstack.08:48
*** Drankis has joined #openstack-nova08:49
cdentslaweq_: I think doffm is right. The rate limiting is handled at a layer above the api code, established by the paste config08:52
*** thorst has quit IRC08:52
slaweq_ok, thx08:55
*** paul-carlton2 has joined #openstack-nova08:56
slaweq_cdent: but about what paste config You are talking?08:57
slaweq_for v2 it was configured in api-paste.ini config file08:57
slaweq_but for v21 it's not there08:57
slaweq_and also in docs for liberty:
*** aswadr has joined #openstack-nova08:58
slaweq_there is info that "Rate limiting is removed from v2.1 api."08:58
cdentslaweq_: yeah, appears you are right, I was confusing the sizelimit and ratelimit middlewares08:58
slaweq_but I still don't know why... :/08:58
*** ljxiash has quit IRC08:58
*** vilobhmm11 has quit IRC08:59
cdentsorry, I don't know the history08:59
slaweq_cdent: no prob :)09:00
cdenthere we go slaweq_: there's discussion of why in this review:
*** haomaiwang has quit IRC09:01
*** haomaiwang has joined #openstack-nova09:01
cdentbasic summary: not fit for purpose09:01
slaweq_I will read it09:01
*** dedery has joined #openstack-nova09:02
bauzasgood morning Nova09:04
*** ljxiash has joined #openstack-nova09:05
*** ljjjustin has quit IRC09:07
*** jlanoux has joined #openstack-nova09:07
*** paul-carlton2 has quit IRC09:08
*** ljjjustin has joined #openstack-nova09:08
openstackgerritMoshe Levi proposed openstack/nova: libvirt: Fix ssh driver to to prevent prompting
*** wanghao has quit IRC09:09
*** wanghao has joined #openstack-nova09:09
johnthetubaguyslaweq_: yeah, cdent is correct, that rate limiting stuff only works if you have one API node, which makes it mostly pointless, most folks use some generic REST API tool to ratelimit, AFAIK.09:10
*** ljxiash has quit IRC09:10
johnthetubaguybauzas: hows the reno stuff we want in going?09:10
*** salv-orlando has joined #openstack-nova09:11
*** sridharg has quit IRC09:11
*** aswadr has left #openstack-nova09:12
slaweq_johnthetubaguy: thx09:12
*** mgoddard has joined #openstack-nova09:12
slaweq_I think that now I understand it better after reading comments on this review :)09:12
doffmbauzas: Morning.09:13
johnthetubaguyslaweq_: do let us know if there are admin docs that could be clearer, might be worth a doc bug detailing the confusion09:14
*** yuywz has quit IRC09:16
bauzasjohnthetubaguy: so we have a reno problem that's not blocking09:16
bauzasjohnthetubaguy: basically, reverting changes doesn't remove the reno note09:16
bauzasjohnthetubaguy: so I discussed with dhellmann and he has a patch that I need to test09:16
bauzasin the meanwhile, I'm verifying that most of the notes that we did yesterday and accepted are mertged09:17
bauzasjohnthetubaguy: I haven't looked at the translations, are you planning a RC2 at least for those ?09:17
johnthetubaguybauzas: ah, I did wonder about that09:17
*** ZZelle has joined #openstack-nova09:17
bauzasthat's my main point before tagging RC109:18
bauzasthat and the reno files09:18
johnthetubaguybauzas: we *have* to do RC2 for translations anyways, as RC1 is the hard string freeze09:18
johnthetubaguybauzas: I should ask that the bot is doing09:18
bauzasdo we have a current zanata change to approve ?09:19
bauzasjohnthetubaguy: so, about the notes, all of them are merged but the prelude section09:19
bauzasjohnthetubaguy: mriedem was having a concern about
bauzasjohnthetubaguy: do you remember when we dropped EC2 ?09:20
bauzasjohnthetubaguy: my concern is that is confusing because 2 notes are saying different things09:21
*** ndipanov has joined #openstack-nova09:22
johnthetubaguybauzas: can we put a note in the bullet to ignore the previous note?09:23
*** ferhat has joined #openstack-nova09:23
ferhathello all09:24
bauzasjohnthetubaguy: actually, I'm about to write a last reno change for fixing that, plus a formatting nit I saw09:24
*** jistr has joined #openstack-nova09:24
bauzasjohnthetubaguy: so, the thing is, we should probably also merge the prelude section before RC109:24
bauzasjohnthetubaguy: because if we merge that once the branch is splitted, it would be a newton thing, right?09:25
*** Daisy has quit IRC09:25
johnthetubaguybauzas: well, we can backport, but I was assuming that would be in RC1, yes09:26
*** Daisy has joined #openstack-nova09:26
johnthetubaguyactually, you can't really backport it, I guess09:26
bauzasjohnthetubaguy: we can surely backport, sure, but the main commit will be newton09:26
johnthetubaguyyeah, which is not what we want09:26
johnthetubaguyhmm, I am starting to dislike bits of this reno stuff09:27
bauzasokay, give me 10 mins, sorting out the relnotes09:27
johnthetubaguyits close, though, better than just me making up some stuff on the last day of the release09:27
bauzasI agree09:28
openstackgerritZhenyu Zheng proposed openstack/python-novaclient: Add changes-since support when list servers
*** Daisy has quit IRC09:30
openstackgerritZhenyu Zheng proposed openstack/nova: Add checks for driver attach_interfaces capability
*** BobBall has joined #openstack-nova09:31
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing sr_uuid_map
BobBallWill check out that change in a bit johnthetubaguy - thanks.09:32
bauzasjohnthetubaguy: you agree with not being correct now, right?09:32
*** dedery_ has joined #openstack-nova09:32
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing sr_uuid_map
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up issues with live-migrate for xenapi
*** derekh has joined #openstack-nova09:34
*** ljxiash has joined #openstack-nova09:35
*** MVenesio has joined #openstack-nova09:35
*** rk4n has joined #openstack-nova09:35
johnthetubaguybauzas: yeah, we removed it now right?09:36
*** yassine__ has joined #openstack-nova09:36
bauzasjohnthetubaguy: yup09:36
bauzasmaybe sdague could help us in the next hour09:36
*** dedery has quit IRC09:37
johnthetubaguybauzas: yeah, it will not work at all this realse09:37
johnthetubaguyso that means folks need to update their paste.ini, I think09:37
johnthetubaguyalthough I think we made that limp on, for now, if I remember correctly09:37
*** mjura has quit IRC09:37
bauzasokay, I'm just nearly done with updating the change for fixing the formatting nits + the removal problem we have09:37
openstackgerritChangBo Guo(gcb) proposed openstack/nova: Make config option daemon's help string more accurate
*** lucas-dinner is now known as lucasagomes09:38
*** MVenesio has quit IRC09:39
*** rk4n has quit IRC09:39
*** rk4n has joined #openstack-nova09:39
*** ljxiash has quit IRC09:40
*** sdague has joined #openstack-nova09:43
*** thorst has joined #openstack-nova09:50
*** dims has joined #openstack-nova09:51
openstackgerritSylvain Bauza proposed openstack/nova: Add a prelude section for Mitaka relnotes
openstackgerritSylvain Bauza proposed openstack/nova: Fix reno for RC1
*** openstackgerrit has quit IRC09:53
*** openstackgerrit_ is now known as openstackgerrit09:53
bauzasjohnthetubaguy: last reno changes I feel ^09:53
*** openstackgerrit has quit IRC09:53
*** openstackgerrit_ has joined #openstack-nova09:54
*** openstackgerrit_ is now known as openstackgerrit09:54
*** openstackgerrit_ has joined #openstack-nova09:55
*** openstackgerrit has quit IRC09:55
*** openstackgerrit has joined #openstack-nova09:56
openstackgerritgaryk proposed openstack/nova: VMware: make the opaque network attachment more robust
*** thorst has quit IRC09:57
openstackgerritZhenyu Zheng proposed openstack/python-novaclient: Add changes-since support when list servers
*** sdague has quit IRC10:00
*** haomaiwang has quit IRC10:01
*** allen_gao has quit IRC10:01
*** haomaiwang has joined #openstack-nova10:01
*** sdague has joined #openstack-nova10:03
*** sridharg has joined #openstack-nova10:03
johnthetubaguybauzas: not sure about the commit message for this change: seems a bit confusing, maybe just go from three paragraphs that describe each of the changes?10:04
johnthetubaguybauzas: the reno bits make sense I think10:04
bauzasjohnthetubaguy: sure, lemme do that10:05
*** sridharg has quit IRC10:05
*** sridharg has joined #openstack-nova10:05
*** mylu has joined #openstack-nova10:08
*** lpetrut has joined #openstack-nova10:09
*** allen_gao has joined #openstack-nova10:10
*** sdague has quit IRC10:11
*** sdague has joined #openstack-nova10:12
*** mylu has quit IRC10:13
markus_zbauzas: Do you have a minute? I don't understand what's happening with It's not listed in any of the zuul queues. Got it "stuck"?10:13
*** hoangcx has quit IRC10:13
bauzasmarkus_z: sec, fixing a last minute problem with reno and I'm done10:13
BobBalljohnthetubaguy: Holding nodes in our RAX CI is harder than I remembered; full logs uploaded to and
johnthetubaguyBobBall: thats showing a syntax error still, I think10:16
johnthetubaguyBobBall: I don't think thats present in the latest version of the patch series10:16
BobBallWhat's worse is that the swift upload is broken10:16
BobBallYes; that was the logs from last night not including the two changes you made this morning.10:17
johnthetubaguyrhr QOULS SO IR10:17
johnthetubaguymarkus_z: we had issues yesterday where due to a gerrit restart you need to recheck changes10:17
johnthetubaguymarkus_z: unsure if thats related10:17
*** ljjjustin has left #openstack-nova10:18
markus_zjohnthetubaguy: ah, could be, the +W from matt was yesterday10:18
*** cdelatte has quit IRC10:19
markus_zjohnthetubaguy: So, I just leave a "recheck" as comment and then zuul will pick it up again?10:19
markus_zdidn't we have a "reverify" once?10:19
sdaguemarkus_z: I poked it with another +A10:19
sdaguerecheck doesn't often work in these cases10:20
sdaguebecause you need the state transition bit10:20
markus_zsdague: Ah, good to know, thanks for poking it.10:20
sdaguemarkus_z: there was about a 12 hour window where a bad regex was in the zuul definition that blocked things from running10:20
openstackgerritSylvain Bauza proposed openstack/nova: Add a prelude section for Mitaka relnotes
openstackgerritSylvain Bauza proposed openstack/nova: Fix reno for RC1
bauzasjohnthetubaguy: sdague^10:21
bauzasjohnthetubaguy: I also added 2 points in the bus change10:21
*** PaulMurray has joined #openstack-nova10:22
markus_zsdague: OK, I see the spike in the zuul job queue in Grafana. I didn't check that before I asked.10:22
sdaguemarkus_z: no worries10:22
bauzasmarkus_z: I guess your point was fixed by sdague, right?10:24
markus_zbauzas: yep, right10:24
johnthetubaguyBobBall: maybe the bigger question is does the latest patchset fix the issue?10:25
*** zenoway has quit IRC10:26
*** claudiub has joined #openstack-nova10:27
BobBalljohnthetubaguy: Indeed; will be checking that just now.10:27
sdaguemarkus_z: I just queried for a bunch of things that were stuck on the nova side and flushed them10:27
sdaguelet me know if you see others10:27
markus_zsdague: thanks, will do10:27
*** daemontool has joined #openstack-nova10:28
sdaguejohnthetubaguy: can we not do things like this please -
sdagueespecially as I think there was a different set of patches earlier in the cycle going the other way10:33
johnthetubaguysdague: sorry, yeah, I thought we went the other way10:33
*** EinstCrazy has quit IRC10:34
johnthetubaguysdague: I was tempted to -2 with saying we don't have a pep8 thing to check this, so lets not bother with these patches, I guess that was the rule at one point10:34
openstackgerritAlexis Lee proposed openstack/nova: Wait for device to be mapped
*** dedery has joined #openstack-nova10:37
*** shaohe_feng has joined #openstack-nova10:37
*** dedery_ has quit IRC10:40
*** gcb has quit IRC10:40
*** dedery has quit IRC10:41
*** zenoway has joined #openstack-nova10:41
sdagueregardless of the pep8 rule, I get somewhat annoyed with "this makes things clearer" when it doesn't.10:41
*** zenoway has quit IRC10:42
*** zenoway has joined #openstack-nova10:42
*** mkrcmari__ has quit IRC10:42
*** phuongnh has quit IRC10:42
*** shaohe_feng has quit IRC10:44
*** shaohe_feng has joined #openstack-nova10:44
*** shaohe_feng_robo has joined #openstack-nova10:45
*** ferhat has quit IRC10:46
openstackgerritAbhishek Talwar proposed openstack/python-novaclient: flavor-id should be only integer or UUID
*** ljxiash has joined #openstack-nova10:51
*** Wenzhi has joined #openstack-nova10:52
*** gcb has joined #openstack-nova10:52
*** boden has joined #openstack-nova10:52
*** dims has quit IRC10:53
*** shaohe_feng has quit IRC10:54
*** shaohe_feng has joined #openstack-nova10:54
*** thorst has joined #openstack-nova10:55
*** ociuhandu has quit IRC10:55
*** ljxiash has quit IRC10:56
*** salv-orlando has quit IRC10:56
*** tbachman has quit IRC10:57
*** sridharg has quit IRC10:58
*** pece has joined #openstack-nova10:59
*** flwang1 has joined #openstack-nova10:59
*** dims has joined #openstack-nova10:59
bauzassdague: in case you could verify that is right10:59
bauzassdague: IIUC, we removed the ec2 commands11:00
bauzassdague: so, now, ec2 is like out of tree11:00
*** haomaiwang has quit IRC11:01
sdaguebauzas: hmmm.... wait so the release notes all build up and we don't have a defined line there?11:01
*** gongysh has quit IRC11:01
bauzassdague: the problem is that removing a reno file doesn't remove it from the relnotes11:01
*** haomaiwang has joined #openstack-nova11:01
*** tbachman has joined #openstack-nova11:01
bauzassdague: since this yaml was merged, it was confusing with another bullet saying that EC2 API is out-of-tree11:02
sdagueis that a thing that's being worked on?11:02
sdagueI guess, my assumption in how this would work is that this would exit, but clearly be referenced to an older branch11:02
*** thorst has quit IRC11:02
openstackgerritChris Dent proposed openstack/nova: Workaround stall in py34 unit tests
sdagueI guess if that's what we have to do, so be it, but this seems pretty confusing11:03
cdentsdague: that ^^ patch is related to the py34 slowdown we were discussing the other day11:03
bauzassdague: not sure I understand you, sorry11:04
cdentI traced it back to an eventlet lockup. There's a fix in that patch, but it is dirty.11:04
*** shaohe_feng has quit IRC11:04
bauzassdague: for Mitaka, we have 2 conflicting bullets about EC211:04
*** shaohe_feng has joined #openstack-nova11:05
sdaguebauzas: right, I get that our published release notes are wrong11:05
bauzas"As of Kilo the in tree EC2 API is marked as deprecated. The in tree EC2 API is now disabled by default. You have to add ec2 to the enabled_apis configuration parameter if you stil want to use it." vs. " All code and tests for Nova’s EC2 and ObjectStore API support which was deprecated in Kilo ( has been completely removed in Mitaka. This has been replaced by th11:05
sdagueand i get this is the way to hack around reno to make them right11:05
sdaguebut it seems like we shouldn't have to do that11:05
sdaguelong term11:05
bauzassdague: yeah, so I discussed that with dhellmann11:06
sdagueyou should never have to delete a release note artifact11:06
bauzassdague: and he provided me to verify11:06
*** Jeffrey4l has joined #openstack-nova11:06
sdagueso, sure, I still don't think we should have to delete it either11:06
bauzassdague: but I'd like to make sure that our relnotes are happy for mitaka and not waiting for reno to be updated11:06
sdaguebauzas: sure11:07
sdaguebut that fix isn't really the fix either, right?11:07
bauzasit's a workaround11:07
sdagueI seems completely wrong to have to go back through and delete release notes11:07
sdaguethe release notes should have some understanding of what release they apply to11:07
bauzassdague: have you seen my commit msg ?11:08
bauzassdague: I agree with you, the relnotes are provided for the milestones11:08
sdaguebauzas: I think we are talking past each other11:09
bauzassdague: I'm really sorry, it's probably me that not understanding your point :(11:09
bauzasI'm just testing locally the reno change dhellmann wrote for making sure that reverts are not leaving the notes there11:10
sdagueok, lets take a new example -
sdaguein Newton what are we going to do with that11:10
bauzassdague: that YAML file won't be rendered in the newton output11:11
moshelendipanov: hi11:11
sdaguebauzas: what is the mechanism for that?11:11
bauzassdague: it's scanning git commits for the release11:11
*** mkrcmari__ has joined #openstack-nova11:11
bauzassdague: tbc, it's scanning git commits for the master branch that haven't been released11:12
bauzassdague: unless you specifically ask the reno scanner to only look at a specific branch11:12
sdagueoh, because it's doing deep git inspection11:12
sdaguethough, if we fix a typo in stable then we have to be divergent forever?11:12
bauzasthe case didn't happen yet, but that's a good question11:13
bauzassay that liberty relnotes are bad11:13
sdagueit seems like it's going to be a lot easier to understand going forward if this stops doing magical git things, and instead uses subdirs11:13
BobBalljohnthetubaguy: Still no :/ and - still getting the Cannot load 'sr_uuid_map' in the base class11:13
bauzasthen, for fixing that, we'd have to provide a specific commit against the stable branch, not something merged in master11:13
bauzassdague: that's indeed a good point11:14
sdaguebecause 3 releases in, figuring out why certain release notes are rendering and not with git magic is going to get insance11:14
*** haukebruno has quit IRC11:14
bauzasyeah I agree11:14
bauzasactually, reno supports that already11:14
bauzasI mean, subdirs11:14
sdagueoh, can we move to that model instead of a branch model?11:14
johnthetubaguyBobBall: OK, should have some real logs this time, unit tests are slowly getting there, some real issues to fix up in those11:14
*** shaohe_feng has quit IRC11:14
johnthetubaguysdague: hmm, that does sound a whole deal easier11:15
bauzassdague: not really, it's just that we can modify the relnotessubdir value in
*** klkumar has quit IRC11:15
*** Wenzhi has quit IRC11:15
bauzasoops, I meant the notesdir11:15
*** shaohe_feng has joined #openstack-nova11:15
bauzassdague: so that would still scan the full list of git commits, but it would have a list of yaml files per release11:16
BobBalljohnthetubaguy: What shall I investigate? (so we don't overlap)11:16
bauzassdague: I can actually try to do this, lemme check11:16
johnthetubaguyBobBall: is the date wrong on your box? the logs still seem old11:17
sdaguebauzas: anyway, I think that this is feedback we should give. I think that the current reno model with 3 or 4 active branches is going to become way too frustrating. And getting off git magic to explicit directories would be way better.11:17
bauzassdague: the only problem I see is backports of changes including a note11:17
sdaguebauzas: problem with what?11:18
BobBallNope: 2016-03-16 11:03:02.303 TRACE oslo_messaging.rpc.dispatcher NotImplementedError: Cannot load 'sr_uuid_map' in the base class11:18
*** openstackgerrit has quit IRC11:18
bauzassdague: ie. say that you have a bugfix change that includes a reno file11:18
johnthetubaguyBobBall: I guess maybe output the migrate_data to the log at the start of the live-migrate method, see what it has11:18
bauzassdague: that bugfix is merged in master11:18
johnthetubaguyBobBall: actually the start of each method, so we can track what is happening11:18
*** openstackgerrit has joined #openstack-nova11:18
ndipanovmoshele, hola11:18
johnthetubaguyBobBall: by the way, which live-migrate types are you testing here?11:18
sdagueok, so you have to retarget the reno11:18
sdagueto a different directory11:18
sdagueyeh, it's not super ideal, but the current magic is a bit weird11:19
bauzassdague: given that currently, the notes directory is the same for all branches, it's perfectly okay to backport (if wanted) that change to eg. liberty and not needing to modify the reno file path11:19
BobBalljohnthetubaguy: Just --block-migrate so far; then will remove that when that works.  I presume specifying it is still the easier one.11:19
sdaguethat being said, we're not supposed to be doing feature backports, so it should be the exception11:19
bauzassdague: that way, the git scan finds that note for both the unreleased (ie. master) file, and for the stable branch file11:19
bauzassdague: security fixes11:19
bauzassdague: not speaking of featurs11:20
sdagueyeh, sure11:20
johnthetubaguyBobBall: its the one I am worried about at least, I just don't see why the data is not getting out of pre-live-migrate, and the logs are playing hard to get on me11:20
BobBalljohnthetubaguy: I guess TTL on CDN is too high...11:20
moshelendipanov: did you see my comment in
sdaguewell, the current stuff seems pretty confusing, because it creates this "can't fix old release notes that aren't clear"11:20
bauzassdague: but I agree, we should prevent such changes that impact upgrade paths or features11:20
sdagueunless you zero them out11:20
sdaguelike you had to do11:21
bauzassdague: ideally, I'd have git rm that yaml, but I was hit by
BobBalljohnthetubaguy: RAX CDN only lets us have a minimum 15 minute TTL - so I'll need to upload to new directories each time.11:21
bauzasso I hope this is temporary11:21
openstackgerritSergey Belous proposed openstack/nova: do not merge, trying to run tempest with enabled os-vif
sdaguebauzas: ok, except that's confusing11:21
moshelendipanov: I failed to understand if something else got broken or the patch work by mistake11:21
sdaguebecause you now have a directory with release notes11:22
bauzasI agree, that's like tribal knowledge11:22
sdaguesome are missing11:22
bauzassdague: some are missing because of the bug11:22
sdaguebauzas: you'd still have to delete this one regardless right?11:22
bauzassdague: but you should have the same comments between the yaml files and the output11:22
sdaguebecause it got modified during this cycle for some reason11:22
bauzassdague: at that point of time, I agree11:22
sdagueI feel like if "deleting renos to be consistent" is a thing, we should git rm -f the whole structure at the beginning of a new branch11:23
bauzassdague: tbc, at a point in time, anyone can build relnotes that should surface the same comments than in the YAML files for that specific point in time11:23
bauzasanyway, I need to drop off for lunch, but interesting talk :)11:23
bauzaswe should loop back with dhellmann11:24
*** haomaiwang has quit IRC11:24
johnthetubaguyBobBall: dang it, I see the issue now...11:24
*** rtheis has joined #openstack-nova11:24
johnthetubaguyBobBall: _call_live_migrate_command gets called from two places, I only saw one of them, you are hitting errors in the other one11:24
BobBallAh ok - no new logs needed yet then? :)11:24
BobBallAhhh joy.11:25
*** shaohe_feng has quit IRC11:25
*** shaohe_feng has joined #openstack-nova11:25
*** Daisy has joined #openstack-nova11:26
johnthetubaguyBobBall: assert_can_migrate and migrate_send should talk the same params right?11:26
johnthetubaguyBobBall: I think we actually have some bugs hiding in here11:27
BobBallyes, confirmed that they do take the same params in XenAPI11:28
*** hieulq has joined #openstack-nova11:29
BobBalland I think they are?11:29
*** klkumar has joined #openstack-nova11:29
*** hieulq has quit IRC11:31
*** smatzek has joined #openstack-nova11:31
*** Daisy has quit IRC11:31
*** robby_ has joined #openstack-nova11:32
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing of sr_uuid_map
johnthetubaguyBobBall: can you review that, while I improve the tests a touch, does that make sense? basically sr_uuid_map is not defined when we call check, but it is for migrate_send11:32
BobBalljohnthetubaguy: That'll be fine.  We're asserting that Nova gets the right SRs defined in sr_uuid_map and that they exist on the destination.11:33
BobBallI pressume that migrate_check could reject it if we requst an SR not on the destination, but that would be a bug in the nova code if we did.11:34
johnthetubaguyBobBall: true11:34
*** shaohe_feng has quit IRC11:35
*** thorst has joined #openstack-nova11:35
*** MVenesio has joined #openstack-nova11:36
*** shaohe_feng has joined #openstack-nova11:36
*** y_sawai has quit IRC11:37
*** rfolco has joined #openstack-nova11:39
*** MVenesio has quit IRC11:41
*** ihrachys has quit IRC11:42
*** ihrachys has joined #openstack-nova11:42
*** ihrachys has quit IRC11:42
*** moshele has quit IRC11:43
*** cdent has quit IRC11:43
*** ishant has quit IRC11:43
*** erlon has joined #openstack-nova11:44
*** cdelatte has joined #openstack-nova11:44
*** shaohe_feng has quit IRC11:45
*** cdelatte has quit IRC11:46
*** shaohe_feng has joined #openstack-nova11:46
*** cdelatte has joined #openstack-nova11:48
*** moshele has joined #openstack-nova11:49
openstackgerritZhengguang Ou proposed openstack/nova: libvirt: fix detach volume race condition
*** allen_gao has quit IRC11:50
*** rk4n_ has joined #openstack-nova11:51
*** allen_gao has joined #openstack-nova11:51
*** dave-mccowan has joined #openstack-nova11:52
*** rk4n has quit IRC11:54
*** shaohe_feng has quit IRC11:55
*** robcresswell has joined #openstack-nova11:56
*** shaohe_feng has joined #openstack-nova11:56
*** mylu has joined #openstack-nova11:56
*** salv-orlando has joined #openstack-nova11:57
robcresswellWondering if anyone could point me in the right direction: Does specifying a port-id when launching an instance hit any particular policy calls other than compute:create? Wondering if attach_interface factors in.11:57
BobBalljohnthetubaguy: Now working with the --block-migrate flag.11:57
johnthetubaguyBobBall: awesome, how are the others looking, I guess they need the aggregate change11:58
johnthetubaguyBobBall: actually, did you try with auto?11:58
BobBallI'm testing all 3 changes combined, with the aggregate change.  Auto failed and uploading the logs now.11:58
*** tiendc has quit IRC12:00
BobBalljohnthetubaguy: - cannot load block_migration in the base class12:00
openstackgerritgaryk proposed openstack/nova: VMware: make the opaque network attachment more robust
*** raildo-afk is now known as raildo12:00
johnthetubaguyBobBall: darn, that figures I guess12:00
BobBalljohnthetubaguy: full logs at
BobBalljohnthetubaguy: It may be because of my aggregate change - I'm setting block_migration = True but the new code was setting destcheck_data.block_migration12:01
*** pece has quit IRC12:01
BobBalljohnthetubaguy: Let me rebase my aggregate change to the bottom of the pile so I can fix it there.12:01
*** dedery has joined #openstack-nova12:01
*** robby_ has quit IRC12:02
johnthetubaguyBobBall: can you put it on the top of the chain12:02
johnthetubaguyBobBall: so the final patch fixes the non_block_migration, that way12:02
BobBalljohnthetubaguy: we have 3 changes: aggregate, sr_map, block_migration.  Currently in that order.12:02
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing of sr_uuid_map
*** mylu has quit IRC12:03
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up issues with live-migrate for xenapi
*** dedery has quit IRC12:04
BobBalljohnthetubaguy: The aggregate change is needed first (really) but the block_migration change modifies my current aggregate fix12:04
*** smatzek has quit IRC12:04
johnthetubaguyBobBall: OK, I was thinking that should come last, let me take a quick peak at that on your branch12:05
*** shaohe_feng has quit IRC12:06
*** shaohe_feng has joined #openstack-nova12:06
*** baoli has joined #openstack-nova12:07
*** smatzek has joined #openstack-nova12:08
johnthetubaguyBobBall: oh, I see, thats a bit nasty...12:08
*** jaosorior has quit IRC12:08
BobBallAggregate fix needs to be the bottom (or second) on the pile, with a slightly modified block_migration fix at the top12:08
*** jaosorior has joined #openstack-nova12:09
*** pece has joined #openstack-nova12:09
*** jaosorior has quit IRC12:09
*** aysyd has joined #openstack-nova12:09
BobBallI'm working on UT for the aggregate fix ATM then will push12:09
*** cdent has joined #openstack-nova12:10
*** pece has quit IRC12:10
*** jaosorior has joined #openstack-nova12:11
johnthetubaguyBobBall: you are right, your fix isn't quite right given the changes at the top of the change, just trying something...12:12
*** pece has joined #openstack-nova12:13
BobBalljohnthetubaguy: Indeed.  Maybe actually we put block_migration change at the base, then yours, then aggregate fix?12:13
BobBallLM wouldn't work until it was all merged ofc12:13
openstackgerritJohn Garbutt proposed openstack/nova: Fix LM aggregate bug
johnthetubaguyso I tried it on the end12:13
johnthetubaguybut, yeah, it all fails till its all there I guess, hmm12:13
BobBallindeed - anything like that will work.  I was just trying to get LM with --block-migration first before tacking on the auto change; but in reality the order doesn't matter.12:14
markus_zmelwitt: Could you have a look at if you find time? I'm unsure about the strategy of printing output after executing a nova CLI command.12:15
openstackLaunchpad bug 1557888 in python-novaclient "Nova CLI - No message on nova-agent deletion" [Undecided,Incomplete] - Assigned to Preeti (pandey-preeti1)12:15
johnthetubaguyBobBall: so I made a one line change in your patch that might have fixed it:
*** shaohe_feng has quit IRC12:16
*** haomaiwang has joined #openstack-nova12:16
BobBalljohnthetubaguy: Yeah - I said that should fix it above :)12:16
BobBallI'll try that series.12:16
*** shaohe_feng has joined #openstack-nova12:17
johnthetubaguyBobBall: awesome12:17
*** haomaiwang has quit IRC12:20
*** sahid has joined #openstack-nova12:22
dimssdague : seen this yet?
*** EinstCrazy has joined #openstack-nova12:24
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing of sr_uuid_map
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: fix the live-migrate aggregate check
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up issues with live-migrate for xenapi
*** ociuhandu has joined #openstack-nova12:25
jaypipesmorning folks.12:26
sdaguedims: I was kind of ignoring it12:26
*** shaohe_feng has quit IRC12:26
BobBalljohnthetubaguy: Both auto mode and specifying --block-migrate on the CLI work with the latest series (from 5 minutes ago... before those uploads just now :P)12:26
*** rodrigods has quit IRC12:26
*** rodrigods has joined #openstack-nova12:26
cdentsdague: I put more comments on: basically "what now?"12:27
*** shaohe_feng has joined #openstack-nova12:27
openstackgerritsahid proposed openstack/nova: network: make nova to handle port_security_enabled=False
sdaguecdent: you are changing the test12:27
johnthetubaguydims: eek12:27
sdaguebut the test runs the same code as the mainpath use of it12:27
*** salv-orlando has quit IRC12:28
johnthetubaguyBobBall: cool, sounds like we are tracking it down12:28
dimsjohnthetubaguy : indeed12:28
*** pcaruana has quit IRC12:28
johnthetubaguyBobBall: I am going for lunch, do you want to work on unit tests on the aggregate thing, and I can do the top patch12:28
johnthetubaguyBobBall: let me add one comment on your patch quickly12:29
cdentsdague: the test is broken. The use of eventlet in tests is broken. This is a way to fix it so that py34 test runs don't timeout while still confirming the thing that the test wants to confirm. The fact that that eventlet is mysterious and behaves weirdly is something that we already know and is neither confirmed nor denied by this test.12:29
BobBallI'm also going to grab lunch - but will happily work on the UT etc for the aggregate patch once I'm refueled12:29
*** akshai has joined #openstack-nova12:29
*** rk4n has joined #openstack-nova12:30
sahidjohnthetubaguy, since number of bug reported against don't you think you think that could be12:30
sahidinteresting to have it in our tree for RC?12:30
cdentI agree that it is entirely confusing that fixing this one thing gets us past the hump without immediately going on to another one, but: eventlet.12:30
johnthetubaguyBobBall: added a comment here:
sdaguecdent: ok but that means - would never work on python3 either12:31
sdagueor anything in the functional tree12:31
johnthetubaguysahid: possibly, not sure I have seen that before now, did you tag the bug?12:31
*** fawadkhaliq has quit IRC12:31
sdaguethe test isn't broken. The test represents the same call path we use when we call this code for real.12:32
cdentsdague: The test I've fixed blocks every single time. Why other things do not block is beyond my ken.12:32
sdaguethe code that we are calling is broken12:32
sahidjohnthetubaguy: no, no - what kind of tag should i set ?12:32
sdaguecdent: because we don't run the functional tests under python312:32
*** rk4n_ has quit IRC12:32
*** mylu has joined #openstack-nova12:32
cdentsdague: we can cross that bridge when we come to it?12:32
sdaguebut that's not going to always be the case. This is unit testing the fixtures we use on that side12:32
cdentwhat we're going to discover is that the use of eventlet to test the api servers in python3 it utterly broken12:33
cdentand we'll have to fix it then12:33
cdentbut in the meantime we can move forward12:33
BobBalljohnthetubaguy: The exceptions cannot be raised there as we have either specified block_migration or it has been determined by one of the exceptions being raised earily.  Now we're detecting block migration we don't need to ensure the host is in the aggregate at 2209, so we can remove that check.12:33
johnthetubaguysahid: we are doing the usual tracking of RC potential bugs by using this list:
BobBalljohnthetubaguy: I think?12:33
cdentsdague: You could send me off to fix the fixtures, but I don't have time for that right now. I do have time to fix the fact that a py34 test blocks.12:33
*** boris-42 has quit IRC12:34
BobBalljohnthetubaguy: And it's not a new exception; just raised in slightly different circumstances12:34
sdaguecdent: ok, that's fine. I do get concerned about the work around because this is a really early warning about a bigger issue that we have to crack.12:35
*** mylu_ has joined #openstack-nova12:35
cdentsdague: yeah, I totally get that12:35
sahidjohnthetubaguy: done12:35
*** lucasagomes is now known as lucas-hungry12:35
johnthetubaguyBobBall: its the other way around, when you users requested block_migrate=false, if its not in an aggregate, we raise, and that bubbles up to the API handing, ish12:35
BobBalljohnthetubaguy: But that same exception could still have bubbled up; and it was the right exception to raise12:36
cdentsdague: hopefully by the time we need to cross that bridge we will have more strategies for extracting eventlet out of the tests or python3 eventlet will have miraculously matured. at the moment we know, from many inputs, that it has not12:36
*** shaohe_feng has quit IRC12:36
openstackgerritGleb Stepanov proposed openstack/nova: Add console log offset
*** shaohe_feng has joined #openstack-nova12:37
BobBalljohnthetubaguy: MigrationPreCheckError is only the right exception if we're checking the _destination_ host with get_host_uuid_from_aggregate12:37
BobBalljohnthetubaguy: But that check is for the source host, making sure that it is in an aggregate in the first place.  At that point the only valid exception would be AggregateHostNotFound not MigrationPreCheckError12:37
*** MVenesio has joined #openstack-nova12:38
*** mylu has quit IRC12:38
BobBalloh no... I see what you mean.12:38
johnthetubaguyBobBall: you are probably right though, that other one might be wrong, when its used like this12:38
BobBallAwesome.  This is running on the dest so dest is the current aggregate.  There is potentially a case where we try and migrate from a host that isn't in an aggregate to a dest that is in an aggregate (but doesn't have src in that aggregate)12:39
johnthetubaguyBobBall: yeah, the exception gets handled here:
johnthetubaguyoops, it doesn't12:40
*** rdopiera has joined #openstack-nova12:40
*** mylu_ has quit IRC12:40
sdaguecdent: yeh, I was just wonder if it might make for a good unit test to push the folks that have been doing eventlet python3 bits into using as an example of a thing to fix12:41
* cdent is starting a posting to the ml12:42
johnthetubaguyBobBall: arg, also here:
BobBalljohnthetubaguy: OK - let's change it.  Set ensure_host_in_aggregate to catch AggregateHostNotFound and raise as MigrationPreCheckError.  That'd be the right move IMO.12:43
johnthetubaguyBobBall: sounds correct12:44
*** shaohe_feng has quit IRC12:47
*** shaohe_feng has joined #openstack-nova12:48
*** pradk has joined #openstack-nova12:51
*** jlvillal has quit IRC12:52
*** jlvillal has joined #openstack-nova12:52
* alex_xu reminders nova api meeting in 8 mins12:53
*** gmann_ has joined #openstack-nova12:53
*** gmann_ has quit IRC12:53
*** gmann_ has joined #openstack-nova12:53
*** gszasz has joined #openstack-nova12:55
*** shaohe_feng has quit IRC12:57
doffmcdent: So is python3 eventlet wierdness a known thing?12:57
cdentwas a recently related changed12:57
*** eharney has joined #openstack-nova12:57
*** shaohe_feng has joined #openstack-nova12:58
BobBalljohnthetubaguy: Actually, I think we don't need AggregateHostNotFound here and the safest thing to do is just use MigratePreCheckError.  AggregateHostNotFound is only caught in one place and it's re-raised.12:59
*** ptm_away has joined #openstack-nova13:00
*** ducttape_ has joined #openstack-nova13:00
cdentdoffm: looks like there's ongoing discussion about related problems:
doffmcdent: Ughh, must have coffee before thinking about this.13:01
cdentyeah, I've gone very deep into the coffee pot this morning13:01
*** rk4n has quit IRC13:01
*** Jeffrey4l has quit IRC13:02
*** rk4n has joined #openstack-nova13:02
*** PaulMurray has quit IRC13:02
openstackgerritSean Dague proposed openstack/nova: WIP: experiments with RST and API docs
*** rcernin has joined #openstack-nova13:03
*** shaohe_feng has quit IRC13:07
*** jwcroppe has joined #openstack-nova13:08
*** shaohe_feng has joined #openstack-nova13:08
*** kaisers1 has quit IRC13:09
*** rlrossit has joined #openstack-nova13:10
*** e0ne has joined #openstack-nova13:10
openstackgerritSergey Belous proposed openstack/nova: Do not merge. Patch exists for testing os-vif with Nova
*** ducttape_ has quit IRC13:11
rdopierapkoniszewski: hi, do you have a moment? I wonder what the status of is... would you have a moment to talk about it?13:12
*** zhenq has joined #openstack-nova13:12
*** zhenq has left #openstack-nova13:13
*** zhenq has joined #openstack-nova13:13
*** Jeffrey4l has joined #openstack-nova13:14
*** kro_sinus_hell is now known as krot_sinus_purga13:15
*** krot_sinus_purga is now known as kro_sniffles13:15
*** ctrath has joined #openstack-nova13:15
*** smcginnis_away is now known as smcginnis13:15
pkoniszewskirdopiera: i need to drop, going to orthodontist, we can talk later or tomorrow13:16
sahidjaypipes: any chance you want to review something related to NUMA/HP?13:17
*** shaohe_feng has quit IRC13:17
*** rk4n_ has joined #openstack-nova13:18
*** haukebruno has joined #openstack-nova13:18
*** shaohe_feng has joined #openstack-nova13:18
jaypipessahid: sure thing.13:18
rdopierapkoniszewski: ok, enjoy your drilling :)13:19
*** kaisers has joined #openstack-nova13:19
doffmjohnthetubaguy: When you have the time could you take a quick look at (spec). Let me know what you think.13:19
doffmPlanning to get some scale testing done on it, if people don't hate it.13:19
sahidjaypipes: cool thanks,
*** gcb has quit IRC13:19
*** xyang1 has joined #openstack-nova13:19
*** rk4n has quit IRC13:20
*** ljxiash has joined #openstack-nova13:22
*** ig0r_ has joined #openstack-nova13:24
*** vladik has joined #openstack-nova13:24
*** lucas-hungry is now known as lucasagomes13:26
*** kaisers has quit IRC13:26
*** ljxiash has quit IRC13:26
*** Jeffrey4l has quit IRC13:27
*** e0ne has quit IRC13:27
*** shaohe_feng has quit IRC13:28
*** rk4n has joined #openstack-nova13:28
*** kaisers has joined #openstack-nova13:28
*** mriedem has joined #openstack-nova13:28
*** fawadkhaliq has joined #openstack-nova13:29
*** shaohe_feng has joined #openstack-nova13:29
*** salv-orlando has joined #openstack-nova13:29
*** haomaiwang has joined #openstack-nova13:29
*** hanrong has joined #openstack-nova13:29
*** hanrong has left #openstack-nova13:30
mriedemjohnthetubaguy: bauzas: are we waiting on anything for rc1?13:31
*** rk4n_ has quit IRC13:32
bauzasmriedem: yup, the prelude section13:32
*** ducttape_ has joined #openstack-nova13:32
bauzasmriedem: and I discussed with john, we'll add the translations into a RC213:32
bauzasmriedem: was reviewing
*** gcb has joined #openstack-nova13:33
mriedembauzas: did you have a separate change for reno about the ec2 api removal?13:33
bauzasmriedem: yup, was merged13:33
mriedem looks weird13:34
bauzasmriedem: see the commit msg13:34
mriedemah ok13:35
*** deepakcs has quit IRC13:35
bauzasmriedem: I actually just noticed that didn't work out13:36
bauzasso, there is a reno change about that13:36
*** Drankis has quit IRC13:36
*** smatzek has quit IRC13:36
bauzasI should also have emptied the above lock_policy yaml in the bus actually...13:36
mriedembauzas: do you want to fix that quick?13:37
mriedemi think we could fast approve13:37
bauzasmriedem: sure13:37
dansmithBobBall: johnthetubaguy: you tested live migration after the object changes.. why is it broken now? are there multiple ways to live migrate in xen or something?13:37
*** ishant has joined #openstack-nova13:37
bauzasmriedem: I'm actually trying to figure out how many yaml files were deleted that we missed13:38
*** shaohe_feng has quit IRC13:38
johnthetubaguydansmith: that was my question to, not sure how this happened13:38
dansmithjohnthetubaguy: yeah, hmm13:38
*** shaohe_feng has joined #openstack-nova13:39
*** doug-fish has quit IRC13:39
*** salv-orlando has quit IRC13:40
BobBalldansmith: I tried to go back and re-test to bisect and find the broken changes, but it's a bit tricky because of various changes such as the scheduler class name which aren't trivial to cherry-pick.13:41
bauzasmriedem: oh terrible I figured out a couple of deleted yamls still in the notes, fixing that13:41
*** edmondsw has joined #openstack-nova13:41
bauzas[sbauza@sbauza nova]$ git log --diff-filter=D --oneline releasenotes/notes/13:41
bauzas83467b8 Add release note for policy sample file update13:41
bauzas56d1293 Revert "virt: reserved hugepages on compute host"13:41
bauzas969c81f Revert "Added new scheduler filter: AggregateTypeExtraSpecsAffinityFilter"13:41
bauzas8b65185 Revert "Enable options for oslo.reports"13:42
*** doug-fish has joined #openstack-nova13:42
*** mdrabe has joined #openstack-nova13:42
bauzasokay, typey typey13:42
bauzasdhellmann: you should probably communicate to all teams that reverts are not removing relnotes13:43
dhellmannbauzas : I have a patch up to fix that, so when the new reno is released with the fix they'll disappear13:43
bauzasdhellmann: we have 5 commits to clean-up13:43
dhellmannbauzas : do what you expect to work, and we'll fix reno13:43
bauzasdhellmann: agreed, I know your change, remember13:43
dhellmannstill, good idea so folks don't get alarmed13:44
bauzasdhellmann: my biggest worry is that the reno package upgrade won't be in time for mitaka13:44
dhellmannit should happen this week13:44
bauzasdhellmann: in the meantime, I'm flushing out the content of those files13:44
bauzasthat's a preventive measure13:44
bauzasand I'd drop those empty files once reno is upgraded13:44
*** ptm_away has quit IRC13:45
bauzasdhellmann: see my workaround here
jichenoomichi: hi, do you have time to talk about
dhellmannbauzas : good plan13:46
*** ducttape_ has quit IRC13:46
*** doug-fish has quit IRC13:46
*** shaohe_feng has quit IRC13:48
*** shaohe_feng has joined #openstack-nova13:50
dansmithBobBall: okay13:51
dansmithBobBall: so there are things up I should help review right?13:51
johnthetubaguydansmith: need some decent unit tests on some of the patches still, just about to start back on those13:52
*** cleong has joined #openstack-nova13:53
auggyjohnthetubaguy: i can follow up on the i18n stuff to see if there are any translation patches13:54
*** mtanino has joined #openstack-nova13:54
*** mriedem is now known as mriedem_meeting13:54
*** gmann_ has quit IRC13:54
*** armax has joined #openstack-nova13:54
auggyi volunteered to be CPL for that with the disclaimer of being fairly new ;)13:55
johnthetubaguydoffm: I did actually try out, I think we a version of that on our cells scheduler at the moment actually, I should take a look post RC1, my main memory is that negative weights blew my mind, and did you see this one where I remove weights:
johnthetubaguyauggy: awesome, that would be great, thank you13:55
johnthetubaguyauggy: I think we changed how we do that at some point13:55
BobBallI think the two main issues were what seems to be a change in behaviour of AggregateList (perhaps if you weren't in an aggregate it would return [None] before but now returns []), and the changes around migrate_send_data being passed all the way through to support auto detecting block_migration13:55
BobBalldansmith: ^^13:55
sdaguedhellmann: I think there are some subtleties of the fact that reno is using the git history and not the contents of a directory that makes for unexpected situations13:55
auggyjohnthetubaguy: lol ok, i just went by what the zanata folks had told me the process was13:56
*** houming_ has quit IRC13:56
dansmithBobBall: oh, I think that latter one came after the original migration patches, so that makes sense13:56
dhellmannsdague : yeah, I noted that bug. that'll have to wait for newton, but I think we can try to address it13:56
*** PaulMurray has joined #openstack-nova13:56
sdaguedhellmann: cool13:56
dhellmannsdague : there's a pretty narrowly focused spot to make the change, it's just a matter of time13:57
johnthetubaguyBobBall: I was going to start tidying this one up now:
BobBalldansmith: We do have a working series of patches though - but as johnthetubaguy says, more work on unit tests going on13:57
dhellmannsdague : I think the main issue was differentiating between a local file on the "current" branch and versions of that file (or not having it at all) on other branches13:58
BobBalljohnthetubaguy: Are your unit tests clean on now?13:58
*** shaohe_feng has quit IRC13:58
*** pushkaru has joined #openstack-nova13:58
*** smatzek has joined #openstack-nova13:58
BobBalljohnthetubaguy: I think you wanted 293085 as the base, right? then 292937 then 293393?13:59
sdaguedhellmann: yeh, though I think as we're managing 3 or 4 active branches with this content, it's going to get more confusing13:59
johnthetubaguyBobBall: I switched it around, I liked your suggested order, in the end, once I understood your patch better13:59
*** shaohe_feng has joined #openstack-nova13:59
johnthetubaguyBobBall: the middle patch should work, I want to do a review of it, because the tests are too messy really14:00
BobBalljohnthetubaguy: So you want me to fix UT at the base with the aggregate check in 293393?14:00
dhellmannsdague : yeah, some of the refactoring work I've done to deal with caching for non-git use will actually make this easier to solve, but that hasn't landed yet14:00
sdaguedhellmann: cool, sounds like you have it under hand14:00
johnthetubaguyBobBall: ideally check out the latest gerrit version of that aggregate fix, yeah14:00
openstackgerritgaryk proposed openstack/nova: Virt: add context to attach and detach interface
*** doug-fish has joined #openstack-nova14:00
*** doug-fish has quit IRC14:01
*** haomaiwang has quit IRC14:01
johnthetubaguyBobBall: I will work on the top one, and I will try not to kill your changes, honest, although not sure how easy that actually is14:01
*** doug-fish has joined #openstack-nova14:01
*** haomaiwang has joined #openstack-nova14:01
*** rk4n_ has joined #openstack-nova14:01
*** mgoddard_ has joined #openstack-nova14:01
*** edtubill has joined #openstack-nova14:02
*** ivase has joined #openstack-nova14:02
*** dmellado is now known as dmellado|lunch14:02
BobBalljohnthetubaguy: That's fine.  I'll work on the bottom one, then you just need to rebase the 2nd one on the bottom one.14:03
*** rk4n has quit IRC14:04
BobBallAny ideas how to avoid race conditions with large (32 CPU) boxes running tox? :) I'm getting TypeError: this TransactionFactory is already started randomly which kills dozens of tests when it occurs14:04
*** mgoddard has quit IRC14:05
*** lhinds_ has joined #openstack-nova14:06
openstackgerritSylvain Bauza proposed openstack/nova: Fix reno reverts that are still shown
bauzasmriedem_meeting: johnthetubaguy: sdague: one last clean-up and we should be good to go14:06
*** mgagne has quit IRC14:07
*** toan has quit IRC14:07
*** mgagne has joined #openstack-nova14:07
*** toan has joined #openstack-nova14:08
*** busterswt has joined #openstack-nova14:08
bauzasjohnthetubaguy: once the above is merged, I think we should be good for RC114:08
bauzasunless someone sees a blocker14:09
*** shaohe_feng has quit IRC14:09
bauzasmarkus_z: any recent bug that which would need to stop RC1 ?14:09
*** shaohe_feng has joined #openstack-nova14:09
*** ducttape_ has joined #openstack-nova14:10
johnthetubaguybauzas: awesome, thank you14:10
johnthetubaguybauzas: the xenapi fix live-migrate would be nice to have, but given we need an RC2, and the code is not yet ready for review, I say its missed RC114:10
bauzasjohnthetubaguy: yeah I'm just reviewing
bauzas4 open bugs14:11
*** pwnall has joined #openstack-nova14:11
johnthetubaguyyeah, we had a few new ones that need a peak14:11
bauzassahid: not sure needs to block RC114:11
openstackLaunchpad bug 1460630 in OpenStack Compute (nova) "nova should not verify "port_security_enabled" according the info from network" [Undecided,In progress] - Assigned to sahid (sahid-ferdjaoui)14:12
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
bauzassahid: at least it's not a mitaka regression14:12
dansmithbauzas: agreed14:13
BobBalld'oh - missed one of my changes... hang on..14:13
*** knikolla has joined #openstack-nova14:13
bauzassahid: that's something we could backport to mitaka if that's merged at newton14:14
openstackgerritSergey Nikitin proposed openstack/nova: Short proof of concept for gabby test framework
robcresswellQuick question, sorry if I'm misunderstanding; does booting with a port-id hit the attach_interface policy, or is it just the create policy that is applied?14:14
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
BobBalljohnthetubaguy: ^^ I think (hope) that's OK for review now.14:15
bauzassahid: I'm also concerned by given it's not a regression14:15
openstackLaunchpad bug 1543149 in OpenStack Compute (nova) "Reserve host pages on compute nodes" [High,In progress] - Assigned to sahid (sahid-ferdjaoui)14:15
openstackgerritSean Dague proposed openstack/nova: WIP: experiments with RST and API docs
*** salv-orlando has joined #openstack-nova14:15
johnthetubaguyBobBall: which line generates the keyerror, that looks a bit odd now14:16
johnthetubaguyBobBall: I guess thats the .metadata[hostname] bit?14:16
BobBalljohnthetubaguy: Line immediately before the exception, with the metadata[hostname]14:16
bauzasBobBall: johnthetubaguy: how things are going with the XenAPI driver ?14:16
BobBallbauzas: It's a source of continued joy and happiness.14:16
BobBallbauzas: (We have working patches - just tidying them up)14:17
johnthetubaguyyeah that14:17
BobBalljohnthetubaguy: Want a "if hostname in .metadata" instead?14:17
johnthetubaguytidy up and quick re-test14:17
johnthetubaguyBobBall: sounds good14:17
dansmithjohnthetubaguy: BobBall noted some issues in the second patch for sr_uuid_map14:18
openstackgerritgaryk proposed openstack/nova: VMware: prevent volume deletion if volume in instance folder
*** shaohe_feng has quit IRC14:19
*** shaohe_feng has joined #openstack-nova14:20
*** mspreitz has joined #openstack-nova14:21
bauzasmriedem_meeting: that sounds a gate problem, hence Critical, right ?14:21
bauzasoh man14:21
bauzasmissing the link14:22
bauzas is much better14:22
openstackLaunchpad bug 1480305 in OpenStack Compute (nova) "DBDeadlock inserting into instance_extra" [Medium,In progress] - Assigned to John Garbutt (johngarbutt)14:22
bauzasjohnthetubaguy: just reviewing the above ^ why are you thinking it's worth waiting for it for RC1 ?14:22
bauzasspeaking of
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
*** jichen has quit IRC14:23
johnthetubaguybauzas: its not now, it was causing gate issues for neutron, and seemed worth doing, but it too late now14:23
bauzasjohnthetubaguy: I'm about to +2 it given some notice about the possible cell impact, but I'd drop mitaka-rc-potential, agreed ?14:24
johnthetubaguybauzas: yeah, thats fine14:25
*** ducttape_ has quit IRC14:25
*** Daisy_ has joined #openstack-nova14:26
dansmithjohnthetubaguy: BobBall just to be clear, the base migratedata object should never be used by your driver, nor received anywhere.. it's just a base class for inheritance purposes, which is why it's not registered14:27
*** zenoway has quit IRC14:27
johnthetubaguydansmith: yeah agreed, the issue we saw what unset attributes triggering a not found in base class error, so we had to use the "in" checks, is that normal?14:28
*** mrkz has joined #openstack-nova14:28
*** ducttape_ has joined #openstack-nova14:28
*** Swami has joined #openstack-nova14:29
*** shaohe_feng has quit IRC14:29
*** zz_dimtruck is now known as dimtruck14:29
dansmithjohnthetubaguy: yeah, if it's possible to be unset, then you need to check14:30
*** fawadkhaliq has quit IRC14:30
*** shaohe_feng has joined #openstack-nova14:30
dansmithjohnthetubaguy: that's a kinda fundamental objects thing, since we can always have unset fields (potentially-lazy-loaded for db-type objects)14:30
*** aswadr_ has quit IRC14:30
*** fawadkhaliq has joined #openstack-nova14:30
johnthetubaguydansmith: yeah, I see your point, you need that for lazy load, got it14:30
dansmithjohnthetubaguy: the other option is to always send it over the wire of course, but I was mirroring the existing code that didn't always send it for whatever reason14:30
*** rk4n has joined #openstack-nova14:30
johnthetubaguydansmith: yeah, that makes sense14:31
*** Daisy_ has quit IRC14:31
*** zenoway has joined #openstack-nova14:31
BobBalldansmith / johnthetubaguy: FYI I think the aggregate issue fixed in may have been introduced when NovaObjectDictCompat was removed14:32
dansmithBobBall: yeah14:32
johnthetubaguyah... I see14:32
dansmithBobBall: I would say you could blame rlrossit, but with no xen live migration tests, he really can't be blamed :)14:32
*** sc68cal has joined #openstack-nova14:33
BobBallWell since rlrossit doesn't seem to be here at the moment, I might blame him anyway.  Just don't tell him I did so.14:33
*** dedery has joined #openstack-nova14:33
*** rk4n_ has quit IRC14:33
BobBallBut yeah... Xen live migration tests do seem to be somewhat required14:33
dansmithwell, only if you want to keep it working I guess :)14:34
auggyjohnthetubaguy: re: translations, I am following up with the i18n folks to make sure, but the gerrit query the zanata team gave me shows zero open translations. it shows the last submission as from sept last year and that has already been merged14:34
BobBallI suppose keeping live migration is kinda important... ;)14:34
johnthetubaguyauggy: OK, I saw similar things, not sure if thats correct or not14:35
BobBallIt's on our backlog of things to add.  Got a new CI env setting up soon, so will hopefully be able to use that.14:35
dansmithbauzas: johnthetubaguy so which things are waiting for for rc1 specifically?14:35
*** ducttape_ has quit IRC14:35
auggyjohnthetubaguy: i sent an email to the i18n mailing list just letting them know we're cutting RC1 so if they have anything for us, we need it like *now*14:35
johnthetubaguyauggy: thanks for digging, I am expecting an RC2 to include translations anyways14:35
bauzasdansmith: only a reno change AFAIK plus the xen live-mig but we could do the latter for RC214:35
johnthetubaguyauggy: mostly because hard string freeze hits with RC114:35
auggyjohnthetubaguy: kk, i'll keep an eye on it14:35
johnthetubaguyauggy: thank you :)14:35
auggyjohnthetubaguy: yeah the hard string freeze14:35
dansmithbauzas: the xen ones should be rc2, IMHO14:36
* auggy is learning things!14:36
*** dedery has quit IRC14:36
bauzasdansmith: okay, so the only blocker is
johnthetubaguydansmith: +114:37
dansmithbauzas: got it14:37
bauzasdansmith: johnthetubaguy: are we planning to do RPC pins during RC2 ?14:37
bauzasare we also planning to bump major RPC versions ?14:38
dansmithbauzas: we bumped last time, I think we're good this time14:38
bauzasfor bumps, gotcha14:38
bauzasbut for the version aliases ?14:38
dansmithwe do that post-release right?14:38
bauzasI know that we have the 'auto' value14:38
bauzasdansmith: that was an open question :)14:38
johnthetubaguydansmith: so the checklist says its the other way around, its before release14:39
bauzasif we do that post-release, it's finr14:39
bauzasjohnthetubaguy: yeah that14:39
bauzaswe should verify14:39
johnthetubaguydansmith: I guess you want N-1 to have the pin for the name of the N-1 release?14:39
johnthetubaguyand use that same config as you roll forward14:39
*** shaohe_feng has quit IRC14:39
dansmithyeah, I think the first few times we backported the alias14:39
dansmithso that's why I think of it as after14:40
dansmithbecause if we had to change rpc in rc3, we'd have to move the alias14:40
*** wanghao has quit IRC14:40
dansmithsince we know we're going to have an rc2, let's hold off14:40
dansmithin case we have to do something stupid for this xen thing14:40
johnthetubaguyyeah, that works14:40
*** shaohe_feng has joined #openstack-nova14:40
*** wanghao has joined #openstack-nova14:41
bauzasdansmith: not sure I understood the point, you said that we should do the aliases during rc214:43
bauzasI mean, at the last release candidate, gotcha14:43
openstackgerritChris Dent proposed openstack/nova: Work around stall in py34 unit tests
bauzasokay, wfm14:43
bauzasjohnthetubaguy: actively monitoring zuul for prepping the release patch, if you agree14:45
bauzasI don't know yet the merged patch sha114:45
bauzasfor topping rc114:45
*** esracelik has quit IRC14:46
*** mgoddard_ has quit IRC14:47
*** mgoddard has joined #openstack-nova14:47
*** shaohe_feng has quit IRC14:50
*** sahid has left #openstack-nova14:50
openstackgerritAlexis Lee proposed openstack/nova: Add FK bw_usage_cache.uuid -> instances.uuid
*** shaohe_feng has joined #openstack-nova14:51
sbelouspkholkin: ping14:53
*** mriedem_meeting is now known as mriedem14:56
mriedembauzas: what gate problem?14:56
mriedembauzas: do you mean the xen live migration thing?14:57
doffmjohnthetubaguy: I had not seen that. Will read through now.14:58
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
bauzasmriedem: no sorry,
openstackLaunchpad bug 1480305 in OpenStack Compute (nova) "DBDeadlock inserting into instance_extra" [Medium,In progress] - Assigned to John Garbutt (johngarbutt)15:00
doffmjohnthetubaguy: I have thought a bunch about negative weights for the probability placment. Haven't put it in the spec yet. I ended up just considering the total weight a measurement of 'emptiness' of a host.15:00
*** shaohe_feng has quit IRC15:00
bauzasmriedem: but that seems not really needing a critical15:00
rlrossitdansmith: BobBall: I love taking blame15:00
*** ctrath has quit IRC15:00
*** haomaiwang has quit IRC15:01
BobBallWin-win then! :)15:01
*** ctrath has joined #openstack-nova15:01
*** shaohe_feng has joined #openstack-nova15:01
*** haomaiwang has joined #openstack-nova15:01
*** AJaeger has joined #openstack-nova15:01
cdentrlrossit: I'm hungry and have a headache, can I blame  you?15:02
rlrossitcdent: go for it15:02
cdentrlrossit: you bastard, why you make me suffer so?15:02
rlrossitit's also my fault that america isn't great, that's why we need someone to make it great again15:02
cdentIf I didn't have a headacahe america would be great again15:03
AJaegernova team, the compute API Guide contains a few sphinx bugs when build, I fixed these with and also enabled warnings as errors to avoid regressions. The job is non-voting currently - could you review and tell me whether it's ok, please?15:03
mriedembauzas: that's been around for a long time15:03
bauzasmriedem: that's what I saw, hence me saying it's not critical15:04
doffmjohnthetubaguy: I took the weight 'as-is' without really worrying how it was normalized or combined.15:04
doffmalaski: Are we having a cells meeting today?15:07
alaskidoffm: yes sir15:07
alaskiin just under two hours15:07
*** sfinucan has joined #openstack-nova15:07
openstackgerritsahid proposed openstack/nova: livirt: handle reserved pages size
openstackgerritsahid proposed openstack/nova: virt: reserved hugepages on compute host
*** READ10 has joined #openstack-nova15:08
*** mdrabe has quit IRC15:09
*** mdrabe has joined #openstack-nova15:09
*** shaohe_feng has quit IRC15:10
*** shaohe_feng has joined #openstack-nova15:11
*** edtubill has quit IRC15:13
*** yamahata has joined #openstack-nova15:15
*** david-lyle has quit IRC15:15
*** david-lyle has joined #openstack-nova15:15
*** dmellado|lunch is now known as dmellado15:15
*** ociuhandu has quit IRC15:16
*** purplerbot has quit IRC15:16
*** purplerbot has joined #openstack-nova15:17
*** ekuris_ has quit IRC15:17
*** knikolla has quit IRC15:17
*** liverpooler has quit IRC15:17
BobBalljohnthetubaguy: What's the next step on the LM bug?  Do you need help in fixing your change or rebasing on ?15:18
*** gjayavelu has joined #openstack-nova15:18
mriedemjaypipes: dansmith: alaski: johnthetubaguy: bauzas: commented on that db deadlock thing
johnthetubaguyBobBall: sorry, yeah, I am working through my change and the little fix ups, doing that full time now, rather than during a meeting, so should be making progress now!15:19
jaypipesmriedem: looking...15:19
BobBalljohnthetubaguy: No worries; just shout if you need my help or when there is a new combined change I can test.15:19
dansmithjohnthetubaguy: mriedem: Is this something to do with recent enginefacade stuff?15:20
bauzasmriedem: so I feel your point looks good, was just +2ing because I thought it was a very smart workaround for that until we have a better solution15:20
dansmithI hadn't heard of this being a problem before15:20
mriedemdansmith: it's been around since last year15:20
mriedemit's not very frequent15:20
johnthetubaguydansmith: yeah, I thought the same honestly, but it seems to have been cropping up for a while15:20
bauzasmriedem: TBH, I'm concerned by a cells v2 world where we would create instances in the DB after we have a BuildRequest15:20
dansmithI'm not really sure why creating new things would deadlock I guess15:20
dansmithbut I wouldn't know15:20
*** shaohe_feng has quit IRC15:20
mriedemi'm just wondering if creating all of the new things at once is causing the deadlock15:21
bauzasmriedem: hoping that kind of situation in a cells v2 world would be nicer to fix15:21
mriedembut this is only 5 hits in 10 days15:21
bauzasmriedem: that was what I was supposing15:21
johnthetubaguymriedem: I am having a similar feeling, I do wonder if we committed in the same order each time we would avoid the deadlock15:21
openstackgerritGleb Stepanov proposed openstack/nova: Console log copy for block live migration
*** shaohe_feng has joined #openstack-nova15:21
johnthetubaguygoogle seems to suggest using select...for update to force an ordering, but that just breaks galera15:22
*** knikolla has joined #openstack-nova15:22
johnthetubaguyall I discovered was that I clearly don't understand how INSERT generated deadlocks :(15:22
mriedemhence my ping to mr mysql15:22
openstackgerritSergey Belous proposed openstack/nova: Do not merge. Patch exists for testing os-vif with Nova
*** ociuhandu has joined #openstack-nova15:23
johnthetubaguymriedem: ah, I meant to ping him, but forgot to do that15:23
*** dedery has joined #openstack-nova15:25
sdagueso, under pymysql we can have multiple db things going on in a single worker, unlike the c lib one15:29
*** moshele has quit IRC15:29
*** browne has joined #openstack-nova15:29
openstackgerritGleb Stepanov proposed openstack/nova: Add console log offset
sdagueso is this an issue where we yielded because something else in a complex transaction, and another API request comes in triggering something similar15:30
dansmithI guess if we lock the table and yield to something else that goes to lock the table15:30
*** shaohe_feng has quit IRC15:31
johnthetubaguyoh... thats possible15:31
dansmithwe had parallelism here before,15:31
sdaguewe should only be row locking, but are there transactions we're in the middle of?15:31
sdaguedansmith: we hard paralleism at the worker level right?15:31
dansmithbut only across workers, which wouldn't cause one to deadlock15:31
*** shaohe_feng has joined #openstack-nova15:31
dansmithwhat I mean is,15:31
sdaguenow we are potentially sharing a connection15:32
dansmithwe had parallelism before from mysql's perspective, but not within our workers which could double acquire15:32
*** dedery has quit IRC15:32
sdaguewhich would also explain the low frequency15:32
sdaguebecause it just is a legit race under load right now15:32
*** dedery has joined #openstack-nova15:32
*** fawadkhaliq has quit IRC15:33
sdagueI would bet if you look at the logs there are 2 simultaneous creates on different API requests in the same API worker15:33
johnthetubaguycould it just be our delay that triggers the deadlock detection for false reasons?15:33
sdaguehonestly, it's all just speculation at this point, going through detailed logs would be required15:34
*** fawadkhaliq has joined #openstack-nova15:34
dansmithjohnthetubaguy: what delay?15:34
sdaguewould also be interesting if we could instrument the exception path to dump more details when this happens15:34
*** david-lyle_ has joined #openstack-nova15:34
johnthetubaguydansmith: eventlet letting another thread do things while we are stalled15:34
*** david-lyle has quit IRC15:34
dansmithso I thought the deadlock comes from the sql client15:35
johnthetubaguysdague: I just wonder how we check we aquire the locks in the same order15:35
dansmithlike, I didn't think it was time-based15:35
dansmithjohnthetubaguy: a guru-meditation dump at the time would show us with two greenthreads in that path at the same time15:35
dansmithwe might be able to instrument that detection in the handler15:36
johnthetubaguyah, right, thats a good idea15:36
dansmithcheck the rest of our siblings to see if they have a stack frame equivalent to ours at the place where we raised or something15:36
johnthetubaguyits tempting to do that for all DB deadlock retries by default, in some ways?15:36
sdaguethat stacktrace is super ass confusing15:36
*** jaosorior has quit IRC15:36
dansmithwell, right, it'd be something we'd want to do generically15:36
dansmithsdague: link?15:37
mriedemsdague: it's wrapped in lots of decorators15:37
sdaguewell that's the logstash one I was looking at15:37
*** vishwanathj_zzz is now known as vishwanathj15:38
dansmithjohnthetubaguy: yeah, see it comes from mysql15:38
*** fawadkhaliq has quit IRC15:38
sdagueright it's definitely from mysql15:38
dansmithso that one is different than instance_extra15:39
dansmithbut similar15:39
johnthetubaguyOK, its good to rule that out15:39
dansmitha second thing we create in the create instance path15:39
dansmither, wait15:39
knikollawhen snapshotting a vm, does glance call back to nova?15:40
mriedemthat's liberty
mriedemthe hit in that logstash query is liberty15:40
mriedemwhich is why the line numbers threw me off15:40
mriedemwhich didn't have the enginefacade stuff15:41
dansmithwe're in instance create15:41
*** shaohe_feng has quit IRC15:41
dansmithwhich calls ec2_instance_create15:41
*** jroll has quit IRC15:41
dansmithwhich saves instances?15:41
*** jroll has joined #openstack-nova15:41
*** jroll has quit IRC15:41
*** jroll has joined #openstack-nova15:41
mriedemi think that commits the transation15:41
johnthetubaguyI think thats where the flush happens15:41
johnthetubaguyyeah, that15:41
dansmithoh, that's the sqla ref, nevermind15:41
*** shaohe_feng has joined #openstack-nova15:41
mriedemwhich is why i was suggesting, insert the instance record first, then insert the foreign key table records15:42
mriedemrather than a mega commit15:42
mriedemb/c it has to sort through all the backrefs15:42
dansmithwell, I don't think that's the right fix15:42
dansmithit'd just be mitigation15:42
mriedemsure, so is retrying on deadlock15:43
dansmithyep, yep15:43
johnthetubaguywell, I do wonder if enforcing the order would avoid the deadlock15:43
johnthetubaguyi.e. force we always acquire the locks in the same order, somehow15:43
johnthetubaguybut I guess thats across all db transactions, which doesn't totally sound that tractable, I guess15:44
dansmithI don't think so15:44
dansmithbecause we could yield at any point during any one of these, AFAIK15:44
bauzasfolks, got to leave soon (wife's birthday), so I'm about to tag RC1 by tomorrow morning (or later tonight, will reconnect)15:45
bauzasjohnthetubaguy: you okay?15:45
*** pwnall has quit IRC15:45
dansmithwe can tag once that other patch lands right?15:45
johnthetubaguybauzas: I think so, are all the reno things in now?15:45
bauzas is 15 mins away from being merged15:45
johnthetubaguyI think we can tag once that lands15:45
dansmithjohnthetubaguy: +115:45
mriedemwe're not waiting for the xen live migration thing?15:46
johnthetubaguycool, I will push up the release patch once it merges15:46
bauzasmriedem: no, planned as rc215:46
bauzasjohnthetubaguy: sounds good to me15:46
johnthetubaguyRC2 is needed for translations anyways15:46
bauzas(interesting thoughts on the db deadlock, need to scrollback)15:46
johnthetubaguyif we get the xen one merged soon ish, then we backport it before final release15:46
johnthetubaguybauzas: have a good evening15:47
bauzaswell, will reconnect later :p15:47
*** mylu has joined #openstack-nova15:47
*** shaohe_feng has quit IRC15:51
*** browne has quit IRC15:51
*** shaohe_feng has joined #openstack-nova15:52
*** apoorvad has joined #openstack-nova15:52
dansmithreno patch just got reset, it's #5 in the gate15:55
johnthetubaguyhey, ho, its moving up15:55
claudiubit's the final countdown!15:57
AJaegerjohnthetubaguy, sdague, nova team: I just pinged on #openstack-infra but want to tell you here as well: 293526 and 292472 and rename gate-nova-python27 to gate-nova-python27-db etc in project-config. So, that in a followup we can remove database setup from the normal python27 jobs (similar for python34 etc). Could you review those, please? This is part of our infra consolidation on one trusty image.15:57
AJaegerif there are questions, let's discuss on #openstack-infra15:58
openstackgerritBalazs Gibizer proposed openstack/nova: Handle race in allocate_for_instance
*** dedery has quit IRC15:59
*** artom_ has joined #openstack-nova15:59
*** haomaiwang has quit IRC16:01
*** haomaiwang has joined #openstack-nova16:01
*** e0ne has joined #openstack-nova16:01
*** shaohe_feng has quit IRC16:01
*** klkumar has quit IRC16:01
*** shaohe_feng has joined #openstack-nova16:02
*** fawadkhaliq has joined #openstack-nova16:04
*** matrohon has quit IRC16:05
*** cdent has quit IRC16:05
*** harlowja_at_home has joined #openstack-nova16:07
mriedemtjones_zzzz: how can i tell why this failed?
mriedemdevstack was fine16:09
mriedemthere are no test results or console log that i can see from a test run16:09
mriedemlooks like n-api didn't start
*** geekinutah has joined #openstack-nova16:10
markus_zbauzas: I was afk. The bugs which came in the last 24 hours were not serious.16:11
markus_zbauzas: It looks like only the xen one is left16:11
*** EinstCrazy has quit IRC16:11
*** belmoreira has quit IRC16:11
*** shaohe_feng has quit IRC16:12
mriedemsdague: shouldn't devstack fail if one of the services in the ENABLED_SERVICES list didn't start?16:12
dansmithBobBall: AggregateList.get_by_host() is returning [None] ?16:12
*** shaohe_feng has joined #openstack-nova16:12
*** allen_gao has quit IRC16:13
auggyjohnthetubaguy: i'll send an email to the i18n list letting them know we plan to include translations in RC216:14
*** harlowja_at_home has quit IRC16:15
*** allen_gao has joined #openstack-nova16:15
johnthetubaguyauggy: OK, thats what they asked us to do with the Hard vs Soft freeze last cycle, at least, really good idea to double check what they want from us there16:16
auggyjohnthetubaguy: np, i am on it16:17
johnthetubaguyauggy: :)16:17
*** kfarr has joined #openstack-nova16:17
dansmithnow #1 in the gate, 20m to merge16:19
*** Swami has quit IRC16:19
*** allen_gao has quit IRC16:20
*** david-lyle_ is now known as david-lyle16:20
*** paul-carlton1 has quit IRC16:20
*** shaohe_feng has quit IRC16:22
*** harlowja_at_home has joined #openstack-nova16:22
*** shaohe_feng has joined #openstack-nova16:22
*** allen_gao has joined #openstack-nova16:23
*** dedery has joined #openstack-nova16:25
openstackgerritsahid proposed openstack/nova: libvirt: handle reserved pages size
openstackgerritsahid proposed openstack/nova: virt: reserved hugepages on compute host
*** mgoddard has quit IRC16:28
*** mgoddard has joined #openstack-nova16:28
*** pwnall has joined #openstack-nova16:30
*** cdent has joined #openstack-nova16:30
*** astupnikov has quit IRC16:31
*** shaohe_feng has quit IRC16:32
*** fawadkhaliq has quit IRC16:32
*** shaohe_feng has joined #openstack-nova16:33
*** dedery has quit IRC16:33
*** fawadkhaliq has joined #openstack-nova16:33
*** mdrabe_ has joined #openstack-nova16:33
*** fawadkhaliq has quit IRC16:33
*** mdrabe has quit IRC16:34
*** ljxiash has joined #openstack-nova16:36
*** Swami has joined #openstack-nova16:37
*** browne has joined #openstack-nova16:38
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix when auto block_migration in the API
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: refactor check_can_live_migrate_destination
auggyjohnthetubaguy: so it looks like the search I was using was outdated :/16:38
*** links has quit IRC16:38
*** dedery has joined #openstack-nova16:38
auggyjohnthetubaguy: i got this from the mailing list -
johnthetubaguyauggy: ah, oops, we should try get that in before we tag, I guess16:39
auggyjohnthetubaguy: the branch is zanata/translations not transifex/translations apparantly16:39
johnthetubaguyah, right16:40
auggyi can't remember where i got that old link from, i thought someone in the i18n channel gave it tme16:40
johnthetubaguyI know I was missing something16:40
johnthetubaguyyeah, I found the old link in their wiki somehwere, so worth updating that I guess16:40
dansmithjohnthetubaguy: that reno patch is out of the queue16:41
dansmithjohnthetubaguy: i.e. merged16:41
johnthetubaguyah, cools16:42
johnthetubaguyI guess, do we want to wait for translations as well, I think we might just want to cut anayways16:42
*** matrohon has joined #openstack-nova16:42
dansmithjohnthetubaguy: I kinda feel like we should just cut rc1.. we could wait forever.. we know we have rc2 coming16:42
*** shaohe_feng has quit IRC16:42
*** Daisy has joined #openstack-nova16:42
johnthetubaguydansmith: yeah, we can do a quick RC2 if we have to16:43
*** mdrabe has joined #openstack-nova16:43
johnthetubaguyI will get that patch up now16:43
*** eil397 has joined #openstack-nova16:43
dansmithjohnthetubaguy: the translations aren't a rush right?16:43
dansmithwhy wouldn't they just wait for the rc2 with the xen fix and whatever else we're waiting for?16:43
*** shaohe_feng has joined #openstack-nova16:43
johnthetubaguyno, rush, really16:43
johnthetubaguythere will be more updates by the time we hit RC2, which is expected16:43
johnthetubaguythere are not loads of them yet, anyways16:43
*** jlanoux has quit IRC16:44
*** mdrabe_ has quit IRC16:45
markus_zAJaeger: The -1 is only because I thing we can use "toctree hidden" in
markus_zAJaeger: Maybe adding the nits in a next patch set and everything should be fine then16:46
*** e0ne has quit IRC16:47
*** e0ne_ has joined #openstack-nova16:47
dansmithjohnthetubaguy: you can line this up right behind the newton open if you want:16:47
*** Daisy has quit IRC16:47
dansmithjohnthetubaguy: we still can't land the sanity check migration until grenade gets updated, but I have a query saved to find and -2 any other migrations and will keep checking for them16:48
auggyjohnthetubaguy: ok cool, so to confirm, translations are going into RC216:48
johnthetubaguydansmith: I am not quite sure how we do the newton open anymore, given we don't have a version in setup.cfg, just working that out16:48
dansmithjohnthetubaguy: ah, okay16:48
dansmithwell, when you do, please un- -2 that patch so we can get it in early16:49
*** scheuran has quit IRC16:49
dansmithjohnthetubaguy: I should get a patch up for the rpc aliases too, for rc2 huh?16:49
*** flwang1 has quit IRC16:49
mriedemi'd tihnk stable/mitaka is just created from the rc1 tag16:49
johnthetubaguywe should do that16:49
*** READ10 has quit IRC16:49
johnthetubaguymriedem: yeah, I think thats how it work, but not 100% sure16:49
johnthetubaguynormally we just pick the commit before landing the open release commit16:50
*** e0ne has joined #openstack-nova16:50
*** dedery has quit IRC16:50
*** fawadkhaliq has joined #openstack-nova16:50
*** e0ne_ has quit IRC16:50
*** fawadkhaliq has quit IRC16:51
*** fawadkhaliq has joined #openstack-nova16:51
*** nic has joined #openstack-nova16:52
*** shaohe_feng has quit IRC16:53
*** mylu has quit IRC16:53
*** e0ne has quit IRC16:53
*** shaohe_feng has joined #openstack-nova16:54
*** nkrinner has quit IRC16:54
openstackgerritDan Smith proposed openstack/nova: Set RPC version aliases for Mitaka
*** pwnall has quit IRC16:55
*** unicell has quit IRC16:56
auggyjohnthetubaguy: i should probably add the translations change to one of our etherpads, which one is the best one? mitaka-nova-priorities?16:57
*** zenoway has quit IRC16:58
*** geekinutah has quit IRC16:58
johnthetubaguyauggy: so its limbo time, as the etherpad still has mitaka in the name16:59
*** pushkaru has quit IRC16:59
*** pushkaru has joined #openstack-nova16:59
*** raildo is now known as raildo-afk17:00
johnthetubaguydansmith: I guess we don't do pins for the ComputeTaskAPI17:00
*** e0ne has joined #openstack-nova17:00
johnthetubaguyauggy: lets stay on this one till release I guess:
*** haomaiwang has quit IRC17:01
dansmithjohnthetubaguy: pins for either conductor api don't really make sense anyway17:01
johnthetubaguydansmith: true17:01
markus_zmriedem: I was thinking about the mitaka-backport-potential tag for bugs.17:01
auggyjohnthetubaguy: alright! i'll just tack it onto the end of the etherpad in its own section17:01
markus_zmriedem: Does it make sense to put that on all open bugs when we release RC?17:01
*** dedery has joined #openstack-nova17:01
*** haomaiwang has joined #openstack-nova17:01
mriedemmarkus_z: all?17:01
johnthetubaguymarkus_z: for finding RC2 bugs I guess?17:01
markus_zmriedem: yeah17:02
mriedemmarkus_z: i think anything that was rc potential but didn't make it is mitaka-backport-potential17:02
markus_zjohnthetubaguy: I though about when someone fixes a bug in Newton|Ocata that this is already there17:02
mriedemi don't know how many open bugs we have, but it seems wrong to just blanket apply that tag17:02
johnthetubaguyyeah, lets only add the tag when we need it17:02
markus_zmriedem: I wasn't sure if that would be overkill to be honest17:02
johnthetubaguyits a judgment call if it needs the backport17:03
markus_zjohnthetubaguy: mriedem: ok, understood17:03
*** shaohe_feng has quit IRC17:03
johnthetubaguyI think we want the rc-potential tag still, as once we do RC2, we need to decide if we want an RC3, for example17:03
mriedemmarkus_z: if there are confirmed/triaged/in-progress bugs that were reported against mitaka (or liberty), then i could see adding the mitaka-backport-potential tag, but that's probably still a lot of bugs17:03
*** shaohe_feng has joined #openstack-nova17:04
dansmithdefinitely not blanket-apply17:04
markus_zmriedem: yep, it's a lot17:05
markus_zjohnthetubaguy: The rc-blocker tag got removed from a few, but they should still be in the etherpad17:06
johnthetubaguydansmith: I think I got the correct commit here:
dansmithjohnthetubaguy: I'm the wrong one to ask.. I've messed that up multiple times I think :D17:06
dansmithdhellmann: ^17:06
dansmithor mriedem17:07
johnthetubaguydansmith: just checking its the correct git hash for that last patch really17:07
dansmithat one point someone told me it was supposed to be the merge commit17:07
dansmithand then not17:07
dansmithand then again17:07
dansmithso I dunno ;P17:07
johnthetubaguycommit a5486b32a3c476f3ad584d8ff7c4eda2bb3e400d17:07
johnthetubaguy2016-03-16 17:01:28.325 | Author: Sylvain Bauza <>17:07
johnthetubaguy2016-03-16 17:01:28.325 | Date:   Wed Mar 16 14:52:21 2016 +010017:07
johnthetubaguy2016-03-16 17:01:28.325 |17:07
johnthetubaguy2016-03-16 17:01:28.326 |     Fix reno reverts that are still shown17:07
*** lpetrut has quit IRC17:07
johnthetubaguyso it has that one in17:08
johnthetubaguyI guess thats the main thing17:08
mriedemshould be 99ad55ad9d0e954b8a175a61271a6f150cc0503917:08
*** jistr has quit IRC17:12
*** shaohe_feng has quit IRC17:13
*** shaohe_feng has joined #openstack-nova17:14
markus_zjohnthetubaguy: I added the "RC2 (potential)" section in the etherpad. It contains the bugs which are not in RC1 but were|are tagged with rc-blocker.17:15
*** paul-carlton2 has joined #openstack-nova17:15
johnthetubaguymarkus_z: coolness17:17
*** mspreitz has quit IRC17:17
markus_zjohnthetubaguy: AJaeger has found some doc issue which might be interesting for a RC2?
*** dedery has quit IRC17:19
johnthetubaguymarkus_z: that just gets uploaded from master, so I think its ok not backporting that at the moment17:20
markus_zjohnthetubaguy: Ah, you're right, I forgot that17:21
*** shaohe_feng has quit IRC17:23
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
*** shaohe_feng has joined #openstack-nova17:24
openstackgerritThomas Maddox proposed openstack/nova: [WIP - DO NOT MERGE] Testing LXC Gate
thomasemjohnthetubaguy: mriedem: Was having much better success with Fedora 23 yesterday, fyi.17:27
*** unicell has joined #openstack-nova17:27
dansmithBobBall: found a couple bugs in that latest patch17:27
thomasemI put up a patch here: and set the WIP test patch where I'm invoking the experimental pipeline to depend on the Change ID:
*** lpetrut has joined #openstack-nova17:28
thomasemThat should pull in the dependent project-config in that experimental pipeline, correct?17:28
*** e0ne has quit IRC17:30
mriedemjohnthetubaguy: in the cells meeting we're talking about lining up priorities for newton, wrt cells v2 specs, do you have any objection to just starting a new newton priorities etherpad now? i'm assuming cellsv2 wll be on that list17:31
*** lucasagomes is now known as lucas-afk17:33
*** shaohe_feng has quit IRC17:34
*** shaohe_feng has joined #openstack-nova17:35
johnthetubaguymriedem: sounds cool to me17:35
openstackgerritThomas Maddox proposed openstack/nova: [WIP - DO NOT MERGE] Testing LXC Gate
johnthetubaguymriedem: just put a link at the top of the mitaka one telling people where to find the new one I guess, then I am happy17:36
*** ndipanov has quit IRC17:36
mriedemjohnthetubaguy: ok17:36
*** armax has quit IRC17:37
*** sfinucan has quit IRC17:38
*** johnthetubaguy_ has joined #openstack-nova17:39
*** atuvenie has quit IRC17:40
*** mylu has joined #openstack-nova17:41
*** johnthetubaguy has quit IRC17:41
*** johnthetubaguy_ is now known as johnthetubaguy17:41
*** lpetrut has quit IRC17:43
*** ctrath1 has joined #openstack-nova17:43
BobBalldansmith: Oops... Indeed - the standard return value must be [] or None I guess otherwise I don't see how it could work...17:43
*** shaohe_feng has quit IRC17:44
dansmithBobBall: you mean for mock?17:44
dansmithBobBall: the standard return value is a MagicMock()17:44
*** gszasz has quit IRC17:44
BobBallHo hum.  I wonder why the code 'worked' with a MagicMock then...17:44
*** ctrath has quit IRC17:44
*** shaohe_feng has joined #openstack-nova17:45
dansmithyeah, that was my question17:45
*** vishwanathj is now known as vishwanathj_chow17:45
rdopieraMagicMock is falsy17:45
BobBallI wonder what len(MagicMock) is :)  I'm hoping it's 017:46
*** sdague has quit IRC17:46
BobBallw00t - it is 017:46
BobBallThat's why the unit tests pass.17:46
johnthetubaguyoh my, joy17:46
BobBallPresumably MagicMock implements all of the things to look like an iterable as well...17:47
rdopierait "implements" all attributes and returns MagicMock from them17:48
mriedemjohnthetubaguy: we're waiting to +W this right?
johnthetubaguymriedem: I think we can +W that right away17:49
*** fawadkhaliq has quit IRC17:49
*** derekh has quit IRC17:50
*** fawadkhaliq has joined #openstack-nova17:50
*** mdrabe has quit IRC17:51
*** mdrabe has joined #openstack-nova17:51
*** sdague has joined #openstack-nova17:52
*** rlrossit has quit IRC17:52
dansmithmriedem: your comment about "the flavor in the server body" -- do you mean the instance attributes like memory_mb and stuff that are actually attached to the instance itself?17:52
dansmiththere's no actual flavor: attribute inside the instance that I know of17:52
*** rcernin has quit IRC17:53
dansmithmriedem: or do you mean this: ?17:53
*** salv-orl_ has joined #openstack-nova17:54
dansmithokay, I'm confused, because we can't stop returning that without breaking the api, which is the whole point of this spec, right?17:54
*** shaohe_feng has quit IRC17:54
sdaguethis is the new flavor links spec?17:55
*** shaohe_feng has joined #openstack-nova17:55
alaskisdague: yeah17:55
dansmithsdague: yeah17:55
sdaguedansmith: a thing I realized when writing that up is that the fields we'd actually expose on this new resource end up being < bytes than the links to the resource17:56
*** salv-orlando has quit IRC17:56
sdagueso there is an argument to be made for just inline them in the server representation17:57
dansmithsdague: but the point here is to keep the api working, not to reduce the size right?17:57
*** lhinds_ has quit IRC17:57
dansmithsdague: and, if we dump flavors later, it seems less bad to me to have a broken or missing link field, than to remove a whole chunk of attributes17:57
sdaguethe point is to stop regularly handing out busted links17:57
dansmithwell, the point for me is not really the existing bugs,17:58
dansmithbut to not break that entirely with cellsv217:58
sdagueand that any graphical tool is basically *always* making the double call17:58
dansmithbecause if we move flavors to the api, we're unable to honor that17:58
sdaguedansmith: or if you delete a flavor today17:58
dansmithyep, understand17:58
sdaguethe delete case exposes it without cells already17:58
dansmithI know, I just said I don't care as much about that17:59
dansmithit's nice to fix it for sure17:59
*** mgoddard_ has joined #openstack-nova17:59
dansmithbut with cellsv2 (which will be all of nova) it'll be permanently broken17:59
dansmithwhich is more of a concern to me17:59
dansmithor rather, delaying cellsv2 as a result of not wanting to break it (or redesign more first)17:59
*** browne has quit IRC17:59
alaskiso rather than debating these two alternatives why not do both? do what sdague has currently proposed as a short term thing with an eventual goal of moving towards jaypipes suggestion18:00
*** PaulMurray has quit IRC18:00
dansmithjaypipes' suggestion is not really doable today I think,18:00
dansmithbecause it requires some more thinking and abstraction of things18:00
dansmithI think sdague's suggestion was to just inline the same thing in the server itself. right sdague ?18:01
*** haomaiwang has quit IRC18:01
jaypipesdansmith: gimme a bit.. on a meeting18:01
sdaguedansmith: right18:01
alaskiyeah. it could be a dependency of the resource provider work in motion and be handled as part of that effrot later18:01
*** haomaiwang has joined #openstack-nova18:01
dansmithsdague: yeah, so if you want to inline the flavor and provide the link, that's fine18:01
dansmithsdague: but I think jaypipes would argue that he'd rather keep the link and inline some other representation later18:01
mriedemdansmith: sdague: that was an alternative proposed,18:01
mriedeminlining it i mean18:01
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: rename dest_check_data to migrate_data
sdagueright, that was an alternative, sorry18:02
mriedemthe spec as written has a link18:02
sdaguedansmith: I don't know what "also provide a link" means18:02
dansmiththe thing is, we already have inlined copies of most of this stuff18:02
sdaguedansmith: we don't18:02
*** browne has joined #openstack-nova18:02
*** mgoddard has quit IRC18:02
dansmithwe do on the instance, but not in the api, I see my fault18:02
sdaguethe spec as written is that we have it in the instance, use the instance data instead and do almost the same thing18:03
sdaguehowever, in walking that through, it actually made me wonder why (as expresssed in an alternative) we don't just return that information inline18:03
dansmithsdague: the spec as writen says we just fix the link18:03
sdaguedansmith: correct18:04
mriedembut sdague likes the alternative inline approach now18:04
*** baoli has quit IRC18:04
dansmithso, when we discussed this in vancouver I think it was,18:04
*** shaohe_feng has quit IRC18:04
mriedemthis came out of a ML thread i thought18:04
dansmiththe plan was (I thought) to cheat and expose the new link even for v2.0 people to avoid breaking them18:04
alaskisdague: for now there is some dissatisfaction with how the data is modeled, so moving it inline might be better of waiting until some of the resource provider rework is done18:04
dansmithalaski: right, see I was thinking that instead of inlining the thing jaypipes hates, keep it separate as a formal flavor resource18:05
mriedempersonally i tend to like the inline idea so i don't have to make 2 calls18:05
dansmithalaski: and then inline the thing he wants after we do the modeling18:05
*** shaohe_feng has joined #openstack-nova18:05
alaskidansmith: agreed18:05
sdaguedansmith: no, I never intended to break our contract on this one18:05
*** lpetrut has joined #openstack-nova18:05
dansmithmriedem: I don't see how that doesn't make jaypipes more mad18:05
dansmithsdague: not sure which way you mean.. we  break our contract with v2.0 people if we don't keep a working link in there, right?18:06
sdagueadding a new resource on v2.0 seems like a bad call18:06
alaskimriedem: ultimately we should inline, but not with the current flavor model, is my thinking18:06
dansmithalaski: +118:06
mriedemdansmith: if not inlining helps with the future resource stuff, then i'm fine with just fixing the link18:06
dansmithmriedem: I think that makes more sense to me18:07
mriedemas written in the spec18:07
openstackgerritsean mooney proposed openstack/nova: stop setting mtu when plugging vhost-user ports
*** geekinutah has joined #openstack-nova18:07
dansmithsdague: well, I was sure we talked about exposing it just so the link would be there, but... okay18:07
dansmithsdague: horizon breaks today if we do that, right?18:07
*** haukebruno has quit IRC18:07
sdaguebecause people are just going to hardcode GET /v2/{foo}/servers/{id}/flavor into their toolkits, and it will break on a bunch of public clouds18:07
dansmithwhich means horizon from today won't work with nova +118:07
mriedemhorizon uses novaclient so i'm not sure what they do18:08
sdaguehorizon today follows links we provide, but exposing resources retroactively causes all these other issues18:08
dansmithsdague: we could store a uuid in the instance flavor and make it unpredictable :D18:08
dansmithsdague: well, that means horizon of today doesn't work with newton18:08
dansmithwhich seems bad to me, and I thought that was the whole point of doing this route instead of just removing the link entirely18:09
*** ccarmack has joined #openstack-nova18:09
sdagueso in newton, the v2 API just doesn't work, is what you are telling me?18:10
dansmithwell, that's how it would be right?18:10
dansmithand not just v2, but v2.N where N is less than 30 or whatever we're at right now18:11
dansmithjust seems like the lesser of two evils to leak a new resource back than to break everyone18:11
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: rename dest_check_data to migrate_data
sdaguewhen you leak a new resource back, what you actually do is create a ton of client software that doesn't work on all the public clouds18:12
dansmithI hear you18:12
dansmithbut newtony client software won't work on them either18:12
dansmither, existing clients on newtony servers18:13
sdaguewhich was exactly the thing we were trying to avoid in doing the microversion approach and being careful about it18:13
dansmithwhatever :)18:13
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix up passing of sr_uuid_map
sdagueso, I guess I don't understand with all the care to other parts of upgrade, how that can be the case :)18:13
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix when auto block_migration in the API
dansmithI just don't see how we can move flavors out of the API as we need to for cellsv2 and not break every existing client right now18:13
sdaguebut, anyway, I think this is going to need a high bw pow wow, because I'm obviously missing bits18:14
mriedemi'm lost too on the 'moving flavors out of the api' thing18:14
*** rlrossit has joined #openstack-nova18:14
dansmithmriedem: moving them out of the cell and into the api18:14
alaskiflavor links will be broken for deleted flavors18:14
mriedemchanging the link in a microversion so it's a sub-resource on server allows us to get around the deleted flavor thing in the flavors api18:14
dansmithperhaps a confusing choice of words18:14
sdaguealaski: s/will be/are/18:15
*** shaohe_feng has quit IRC18:15
mriedembut with the subresource we get the flavor out of instance.flavor18:15
alaskisdague: fair18:15
*** shaohe_feng has joined #openstack-nova18:15
openstackgerritAndreas Jaeger proposed openstack/nova: Error on API Guide warnings
dansmithmriedem: right, but if we don't generate the flavor link for 2.0 people because they don't have the resource, then the link is missing or empty18:16
johnthetubaguyhonestly, I feel we need to think about a flavor "lifecycle" a little bit more18:16
alaskiby not exposing this throught all api versions we're saying that we're breaking the api for all versions less than the introduction of this new flavor thing18:16
openstackgerritBob Ball proposed openstack/nova: xenapi: Fix the live-migrate aggregate check
*** READ10 has joined #openstack-nova18:16
dansmithso we can break the api a little with a leaky thing,18:16
mriedem"For clarity, you might want to mention whether or not the server response body will have a flavor entry with the existing flavor id and links to the flavors API or if you plan on removing that as part of this (which I don't think we should do)."18:16
dansmithor we can break it completely by not letting the thing leak18:16
mriedemis what i said in bristol18:16
mriedemand -1ed the spec18:16
*** vilobhmm11 has joined #openstack-nova18:17
melwittjohnthetubaguy: +118:17
sdagueso, I'm happy to do a high bw thing at some point, because I don't understand why the existing flavor links are unsupportable. But I think that we're not going to sort it in irc today18:17
alaskisdague: it takes them from sometimes works to never works, when deleted18:18
dansmithalaski: so I've lost my context on one point actually, and maybe I'm completely crazy and making this harder than it needs to be18:18
openstackgerritThierry Carrez proposed openstack/nova: Update reno for stable/mitaka
openstackgerritAndreas Jaeger proposed openstack/nova: Error on API Guide warnings
jaypipesdansmith, sdague, mriedem: ok, sorry guys, my meeting is done now. I can talk if you'd like.18:18
johnthetubaguyI do prefer the always breaks when flavor is deleted18:18
dansmithalaski: for cellsv2, we move flavors from the cell to the api database, which means we effectively lose the ability to even lie about the link between the instance and the flavor right?18:18
*** tesseract- has quit IRC18:19
sdaguealaski: is the delete case the only issue?18:19
alaskisdague: I believe so18:19
dansmithalaski: this all came about as a solution to our needing to move flavors, but I've lost context on some of that18:19
openstackgerritThierry Carrez proposed openstack/nova: Update reno for stable/mitaka
AJaegermarkus_z: updated, addressed all your suggestions18:19
sdagueright, so I think that's probably acceptable casualty to not break some other stuff. If they want the right answer, they can use a microversion.18:20
johnthetubaguyso bigger picture, there could be an "active" flavor thing, where you have image_meta + on boot override + flavor at time of boot18:20
*** zenoway has joined #openstack-nova18:20
alaskidansmith: I don't think we lose the ability to link, or lie about it :), it's really that we lose the ability to link to deleted flavors18:20
johnthetubaguyso flavors could get modified after the instance booted, ish18:20
mriedemalaski: dansmith: because the flavors API today reads soft deleted flavors,18:20
mriedemwhich goes away with the flavors in the api db18:21
dansmithalaski: mriedem: remember 20 minutes ago when I asked about what was blocking the moving of flavors to the api db and you pointed at this as why?18:21
mriedemwhich is why thiscame up18:21
dansmithmriedem: ahh, right, that's it18:21
mriedembut we said that api is already broken if you hard delete the flavors18:21
alaskiright now it sometimes works, in the future it will never work18:21
mriedemso while we have the instance flavor stuff in the instance itself, we said, why not just show that as a subresource on the server18:21
johnthetubaguywhich people do when then prune their DB18:22
sdagueright, so we can't retroactively fix that issue. We can say, we suggest you use this new version of our API, and things will be better18:22
mriedemjohnthetubaguy: right18:22
sdagueand I think that's fine18:22
mriedemsdague: +118:22
sdaguefor non deleted flavors, all is still fine18:22
sdagueeven on the old API18:22
dansmithokay, so I may be wrapping some of this up in resource provider stuff where we lose linkages for non-uuid' things or whatever18:22
dansmithbecause there are duplicates in places18:23
mriedemmy only question is why we can't still provide the link to the flavor sub-resource so we don't have to drop server['flavor']['links'], if we're worried about that18:23
mriedemwe just change the lnk18:23
sdaguemriedem: we could, it seems kind of weird18:23
dansmithwell, so, my question is this18:23
alaskidansmith: that would affect current v1 users, except that flavors in a cell need to match the api already or other breakages happen so there won't be two different flavors with the same id18:23
sdaguemriedem: because         "OS-FLV-DISABLED:disabled": false,18:23
melwittare we saying we want links to deleted flavors after flavors turn to hard deleted? I didn't think we want that18:23
sdagueis kind of pointless18:24
dansmithfor older clients, are we going to try to maintain the link if the flavor looks like it's still there?18:24
mriedemmelwitt: no, that's what we're trying to solve18:24
sdagueas is "os-flavor-access:is_public": true,18:24
mriedemdansmith: i suggestd that in the spec too18:24
mriedem"I think we need to be clear about what we're going to do with the existing part of the server response body that has the flavor in it. FWIW I think we should return it, at least if it still exists."18:24
sdagueand I don't think we wanted bookmark links on the subresource18:24
*** shaohe_feng has quit IRC18:25
*** ishant has quit IRC18:25
*** akshai has quit IRC18:25
*** shaohe_feng has joined #openstack-nova18:26
dansmithalaski: so does this really need to block cells stuff if we migrate flavors and maintain the links? soft deleted flavors will appear to have been purged, which they will for anyone before the microversion that adds this anyway18:26
jaypipesso did you guys want to have a high bw discussion about this?18:27
mriedempersonally, i want some coffee18:28
alaskidansmith: if we're all okay with that then no it doesn't need to block18:28
dansmithjaypipes: no18:28
mriedemi don't think this is blocking really, i never really did,18:28
mriedembut i think the spec is still useful18:28
mriedemon it's own18:28
dansmithmriedem: except you brought it up as the blocking thing :)18:28
alaskibut some people do rely on that so I think we just wanted to not break it knowingly18:29
mriedemdansmith: because we soft blocked it months ago18:29
BobBalldansmith: Looking better?
alaskijaypipes: to summarize what I think you're interested in, some of us agree that we should remodel resources and drop them inline on the instance, but not hold up the current proposal if it blocks cells18:30
alaskiso do it in two steps rather than either/or18:30
openstackgerritMark Doffman proposed openstack/nova-specs: Move aggregates tables to the API database
jaypipesalaski: and then, what, deprecate the flavors/ subresource from the API some time in the future?18:30
alaskijaypipes: wasn't discussed, but that's what I would want to see18:31
jaypipesalaski: I'd be cool with that.18:31
openstackgerritJohn Garbutt proposed openstack/nova: xenapi: Fix when auto block_migration in the API
*** READ10 has quit IRC18:32
dansmithBobBall: yeah18:32
alaskijaypipes: cool. I think we're all mostly on the same page then18:32
jaypipesalaski: as long as dansmith is happy, I'm happy.18:32
dansmithjaypipes: oh stop it18:33
*** akshai has joined #openstack-nova18:33
* jaypipes giggles18:33
*** pwnall has joined #openstack-nova18:33
* cfriesen__ is scared18:33
melwittalaski: by deprecate the flavors API does that mean just the bookmark part or lose the ability to list flavors etc?18:33
jaypipescfriesen__: be afraid. be very afraid. :P18:33
alaskimelwitt: just the link in the instance view, and subresource of instance. flavors api would remain unchanged18:34
melwittalaski: gotcha18:34
diana_clarkedansmith: I've been asked to revisit metadata cache performance. Is there anything you think I should know before I dive in?18:34
diana_clarkedansmith: I'm asking because you have -1 & -2 votes on the existing patches, and the launchpad issue is assigned to you.18:34
diana_clarkedansmith: So far, I've reviewed these:
*** mriedem has quit IRC18:35
*** shaohe_feng has quit IRC18:35
*** shaohe_feng has joined #openstack-nova18:36
*** paul-carlton2 has quit IRC18:36
*** pwnall has quit IRC18:36
dansmithdiana_clarke: back in a bit18:37
*** mriedem has joined #openstack-nova18:38
*** rk4n_ has joined #openstack-nova18:38
diana_clarkedansmith: Thanks, no rush.18:39
*** cdent has quit IRC18:40
*** Marga_ has quit IRC18:40
AJaegermarkus_z: no need to backport my change, we only publish from master... It should go into master soonish so that we have great documentation...18:40
sdaguejohnthetubaguy: are we back to having an open master?18:41
*** mylu has quit IRC18:41
johnthetubaguysdague: yes18:41
johnthetubaguysdague: I should send an ML note18:41
*** rk4n has quit IRC18:42
sdaguejohnthetubaguy: great18:43
johnthetubaguysdague: so ttx totally beat me to it on the email front18:44
johnthetubaguysdague: the one thing is we need to block DB migrations till we get the placeholders in18:44
sdaguejohnthetubaguy: that email should also probably say we can't land db migrations18:44
dansmithjohnthetubaguy: I'll send that mail18:45
johnthetubaguydansmith: OK, cools18:45
*** shaohe_feng has quit IRC18:45
*** rlrossit has quit IRC18:46
*** shaohe_feng has joined #openstack-nova18:46
*** BobBall is now known as BobBall_AWOL18:47
*** sfinucan has joined #openstack-nova18:47
*** baoli has joined #openstack-nova18:49
*** fawadkhaliq has quit IRC18:49
openstackgerritMark Doffman proposed openstack/nova-specs: Move networking tables to the API db.
*** mylu has joined #openstack-nova18:53
*** mgoddard_ has quit IRC18:54
*** mgoddard has joined #openstack-nova18:54
*** moshele has joined #openstack-nova18:55
*** shaohe_feng has quit IRC18:56
*** pece has quit IRC18:56
*** shaohe_feng has joined #openstack-nova18:56
*** rlrossit_ has joined #openstack-nova18:57
*** Daisy has joined #openstack-nova18:58
*** haomaiwang has quit IRC19:01
mriedemsean-k-m1: comments inline
*** vishwanathj_chow is now known as vishwanathj19:01
*** haomaiwa_ has joined #openstack-nova19:01
*** Swami has quit IRC19:02
*** claudiub has quit IRC19:03
*** Swami has joined #openstack-nova19:03
*** Daisy has quit IRC19:03
mriedemedleafe: can you take a look at this bug? confirm it or other?
openstackLaunchpad bug 1477098 in OpenStack Compute (nova) "When you unshelve an instance, AggregateImagePropertiesIsolation filter doesn't work." [Undecided,New]19:03
mriedemit's been sitting in New for awhile19:03
*** pumarani__ has joined #openstack-nova19:03
*** jichen has joined #openstack-nova19:04
*** harlowja has joined #openstack-nova19:04
mriedemrlrossit_: did you want to look at this?
rlrossit_mriedem: is that the one I got thrown under the bus for earlier today?19:05
dansmithdiana_clarke: so.. is there a specific question?19:05
*** mkrcmari__ has quit IRC19:05
*** wanghao has quit IRC19:05
mriedemrlrossit_: yes19:05
*** shaohe_feng has quit IRC19:06
*** wanghao has joined #openstack-nova19:06
*** shaohe_feng has joined #openstack-nova19:07
*** pushkaru has quit IRC19:07
diana_clarkedansmith: I get the impression (but I could be wrong), that you think the metadata cache should be fine as-is now that it's not doing a bunch of lazy loads for each request, and that otherwise config-drive should be used.19:09
*** sfinucan has quit IRC19:10
dansmithdiana_clarke: well, yes on config drive, because that's our preferred thing as a project19:10
dansmithbut yeah, I think the majority of the performance issues with metadata as it was in liberty are probably resolved19:10
dansmithat least, the ones related to the lazy-loads I mean19:10
dansmithI think that not stashing the whole instance in the cache is probably a reasonable change to make, although I wouldn't prioritize it unless there's something compelling19:11
*** moshele has quit IRC19:12
*** moshele has joined #openstack-nova19:12
*** cdent has joined #openstack-nova19:13
*** salv-orl_ has quit IRC19:15
*** shaohe_feng has quit IRC19:16
*** shaohe_feng has joined #openstack-nova19:17
*** ssurana has joined #openstack-nova19:17
diana_clarkedansmith: Okay, I'll start by profiling master (vs liberty) and perhaps increase test coverage if need be.19:17
diana_clarkedansmith: The launchpad ticket is assigned to you. Do you plan on working in this area too? I don't want to get in your way.19:18
dansmithdiana_clarke: I didn't assign it to myself, so feel free to take it19:19
*** browne has quit IRC19:19
*** ircuser-1 has quit IRC19:19
*** ssurana has quit IRC19:22
*** browne has joined #openstack-nova19:22
*** pece has joined #openstack-nova19:25
*** irenab_ has joined #openstack-nova19:26
*** shaohe_feng has quit IRC19:26
doffmmriedem, edleafe, bauzas: There is a very good chance that was fixed by the request spec work right?19:26
openstackLaunchpad bug 1477098 in OpenStack Compute (nova) "When you unshelve an instance, AggregateImagePropertiesIsolation filter doesn't work." [Undecided,New]19:26
*** kro_sniffles is now known as kro_mnckn_doc19:27
*** shaohe_feng has joined #openstack-nova19:27
*** irenab has quit IRC19:28
*** irenab_ is now known as irenab19:28
*** daemontool has quit IRC19:28
*** eliqiao has quit IRC19:29
*** eliqiao has joined #openstack-nova19:29
*** akshai has quit IRC19:30
*** yamahata has quit IRC19:34
*** ociuhandu has quit IRC19:35
*** Marga_ has joined #openstack-nova19:36
*** mylu has quit IRC19:36
*** shaohe_feng has quit IRC19:37
*** shaohe_feng has joined #openstack-nova19:37
*** akshai has joined #openstack-nova19:39
*** moshele has quit IRC19:39
*** mylu has joined #openstack-nova19:41
*** browne has quit IRC19:42
*** gjayavelu has quit IRC19:43
*** ssurana has joined #openstack-nova19:45
dansmithmelwitt: wanna +W this?
dansmithneeds to land firstly in newton now that it's open19:45
dansmiththe newton PTL has already +2d19:45
melwitton it19:46
*** e0ne has joined #openstack-nova19:46
*** harlowja has quit IRC19:46
*** shaohe_feng has quit IRC19:47
*** gjayavelu has joined #openstack-nova19:48
*** shaohe_feng has joined #openstack-nova19:48
melwittdansmith: actually, shouldn't it be range(320, 330) in ?19:52
dansmithbecause rebase19:53
*** ig0r_ has quit IRC19:55
openstackgerritDan Smith proposed openstack/nova: Add placeholder migrations for Mitaka backports
openstackgerritDan Smith proposed openstack/nova: Add Newton sanity check migration
dansmithI actually want to make that test file able to ignore _placeholder migrations so we don't have to do this error-prone thing19:55
melwittah, yeah19:55
*** ociuhandu has joined #openstack-nova19:56
*** shaohe_feng has quit IRC19:57
*** shaohe_feng has joined #openstack-nova19:58
openstackgerritjichenjc proposed openstack/nova: Change REGEX for instance display name filter
*** sfinucan has joined #openstack-nova19:58
*** ericksonsantos has joined #openstack-nova19:58
melwittmriedem: we had to respin because of the unit test if you want to take a look ^19:59
mriedemi'm looking19:59
*** haomaiwa_ has quit IRC20:01
*** moshele has joined #openstack-nova20:01
*** haomaiwang has joined #openstack-nova20:01
*** salv-orlando has joined #openstack-nova20:01
*** claudiub has joined #openstack-nova20:03
*** gjayavelu has quit IRC20:04
*** claudiub|2 has joined #openstack-nova20:05
*** sfinucan has quit IRC20:06
*** rk4n_ has quit IRC20:06
*** gjayavelu has joined #openstack-nova20:06
*** fhermeni has joined #openstack-nova20:07
*** shaohe_feng has quit IRC20:07
*** shaohe_feng has joined #openstack-nova20:08
*** claudiub has quit IRC20:08
openstackgerritJoe Cropper proposed openstack/nova: Add get_host_memory_stats virt driver API
openstackgerritJoe Cropper proposed openstack/nova: Add memory metric monitor
*** moshele has quit IRC20:10
*** pc_m has joined #openstack-nova20:12
*** smatzek has quit IRC20:13
*** cdent has quit IRC20:15
mriedemjaypipes: i'm cleaning up old specs so dropped just fyi20:16
mriedemit's not personal :)20:16
mriedemrlrossit_: gibi: you might be interested in
jaypipesmriedem: np :)20:16
mriedemmordred: are you planning on pursuing this?
*** AJaeger has left #openstack-nova20:17
*** shaohe_feng has quit IRC20:18
*** penick has joined #openstack-nova20:18
*** shaohe_feng has joined #openstack-nova20:18
*** ssurana has quit IRC20:19
*** mylu has quit IRC20:19
*** rtheis has quit IRC20:20
*** mylu has joined #openstack-nova20:20
mriedemalaski: should we pursue this? or is anyone else planning on working on quotas?20:22
mriedemi saw the quota service/lib ML explosion this morning20:22
*** ssurana has joined #openstack-nova20:22
alaskiI am not planning on working on that for now.  someone mentioned ccarmack might be interested in it and if not I'll abandon20:23
mriedemi was just going to call out ccarm20:24
ccarmackI thought we were investigating root causes of quota out-of-sync20:24
melwittyeah, we had a quick chat here about quotas the other day and I think ccarmack and doffm had some interest in working on getting some functional tests together to get us a real baseline on what quotas are doing as we go toward a design20:25
ccarmackmelwitt had an etherpad for that20:25
ccarmackmelwitt: yes that's right20:25
melwittyeah, was grabbing the link. I added it to the quotas subteam section of the priorities etherpad to help capture the work and ideas as we go20:26
mriedemi'll hook that back into the review etherpad20:26
mriedemmelwitt: ahh good20:26
alaskithat would indicate the spec is early at this point so I can abandon until testing/research is done20:26
*** sbezverk has quit IRC20:26
melwittwe kept talking about it on ML etc about how we don't *really* know what's going on with quotas right now and it would be good to know that and also be able to test new designs and it would help prioritize any such quota work20:27
ccarmackmriedem: so I'm ok with abadon with "quotas reimagined"20:27
mriedemyeah, agree on abandoning that until we have some functional tests20:28
*** shaohe_feng has quit IRC20:28
*** shaohe_feng has joined #openstack-nova20:28
*** penick has quit IRC20:29
*** toure_ is now known as toure20:30
*** baoli has quit IRC20:32
*** diogogmt has joined #openstack-nova20:33
*** penick has joined #openstack-nova20:33
*** ircuser-1 has joined #openstack-nova20:33
*** shaohe_feng has quit IRC20:38
*** shaohe_feng has joined #openstack-nova20:39
*** andymaier has quit IRC20:41
openstackgerritMark Doffman proposed openstack/nova: Add database tables to CellsV2 database split docs
*** lpetrut has quit IRC20:44
bauzasalaski: melwitt: sorry, I was out tonight20:46
bauzasso I missed the cells v2 meeting20:46
alaskibauzas: no worries. as always it just means you've volunteered for everything20:46
*** shaohe_feng has quit IRC20:48
*** ianw has joined #openstack-nova20:48
*** cleong has quit IRC20:50
*** shaohe_feng has joined #openstack-nova20:50
mriedemeasy +W for specs core
*** aysyd has quit IRC20:52
*** mylu has quit IRC20:53
*** zhenq has quit IRC20:57
*** zhenq has joined #openstack-nova20:57
*** e0ne has quit IRC20:58
*** shaohe_feng has quit IRC20:59
*** shaohe_feng has joined #openstack-nova20:59
*** pumarani__ has quit IRC21:00
*** woodster_ has joined #openstack-nova21:00
*** thorst has quit IRC21:01
*** haomaiwang has quit IRC21:01
*** haomaiwa_ has joined #openstack-nova21:01
*** vilobhmm11 has quit IRC21:02
*** vilobhmm11 has joined #openstack-nova21:02
*** thorst has joined #openstack-nova21:03
*** vilobhmm11 has quit IRC21:03
*** vilobhmm11 has joined #openstack-nova21:03
*** pushkaru has joined #openstack-nova21:04
*** achanda has joined #openstack-nova21:05
*** penick has quit IRC21:05
*** READ10 has joined #openstack-nova21:07
*** thorst has quit IRC21:07
openstackgerritjichenjc proposed openstack/nova: WIP: add regression test for 1546396
*** toure is now known as toure_21:08
openstackgerritSean Dague proposed openstack/nova: WIP: attempt to parse content inside table cells
*** shaohe_feng has quit IRC21:09
*** shaohe_feng has joined #openstack-nova21:09
mriedemnova specs dashboard if anyone wants one
*** doug-fish has quit IRC21:11
*** doug-fish has joined #openstack-nova21:12
auggycan i get a dashboard of dashboards in my dashboard?21:12
*** Daisy has joined #openstack-nova21:14
*** toure_ is now known as toure21:14
*** rfolco has quit IRC21:14
*** browne has joined #openstack-nova21:16
*** rlrossit_ has quit IRC21:17
melwittmriedem: nice, added it to My Menu21:18
mriedemone thing i don't get is i +2ed and now it doesn't show up in my dashboard21:18
*** Daisy has quit IRC21:18
mriedemoh i suppose b/c i +2ed it so i don't need to see it now21:19
*** kro_mnckn_doc has quit IRC21:19
*** shaohe_feng has quit IRC21:19
*** anteaya has quit IRC21:19
*** samueldmq has quit IRC21:19
*** shaohe_feng has joined #openstack-nova21:20
*** mdrabe has quit IRC21:23
*** achanda has quit IRC21:24
*** gjayavelu has quit IRC21:24
*** nickchase has joined #openstack-nova21:25
dansmithmelwitt: so what is the state of db and mq connection switching?21:27
*** achanda has joined #openstack-nova21:28
*** rdopiera has quit IRC21:28
*** shaohe_feng has quit IRC21:29
*** shaohe_feng has joined #openstack-nova21:30
*** eharney has quit IRC21:30
*** darrenc is now known as darren_afk21:31
*** vilobhmm11 has quit IRC21:32
alaskidansmith: db switching merged, though nothing is using it21:32
*** pradk has quit IRC21:32
*** vilobhmm11 has joined #openstack-nova21:32
*** jichen has quit IRC21:33
*** knikolla has quit IRC21:34
dansmithalaski: okay I thought so21:35
dansmithalaski: and the actual flavor migration is still pending somewhere?21:35
melwittdansmith: also I have a WIP patch up for the mq connection switching where I tested it locally with only build_and_run_instance and so far so good21:35
*** gjayavelu has joined #openstack-nova21:35
*** achanda has quit IRC21:36
melwittit needs a rebase and more work21:36
*** e0ne has joined #openstack-nova21:37
dansmithalaski: okay I thought the code was owned by someone else.. did it shift?21:37
alaskinot that I know of. But someone else may need to take over. I need to reach out to CERN to see what sort of involvement they'll have this cycle21:38
dansmithwell, I was wondering if I should just ask to take it over21:39
dansmithsince I have a history of screwing up, er, pulling off, these things21:39
alaskiyeah, you should definitely see about taking over21:39
*** shaohe_feng has quit IRC21:40
*** e0ne has quit IRC21:40
dansmithI feel like alaski just said "you should definitely see about getting kicked in the nuts"21:40
*** shaohe_feng has joined #openstack-nova21:41
alaskipretty much, though you did basically volunteer for it21:41
*** nickchase has quit IRC21:41
melwitthey, you said something crazy and we're like...roll with it!21:41
alaskiI mean if that's what you're into who's to judge21:42
dansmithalaski: I am into some kinky shit21:43
bauzaseh, nice, RC1 is on the shelf21:44
*** ccarmack has left #openstack-nova21:44
*** matrohon has quit IRC21:45
*** darren_afk is now known as darrenc21:46
*** achanda has joined #openstack-nova21:47
alaskidansmith: we all know you're a masochist for taking on the online migration instance_extra thing, among other horrible projects21:48
dansmithalaski: I still have nightmares21:48
melwittis something afoot with the py34 job? failed it21:49
melwittbuild time out21:49
*** permalac has quit IRC21:50
*** shaohe_feng has quit IRC21:50
dansmithmelwitt: I give that afoot usage a 5/10.. weak justification21:50
dansmithand yes21:50
dansmiththere is a thread on the list21:50
*** shaohe_feng has joined #openstack-nova21:51
dansmithmelwitt: alaski: so, this patch is not something we can merge right now because it assumes flavors are all migrated when they're not21:51
bauzasoh man, that deprecation warning for nova.rpc is terrible for the logs :p21:51
mriedemmelwitt: see cdent's eventlet + py34 thread21:53
melwittdansmith: the placeholders?21:53
*** kfarr has quit IRC21:53
dansmithmelwitt: no the flavors patch21:54
alaskidansmith: it's migrating the flavors right?21:54
*** MVenesio has quit IRC21:54
dansmithalaski: FlavorList.get_all() assumes they're all in the api_db, AFAIC21:55
melwittdansmith: the one we were just talking about with active migration? it intends to online migrate flavors but there are problems in it21:55
dansmithI don't think it does intend to do that21:55
*** alex_xu has quit IRC21:56
dansmithexcept maybe by saying "this intends to migrate live"21:56
alaskiI think it's the intent, though the implementation may not back that up21:56
*** mriedem has quit IRC21:57
melwittyeah live migrate flavors. it does things like try lookup in api db, if not found, get it from main db, then only write to api db. that sort of thing21:57
*** ctrath1 has quit IRC21:58
*** alex_xu has joined #openstack-nova21:58
dansmithmelwitt: I don't think so21:59
melwittI wrote a lot of comments in there21:59
dansmithat least, not across the board21:59
melwittright. I wrote comments about the gaps I saw21:59
*** shaohe_feng has quit IRC22:00
*** haomaiwa_ has quit IRC22:01
melwittnamely L713
*** shaohe_feng has joined #openstack-nova22:01
*** haomaiwang has joined #openstack-nova22:01
dansmithalaski: melwitt: and your thought is that we take advantage of some cycle in which we can only ever have a cell of one to do the migration lively, knowing we can easily connect to both databases and not need to iterate over N cells?22:01
* dansmith is already starting to regret volunteering for this22:02
melwitttoo late!22:02
alaskidansmith: yes. even in a multiple cells case all cells must match flavors so it shouldn't matter, but we assume a cell of one for now22:03
dansmithalaski: how many flavors at RAX? can you tell me in public?22:03
*** boden has quit IRC22:03
alaskitbh I don't know, but they're all in the public listing. maybe 30 or so?22:03
dansmithseriously? that's it?22:04
alaskiyeah, nothing too crazy22:04
*** vilobhmm111 has joined #openstack-nova22:04
dansmithso I mean22:04
melwittthat's an order of magnitude lower than I expected22:04
dansmithif we're talking about 30ish flavors, it's almost worth something kinda crazy to just migrate the whole she'bang in one go22:05
dansmithcorrect me if I'm wrong, but shouldn't we nearly be able to reuse all the existing db machinery and control what we're talking to just with the db connection in the context?22:05
*** alex_xu has quit IRC22:06
*** vilobhmm11 has quit IRC22:06
alaskithere were slight changes in the model so I don't think that'll work22:07
dansmithokay and we've already got the tables in the api db I guess?22:07
alaskiI do think it's worth looking at doing it in one go, but who knows what crazy flavor explosion someone has created for themselves22:08
alaskidansmith: yes22:08
*** rfolco has joined #openstack-nova22:08
dansmithyeah, I know IBM was auto-creating flavors by the pantload at one point22:08
*** davideagnello has quit IRC22:08
*** davideagnello has joined #openstack-nova22:09
*** alex_xu has joined #openstack-nova22:09
alaskiit should still be a quick migration, but we sort of set the precedent of doing everything online at this point22:10
dansmithalaski: so flavor has hardly ever changed. I wonder if it might be easier to set the current flavor aside and build a new flavor class at the current version or something22:10
*** shaohe_feng has quit IRC22:10
dansmithalaski: I didn't mean do it offline ahead of time22:10
melwittI also know of a case where there are a lot of flavors in a deployment22:10
alaskidansmith: ahh, gotcha22:11
dansmithI meant something online that just says "I'm your huckleberry" and just takes the bullet for the cloud and does the migration22:11
*** rfolco has quit IRC22:11
alaskidansmith: I would support building a new flavor22:11
dansmithactual doc quote optional22:11
alaskidansmith: +3 for Tombstone reference22:11
melwitthaha, nice22:11
*** pushkaru has quit IRC22:12
dansmithif matt is going to be ptl, he's going to be a lot less fun, so someone has to do it22:12
alaskiI'm onboard with both of those ideas22:12
alaskiheh, you need to be quick with the youtube links though22:12
alaskidinner has arrived, gotta run22:13
dansmithnobody commented on my final countdown link above22:13
* alaski actually goes22:14
melwittI didn't see a final countdown link22:14
*** pcm_ has joined #openstack-nova22:14
*** mattymo_ has joined #openstack-nova22:15
*** shaohe_feng_ has joined #openstack-nova22:16
*** mattymo has quit IRC22:16
*** pc_m has quit IRC22:16
*** dims_ has joined #openstack-nova22:16
*** shaohe_feng_ is now known as shaohe_feng22:16
*** erlon has quit IRC22:17
*** geekinutah has quit IRC22:17
auggydansmith: sorry, i was late to the game when i saw it but i appreciated it :D22:17
auggythis was pretty epic, I totally saw it -
*** dims has quit IRC22:18
auggyseeing that show made me want to learn to play the keytar22:19
*** fhermeni has quit IRC22:20
*** yamahata has joined #openstack-nova22:20
*** shaohe_feng has quit IRC22:21
*** shaohe_feng has joined #openstack-nova22:21
*** vilobhmm111 has quit IRC22:22
*** vilobhmm11 has joined #openstack-nova22:23
jaypipesmmm, my deck82 just arrived.22:23
auggyooh fancy :) I use a HHKB22:26
auggyi like the clackity clackity key sounds22:26
jaypipes^ typed on my deck82 ;)22:27
melwittdansmith: to your earlier question, the db connection in the context is only for main dbs, not the api db. unless you set the attr in the context to be the api db connection22:29
dansmithmelwitt: okay, well, the models differ a little anyway, so..22:30
*** shaohe_feng has quit IRC22:31
*** nic has left #openstack-nova22:31
*** nic has joined #openstack-nova22:31
*** shaohe_feng has joined #openstack-nova22:32
*** READ10 has quit IRC22:33
*** penick has joined #openstack-nova22:34
*** busterswt has quit IRC22:34
*** pushkaru has joined #openstack-nova22:35
dansmithno idea what that was22:35
dansmithmelwitt: what would you think about making the migration happen only on save(),22:35
*** penick has quit IRC22:36
dansmithand make save (a) write the new thing to do the api database and delete the thing from the original database before it returns/commits the transaction22:36
dansmithwe could make all the get commands look in the old db first, which means they might see an old flavor briefly if they race with the update22:37
dansmithbut that's okay because eventual consistency22:37
dansmithand then we can cache the flavors we've looked for and not found in the old db, so we don't look there anymore,22:37
dansmithwith some per-process startup flag of "are there any flavors in the old db" flag to know if we even need to check anymore22:37
dansmithquickly that flag will be False and we remove it in O22:38
dansmithhmm, there was no (b) in there22:38
melwittmakes sense to me. I like the idea of migrating on save only22:39
*** penick has joined #openstack-nova22:40
*** shaohe_feng has quit IRC22:41
*** shaohe_feng has joined #openstack-nova22:42
melwittdansmith: startup flag will just do a sweep to see if it gets anything for a flavor get all?22:43
*** edmondsw has quit IRC22:43
dansmithmelwitt: just a SELECT count() FROM flavors22:44
dansmithwhich should be super fast22:44
*** pumarani__ has joined #openstack-nova22:45
melwittdansmith: much better. that knows to ignore the soft deleted ones then22:46
*** doug-fis_ has joined #openstack-nova22:47
*** pushkaru has quit IRC22:47
*** yamahata has quit IRC22:48
*** ivase has quit IRC22:48
melwittor you just have the read_deleted='no'. anyway22:49
*** ivase has joined #openstack-nova22:49
*** Jeffrey4l has joined #openstack-nova22:49
dansmithoh god22:50
dansmithlots of conflicts22:50
*** doug-fish has quit IRC22:50
*** doug-fis_ has quit IRC22:51
*** shaohe_feng has quit IRC22:51
*** shaohe_feng has joined #openstack-nova22:52
*** xyang1 has quit IRC22:52
dansmithmelwitt: okay, you can have this one and I'll go find something else22:55
* melwitt runs for the hills22:56
*** yamahata has joined #openstack-nova22:57
*** mriedem has joined #openstack-nova22:57
*** jwcroppe has quit IRC22:59
*** pushkaru has joined #openstack-nova22:59
*** jwcroppe has joined #openstack-nova22:59
alaskidansmith: if you're just migrating on save() I assume you're planning some task to go through and save() each flavor?23:00
*** haomaiwang has quit IRC23:01
dansmithalaski: sure, just like we have for the other migrations23:01
dansmithalaski: you saw the generic command on nova-migrate we have now right?23:01
*** haomaiwa_ has joined #openstack-nova23:01
*** shaohe_feng has quit IRC23:02
alaskiflavors aren't saved that much compared to other objects so something like that would be needed23:02
*** karlamrhein has joined #openstack-nova23:02
dansmithalaski: you saw my half baked idea above then?23:02
*** davideagnello has quit IRC23:02
*** pumarani__ has quit IRC23:02
* dansmith waits for alaski to make poop on it23:03
*** shaohe_feng has joined #openstack-nova23:03
alaskiI think my only question is why look for flavors in the old db first?23:03
*** davideagnello has joined #openstack-nova23:04
*** pushkaru has quit IRC23:04
alaskiif you look in the api db first you don't need a flag to know when to switch23:04
dansmithalaski: well, I was thinking it would help in the case where you might need to abort23:04
*** jwcroppe has quit IRC23:04
*** pushkaru has joined #openstack-nova23:04
dansmithalaski: so, you'd never look in the new db unless it was missing from the old one, which means it's already written fully to the new one23:05
dansmithbecause the thing was brought up about having half-written the projects but not the flavor or whatever23:05
dansmithbut maybe that's silly23:05
*** vishwanathj has quit IRC23:06
alaskiI guess it depends on the complexity of writing the flavor and all it's associated rows atomically23:06
dansmithI guess if you can write the whole thing, flavor, projects, extra_specs in one transaction and make sure it lands, then it's okay23:06
dansmithI was just thinking that we kinda need atomicity over two databases,23:07
dansmithand while something could be half-created, it's easy to make sure it's deleted23:07
alaskiif you write to the api db first and only delete when done I think that could be two transactions23:08
alaskibut we would have to ensure it can't be half created23:08
alaskiI'm fine either way, just asking really23:08
dansmithsure, across the two, but depends on .. yeah, atomic create in the api23:08
*** yassine__ has quit IRC23:08
alaskilooking up in the cell and then slowly switching to the api db is surely easier than going the other way23:09
alaskiwith just the cost of a flag23:09
* alaski alaski hovered but did not poop on the idea23:10
dansmithwell, I think it's less complex to hit the api db first,23:10
dansmithassuming the create is atomic23:11
dansmithanyway, whatever23:11
dansmithI think we should migrate on save()23:11
*** krotscheck has joined #openstack-nova23:11
dansmithtake the hit there, which is infrequent and likely to be done completely by the task anyway23:11
dansmithand optimize for it being done quickly23:11
dansmithin fact,23:11
dansmithyou could also argue that the only thing we should do is:23:11
dansmithcreate in new db23:12
dansmithcheck first in new, fallback to old when getting via id23:12
*** shaohe_feng has quit IRC23:12
dansmithand only migrate from the tool23:12
alaskiright, that's what I was thinking originally23:12
dansmithI guess the deal is you could technically race with some update23:12
dansmithbut it's also probably reasonable to expect that admins could not update flavors for a minute while they run their tool23:13
*** shaohe_feng has joined #openstack-nova23:13
*** sdague has quit IRC23:14
dansmitheven with 300 flavors, you should be able to do that in nearly zero time23:14
alaskiagreed. it's not much data23:14
*** dimtruck is now known as zz_dimtruck23:15
openstackgerritDan Smith proposed openstack/nova: Flavor object which support both DB with active migration
openstackgerritDan Smith proposed openstack/nova: Use the api database fixture in tests
dansmithI rebased that, addressed a couple of the nits and split out the test base change ^23:16
alaskiawesome. I'm heading out for the evening but I'll take a look in the morning23:17
melwittdansmith: I just -1 the test change, that patch goes way back to before it already got added to the base test case23:17
dansmithrecess approved23:17
dansmithoh is it already there?23:17
melwittthat sentence barely made sense. it is now. it wasn't when this patch first started23:18
*** zenoway has quit IRC23:18
*** yamahata has quit IRC23:18
dansmithah I see it23:18
dansmithit didn't conflict and I didn't see it in the immediate vicinity23:19
*** fawadkhaliq has joined #openstack-nova23:19
*** gjayavelu has quit IRC23:19
dansmithcool, one more thing out of that mondo patch23:19
openstackgerritDan Smith proposed openstack/nova: Flavor object which support both DB with active migration
melwittyeah, it got caught up in the mix. I just realized it immediately when I saw the patch subject line23:19
*** samueldmq has joined #openstack-nova23:20
mriedemhmm, melwitt: dansmith: that bp was marked complete
dansmithalso, if we delete from the old db when we're done, we don't need to do much for the union operation I think23:21
*** anteaya has joined #openstack-nova23:21
dansmithmriedem: lol, very much not complete23:21
mriedemi'm going to re-open and then we need to re-propose the spec for newton23:22
melwitthm, that's odd23:22
*** shaohe_feng has quit IRC23:22
mriedemdansmith: should i assign the bp to you now?23:22
*** shaohe_feng has joined #openstack-nova23:23
melwittohh, I see the title was renamed to "(partial)"23:23
dansmithmriedem: I dunno, I asked the person if I could steal it from them, but no answer in the last 45 minutes :)23:23
melwittso I guess johnthetubaguy was thinking to open a new one for the remainder of the work?23:24
mriedemwell, the spec still points at this bp,23:24
mriedemand the spec has to be re-proposed23:24
*** akshai has quit IRC23:24
mriedemso i think we're ok with just keeping this one, please i just re-opened it23:24
*** karlamrhein has quit IRC23:25
*** shaohe_feng has quit IRC23:32
*** shaohe_feng has joined #openstack-nova23:33
*** nic has quit IRC23:34
*** rook-lappy has joined #openstack-nova23:34
*** mriedem1 has joined #openstack-nova23:35
*** achanda has quit IRC23:36
*** mriedem has quit IRC23:37
*** vilobhmm11 has quit IRC23:37
*** diogogmt has quit IRC23:38
*** vilobhmm11 has joined #openstack-nova23:38
*** tjones_zzzz has left #openstack-nova23:39
*** pushkaru has quit IRC23:40
*** ducttape_ has joined #openstack-nova23:41
*** shaohe_feng has quit IRC23:43
*** doug-fish has joined #openstack-nova23:43
*** shaohe_feng has joined #openstack-nova23:43
*** pece has quit IRC23:44
openstackgerritMatt Riedemann proposed openstack/nova-specs: Re-propose: Add Flavor tables to API Database
*** jwcroppe has joined #openstack-nova23:44
*** mylu has joined #openstack-nova23:44
*** ducttape_ has quit IRC23:47
*** eil397 has quit IRC23:49
*** vilobhmm11 has quit IRC23:53
*** shaohe_feng has quit IRC23:53
*** mylu has quit IRC23:53
*** salv-orl_ has joined #openstack-nova23:54
*** shaohe_feng has joined #openstack-nova23:54
*** MVenesio has joined #openstack-nova23:55
*** salv-orlando has quit IRC23:57
*** penick has quit IRC23:57
*** MVenesio has quit IRC23:59

Generated by 2.14.0 by Marius Gedminas - find it at!