Wednesday, 2015-09-23

*** hodos has quit IRC00:01
*** tsekiyam_ has joined #openstack-cinder00:04
*** jerrygb_ has joined #openstack-cinder00:05
*** mtanino has quit IRC00:07
*** annasort has joined #openstack-cinder00:08
*** tsekiyama has quit IRC00:08
*** jerrygb_ has quit IRC00:08
*** tsekiyam_ has quit IRC00:09
*** jerrygb has quit IRC00:09
*** rushil has quit IRC00:10
*** Guest18166 is now known as dims_00:17
*** RA has joined #openstack-cinder00:19
*** RA is now known as Guest3588600:19
*** _cjones_ has quit IRC00:22
*** leeantho has quit IRC00:26
*** garthb has quit IRC00:30
*** stevemar has joined #openstack-cinder00:33
*** stevemar has quit IRC00:36
*** takedakn has joined #openstack-cinder00:36
*** julim has quit IRC00:39
*** markvoelker has joined #openstack-cinder00:40
*** xyang1 has quit IRC00:42
*** p0rtal_ has joined #openstack-cinder00:48
*** p0rtal has quit IRC00:51
*** p0rtal_ has quit IRC00:52
*** p0rtal has joined #openstack-cinder00:52
*** Apoorva_ has joined #openstack-cinder00:56
*** p0rtal has quit IRC00:57
*** EinstCrazy has joined #openstack-cinder00:57
*** Apoorva has quit IRC00:59
*** Apoorva_ has quit IRC00:59
*** takedakn has quit IRC01:00
*** Apoorva has joined #openstack-cinder01:00
*** Apoorva has quit IRC01:03
*** Apoorva_ has joined #openstack-cinder01:03
*** Apoorva_ has quit IRC01:05
*** Apoorva has joined #openstack-cinder01:05
*** DericHorn-HP has joined #openstack-cinder01:07
*** Apoorva has quit IRC01:09
*** Apoorva has joined #openstack-cinder01:09
*** stevemar has joined #openstack-cinder01:11
*** DericHorn-HP has quit IRC01:17
*** p0rtal has joined #openstack-cinder01:20
*** p0rtal has quit IRC01:25
*** DericHorn-HP has joined #openstack-cinder01:27
*** Lee1092 has joined #openstack-cinder01:28
*** rushil has joined #openstack-cinder01:38
*** theanalyst has quit IRC01:40
*** theanalyst has joined #openstack-cinder01:42
*** jerrygb has joined #openstack-cinder01:44
*** jerrygb has quit IRC01:45
*** jerrygb has joined #openstack-cinder01:45
*** p0rtal has joined #openstack-cinder01:45
*** Apoorva has quit IRC01:46
*** salv-orlando has joined #openstack-cinder01:46
*** akerr has joined #openstack-cinder01:48
*** rushil has quit IRC01:48
*** davechen has joined #openstack-cinder01:49
*** salv-orlando has quit IRC01:56
*** dims_ has quit IRC01:57
*** dims_ has joined #openstack-cinder01:57
*** haomaiwang has joined #openstack-cinder01:58
*** rushil has joined #openstack-cinder01:58
*** rushil_ has joined #openstack-cinder01:59
*** haomaiwang has quit IRC02:01
*** haomaiwang has joined #openstack-cinder02:01
*** rushil has quit IRC02:03
*** akerr has quit IRC02:04
*** DericHorn-HP has quit IRC02:04
*** DericHorn-HP has joined #openstack-cinder02:05
*** DericHorn-HP has quit IRC02:06
*** DericHorn-HP has joined #openstack-cinder02:07
*** dims_ has quit IRC02:10
*** dims_ has joined #openstack-cinder02:13
*** dims__ has joined #openstack-cinder02:18
*** dims_ has quit IRC02:18
*** bswartz has joined #openstack-cinder02:19
*** dims__ has quit IRC02:23
openstackgerritLisaLi proposed openstack/cinder: Copy encryptors from Nova to Cinder
openstackgerritLisaLi proposed openstack/cinder: Copy encryptors from Nova to Cinder
*** DericHorn-HP has quit IRC02:34
*** DericHorn-HP has joined #openstack-cinder02:37
tbarronjgriffith: jungleboyj: just catching up with backlog.  Thanks for the help.02:39
openstackgerritWilson Liu proposed openstack/cinder: Fix capacity report error in Huawei driver
*** dobson has quit IRC02:47
*** p0rtal_ has joined #openstack-cinder02:49
*** p0rtal has quit IRC02:52
*** dobson has joined #openstack-cinder02:53
openstackgerritWilson Liu proposed openstack/cinder: Fix capacity report error in Huawei driver
*** haomaiwang has quit IRC03:01
*** markvoelker has quit IRC03:01
*** haomaiwa_ has joined #openstack-cinder03:01
openstackgerritVincent Hou proposed openstack/cinder: Remove the destination volume check in delete_volume
openstackgerritWilson Liu proposed openstack/cinder: Change "Storage" to "LUN" in Huawei driver
*** DericHorn-HP has quit IRC03:04
*** DericHorn-HP has joined #openstack-cinder03:05
openstackgerritWilson Liu proposed openstack/cinder: Encode username and password in Huawei driver
*** boris-42 has quit IRC03:09
*** david-lyle has joined #openstack-cinder03:13
openstackgerritchenying proposed openstack/python-cinderclient: Fix three resources not being deleted by using name
*** markvoelker has joined #openstack-cinder03:16
*** zhangjn_ has joined #openstack-cinder03:18
*** EinstCrazy has quit IRC03:19
*** zhangjn has quit IRC03:20
*** DericHorn-HP has quit IRC03:38
*** DericHorn-HP has joined #openstack-cinder03:41
openstackgerritWilson Liu proposed openstack/cinder: Fix URL format in Huawei driver
*** dims_ has joined #openstack-cinder03:47
*** diogogmt has quit IRC03:48
*** rushil_ has quit IRC03:53
*** dims_ has quit IRC03:54
*** jerrygb has quit IRC03:56
*** garthb has joined #openstack-cinder03:57
*** DericHorn-HP has quit IRC03:59
*** haomaiwa_ has quit IRC04:01
*** haomaiwang has joined #openstack-cinder04:01
*** changbl has quit IRC04:04
*** changbl has joined #openstack-cinder04:09
*** jwcroppe has joined #openstack-cinder04:14
*** shyama has joined #openstack-cinder04:16
openstackgerritCedric Zhuang proposed openstack/cinder: Add retype logic in manage_existing for VNX
*** salv-orlando has joined #openstack-cinder04:17
*** zhangjn_ has quit IRC04:24
*** zhangjn has joined #openstack-cinder04:25
*** zhangjn_ has joined #openstack-cinder04:28
*** zhangjn has quit IRC04:29
*** zhangjn_ has quit IRC04:32
*** zhangjn has joined #openstack-cinder04:33
*** markstur_ has joined #openstack-cinder04:33
*** salv-orlando has quit IRC04:35
*** zhangjn_ has joined #openstack-cinder04:39
*** markstur_ has left #openstack-cinder04:41
*** zhangjn has quit IRC04:42
*** zhangjn_ has quit IRC04:42
*** zhangjn has joined #openstack-cinder04:43
*** BharatK has joined #openstack-cinder04:45
*** harlowja_at_home has joined #openstack-cinder04:51
*** garthb has quit IRC04:55
*** deepakcs has joined #openstack-cinder04:56
*** zhangjn has quit IRC04:57
*** zhangjn has joined #openstack-cinder04:59
*** Zhongjun_ has joined #openstack-cinder04:59
*** haomaiwang has quit IRC05:01
*** haomaiwang has joined #openstack-cinder05:01
*** asselin_ has quit IRC05:08
*** harlowja_at_home has quit IRC05:14
*** edtubill has joined #openstack-cinder05:15
*** p0rtal_ has quit IRC05:17
*** p0rtal has joined #openstack-cinder05:17
*** p0rtal has quit IRC05:22
*** BharatK has quit IRC05:24
*** Zhongjun_ is now known as Zhongjun05:28
*** BharatK has joined #openstack-cinder05:40
*** salv-orlando has joined #openstack-cinder05:45
*** boris-42 has joined #openstack-cinder05:47
*** dims_ has joined #openstack-cinder05:50
*** edtubill has quit IRC05:50
*** salv-orlando has quit IRC05:52
*** dims_ has quit IRC05:55
openstackgerritAnkit Agrawal proposed openstack/cinder: Remove duplicate keys from dictionary
*** haomaiwang has quit IRC06:01
*** haomaiwang has joined #openstack-cinder06:01
*** chenying has joined #openstack-cinder06:03
*** jerrygb has joined #openstack-cinder06:08
*** jerrygb has quit IRC06:12
*** mudassirlatif has quit IRC06:12
*** bswartz has quit IRC06:18
*** Guest35886 has quit IRC06:18
*** bswartz has joined #openstack-cinder06:21
*** nkrinner has joined #openstack-cinder06:21
*** stevemar has quit IRC06:23
*** jwcroppe has quit IRC06:23
*** jwcroppe has joined #openstack-cinder06:23
*** jwcroppe has quit IRC06:28
*** vlaza has joined #openstack-cinder06:31
*** EinstCrazy has joined #openstack-cinder06:32
openstackgerritCedric Zhuang proposed openstack/cinder: Add retype logic in manage_existing for VNX
*** haomaiwang has quit IRC07:01
*** haomaiwang has joined #openstack-cinder07:01
*** lpetrut has joined #openstack-cinder07:04
openstackgerritWilson Liu proposed openstack/cinder: Huawei driver add check before use a QoS
*** vgridnev has quit IRC07:07
*** Zhongjun has quit IRC07:07
*** markvoelker has quit IRC07:15
*** stevemar has joined #openstack-cinder07:23
*** jwcroppe has joined #openstack-cinder07:23
*** stevemar has quit IRC07:27
*** cfriesen__ has quit IRC07:28
*** ronis has joined #openstack-cinder07:32
*** sayali has quit IRC07:34
*** lpetrut has quit IRC07:36
*** sayali has joined #openstack-cinder07:38
*** chenying has quit IRC07:38
*** chenying has joined #openstack-cinder07:39
*** anshul has joined #openstack-cinder07:40
*** salv-orlando has joined #openstack-cinder07:40
*** jordanP has joined #openstack-cinder07:41
*** salv-orl_ has joined #openstack-cinder07:47
*** salv-orlando has quit IRC07:48
*** dims_ has joined #openstack-cinder07:52
*** bluex-pl has joined #openstack-cinder07:52
*** bluex-pl has quit IRC07:53
*** bluex-pl has joined #openstack-cinder07:53
*** bluex-pl has quit IRC07:58
*** salv-orl_ has quit IRC07:59
*** salv-orlando has joined #openstack-cinder07:59
*** dims_ has quit IRC07:59
*** haomaiwang has quit IRC08:01
*** haomaiwang has joined #openstack-cinder08:01
*** markus_z has joined #openstack-cinder08:06
*** haomaiwang has quit IRC08:10
*** haomaiwa_ has joined #openstack-cinder08:12
*** aix has joined #openstack-cinder08:15
*** markvoelker has joined #openstack-cinder08:15
*** jwcroppe has quit IRC08:19
openstackgerritWilson Liu proposed openstack/cinder: Encode username and password in Huawei driver
*** markvoelker has quit IRC08:20
*** vgridnev has joined #openstack-cinder08:22
*** jith_ has joined #openstack-cinder08:53
*** haomaiwa_ has quit IRC09:01
*** 16WAAB8UT has joined #openstack-cinder09:01
*** BharatK has quit IRC09:06
*** jerrygb has joined #openstack-cinder09:08
*** lpetrut has joined #openstack-cinder09:09
*** dims_ has joined #openstack-cinder09:11
*** jerrygb has quit IRC09:13
*** e0ne has joined #openstack-cinder09:15
*** jwcroppe has joined #openstack-cinder09:16
*** BharatK has joined #openstack-cinder09:20
*** stevemar has joined #openstack-cinder09:25
*** stevemar has quit IRC09:28
*** alexpilotti has joined #openstack-cinder09:32
*** ociuhandu has quit IRC09:36
*** dims_ has quit IRC09:44
*** jwcroppe_ has joined #openstack-cinder09:47
*** davechen has left #openstack-cinder09:48
*** e0ne has quit IRC09:49
*** jwcroppe has quit IRC09:49
*** e0ne has joined #openstack-cinder09:50
openstackgerritCedric Zhuang proposed openstack/cinder: Add retype logic in manage_existing for VNX
openstackgerritTina Tang proposed openstack/cinder: Create volume in cg enhancement in VNX driver
*** bluex-pl has joined #openstack-cinder09:55
*** bluex-pl has quit IRC09:56
*** bluex-pl has joined #openstack-cinder09:56
*** 16WAAB8UT has quit IRC10:01
*** 6JTAB3MXI has joined #openstack-cinder10:01
*** BharatK has quit IRC10:04
*** dims_ has joined #openstack-cinder10:09
*** ndipanov has quit IRC10:09
*** ndipanov has joined #openstack-cinder10:11
*** IanGovett has joined #openstack-cinder10:16
*** markvoelker has joined #openstack-cinder10:17
*** BharatK has joined #openstack-cinder10:17
*** markvoelker has quit IRC10:21
openstackgerritWilson Liu proposed openstack/cinder: Encode username and password in Huawei driver
*** chenying has quit IRC10:26
*** ociuhandu has joined #openstack-cinder10:29
*** jwcroppe_ has quit IRC10:30
*** jwcroppe has joined #openstack-cinder10:31
*** jwcroppe has quit IRC10:35
dulekgeguileo: Hi! I've been out of sync for a while, but I'm getting the feeling that I understand the approach to c-vol A/A in Mitaka.10:43
dulekgeguileo: Do we have an idea how to tackle cleanup phase in init_host correctly?10:43
dulekgeguileo: Right now on restart we'll get our volume deleted even if other node is taking care of it.10:44
geguileodulek: I have an idea, yes10:44
dulekgeguileo: Are we jumping into your "simpler" approach, or there's something else on the wall?10:44
geguileodulek: I don't know if anyone has a better idea10:44
geguileodulek: But I haven't heard any other thoughts on it10:45
* dulek is getting into careful review of the idea then.10:45
dulekA spec is probably needed for that, isn't it?10:46
geguileodulek: The cleanup should not add cluster@backend when queuing the job10:48
geguileodulek: As this would conflict with default cluster value (equals host)10:48
*** akerr has joined #openstack-cinder10:51
*** EinstCrazy has quit IRC10:51
*** zhangjn has quit IRC10:51
*** akerr_ has joined #openstack-cinder10:55
*** akerr has quit IRC10:58
*** 6JTAB3MXI has quit IRC11:01
*** haomaiwang has joined #openstack-cinder11:01
*** aix has quit IRC11:03
openstackgerritWilson Liu proposed openstack/cinder: Change "Storage" to "LUN" in Huawei driver
*** salv-orlando has quit IRC11:12
*** salv-orlando has joined #openstack-cinder11:13
dulekgeguileo: Sorry, needed to switch context. You're referring to some changes in approach described in the blogpost? Or directly to it?11:17
geguileodulek: Something that the post says I'd do, but later realized that it would create problems11:17
*** markvoelker has joined #openstack-cinder11:17
dulekgeguileo: Ok, I'll have that in mind while rereading it.11:18
geguileodulek: Ok11:18
*** jwcroppe has joined #openstack-cinder11:21
*** markvoelker has quit IRC11:22
*** haomaiwang has quit IRC11:22
*** shakamunyi has quit IRC11:25
*** akerr_ has quit IRC11:25
*** akerr has joined #openstack-cinder11:28
*** akerr_ has joined #openstack-cinder11:30
*** akerr has quit IRC11:33
*** aix has joined #openstack-cinder11:34
*** jerrygb has joined #openstack-cinder11:35
*** jerrygb has quit IRC11:39
*** jith_ has quit IRC11:46
*** jerrygb has joined #openstack-cinder11:48
*** cristicalin has joined #openstack-cinder11:49
*** EinstCrazy has joined #openstack-cinder11:49
*** sayali has quit IRC11:50
*** sayali has joined #openstack-cinder11:55
*** EinstCrazy has quit IRC11:59
*** krtaylor has quit IRC12:01
*** zhangjn has joined #openstack-cinder12:04
*** EinstCrazy has joined #openstack-cinder12:04
*** ankit_ag has joined #openstack-cinder12:04
*** EinstCrazy has quit IRC12:06
*** Zhongjun has joined #openstack-cinder12:12
*** krtaylor has joined #openstack-cinder12:13
*** e0ne has quit IRC12:13
*** deepakcs has quit IRC12:15
*** markvoelker has joined #openstack-cinder12:20
*** timcl has joined #openstack-cinder12:21
*** raildo-afk is now known as raildo12:22
*** jwcroppe_ has joined #openstack-cinder12:22
*** zhenguo has quit IRC12:24
*** jwcroppe has quit IRC12:25
*** shyama_ has joined #openstack-cinder12:26
*** takedakn has joined #openstack-cinder12:27
*** shyama has quit IRC12:28
*** shyama_ is now known as shyama12:28
*** jwcroppe_ is now known as jwcroppe12:29
*** porrua has joined #openstack-cinder12:30
*** bswartz has quit IRC12:32
*** edmondsw has joined #openstack-cinder12:32
*** diogogmt has joined #openstack-cinder12:32
*** diogogmt has quit IRC12:33
openstackgerritwanghao proposed openstack/cinder: CG creation should be scheduled on backend level
*** takedakn has quit IRC12:41
*** takedakn has joined #openstack-cinder12:42
*** takedakn has quit IRC12:44
dulekgeguileo: - I think you've fixed that accidentally. Should we think of a Kilo backport?12:46
openstackLaunchpad bug 1498886 in Cinder "detail snapshot list API return empty object for metadata" [Undecided,New]12:46
*** sayali has quit IRC12:47
*** martyturner has joined #openstack-cinder12:47
dulekgeguileo: Ah, maybe not, you've just moved isinstance check somewhere else. Anyway I cannot reproduce that in Liberty.12:47
geguileodulek: If I accidentaly fixed it moving it to the view builder I don't think that will be allowed to be backported...12:48
*** takedakn has joined #openstack-cinder12:49
*** sayali has joined #openstack-cinder12:50
geguileodulek: That actually may have been fixed by the snapshot versioned object12:50
dulekgeguileo: Snapshot object got merged in Kilo.12:51
geguileodulek: Mmmm, that's weird then (that you cannot reproduce it)12:51
geguileoMore info is needed to reproduce it12:51
dulekgeguileo: I cannot in Liberty, maybe it is reproducible in Kilo.12:52
geguileoMaybe, but that is a little bit weird bug r:-?12:53
geguileodulek: By the way, thanks for catching that bug in the conditional update patch!!12:54
geguileoWell, it's not a bug because it doesn't fail, but it's bad coding anyway  XD12:54
*** gouthamr has joined #openstack-cinder12:56
*** tpsilva has joined #openstack-cinder12:57
*** bswartz has joined #openstack-cinder12:57
*** timcl has quit IRC12:59
*** martyturner has quit IRC13:04
*** martyturner has joined #openstack-cinder13:07
*** salv-orlando has quit IRC13:07
*** haomaiwang has joined #openstack-cinder13:08
*** Yogi1 has joined #openstack-cinder13:10
*** julim has joined #openstack-cinder13:11
*** mdenny has quit IRC13:19
*** dims_ has quit IRC13:21
*** dims_ has joined #openstack-cinder13:21
*** alexpilotti has quit IRC13:22
*** ankit_ag has quit IRC13:23
*** jungleboyj has joined #openstack-cinder13:24
*** takedakn has quit IRC13:25
*** jwcroppe has quit IRC13:26
*** stevemar has joined #openstack-cinder13:26
*** jgregor has joined #openstack-cinder13:28
*** dustins has joined #openstack-cinder13:28
*** merooney has joined #openstack-cinder13:28
*** stevemar has quit IRC13:29
*** martyturner has quit IRC13:31
*** jaypipes has joined #openstack-cinder13:36
*** lprice has joined #openstack-cinder13:39
*** xyang1 has joined #openstack-cinder13:39
*** cfriesen__ has joined #openstack-cinder13:41
*** lprice1 has joined #openstack-cinder13:45
*** mc_nair has joined #openstack-cinder13:45
*** lprice has quit IRC13:46
*** jwcroppe has joined #openstack-cinder13:49
*** e0ne has joined #openstack-cinder13:51
*** BharatK has quit IRC13:52
*** stevemar has joined #openstack-cinder13:55
*** delatte has quit IRC13:56
*** delattec has quit IRC13:58
*** haomaiwang has quit IRC14:01
*** haomaiwang has joined #openstack-cinder14:01
*** jerrygb has quit IRC14:03
*** Yogi11 has joined #openstack-cinder14:03
*** martyturner has joined #openstack-cinder14:04
*** Yogi1 has quit IRC14:07
*** jerrygb has joined #openstack-cinder14:07
*** mdenny has joined #openstack-cinder14:07
*** takedakn has joined #openstack-cinder14:08
*** breitz has quit IRC14:10
*** breitz has joined #openstack-cinder14:10
*** rhagarty__ has joined #openstack-cinder14:12
*** lixiaoy1 has joined #openstack-cinder14:12
*** mtanino has joined #openstack-cinder14:13
*** rhagarty_ has quit IRC14:15
*** rhagarty_ has joined #openstack-cinder14:15
*** rhagarty has quit IRC14:16
*** rushil has joined #openstack-cinder14:22
*** edtubill has joined #openstack-cinder14:23
*** eharney has joined #openstack-cinder14:26
*** asselin_ has joined #openstack-cinder14:27
*** zhipeng has joined #openstack-cinder14:33
*** cdelatte has joined #openstack-cinder14:36
*** delattec has joined #openstack-cinder14:36
*** takedakn has quit IRC14:37
*** wilson1 has joined #openstack-cinder14:38
*** anshul has quit IRC14:38
*** hodos has joined #openstack-cinder14:39
*** salv-orlando has joined #openstack-cinder14:40
*** stevemar has quit IRC14:40
*** Guest75391 has quit IRC14:41
dulekgeguileo: I think "simpler" solution is missing a part on reader locks without a DLM.14:43
*** r-daneel has joined #openstack-cinder14:43
*** aarefiev has quit IRC14:43
geguileodulek: I don't remember right now, but since locks will be pushed down to drivers14:44
geguileodulek: We just need to remove 4 locks from manager (if I remember correctly)14:44
dulekgeguileo: Yeah, these are the problematic ones.14:44
geguileodulek: And I looked at several options some time ago, but can't remember specifics right now14:44
dulekgeguileo: How about just having a constraint that we cannot remove a snapshot if there's a volume in creating state that are originating from it?14:44
dulekgeguileo: With your cleanup solution we get that cleared into error eventually.14:45
*** salv-orlando has quit IRC14:45
dulekgeguileo: That would mean - we don't have 'reading' state and don't need to worry about cleaning it up.14:45
geguileodulek: Yeah, probably we'll handle half of that in API and we'll have to check if there are more cases14:46
*** delattec has quit IRC14:46
*** cdelatte has quit IRC14:46
dulekgeguileo: That would be for the delete_volume and delete_snapshot stuff. I'm not sure on detach/attach locks.14:46
geguileoI have to go back and have a look at the whole solution14:46
geguileoThose have no problem when comming from Nova14:46
*** stevemar has joined #openstack-cinder14:46
geguileoThe problem is when we do attach/detach inside Cinder14:47
geguileoI have a couple of ideas, but have to look more in depth into the code14:47
*** haomaiwang has quit IRC14:47
openstackgerritScott DAngelo proposed openstack/cinder-specs: cinder-api-microversions
*** cdelatte has joined #openstack-cinder14:48
*** delattec has joined #openstack-cinder14:48
*** baumann has joined #openstack-cinder14:48
dulekgeguileo: Are you planning to organize that into a spec?14:48
*** aarefiev has joined #openstack-cinder14:49
geguileodulek: Yes14:49
dulekgeguileo: Discussion would be probably easier then.14:49
geguileodulek: But I want to finish with a sqlalchemy bug that is blocking my conditional updates14:49
*** garthb has joined #openstack-cinder14:50
*** porrua has quit IRC14:50
dulekgeguileo: Thanks! Great to be in sync again! :)14:50
eharneydoes this error message ring a bell for anyone?14:51
eharneyCaught error: (MissingDependencies(u"'taskflow.patterns.linear_flow.Flow: volume_create_api(len=5)' requires ['reservations', 'volume_id', 'volume_properties'] but no other entity produces said requirements"14:51
dulekeharney: It doesn't, but can you reveal line number?14:51
dulekeharney: Is it on master?14:52
eharneysomeone filed a bug about this from our liberty packages, but it looks like a bug in our use of taskflow etc14:53
eharneyi don't have a line number14:53
eharney(our liberty packages downstream)14:53
dulekeharney: Ah, so it's unlikely same happens on master?14:54
eharneydulek: well, these packages are a couple days old at best, so... i bet there's some way to make it happen14:55
eharneyi'll try to find some more useful info, was just hoping someone recognized the error signature14:55
*** porrua has joined #openstack-cinder14:55
dulekeharney: Okay, I'll look for the problem.14:55
*** merooney has quit IRC14:57
*** merooney has joined #openstack-cinder15:01
*** zhangjn has quit IRC15:05
dulekeharney: In current state of master such bug shouldn't be possible to happen.15:05
dulekeharney: Every requirement for a task seems satisfied.15:05
dulekeharney: Does it happen on volume creation?15:07
openstackgerritKendall Nelson proposed openstack/cinder: WIP:(until Mitaka opens) Missing configuration opts from cinder.sample.conf
eharneydulek: yes -- i'm trying to figure out what Cinder git sha this correlates to... not really familiar w/ delorean packages15:09
*** ronis has quit IRC15:09
dulekeharney: I've got to go away for a moment, but feel free to ask questions anytime. I've refactored all the TaskFlow stuff related to vol creation in L, so I *should* understand that (am I really? ;)).15:10
smcginnishemna: Morning15:10
smcginniseharney: Is there a bug link for this?15:10
eharneysmcginnis: just the RDO one at bug 1265690 in openstack-cinder "Failed to create cinder volume using Delorean Trunk packages" [Unspecified,New] - Assigned to eharney15:11
hemnaDelorean ?15:11
hemnaI thought they stopped making those cars15:11
smcginnisWow, how many different bug trackers is the OpenStack bot handling. :)15:11
dulek"Delorean: Time Travel Made Easy", I really like that. :D15:12
eharneysmcginnis: yeah, i wasn't really expecting it to yell on that one15:12
hemnadulek, hah!15:12
dulekThough I'm not sure if that's what eharney meant. ;)15:12
smcginnisSo kilo code but an uncapped library?15:12
smcginnisOh, nope, now I see liberty.15:13
*** Apoorva has joined #openstack-cinder15:13
*** akerr_ is now known as akerr15:15
smcginnishemna: Done.15:16
*** ronis has joined #openstack-cinder15:17
hemnadulek, ping15:18
*** garthb has quit IRC15:20
*** garthb has joined #openstack-cinder15:20
*** mudassirlatif has joined #openstack-cinder15:22
*** salv-orlando has joined #openstack-cinder15:22
*** salv-orlando has quit IRC15:22
*** martyturner has quit IRC15:23
*** salv-orlando has joined #openstack-cinder15:23
*** tsekiyama has joined #openstack-cinder15:24
*** pots has quit IRC15:25
openstackgerritxing-yang proposed openstack/cinder: Fix VMAX live migration problem
*** vlaza1 has joined #openstack-cinder15:27
*** ronis has quit IRC15:27
*** Apoorva has quit IRC15:30
*** rushil has quit IRC15:31
*** delattec has quit IRC15:31
*** cdelatte has quit IRC15:31
*** vlaza has quit IRC15:31
*** jdurgin1 has joined #openstack-cinder15:31
*** alejandrito has joined #openstack-cinder15:33
openstackgerritThierry Carrez proposed openstack/cinder: Open Mitaka development
*** vlaza1 has left #openstack-cinder15:38
*** zhipeng has quit IRC15:38
*** thingee has quit IRC15:40
*** cristicalin has quit IRC15:42
*** thingee has joined #openstack-cinder15:42
*** rushil has joined #openstack-cinder15:45
*** e0ne has quit IRC15:47
*** rushil has quit IRC15:47
*** martyturner has joined #openstack-cinder15:48
*** e0ne has joined #openstack-cinder15:48
*** merooney has quit IRC15:49
*** adurbin__ has joined #openstack-cinder15:50
*** _cjones_ has joined #openstack-cinder15:52
openstackgerritxing-yang proposed openstack/cinder: Fix VMAX live migration problem
*** lixiaoy1 has quit IRC15:52
*** _cjones_ has quit IRC15:52
*** _cjones_ has joined #openstack-cinder15:52
openstackgerritxing-yang proposed openstack/cinder: Fix VMAX live migration problem
*** sayali has quit IRC15:55
*** rajinir has joined #openstack-cinder15:56
*** esker has joined #openstack-cinder15:56
*** dulek_ has joined #openstack-cinder15:57
*** dulek_ has quit IRC15:57
*** vgridnev has quit IRC15:57
*** nkrinner has quit IRC15:57
dulekhemna: pong15:59
*** garthb_ has joined #openstack-cinder16:00
*** garthb has quit IRC16:02
*** Apoorva has joined #openstack-cinder16:03
*** p0rtal has joined #openstack-cinder16:03
*** p0rtal has quit IRC16:04
*** p0rtal has joined #openstack-cinder16:04
*** martyturner has quit IRC16:05
*** esker has quit IRC16:06
*** leeantho has joined #openstack-cinder16:07
*** sayali has joined #openstack-cinder16:08
*** martyturner has joined #openstack-cinder16:10
*** jungleboyj_ has joined #openstack-cinder16:11
*** resker has joined #openstack-cinder16:12
*** jungleboyj has quit IRC16:12
*** martyturner1 has joined #openstack-cinder16:14
*** martyturner has quit IRC16:15
*** resker has quit IRC16:16
*** amit213 has quit IRC16:17
*** amit213 has joined #openstack-cinder16:17
*** jdurgin1 has quit IRC16:19
*** pv_ has joined #openstack-cinder16:22
*** jwcroppe has quit IRC16:25
*** p0rtal has quit IRC16:25
*** jordanP has quit IRC16:29
*** garthb_ has quit IRC16:30
*** garthb has joined #openstack-cinder16:30
*** Yogi1 has joined #openstack-cinder16:32
*** ronis has joined #openstack-cinder16:34
*** Yogi11 has quit IRC16:34
*** rhedlind-m has joined #openstack-cinder16:38
*** diogogmt has joined #openstack-cinder16:40
*** daneyon has joined #openstack-cinder16:43
*** daneyon_ has quit IRC16:43
pv_ hey i keep running into this error while testing the copy_image_to_volume method16:47
pv_has anyone seen this?16:47
pv_sorry for the weird log formatting16:47
*** delattec has joined #openstack-cinder16:47
*** cdelatte has joined #openstack-cinder16:47
*** delattec has quit IRC16:50
*** cdelatte has quit IRC16:50
*** vgridnev has joined #openstack-cinder16:50
*** salv-orlando has quit IRC16:51
*** rhagarty__ has quit IRC16:51
*** salv-orlando has joined #openstack-cinder16:52
*** rhagarty_ has quit IRC16:52
*** aix has quit IRC16:53
*** merooney has joined #openstack-cinder16:54
*** jwcroppe has joined #openstack-cinder16:55
*** rushil has joined #openstack-cinder16:58
*** salv-orl_ has joined #openstack-cinder16:59
*** baumann has left #openstack-cinder17:00
*** salv-orlando has quit IRC17:01
jgriffithpv_: no, but make sure your filesystem isn't full17:01
*** salv-orlando has joined #openstack-cinder17:02
jgriffithalthough... hmm, that's probably not it17:02
*** angela-s has joined #openstack-cinder17:03
smcginnispv_: We had some weird CI failures that appeared to be caused by the network.17:04
smcginnispv_: I think it caused IO failures that would bubble up to odd errors like saying the volume was too small.17:04
smcginnispv_: Maybe something similar?17:04
*** jungleboyj_ has quit IRC17:04
*** salv-orl_ has quit IRC17:04
pv_hmm ok, yeah it may be17:04
pv_i originally though the volumes were too small17:04
pv_but that doesnt seem to be it17:05
*** amoturi has joined #openstack-cinder17:05
patrickeastcould be that something is happening to the temporary image file while the convert is going on17:05
pv_i see this "NOTE(avishay): I'm not crazy about creating temp files which may be large and cause disk full errors which would confuse users. Unfortunately it seems that you can't pipe to 'qemu-img convert' because it seeks.17:06
pv_in the comments of the fetch_to_raw method which is part of my traceback17:06
pv_could that be it?17:06
smcginnispv_: Maybe run df while it's happening and see if you are running low?17:07
*** timcl has joined #openstack-cinder17:07
jgriffithpv_: just run df now17:10
jgriffithpv_: how much free-space do you have?17:10
jgriffithpv_: and how big of a volume/image are you trying to create?17:10
pv_df output
*** jungleboyj_ has joined #openstack-cinder17:11
pv_and 100gb recognized by cinder, but in actuality only 100mb because of testing i havent made the units from cinder trickle down into our backend17:12
pv_so i type in cinder create --image-id <image id> 100, but thats because i need a 100 MB volume created on our backend17:12
pv_for the image i usually use cirros-0.3.4-x86_64-uec17:12
jgriffithpv_: oh... so this isn't necessarily a bug, but you can't get your driver that you're developing to work17:12
jgriffiththat's different :)17:12
jgriffithpv_: in that case, put a break in that conversion method and make SURE that your volume is actually attached17:13
jgriffithpv_: and that you can dd to it17:13
pv_yeah i guess? haha i'm using the base driver's copy_image_to_volume17:13
pv_yeah i can dd to it17:13
asselinpatrickeast, we've been seeing issues with FC passthrough17:14
patrickeastasselin: uh oh, new problems?17:14
asselinpatrickeast, somehow the virst command fails and returns error code 0, previously it was error code 117:14
patrickeastoh weird17:15
*** ociuhandu has quit IRC17:15
patrickeastjust new return code? or the passthrough isn't working?17:15
asselinpatrickeast, well, after the 1st one, the others fail17:15
asselinb/c the passthrough script expects non-zero to look at the 2nd hba17:15
*** esker has joined #openstack-cinder17:15
*** e0ne has quit IRC17:16
asselinI though it was new version of virsh b/c we updated our blades, but they're all 1.2.217:16
mriedemi'm curious as to what the expected behavior would be here?17:16
openstackLaunchpad bug 1496235 in OpenStack Compute (nova) "Boot from volume faild with availability_zone option, in case of cinder do not have availability_zone" [Undecided,New] - Assigned to Ikuo Kumagai (i-kumagai)17:16
patrickeasthmm so why would it start returning different codes?17:16
patrickeastlemmie go check mine and see if its doing the same thing17:16
asselinpatrickeast, no idea....and why would it return success 0 when it spits out an error ?17:17
smcginnisasselin: Did you change it to pipe through something else. Shot myself in the foot once because I piped through tee that will always return 0.17:17
asselinsmcginnis, no, we pin to a version of the script17:17
patrickeastasselin: mine looks ok,
patrickeastasselin: is this happening consistently for you guys?17:18
asselinpatrickeast, yes17:18
patrickeastalthough i am using a slightly different passthrough script17:18
* patrickeast remembers that i still have to go update that review that has been sitting there for weeks17:19
asselinsmcginnis, if you have ideas to help, le me konw ^^17:21
smcginnisasselin: Sure, I'll take a look.17:21
pv_my attach and detach methods seem to be working fine17:21
mriedemif you're going to boot an instance with a volume created from an image (nova does the volume create request), and you're specifying an az and that az isn't in cinder, the volume create is going to fail17:22
scottdamriedem: Others could answer better than I, but I think if you can create a c-vol without a Cinder AZ, you should be able to boot-from-vol without it.17:22
mriedemscottda: well the problem is knowing if the az exists in cinder17:22
jgriffithmriedem: you're correct17:22
scottdayeah, you're right.17:22
mriedemso that's going to barf17:22
mriedemand hten nova barfs17:22
mriedemand then i reject your bug17:22
jgriffithmriedem: scottda the root of the problem is Cinder never really did AZ's17:23
jgriffithmriedem: scottda we've always just had one AZ and played dumb17:23
jgriffithwhich used to work ok, because we basically ignored it on both sides17:24
patrickeastasselin: weird, so that part of the code is exactly the same between the script i'm using and the one checked in17:24
asselinpatrickeast, actually I think I know what could have cause it17:25
patrickeastasselin: but the logs are showing different return codes for basically the same error17:25
*** bluex-pl has quit IRC17:25
asselinpatrickeast, I added some ssh restrictions for that key17:25
*** raildo is now known as raildo-afk17:25
scottdaIt seems we could/should do something on the Cinder side for the AZ issue, like possibly ignore the parameter if cinder doesn't have an AZ17:25
*** garthb has quit IRC17:26
*** garthb has joined #openstack-cinder17:26
pv_hmm it seems [01;31m2015-09-23 09:42:09.816 TRACE cinder.volume.flows.manager.create_volume [01;35m[00mProcessExecutionError: Unexpected error while running command. [01;31m2015-09-23 09:42:09.816 TRACE cinder.volume.flows.manager.create_volume [01;35m[00mCommand: sudo cinder-rootwrap /etc/cinder/rootwrap.conf qemu-img convert -t none -O raw /opt/stack/data/cinder/conversion/tmpSFP7fM /dev/disk/by-path/pci-0000:02:00.0-fc-0x1017:27
pv_the rootwrap command is throwing an error17:27
eharneypv_: qemu-img was throwing an error indicating an I/O error writing to your LUN17:28
asselinpatrickeast, yup that was it!17:28
harlowjathingee hey, i put up fyi17:28
harlowjaif we both want to do that, i'm ok17:29
asselinsorry, but thanks for helping me figure it out. I should submit that as a patch now that I think of it...and that it works ;)17:29
xyang1scottda: there is a fishbowl session for AZ17:29
scottdaxyang1: That's right, I forgot17:30
patrickeastasselin: np, i'm glad that others run into these problems before me and can add documentation :D17:30
asselinpatrickeast, yeah I was working on limiting what commands you can run via ssh using the passthrough key17:30
patrickeastahh cool17:30
patrickeasti had been wondering about that17:30
patrickeastsince i am basically giving a root key for the hypervisor to the ci slave17:31
asselinI had only tested one fc passthrough17:31
asselinpatrickeast, right lol17:31
*** ociuhandu has joined #openstack-cinder17:32
mriedemjgriffith: scottda: thanks for the input re: - i just marked it as invalid17:35
openstackLaunchpad bug 1496235 in OpenStack Compute (nova) "Boot from volume faild with availability_zone option, in case of cinder do not have availability_zone" [Undecided,Invalid] - Assigned to Ikuo Kumagai (i-kumagai)17:35
mriedemnova could filter the az from the volume create request, but i'm not sure if that's something nova should do17:35
jgriffithmriedem: yeah, unfortunately we need to figure out how to fix AZ's on the Cinder side17:36
thingeeharlowja: thanks17:36
scottdaFor now maybe it should, after discussing at the Summit we may decide to support Cinder AZs better .17:36
mriedemyou guys have a lot on your plate for the summit it sounds like :)17:37
mriedemoptimists for sure17:37
smcginnisI pity whoever is going to be the next PTL. Hah!17:38
scottdamriedem: How should we deal with closing Nova [volume] bugs?17:38
openstackLaunchpad bug 1208799 in OpenStack Compute (nova) "after detach vol from VM the multipath device is in failed state" [High,Confirmed]17:38
scottdahemna: Just stated he thinks it's fixed. Should he close it?17:39
hemnascottda, I'd like to see this retested and make sure it's fixed.  I believe it is.17:39
mriedemif there is a bug that fixed it (or a commit even) then i'd dupe17:39
scottdait's fixed in Brick (we think)17:40
mriedemif we think it's fixed but needs a retest on latest code, i typically mark as incomplete17:40
mriedembut incomplete leaves it on the books,17:40
mriedemso maybe just mark as invalid and say we think this is fixed in latest liberty, please retry and re-open if still hitting the issue17:40
hemnathat bug looks like it's an IBM based volume.17:40
mriedemoh gross17:40
mriedemthat's jungleboyj_'s fault then17:40
hemnajungleboyj_, do you know the guy that filed the bug ?17:40
jungleboyj_Is there a bus coming?17:41
hemnajungleboyj_, bug url ^^17:41
*** bill_az has joined #openstack-cinder17:41
*** raildo-afk is now known as raildo17:41
* mriedem moves room, bbiab17:41
hemnajungleboyj_, I believe that bug has since been fixed.   The details of the bug description looks like they were using an XIV array at the time17:41
scottdahemna: Are you OK with closing "probably fixed with Brick" bugs as invalid and "please re-try and re-open if still hitting..."17:41
thingee"if we both want to do that" ?17:41
thingeeharlowja: ^17:42
hemnajungleboyj_, so I was curious if you happened to know the filer of the bug.17:42
jungleboyj_Know, don't recognize that person.17:42
hemnascottda, I think at a minimum putting the bug in incomplete status is warranted.17:42
harlowjathingee well if it gets selected, we can both talk about it, idk17:42
jungleboyj_hemna: s/know/no17:42
thingeeharlowja: I would be interested. Mostly to make sure the conversation stays on track17:42
harlowjathats fine with me :)17:43
jungleboyj_hemna: Yeah, that looks like XIV.17:43
scottdaok, that's fine. Just wanted to figure out how to go about triaging 100+ bugs efficiently17:43
harlowjathingee wrangle me in to, lol17:43
hemnajungleboyj_, the bug isn't specific to XIV fwiw.17:43
hemnawe didn't have multipath -f <device> happening back then after detach17:43
hemnawhich would cause that17:43
*** esker has quit IRC17:43
xyang1Can we backport a patch that has added a new string, like a warning msg?17:44
thingeeharlowja: I will be commenting on your spec soon. Just have some loose ends I'm finishing up17:44
harlowjathingee cool17:44
xyang1jgriffith: jungleboyj_ ?17:44
jungleboyj_hemna: Ok.  Sorry, I am in another meeting right now.  So, what should we do with that bug?17:44
hemnajungleboyj_, mark it incomplete.  needs to be retested17:45
jungleboyj_hemna: Ok, makes sense.17:45
smcginnisxyang1: I believe so. It just might not get a translation right away. Or ever.17:45
jungleboyj_xyang1: Yes, I think so.  It just may not be translated.17:45
xyang1smcginnis: jungleboyj_ ok, thanks17:45
*** markus_z has quit IRC17:45
jungleboyj_If there are bugs, they should be fixed.17:45
hemnascottda, jungleboyj_
*** mriedem has quit IRC17:46
cebrunsDuncanT: In the Mitaka Summit topic etherpad - what is the "bare metal attach" that is on the way?17:46
jungleboyj_hemna: Oh, I remember that being fixed now that you point it out.17:47
smcginniscebruns: I believe that's related to the keystone-less POC e0ne is working on. To be able to provision storage via Cinder, one consumer of which would be bare metal (ironic) outside of full OpenStack deployment.17:47
hemnacebruns, it's buried in there17:47
cebrunsxyang1: smcginnis hemna - Awesome - thanks!17:48
hemnacebruns, it's not explicitly called out, but it should be actually.17:48
hemnaI can add it, I think it's worth noting at least and a sprint discussion.17:49
*** mriedem has joined #openstack-cinder17:49
*** esker has joined #openstack-cinder17:50
*** p0rtal has joined #openstack-cinder17:50
hemnacebruns, ok it's in the etherpad at the bottom now.17:51
xyang1hemna: i thing that one is good candidate for fishbowl17:52
lpetrutthingee: Hi17:54
*** mriedem1 has joined #openstack-cinder17:54
*** mriedem has quit IRC17:54
*** timcl has quit IRC17:55
*** thingee has quit IRC17:56
lpetrutthingee: We were wondering if the Microsoft CI can be re-enabled as it's stable now. ociuhandu sent some mails related to this on the ML, providing recent run info, but did not get any reply17:56
*** e0ne has joined #openstack-cinder17:56
*** krtaylor has quit IRC17:58
*** diogogmt has quit IRC17:58
hemnaxyang1, so that one.17:59
*** adurbin__ has quit IRC17:59
xyang1hemna: hi17:59
hemnaxyang1, we don't know why initialize_connection timed out, so the volume may be exported and it may not be.17:59
*** diogogmt has joined #openstack-cinder17:59
*** shyama has quit IRC17:59
*** thingee has joined #openstack-cinder17:59
xyang1hemna: yes, so it is a best attempt,17:59
hemnaif we call terminate_connection, what happens in drivers when the volume isn't exported ?18:00
thingeelpetrut: hey, let me take a look at the results. sorry forgot to add that to my list of things to do.18:00
lpetrutthingee: np, thanks!18:00
xyang1hemna: driver has to handle that any way, there is try except18:00
hemnamultiattach makes this a bit more complex as well fwiw18:00
hemnaxyang1,  I have a meeting to run to....bbiab.18:01
xyang1hemna: ok18:01
*** hemna is now known as hemnafk18:01
thingeelpetrut: you mind giving me the result links again?18:02
*** esker has quit IRC18:03
*** vgridnev has quit IRC18:03
*** garthb has quit IRC18:03
*** garthb has joined #openstack-cinder18:03
lpetrutthingee: sure,
jgriffithhemnafk: I've been meaning to talk to you about multi-attach when you have some time18:04
thingeelpetrut: thanks checking18:04
*** e0ne has quit IRC18:06
*** jwcroppe has quit IRC18:08
*** mriedem1 has quit IRC18:10
*** krtaylor has joined #openstack-cinder18:11
thingeelpetrut: where are iscsi driver tests?18:13
*** e0ne has joined #openstack-cinder18:13
*** annegentle has joined #openstack-cinder18:13
*** mriedem has joined #openstack-cinder18:13
*** zhangjn has joined #openstack-cinder18:14
lpetrutthingee: there's one folder for each driver: iscsi, smb3_windows, smb3_windows. eg:
*** rushil has quit IRC18:15
*** vgridnev has joined #openstack-cinder18:16
*** Zhongjun has quit IRC18:17
*** vgridnev has quit IRC18:18
hogepodgethingee: (and anyone else on the team), I've updated this with new capabilities. Let me know if there are any glaring errors in the categorization. Also, input on the scoring would be appreciated. Is everything there essential?18:19
*** zhangjn has quit IRC18:19
*** vgridnev has joined #openstack-cinder18:21
hogepodgethingee: oh yeah
asselinpatrickeast, but I already see that it needs to be customized a bit18:22
*** rushil has joined #openstack-cinder18:22
*** vgridnev has quit IRC18:23
*** rushil has quit IRC18:23
*** delattec has joined #openstack-cinder18:24
*** cdelatte has joined #openstack-cinder18:24
thingeehogepodge: ack18:25
thingeelpetrut: ah sorry missed that.18:26
thingeeasselin: you mind looking at these Microsoft CI logs ... looking for being reenabled18:26
thingeeasselin: I seem some basic stuff I look for18:26
asselinthingee, ok18:27
*** dims__ has joined #openstack-cinder18:28
asselin*sigh* microsoft ci has to be different from everyone else18:29
*** vgridnev has joined #openstack-cinder18:30
*** vgridnev has quit IRC18:31
*** dims_ has quit IRC18:31
*** aix has joined #openstack-cinder18:33
*** rushil has joined #openstack-cinder18:35
mriedemhemnafk: there were two bugs that looked very similar for this, reported by the same person, i duped the newer against the older18:36
openstackLaunchpad bug 1454621 in OpenStack Compute (nova) "Multipath device descripter is not deleted while the device pathes are removed after detach last volume form VM on a host" [Undecided,New]18:36
mriedemhemnafk: looks like that was reported for kilo, i'm wondering if it's fixed in os-brick in liberty18:36
*** bauzas has joined #openstack-cinder18:37
*** tpsilva has quit IRC18:37
thingeeasselin, lpetrut: you mean because of the folder names? I think lpetrut can fix that...18:37
*** tpsilva has joined #openstack-cinder18:37
asselinthingee, there's a lot that could be made more common, but the basics are there18:38
*** vgridnev has joined #openstack-cinder18:39
*** diogogmt has quit IRC18:39
*** rhagarty__ has joined #openstack-cinder18:39
*** rhagarty_ has joined #openstack-cinder18:39
*** sgotliv has joined #openstack-cinder18:40
*** amoturi has left #openstack-cinder18:42
*** Lee1092 has quit IRC18:43
*** esker has joined #openstack-cinder18:44
*** vgridnev has quit IRC18:49
*** e0ne has quit IRC18:49
*** vgridnev has joined #openstack-cinder18:52
*** delattec has quit IRC18:54
*** cdelatte has quit IRC18:54
mriedemwho works on iser volume stuff?18:56
bauzasheya, could someone explain me why AZs are in use with Cinder?18:56
*** baumann has joined #openstack-cinder18:56
bauzasI mean, I can understand that you create a volume with an AZ, just wonder what it relates to18:56
*** lpetrut has quit IRC18:57
bauzaslooking at bug
openstackLaunchpad bug 1496235 in OpenStack Compute (nova) "Boot from volume faild with availability_zone option, in case of cinder do not have availability_zone" [Undecided,Invalid] - Assigned to Ikuo Kumagai (i-kumagai)18:57
bauzasI don't get the relationship between Nova and Cinder18:57
*** baumann has left #openstack-cinder18:58
bauzasmostly because Nova AZs are actually aggregate metadata, ie. attached to hosts, not exactly corresponding to failure domains18:58
*** diogogmt has joined #openstack-cinder18:58
*** diogogmt_ has joined #openstack-cinder19:02
*** thingee has quit IRC19:02
*** cdelatte has joined #openstack-cinder19:03
*** delattec has joined #openstack-cinder19:03
scottdabauzas: I think the answer is that there is no relationship for AZs between Cinder and Nova19:03
scottdaCinder team is going to discuss AZs at the Tokyo summit...19:03
bauzasscottda: so I wonder why Cinder needs to get that info19:03
bauzasscottda: oh coolness19:03
scottdaCinder does not need the info from Nova19:03
bauzasscottda: I'll be in Tokyo too, so I could attend any session related to that19:04
scottdamriedem: had thoughts on how to fix from the Nova side.19:04
*** diogogmt has quit IRC19:04
*** diogogmt_ is now known as diogogmt19:04
bauzasscottda: well, I've been discussing that with mriedem and MHO is that I wonder why we need to give Cinder that info19:04
mriedemif there is no relationship between AZs in cinder and nova,19:04
scottdaFrom earlier today:19:05
scottda11:36 AM <mriedem> nova could filter the az from the volume create request, but i'm not sure if that's something nova should do19:05
mriedemthen we could just not pass the instance az to the volume create19:05
bauzasmriedem: yup19:05
mriedemif there is no relationship between az's in nova and cinder,19:05
mriedemthen why do we have this in nova?
bauzasfrom my perspective, there is a big misunderstanding about what is a Nova AZ19:06
scottdamriedem: good question.19:06
bauzasit's like mapping between apples and bananas19:07
mriedemoldest related commit i see is
mriedemso i bet in the long long ago when nova-volume was still a thing in tree, az's for compute and volumes were probably considered19:07
mriedemndipanov: i don't suppose you're around19:08
bauzasmeh, I need to see how AZs are handled in cinder19:08
ndipanovmriedem, sup19:08
mriedemndipanov: why do we have this in nova? do you know?
mriedemndipanov: i'm wondering if that's an artifact from nova-volume19:09
ndipanovseems useless tbh19:10
mriedemlooks like added in grizzly
jgriffithbauzas: mriedem scottda since this is the hot topic of the day:
openstackLaunchpad bug 1377485 in openstack-manuals "Use cinder availability-zone for creating a volume, NOT nova availability-zone" [Medium,Fix released] - Assigned to Darren Chan (dazzachan)19:10
bauzasjgriffith: but why AZs are used in cinder?19:10
openstackLaunchpad bug 1380780 in OpenStack Compute (nova) "Boot from image and create a new volume ignores availability zone" [Low,Fix released] - Assigned to Andrey Kurilin (andreykurilin)19:11
jgriffithand the list goes on and on19:11
jgriffithbauzas: because when I ported them over from nova-volume the plan was to eventually make something work :)19:11
jgriffithbauzas: becuase in nova-volume they actually "meant" A LOT19:11
bauzasjgriffith: just looking at the cinder scheduler, don't see anything related to AZs19:11
jgriffithbauzas: correct19:11
mriedemand nova-volume was dropped in grizzly19:12
jgriffithbauzas: because there's nothing there19:12
jgriffithmriedem: +119:12
mriedemso i'm not sure what was for19:12
bauzasheh, I remember the old stories19:12
jgriffithmriedem: DuncanT and his team have all sorts of crazy stuff :)19:12
*** e0ne has joined #openstack-cinder19:12
bauzasjgriffith: so, when creating a volume attached to an AZ, what happens ?19:13
*** rushil has quit IRC19:13
jgriffithmriedem: IIRC that was going to be 'step 1' in making cross AZ's work between Cinder and Nova19:13
jgriffithbauzas: you mean if you specify a random AZ at create?19:13
bauzasjgriffith: TBH, I have very little knowledge of Cinder since Grizzly..19:14
bauzasjgriffith: I even played with nova-volume very long time ago19:14
jgriffithbauzas: probably because it didn't really exist til grizzly19:14
bauzasjgriffith: but I'm not seasoned19:14
jgriffithbauzas: so the bottom line is we don't honor/handle AZ's in Cinder the way we should and it needs to be fixed IMO19:15
*** thingee has joined #openstack-cinder19:15
jgriffithusing anything other than the default zone is likely to cause issues19:15
bauzasjgriffith: okay, my take on that is that Nova AZs are misunderstood19:15
jgriffithbauzas: well, that's a whole seperate conversation :)19:15
bauzasjgriffith: most people think it's for having different failure domains at the datacenter granularity19:15
jgriffithbauzas: YES nova AZ's are a bit misunderstood19:16
jgriffithbauzas: cells make things worse19:16
bauzasjgriffith: bu t that's not19:16
bauzasjgriffith: yup, cells mapping to AZs are the special case where AZs are indeed datacenter failure domains19:16
bauzasjgriffith: all the other deployments aren't19:16
jgriffithbauzas: My recollection is that one of the problems is that Nova AZ's has morphed into something not entirely AZ like19:16
mriedemsomeone in the ML pointed out this
scottdabauzas: How would you describe Nova AZs?19:16
bauzasscottda: sec, I wrote something on that19:17
mriedemscottda: bauzas:
jgriffithmriedem: Oh.. nice19:17
mriedemjgriffith: that is a hack though, that cinder change, it seems like a hack anyway19:17
jgriffithI forgot about that19:17
scottdacool, thx19:18
*** rushil has joined #openstack-cinder19:18
jgriffithmriedem: no matter how you look at it, AZ's in Cinder are a hack :(19:18
bauzasmriedem: oh cool, the upstream docs have been updated19:18
bauzasI was thinking that the merge would need a day to get the upstream docs updated19:18
mriedemso is merged but should probably not exist19:19
smcginnisWould be an interesting question for operators, to see how they're using (or trying to use) AZs.19:19
mriedemmaybe HP has some special out of tree sauce for all of this?19:20
guitarzanmost of us probably aren't using them at all for block19:20
scottdaWell, it is a fish bowl sesssion at the summit, so hopefully we'll get ops feedback19:20
guitarzansince they don't really work :)19:20
scottdano, HP is not using out-of-tree code19:20
bauzasscottda: like I said, ping me when you know when the session is planned so I could attend it19:21
jgriffithscottda: HP Cloud?19:21
smcginnisDoesn't the public Helion cloud run through some scripts on a machine under DuncanT's desk??19:21
jgriffithsmcginnis: LOL19:21
scottdaWe just divide the data centre into AZs and use the same AZ name for Cinder and Nova19:21
scottdayes, HP cloud19:21
jgriffithscottda: +119:21
guitarzanthat seems like the obvious way to use them19:21
bauzasscottda: but there is no isolation from a dataplane perspective, right?19:22
scottdaI guess be *believed* that Cinder and Nova AZs work as failure domains, and implemented it that way19:22
bauzasNova AZs aren't failure domains19:22
bauzasI can't speak of Cinder19:22
scottdabauzas: There is some a the subnet level19:22
*** esker has quit IRC19:22
jgriffithand here we goooo... :)19:22
*** cdelatte has quit IRC19:22
*** delattec has quit IRC19:22
scottdabauzas: I get that, but I'm not so sure that was understood when the cloud was designed.19:22
guitarzanbauzas: why are you saying what they are and aren't? they certainly could be19:23
guitarzanunless I'm misunderstanding something19:23
bauzasguitarzan: they are the worst abstraction in Nova for having failure domains19:23
bauzasguitarzan: unless you map them with cells19:23
guitarzanby that you mean the *only* abstration no?19:23
scottdaSo, I think HP cloud pre-dates cells, and certainly pre-dates cinder...19:24
bauzasguitarzan: well, there are 4 ways to segregate your cloud19:24
bauzasguitarzan: regions (but that's keystone), cells, aggregates and AZS19:24
bauzaswe recently added server groups19:24
*** dims__ has quit IRC19:24
*** e0ne has quit IRC19:24
jgriffithif anyone cares russell did a nice write up on this a couple years ago:
bauzasso, cells are the only way to have a datacenter level of isolation19:25
bauzasaggregates are for grouping a list of hosts implicitely19:25
*** delattec has joined #openstack-cinder19:25
*** cdelatte has joined #openstack-cinder19:25
bauzasand AZs are just aggregates that are explicit19:25
*** dims_ has joined #openstack-cinder19:25
jgriffithbauzas: but to guitarzan 's point you can set those up and deploy them to use them in what ever manner you like19:25
bauzasjgriffith: yup, it was the first time I was understanding the right model19:26
*** mriedem has quit IRC19:26
bauzasjgriffith: well, you still have one shared MQ and one shared database from a controlplane perspective19:26
bauzaswhatever you deploy19:26
jgriffithbauzas: possibly... but there are ways around that as well19:27
bauzasjgriffith: you could use neutron and do tenant matching with AZs but that's a very specific deployment19:27
guitarzanbauzas: ah, I thought you were still talking dataplane like above19:27
bauzasjgriffith: well, there are 2 levels of failure19:27
bauzasif we speak of failure domains19:27
jgriffithha... "you could use neutron"19:27
bauzastbc, AWS AZs are both for failure domains and dataplane isolation19:28
jgriffithbauzas: true statement19:28
bauzasby speaking of network isolation, I mean you can't ping from one AZ to another AZ without using an external network19:28
guitarzanhow does the first not imply the second?19:28
*** salv-orlando has quit IRC19:29
* guitarzan thinks this isn't going to interest me :)19:29
jgriffithguitarzan: shame on you19:29
bauzasin Nova, two AZs are not following that assumption unless you do a very specific setup19:29
*** salv-orlando has joined #openstack-cinder19:29
bauzasthat's for the dataplane perspective19:29
* guitarzan misread the concept of data plane isolation19:30
guitarzanso it's not something I really care about :)19:30
bauzasnow, speaking of failure domain at the datacenter granularity, I consider that if my datacenter goes down, the others bays are not impacted19:30
bauzasthat's not true also for Nova, because you share at least the controller19:31
bauzashence the big misunderstanding19:31
*** rhagarty_ has quit IRC19:31
*** tpsilva has quit IRC19:32
bauzaswhat Nova promises you, is that if you boot one instance to AZ1 and one other to AZ2, you'll be 100% sure that they are on two separate nodes19:32
*** annegentle has quit IRC19:32
bauzasthat's the only thing => host failure domain, not datacenter19:32
*** rhagarty__ has quit IRC19:32
*** mriedem has joined #openstack-cinder19:32
*** annegentle has joined #openstack-cinder19:32
scottdamakes sense19:33
*** hemnafk is now known as hemna19:33
bauzasso, Cinder is the only external consumer of our AZs AFAIK19:34
mriedemsorry i got disconnected,19:34
mriedemi hope i didn't miss any revelations19:34
bauzasmriedem: santa doesn't exist19:34
mriedemyou already told me that last week19:34
bauzasmriedem: I'm a bad guy, I know19:34
mriedembtw, they are making a movie about the krumpus19:34
bauzasmriedem: so I was just explaining that Nova AZs are host failure domains19:35
bauzasmriedem: but you know that story19:36
mriedemi tend to forget19:36
mriedembut ok19:36
bauzasso, given Cinder is the only external consumer of Nova AZs, I'd be very happy to know if we still need to provide those to Cinder :)19:36
mriedemyeah, that's what i just said in the ML19:37
mriedemcan we deprecate the cinder.cross_az_attach option in nova19:37
hemnamriedem, that looks like a dupe
openstackLaunchpad bug 1454621 in os-brick "Multipath device descripter is not deleted while the device pathes are removed after detach last volume form VM on a host" [Undecided,New]19:37
openstackLaunchpad bug 1208799 in OpenStack Compute (nova) "after detach vol from VM the multipath device is in failed state" [High,Confirmed]19:37
bauzasmriedem: you're my public voice19:37
hemnaboth of which should be fixed now that we've switched to os-brick19:37
mriedemhemna: ok i duped 2 against that then19:38
scottdaI'll talk to DuncanT about his AZ use cases when he gets back. It sounds like, if there is something Cinder wants for failure domains, it should have a new name to avoid confusion.19:39
bauzasscottda: huge +119:39
* scottda remembers to schedule a separate Summit session for bike-shedding about names19:40
* bauzas could propose ABAZ = All But AZ19:41
smcginnisThe "Open Mitaka Development" patch has merged.19:41
smcginnisTATAZYLF = These aren't the availability zones you're looking for.19:42
scottdaI like that, smcginnis19:42
*** mriedem has quit IRC19:43
*** merooney has quit IRC19:43
smcginnisSo missing part of the conversation and just skimming quick - but if nova<>cinder az's were an idea that never materialized and cause confusion - it does seem like we should just deprecate it.19:44
*** mriedem has joined #openstack-cinder19:44
mriedemndipanov: can we just deprecate and microversion out boot from volume with source=image?19:45
mriedemseems that causes lots of issues for us19:45
mriedemnova should get out of the business of orchestration19:45
guitarzanmriedem: I wish19:45
mriedemwell because now people want to be able to pass the volume type to nova boot19:45
mriedemas a proxy to the cinder api19:46
guitarzanto be fair, wanting volume type is 100% reasonable in that situation19:46
hemnaah yeah, yet another microversion dep :)19:46
hemnascottda, you got that done yet ?19:46
hemnaship it!19:47
scottdaAnd since Mitaka is open, I've un-WIP'ed the spec...19:47
hemnascottda, absolutely no pressure whatsoever.  none.  zero.19:47
jgriffithmriedem: nova boot from volume creates-volume should die!!!!19:48
scottdahemna: What about my PM? Can we take NOva + Cinder specs for returning VolumeIsBusy to the NOva meeting tomorrow?19:48
mriedemhemna: not a cinder microversion19:48
guitarzanjgriffith: mriedem that makes it official, we should rip it out :)19:48
mriedemhemna: a nova microversion19:48
jgriffithmriedem: it doesn't work in most real world scenarios anyway19:48
mriedemas in before 2.x we deprecate boot from volume with source=image19:48
jgriffithmriedem: you'll hit an RPC timeout on most backends any time you have an image of any significant size19:48
mriedem>= 2.x we fail with a 400 in nova-api19:49
jgriffithmriedem: :)19:49
smcginnisjgriffith: We've hit that.19:49
guitarzanjgriffith: an rpc timeout?19:49
mriedemjgriffith: yeah so you have to then tweak config settings in nova to account for that, which is also the suck19:49
guitarzanwhich part of it is inside an rpc call?19:49
smcginnisguitarzan: Takes too long to copy over the image.19:49
jgriffithguitarzan: yeah, the rpc call over to Cinder times out IIRC19:49
jgriffithguitarzan: which makes sense19:49
guitarzanbut that's not an rpc call19:49
hemnaif only we had image caching to volumes in the backend.....19:49
mriedemnova's making a rest call19:49
guitarzanit's just a rest call, which doesn't timeout19:50
mriedemand then nova polls the status19:50
jgriffithguitarzan: ok.. the rest call times out then19:50
smcginnisLeaves behind crud too if I remember right.19:50
guitarzanah, I see19:50
mriedemnova has a polling loop in the compute manager and the timeouts are configurable19:50
guitarzanyeah, they added a crazy timeout/poll thing to make it work19:50
mriedemand in liberty at least there was code added to terminate the instance on timeout19:50
jgriffithmriedem: not the last time I looked.. but that's cool if they are now19:50
jgriffithmriedem: so ignore me19:50
jgriffithmriedem: do what you like :)19:50
mriedemthe polling was in the compute manager for awhile19:50
mriedemthe delete on timeout wasn't19:50
smcginnismriedem: That's right, that's what I hit. It timed out with the default on any decent sized image.19:50
mriedemthis is the polling method
jgriffithOpenstack... shoving 12 pounds of poop in a 5 pound sack since Grizzly!19:51
mriedemwaits 3 minutes by default19:52
hemna3 minutes isn't very long for a big image and a busy iSCSI channel19:52
jgriffithhemna: it's usualy not iscsi FYI19:53
jgriffithhemna: it's almost always the fetch from Glance19:53
jgriffithhemna: combined with the qemu-convert19:53
hemnanot the dd ?19:53
smcginnishemna: Mine never got that far.19:53
jgriffithhemna: the process is fetch from glance --> temp-file---> qemu-convert the tempfile --> dd to volume19:53
guitarzanqemu-convert is worse than dd since it has to do the whole virtual size19:53
guitarzanby dd I mean copy from glance19:54
jgriffithguitarzan: +1, including if you have a 24Mb image and a 100G volume, it zeros out the entire 100G19:54
guitarzanjgriffith: yep, has to19:54
hemnaso wouldn't this work if we had the image cache enabled and the image happened to be a cache hit on the backend ?19:54
smcginnisIf you can get far enough to have the image cached, then I think it would be quick enough.19:54
jgriffithhemna: Well YEAH!!  Why do you think everybody wanted that featuure :)19:54
*** rhagarty__ has joined #openstack-cinder19:55
*** rhagarty_ has joined #openstack-cinder19:55
jgriffithhemna: and that's also why it's important to use min size and extend the volume19:55
guitarzanyeah, the image cache can be a big win for folks19:55
jgriffithguitarzan: if your backend can do anything with it19:55
jgriffithguitarzan: and as long as the first time you do the create-from-image you're not doing it from nova19:56
guitarzanjgriffith: ok fine, so s/folks/you/ :D19:56
guitarzanjgriffith: true19:56
jgriffithguitarzan: no no... I didn't mean it like that19:56
guitarzanyeah, it doesn't help me either :)19:56
eharneyi'm pretty sure we can optimize qemu-img convert to write sparse data like we recently did in other paths with dd, if writing the whole volume is an issue19:56
jgriffithguitarzan: I meant it as  we shouldn't say "problem solved"19:56
guitarzanjgriffith: yeah, no question19:56
jgriffitheharney: ya, if we could figure out how to do that it would be a help for sure19:56
jgriffitheharney: but when I looked my understanding was you couldn't do that19:57
jgriffitheharney: anyway... there's still the fetch from Glance too19:57
*** delattec has quit IRC19:57
*** cdelatte has quit IRC19:57
jgriffithall good things to look at19:57
jgriffitheharney: the other thing I wanted to do was change the process to create a minimal sized volume and then extend it19:58
guitarzanjgriffith: I'm not sure we want cinder in the business of resizing a fs19:58
jgriffitheharney: but it became rather ugly  and hacked WAYYY more code than I was comfortable with19:58
*** delattec has joined #openstack-cinder19:58
*** cdelatte has joined #openstack-cinder19:58
jgriffithguitarzan: well... for a create from image it's ok19:58
jgriffithguitarzan: because it's not actually resizing the FS19:59
jgriffithguitarzan: know what I mean?19:59
guitarzanjgriffith: I don't follow19:59
guitarzanno, I don't get it :)19:59
eharneythis sounds familiar but i feel like i'm missing why that's beneficial19:59
eharneyor forgot19:59
*** Yogi1 has quit IRC19:59
jgriffithguitarzan: if the raw glance image is 20MB, I can safely write it to a 20MB volume19:59
*** esker has joined #openstack-cinder19:59
guitarzanyou're going to have superblocks out there in your etended volume that need written no?19:59
jgriffithguitarzan: and avoid dd'ing zeros to the remaining blocks (say a 200G volume)20:00
*** hodos|2 has joined #openstack-cinder20:00
jgriffithguitarzan: then extend the volume to whatever size I want....20:00
jgriffiththat might be a non-issue as eharney pointed out by sparse copy in dd now20:00
jgriffithguitarzan: ?  It's an image20:00
guitarzanyeah, if you're writing to a zero device, like lvm thin, we should be able to hack up qemu to do what we want20:00
jgriffithIt's not really a FS20:00
guitarzanjgriffith: I know, that's what I'm not following20:01
guitarzanif you want to boot it, it's a filesystem20:01
guitarzanand if the raw image is 20MB, then it's a 20MB filesystem20:01
jgriffithguitarzan: but cloudinit does the fs resize stuff20:01
guitarzanI don't know anything about that :)20:01
*** salv-orlando has quit IRC20:01
guitarzanisn't that after boot?20:01
*** esker has quit IRC20:02
guitarzanI mean maybe it does work20:02
jgriffithguitarzan: it certainly works20:02
guitarzanI'm just not quite clear on all the pieces20:02
jgriffithextend is safe on image volumes... reduce.. not so much :)20:02
*** hodos has quit IRC20:02
guitarzanI was just pointing out that someone has to extend it, and if this cloudinit thing does that, that's pretty great20:02
guitarzanI've never been clear on what cloudinit even is to be honest20:03
jgriffithguitarzan: well, it does it with any image already20:03
jgriffithguitarzan: for example you say nova boot flavor 100 vs 20020:03
guitarzanI thought it was nova itself that did that20:03
jgriffithone has a root disk of 1G the other 20G20:03
jgriffithit's the same concept20:03
guitarzanyeah, I'm not disagreeing with that part either20:04
jgriffithguitarzan: well.. still it will do the same thing with an external volume20:04
jgriffithI should shut up now :)20:04
guitarzanit will? cool20:05
guitarzanI can only plead ignorance20:05
jgriffithguitarzan: nothing wrong with that20:05
jgriffithguitarzan: now... if you want to talk about multi-attach, I can go on and on about why I was totally wrong on that one and we shouldn't do it :(20:05
guitarzanso right now if you copy image to volume20:06
guitarzan20 gig image, 100 gig volume20:06
smcginnisjgriffith: OK, you got my interest.20:06
guitarzanthen boot it20:06
smcginnisjgriffith: Why no multiattach?20:06
guitarzannova won't extend it right now, right?20:06
jgriffithguitarzan: it should I thought20:06
jgriffithguitarzan: but here's the thing... regardless if it doesn't20:06
guitarzanhmm, interesting20:06
guitarzanI do agree that it could20:06
jgriffithguitarzan: you can do an fdisk and you'll see "whatever" size the volume is20:06
guitarzanbut I'm not sure how it knows whether or not it should20:06
jgriffithsmcginnis: because it's misleading20:07
jgriffithsmcginnis: without a good shared/clustered FS it's flat out disastrous20:07
smcginnisjgriffith: Oh, yeah. I've seen that one before.20:07
*** gouthamr has quit IRC20:07
smcginnis"WHat do you mean I can't just write from both?"20:07
eharneypeople using multi-attach need to understand those kinds of things... it's not exactly something that's specific to Cinder20:08
*** gouthamr has joined #openstack-cinder20:08
jgriffithsmcginnis: and if we don't have STRONG documentation and a recommended/tested FS  it's my opinion that we shouldn't do it.20:08
jgriffitheharney: I thought that too....20:08
*** jwcroppe has joined #openstack-cinder20:08
jgriffitheharney: until I received not one but 2 support calls in the last month on the topic of "why aren't my files showing up"20:08
jgriffitheharney: it seems that once it landed in Cinder some folks hacked things up a bit to make it work in Nova on their own :(20:09
*** pv_ has quit IRC20:09
smcginnisjgriffith: It does have its use cases though. But their is some knowledge required.20:09
eharneywell i don't think we can just skip implementing multi-attach because people do crazy things...20:09
jgriffitheharney: I told them they should look at Gluster :)20:09
*** gouthamr has quit IRC20:09
jgriffitheharney: sure20:09
hemnasmcginnis, yup.20:09
smcginnisI wonder if the docs team allows the <blink> tag? :)20:10
jgriffitheharney: and I'm certainly not objecting to any patches or anything20:10
*** gouthamr has joined #openstack-cinder20:10
hemnasmcginnis, :)20:10
jgriffitheharney: just stating my opinion combined with a fact in channel here... that's all20:10
jgriffitheharney: I don't believe I said we should stop it.  I believe what I said was "I was wrong and it wasn't such a good idea", that's all20:10
eharneyyeah, i know20:11
*** stevemar has quit IRC20:11
jgriffitheharney: mostly because when it first came up I was very vocal that is was going to be the greatest thing since sliced bread20:11
guitarzanalso, can we have a big off switch in the conf file? :)20:11
eharneyi just kinda worry about the warning label thinking20:11
jgriffitheharney: fair enough20:11
*** stevemar has joined #openstack-cinder20:11
guitarzanbecause the 56th support ticket I get I'm turning it off :)20:11
eharneyafter some folks explained to me a bit ago that they thought crash consistent snapshots needed a big warning label too20:11
jgriffitheharney: honestly I plan to stay out of it and certainly won't cause trouble20:11
jgriffithI promise!20:11
guitarzanit's leading to a lot of interesting discussion about the attach/detach code paths though20:12
jgriffitheharney: oh... sigh, yeah, that whole thing20:12
*** stevemar has quit IRC20:13
*** salv-orlando has joined #openstack-cinder20:14
*** rushil has quit IRC20:17
*** delattec has quit IRC20:22
*** cdelatte has quit IRC20:22
*** jungleboyj_ has quit IRC20:23
*** cdelatte has joined #openstack-cinder20:26
*** delattec has joined #openstack-cinder20:26
*** vgridnev has quit IRC20:27
*** vgridnev has joined #openstack-cinder20:31
*** hemna is now known as hemnafk20:37
*** vgridnev has quit IRC20:38
*** Apoorva has quit IRC20:41
*** hemnafk is now known as hemna20:45
*** IanGovett has quit IRC20:45
*** salv-orlando has quit IRC20:45
*** bswartz has quit IRC20:47
*** dustins has quit IRC20:50
*** raildo is now known as raildo-afk20:53
openstackgerritEric Harney proposed openstack/cinder: LVM: Make sparse_copy_volume private, use for capabilities
openstackgerritEric Harney proposed openstack/cinder: Enable sparse image conversion
hemnascottda, looks like we might be able to just use novaclient's or should at least.21:00
*** thingee has quit IRC21:03
*** hemna is now known as hemnafk21:05
*** ronis has quit IRC21:09
*** salv-orlando has joined #openstack-cinder21:11
ndipanovmriedem, not sure I get what you mean21:12
*** julim has quit IRC21:12
ndipanovah volume with source=image21:12
ndipanovah well21:12
ndipanovit's really a shortcut21:13
ndipanovI'm ok with it going away21:13
smcginniseharney: UT failures in those patches.21:13
*** bswartz has joined #openstack-cinder21:13
mriedemndipanov: yeah, source=(image/blank/snapshot)21:13
mriedemndipanov: basically anything that makes nova proxy a volume create call to cinder21:14
mriedemndipanov: but at least one person said in the ML that they don't want to ever lose that b/c they like the orchestration nova is doing21:14
smcginnismriedem: Doesn't seem like the correct place for orchestration to me, but I suppose folks are used to it now.21:15
*** delattec has quit IRC21:15
*** cdelatte has quit IRC21:15
mriedemsmcginnis: i agree21:15
*** mc_nair has quit IRC21:16
*** porrua has quit IRC21:16
mriedemsmcginnis: so our solution has just been an act of attrition instead
smcginnismriedem: Probably the best course, given the situation. :|21:17
mriedemwhich is why we don't want to add volume_type to the boot from volume request for nova21:17
smcginnismriedem: Good to see it's documented out there!21:18
*** cdelatte has joined #openstack-cinder21:19
xyang1mriedem: good to have operators input if you want to get rid of that21:20
*** esker has joined #openstack-cinder21:20
mriedemxyang1: yeah i posted to the ops list21:21
mriedemxyang1: we can revert if there is a problem21:21
xyang1mriedem: great21:21
xyang1mriedem: I wonder if people want more not less21:21
mriedemmore complex stuff that doesn't have a clear use case?21:22
mriedemi'm sure they do :)21:22
mriedemi don't21:22
xyang1mriedem: since nova has knowledge of storage and network21:22
xyang1mriedem: :)21:22
mriedemlots of people would like nova to orhestrate more things21:22
mriedembut it's usually half baked and not tested well so we have lots of technical debt21:22
mriedemand in 3 years you have someone asking what it was added for in the first place21:22
xyang1mriedem: that az thing seems to be left out because of the split21:23
mriedemsort of yeah, but that cross_az_attach option was added in grizzly, and nova-volume wasn't in grizzly21:24
xyang1mriedem: I didn't even know that one, sorry21:24
*** ndipanov has quit IRC21:25
mriedemdon't feel bad, no one does :)21:25
xyang1mriedem: :)21:25
xyang1mriedem: good to remove it, probably21:25
*** gouthamr has quit IRC21:28
guitarzanmriedem: alaski and I both fought this battle here at rax and lost... it'll be great if openstack just takes it out for us :)21:31
*** Apoorva has joined #openstack-cinder21:32
mriedemguitarzan: removing that option?21:32
guitarzanmriedem: sorry, I mean the source=image stuff :)21:32
mriedemyeah, that's a harder change21:32
*** jgregor has quit IRC21:32
*** hemnafk is now known as hemna21:40
*** martyturner1 has quit IRC21:47
*** jaypipes has quit IRC21:52
*** gouthamr has joined #openstack-cinder21:53
*** cdelatte has quit IRC21:53
*** esker has quit IRC21:55
*** esker has joined #openstack-cinder21:59
*** esker has quit IRC21:59
openstackgerritxing-yang proposed openstack/cinder: Check sio_storage_pools in check_for_setup_error
hemnathe orchestration stuff embedded in nova seems problematic22:01
hemnain the case of booting a volume from an image, an external orchestration that's not bound by rpc timeouts would be better suited22:01
*** eharney has quit IRC22:01
*** yrabl has quit IRC22:03
*** alexpilotti has joined #openstack-cinder22:04
*** esker has joined #openstack-cinder22:04
*** yrabl has joined #openstack-cinder22:04
*** garthb has quit IRC22:10
*** garthb has joined #openstack-cinder22:10
xyang1Is it time to remove -2 now since Mitaka is open?  Or do we still need to wait?22:12
*** stevemar has joined #openstack-cinder22:13
*** annegent_ has joined #openstack-cinder22:14
openstackgerritYuriy Nesenenko proposed openstack/cinder: Small optimization in Block Device driver
*** stevemar has quit IRC22:15
*** esker has quit IRC22:15
*** openstackgerrit has quit IRC22:16
xyang1jungleboyj: :)22:16
*** openstackgerrit has joined #openstack-cinder22:17
*** _cjones_ has quit IRC22:17
*** _cjones_ has joined #openstack-cinder22:17
*** annegentle has quit IRC22:18
*** esker has joined #openstack-cinder22:19
*** salv-orlando has quit IRC22:20
*** asselin_ has quit IRC22:20
*** diogogmt has quit IRC22:21
*** salv-orlando has joined #openstack-cinder22:22
*** esker has quit IRC22:24
*** mriedem is now known as mriedem_away22:27
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder-specs: Remove Volume Manager Locks
*** abhilash has joined #openstack-cinder22:35
*** jamielennox is now known as jamielennox|away22:35
*** alejandrito has quit IRC22:36
*** dims__ has joined #openstack-cinder22:43
*** alexpilotti has quit IRC22:44
*** dims_ has quit IRC22:46
*** annegent_ has quit IRC22:49
patrickeasttsekiyama: hemna: whats the latest with ?22:51
patrickeastis it stuck on the xen server ci?22:51
patrickeastor something else?22:51
tsekiyamapatrickeast: I was trying to make xenserver CI recheck. but looks like failed22:52
hemnapatrickeast, I haven't looked at it in a few days.22:53
*** edtubill has quit IRC22:55
hemnatsekiyama, what did we end up resolving about my questions in patchset 5 ?22:55
*** jerrygb has quit IRC22:55
*** jwcroppe has quit IRC22:56
tsekiyamahemna: I have added a code to lookup wwn based device name22:57
*** jwcroppe has joined #openstack-cinder22:57
tsekiyamahemna: the wwn is fetched from connection_info['data']multipath_id22:57
hemnahrmm yah the multipath_id22:57
hemnawhich might go away22:58
hemnaalso if friendly names are enabled w/ multipath then we'll never get a /dev/mapper/<WWN> entry :(22:59
tsekiyamahemna: That would be problematic only when the operator changes the multipathd settings when they update nova, right?23:01
*** jwcroppe has quit IRC23:01
hemnatsekiyama, well, some distros default friendly names being on23:02
hemnafriendly names = evil23:02
*** annegentle has joined #openstack-cinder23:04
*** annegentle has quit IRC23:05
tsekiyamahemna: what does connection_info['data']['device_path'] return when friendly names are enabled? Doesn't it point the valid multipath device?23:08
hemnatsekiyama, for FC devices, it's documented here:
*** mudassirlatif_ has joined #openstack-cinder23:09
hemnaI have to run (soccer coaching)23:10
*** hemna is now known as hemnafk23:10
*** xyang1 has quit IRC23:12
*** mudassirlatif has quit IRC23:12
*** mudassirlatif_ is now known as mudassirlatif23:12
*** garthb has quit IRC23:13
*** stevemar has joined #openstack-cinder23:13
*** mudassirlatif has quit IRC23:17
*** mudassirlatif has joined #openstack-cinder23:21
*** mudassirlatif has quit IRC23:23
*** angela-s has quit IRC23:23
*** zhangjn has joined #openstack-cinder23:24
*** zhangjn has quit IRC23:24
*** mudassirlatif has joined #openstack-cinder23:31
*** takedakn has joined #openstack-cinder23:35
*** mudassirlatif has quit IRC23:36
*** hodos|2 has quit IRC23:36
*** jwcroppe has joined #openstack-cinder23:37
*** mudassirlatif has joined #openstack-cinder23:41
*** jwcroppe has quit IRC23:45
*** mudassirlatif has quit IRC23:49
*** mudassirlatif has joined #openstack-cinder23:50
tsekiyamahemnafk: I'm confused.. I'm starting to think encryption+FC+multipath haven't been working if Cinder was returning '/dev/mapper/*' previously.23:51
*** jamielennox|away is now known as jamielennox23:52
tsekiyamahemnafk: because it conflict with encryption device name nova was going to create.23:52
*** takedakn1 has joined #openstack-cinder23:53
*** gouthamr has quit IRC23:53
tsekiyamahemnafk: If so, we don't have to care about upgrading from non-working version..23:55
*** takedakn1 has quit IRC23:55
*** takedakn1 has joined #openstack-cinder23:55
*** takedakn has quit IRC23:56
*** takedakn1 is now known as takedakn23:56
*** takedakn has quit IRC23:59
*** zhangjn has joined #openstack-cinder23:59

Generated by 2.14.0 by Marius Gedminas - find it at!