Tuesday, 2015-03-10

*** theanalyst has joined #openstack-cinder00:00
*** tbarron has quit IRC00:02
*** tsekiyam_ has joined #openstack-cinder00:08
*** tsekiyama has quit IRC00:11
*** mtanino has quit IRC00:12
*** tsekiyam_ has quit IRC00:12
*** tbarron has joined #openstack-cinder00:16
*** tbarron has quit IRC00:16
*** rushiagr_away has joined #openstack-cinder00:16
*** tbarron has joined #openstack-cinder00:17
*** annashen has quit IRC00:27
*** Apoorva has quit IRC00:28
*** ho has joined #openstack-cinder00:28
openstackgerritAnish Bhatt proposed openstack/cinder: Improve error handling in refactored Tgt driver  https://review.openstack.org/15471300:29
anishhemnafk: done ^00:29
*** [1]Thelo has joined #openstack-cinder00:30
*** Thelo has quit IRC00:33
*** [1]Thelo is now known as Thelo00:33
openstackgerritJohn Griffith proposed openstack/python-cinderclient: Move unit tests into test directory  https://review.openstack.org/16223000:35
*** dannywil_ has quit IRC00:37
*** _cjones_ has quit IRC00:46
*** r-daneel has quit IRC00:49
*** Mandell has quit IRC00:52
*** leeantho_ has quit IRC00:56
*** Lee1092 has joined #openstack-cinder01:00
*** Mandell has joined #openstack-cinder01:01
*** Mandell has quit IRC01:17
*** asselin_ has joined #openstack-cinder01:19
*** asselin_ has quit IRC01:19
*** vilobhmm1 has quit IRC01:20
*** Mandell has joined #openstack-cinder01:22
*** akerr has joined #openstack-cinder01:23
*** bkopilov has joined #openstack-cinder01:24
*** annashen has joined #openstack-cinder01:27
*** annashen has quit IRC01:32
openstackgerritVincent Hou proposed openstack/cinder: Delete the temporary volume if migration fails  https://review.openstack.org/16068201:34
*** scottda_ has joined #openstack-cinder01:36
*** scottda_ has quit IRC01:43
*** Mandell has quit IRC01:47
*** rwsu has quit IRC01:47
openstackgerritwanghao proposed openstack/cinder: Fix response when query host detail by host name  https://review.openstack.org/16260101:50
*** _cjones_ has joined #openstack-cinder01:56
*** esker has joined #openstack-cinder02:00
*** esker has quit IRC02:00
*** esker has joined #openstack-cinder02:01
*** diegows has quit IRC02:02
*** yamada-h has joined #openstack-cinder02:02
*** mriedem has quit IRC02:03
*** Guest82873 has joined #openstack-cinder02:10
*** tellesnobrega_ has joined #openstack-cinder02:14
*** jungleboyj has joined #openstack-cinder02:14
*** tellesnobrega_ has quit IRC02:14
openstackgerritPeter Wang proposed openstack/cinder: Fixes VNX NotImplementedError of unmanage  https://review.openstack.org/16253202:15
*** kaisers1 has joined #openstack-cinder02:18
*** kaisers has quit IRC02:20
*** anki80 has quit IRC02:24
*** coolsvap has joined #openstack-cinder02:33
*** patrickeast has quit IRC02:35
*** scottda_ has joined #openstack-cinder02:35
openstackgerritSam Morrison proposed openstack/cinder: Set attach_status to detached when resetting status to available  https://review.openstack.org/16284502:38
*** scottda_ has quit IRC02:39
*** Anticimex has quit IRC02:43
*** coolsvap has quit IRC02:45
*** coolsvap has joined #openstack-cinder02:46
*** coolsvap has quit IRC02:46
*** coolsvap has joined #openstack-cinder02:46
*** markvoelker has joined #openstack-cinder02:51
*** ho has quit IRC02:55
*** ho has joined #openstack-cinder02:55
*** _cjones_ has quit IRC03:00
*** _cjones_ has joined #openstack-cinder03:00
*** akerr has quit IRC03:03
*** akerr has joined #openstack-cinder03:10
*** akerr has quit IRC03:11
*** yamada-h has quit IRC03:18
*** haomaiwang has joined #openstack-cinder03:19
*** bkopilov has quit IRC03:30
*** esker has quit IRC03:33
*** theanalyst has quit IRC03:48
*** rushiagr_away has quit IRC03:49
*** jcru has joined #openstack-cinder03:52
*** markvoelker has quit IRC03:53
*** markvoelker has joined #openstack-cinder03:53
*** jcru has quit IRC03:54
openstackgerritwanghao proposed openstack/cinder: Implement function of import/export snapshots  https://review.openstack.org/14459003:57
*** markvoelker has quit IRC03:57
*** Longgeek has joined #openstack-cinder03:59
*** Ilja has quit IRC03:59
*** vilobhmm has joined #openstack-cinder04:00
*** yamada-h has joined #openstack-cinder04:04
openstackgerritrajiv proposed openstack/python-cinderclient: cinderclient accepts arguments after metadata without -- separator  https://review.openstack.org/15649904:15
*** theanalyst has joined #openstack-cinder04:20
*** rushiagr_away has joined #openstack-cinder04:22
*** markvoelker has joined #openstack-cinder04:24
*** pradipta has joined #openstack-cinder04:26
*** pradip_vedams has quit IRC04:31
*** BharatK has joined #openstack-cinder04:33
*** pradip_vedams has joined #openstack-cinder04:34
*** Mandell has joined #openstack-cinder04:37
openstackgerritMike Perez proposed openstack/cinder: Update Datera's export to avoid deprecated keys  https://review.openstack.org/16277404:39
*** markvoelker has quit IRC04:41
*** alexpilotti has joined #openstack-cinder04:48
*** pradipta has quit IRC04:49
*** lpetrut has joined #openstack-cinder05:00
*** Ilja has joined #openstack-cinder05:05
*** pradipta has joined #openstack-cinder05:06
*** topshare has joined #openstack-cinder05:12
*** bkopilov has joined #openstack-cinder05:16
*** _cjones_ has quit IRC05:29
openstackgerrityogeshprasad proposed openstack/cinder: Added provider_id to volume and snapshot object  https://review.openstack.org/16261105:29
*** yamada-h has quit IRC05:34
*** Tross1 has joined #openstack-cinder05:35
*** Tross has quit IRC05:36
*** markvoelker has joined #openstack-cinder05:38
openstackgerritSam Morrison proposed openstack/cinder: Set attach_status to detached when resetting status to available  https://review.openstack.org/16284505:44
*** Mandell has quit IRC05:53
*** Mandell has joined #openstack-cinder05:55
*** yamada-h has joined #openstack-cinder05:56
*** ho__ has joined #openstack-cinder05:56
*** harlowja_ is now known as harlowja_away05:57
*** lpetrut has quit IRC05:57
*** ho has quit IRC05:58
*** Longgeek has quit IRC06:04
*** BharatK has quit IRC06:06
openstackgerritOpenStack Proposal Bot proposed openstack/cinder: Imported Translations from Transifex  https://review.openstack.org/15240106:09
*** yamada-h has quit IRC06:20
*** BharatK has joined #openstack-cinder06:20
*** vilobhmm1 has joined #openstack-cinder06:22
*** yamada-h has joined #openstack-cinder06:22
*** Longgeek has joined #openstack-cinder06:23
*** vilobhmm has quit IRC06:23
*** david-lyle has quit IRC06:30
*** _cjones_ has joined #openstack-cinder06:31
*** rushiagr_away is now known as rushiagr06:37
*** david-lyle has joined #openstack-cinder06:39
*** ankit_ag has joined #openstack-cinder06:45
*** coolsvap is now known as coolsvap|afk06:47
*** rj_ has joined #openstack-cinder06:50
*** sgotliv has quit IRC06:53
*** vilobhmm1 has quit IRC06:53
*** vilobhmm has joined #openstack-cinder06:54
*** anshul has joined #openstack-cinder06:54
*** coolsvap|afk is now known as coolsvap06:58
*** _cjones_ has quit IRC07:02
*** _cjones_ has joined #openstack-cinder07:03
*** _cjones_ has quit IRC07:03
*** Ilja has quit IRC07:05
*** bkopilov has quit IRC07:07
*** david-lyle is now known as david-lyle_afk07:11
*** chlong has quit IRC07:21
*** bkopilov has joined #openstack-cinder07:22
*** nshaikh has joined #openstack-cinder07:26
*** Ilja has joined #openstack-cinder07:31
*** jistr has joined #openstack-cinder07:31
*** openstackgerrit has quit IRC07:35
*** TobiasE has joined #openstack-cinder07:35
*** openstackgerrit has joined #openstack-cinder07:36
*** BharatK has quit IRC07:39
*** ronis has joined #openstack-cinder07:45
*** lpetrut has joined #openstack-cinder07:47
*** BharatK has joined #openstack-cinder07:52
*** coolsvap is now known as coolsvap|afk07:53
*** rj_ has quit IRC07:57
*** alecv has joined #openstack-cinder07:58
*** coolsvap|afk is now known as coolsvap08:00
*** markus_z has joined #openstack-cinder08:03
openstackgerritwanghao proposed openstack/cinder: Implement function of import/export snapshots  https://review.openstack.org/14459008:13
*** liusheng has joined #openstack-cinder08:14
*** sgotliv has joined #openstack-cinder08:15
*** Anticimex has joined #openstack-cinder08:20
*** sgotliv has quit IRC08:20
*** ronis has quit IRC08:24
*** [1]Thelo has joined #openstack-cinder08:25
*** ho__ has quit IRC08:27
*** Mandell has quit IRC08:27
*** Thelo has quit IRC08:27
*** [1]Thelo is now known as Thelo08:27
openstackgerritMarkus Zoeller (markus_z) proposed openstack/os-brick: Adjust os-brick to support FCP on System z systems  https://review.openstack.org/16267808:28
*** dulek has joined #openstack-cinder08:30
*** haomaiwang has quit IRC08:32
*** sgotliv has joined #openstack-cinder08:33
openstackgerritGrzegorz Grasza (xek) proposed openstack/cinder: Service object  https://review.openstack.org/16041708:33
*** haomaiwang has joined #openstack-cinder08:37
openstackgerritMarkus Zoeller (markus_z) proposed openstack/os-brick: Adjust os-brick to support FCP on System z systems  https://review.openstack.org/16267808:46
*** topshare_ has joined #openstack-cinder08:51
*** topshare_ has quit IRC08:51
*** alexpilotti_ has joined #openstack-cinder08:52
*** topshare_ has joined #openstack-cinder08:52
*** topshare has quit IRC08:53
*** alexpilotti has quit IRC08:54
*** alexpilotti_ is now known as alexpilotti08:54
*** ho has joined #openstack-cinder09:01
*** pradipta has quit IRC09:06
*** nellysmitt has joined #openstack-cinder09:06
*** dulek has quit IRC09:09
flip214DuncanT: do you have 10 minutes for me, please?09:10
*** aix has joined #openstack-cinder09:19
*** nshaikh has quit IRC09:20
*** markvoelker has quit IRC09:22
*** yuriy_n17 has joined #openstack-cinder09:25
openstackgerritYAMADA Hideki proposed openstack/cinder: Add missing backup/restore methods to Sheepdog driver  https://review.openstack.org/16292709:27
*** david-lyle_afk has quit IRC09:30
*** david-lyle_afk has joined #openstack-cinder09:30
*** Ilja has quit IRC09:32
*** dulek has joined #openstack-cinder09:33
*** jistr has quit IRC09:34
*** Ilja has joined #openstack-cinder09:41
*** jistr has joined #openstack-cinder09:46
*** vilobhmm has quit IRC09:48
*** Ilja has quit IRC09:50
*** afazekas has joined #openstack-cinder09:52
*** Ilja has joined #openstack-cinder09:57
*** _cjones_ has joined #openstack-cinder10:03
*** _cjones_ has quit IRC10:04
*** _cjones_ has joined #openstack-cinder10:05
*** sgotliv has quit IRC10:07
*** p01s0n has joined #openstack-cinder10:08
*** yamada-h has quit IRC10:08
*** _cjones_ has quit IRC10:09
*** ronis has joined #openstack-cinder10:11
*** bkopilov has quit IRC10:15
*** e0ne has joined #openstack-cinder10:15
*** chlong has joined #openstack-cinder10:18
*** sgotliv has joined #openstack-cinder10:20
*** topshare_ has quit IRC10:25
*** topshare has joined #openstack-cinder10:26
*** topshare has quit IRC10:28
*** aix has quit IRC10:32
DuncanTflip214: Sure10:33
DuncanTflip214: Sorry, only just got in from yet more time in the dentist chair10:33
*** coolsvap is now known as coolsvap|afk10:33
flip214Hi DuncanT, thank for answering!10:33
flip214I hope it'll be a long-term advantage to you, at least, even if harmful in the short-time scale.10:34
flip214please imagine a cinder controller node (c-sch, c-vol, ...) that uses DRBDmanage, with a few more storage nodes.10:34
flip214now, when cinder asks drbd to allocate storage, this can be done on _any_ of the nodes....10:35
*** tshefi has joined #openstack-cinder10:35
flip214so it's possible that the allocated storage in not on the c-vol host.10:35
flip214so there couldn't be an iscsi export of the data.10:35
*** annashen has joined #openstack-cinder10:36
*** pradip_vedams has quit IRC10:36
flip2141) force allocation on the c-vol host, or at least make a drbd-client available there to export via iscsi?10:36
openstackgerritYuriy Nesenenko proposed openstack/cinder: Fix Cinder logs to show authentication error in RBD driver  https://review.openstack.org/16294710:36
flip2142) put c-vol (or whatever service is necessary) on _all_ drbd storage hosts?10:36
flip2143) can c-vol tell cinder to create the iscsi export on "that node over there"? which services would need to be installed then?10:37
flip2144) any other ideas?10:37
flip214-- that's the short version, I'm happy to answer any questions, of course.10:37
*** annashen has quit IRC10:40
*** nellysmitt has quit IRC10:41
*** pradipta has joined #openstack-cinder10:48
*** chlong has quit IRC10:51
*** ho has quit IRC10:52
*** haomaiwang has quit IRC10:53
*** markvoelker has joined #openstack-cinder10:53
openstackgerritIvan Kolodyazhny proposed openstack/cinder: Allow archiving deleted rows to shadow tables, for performance  https://review.openstack.org/13118210:56
flip214sorry, gotta go... I'll be back in ~1h.10:56
*** aix has joined #openstack-cinder10:57
*** alexpilotti_ has joined #openstack-cinder11:08
DuncanTflip214: c-vol can certainly create an export on a remote host, just ssh over to it and run the  iscsi export stuff11:09
*** e0ne is now known as e0ne_11:09
*** e0ne_ is now known as e0ne11:09
*** nellysmitt has joined #openstack-cinder11:09
*** Krast has quit IRC11:09
*** yamada-h has joined #openstack-cinder11:10
*** alexpilotti has quit IRC11:10
*** alexpilotti_ is now known as alexpilotti11:10
*** Longgeek has quit IRC11:14
*** yamada-h has quit IRC11:14
eikkeour CI just finished job # 1000 ^_^11:15
e0neeikke: waiting for 1024 :)11:15
DuncanTeikke: Time to crack open a beer in celibration?11:15
eikkeDuncanT: that's planned for later today :)11:15
eikkemust say, the success rate is higher than I'd have expected, so Cinder proposed patches must be of high quality from the start in general11:16
eikkemost 'failures' are jobs cancelled because of a new patchset being pushed11:16
*** diegows has joined #openstack-cinder11:20
*** alexpilotti has quit IRC11:21
openstackgerritBharat Kumar Kobagana proposed openstack/cinder: GlusterFS: Support over subscription in thin provisioning  https://review.openstack.org/16271811:21
openstackgerritKallebe Monteiro proposed openstack/cinder: Replace assertEqual(True, *) -> assertTrue(*)  https://review.openstack.org/16272511:23
*** bkopilov has joined #openstack-cinder11:25
*** IanGovett has joined #openstack-cinder11:27
*** pradipta has quit IRC11:29
openstackgerritLena Novokshonova proposed openstack/cinder: Add notifications about snapshot.update.*  https://review.openstack.org/13304111:34
*** kallebe has joined #openstack-cinder11:35
*** Longgeek has joined #openstack-cinder11:36
openstackgerritLena Novokshonova proposed openstack/cinder: Add notifications about snapshot.update.*  https://review.openstack.org/13304111:37
openstackgerritLena Novokshonova proposed openstack/cinder: Add notifications about snapshot.update.*  https://review.openstack.org/13304111:40
flip214DuncanT: thank you...11:42
flip214do I understand correctly that it already does that?11:42
flip214so, from which function would I need to return the remote host to use for exporting via iscsi?11:43
nikesh_vedamsis this approach correct for retype http://paste.openstack.org/show/191256/11:52
*** coolsvap|afk is now known as coolsvap11:54
*** shyama has joined #openstack-cinder11:56
*** TobiasE1 has joined #openstack-cinder11:57
*** TobiasE has quit IRC11:58
*** karimb has joined #openstack-cinder12:01
e0neDuncanT: hi! i'll ask other cores tonight too. could you please take a look on my patch for devtack https://review.openstack.org/#/c/162488/?12:03
e0nei'm not sure how much space we need there12:04
openstackgerritBharat Kumar Kobagana proposed openstack/cinder: GlusterFS: Support over subscription in thin provisioning  https://review.openstack.org/16271812:10
*** nlevinki has joined #openstack-cinder12:13
*** BharatK has quit IRC12:14
*** bswartz has quit IRC12:15
*** p01s0n has quit IRC12:15
*** sgotliv has quit IRC12:16
*** kaisers has joined #openstack-cinder12:26
*** kaisers1 has quit IRC12:27
*** Mandell has joined #openstack-cinder12:28
*** sgotliv has joined #openstack-cinder12:29
*** victorfeitosa has joined #openstack-cinder12:32
*** Mandell has quit IRC12:32
openstackgerritwanghao proposed openstack/cinder: Implement function of import/export snapshots  https://review.openstack.org/14459012:33
*** Ilja has quit IRC12:34
DuncanTnikesh_vedams: No, that is not the right approach12:37
DuncanTnikesh_vedams: Your driver needs to looks at the difference between the types, and arrange for any significant parts to be applied12:37
openstackgerritYuriy Nesenenko proposed openstack/cinder: Fix Cinder logs to show authentication error in RBD driver  https://review.openstack.org/16294712:38
*** markvoelker has quit IRC12:38
*** markvoelker has joined #openstack-cinder12:39
*** markvoelker has quit IRC12:43
DuncanTflip214: In the end, initialise_connection needs to return the right info to enable nova to connect12:44
DuncanTflip214: Where it gets that from is totally up to your driver12:45
flip214ah, thanks.12:45
flip214I'll take a look at that.12:46
*** TobiasE1 has quit IRC12:47
*** jistr has quit IRC12:47
*** jistr has joined #openstack-cinder12:48
*** nellysmitt has quit IRC12:49
*** TobiasE1 has joined #openstack-cinder12:50
*** annegentle has joined #openstack-cinder12:51
openstackgerritrakesh mishra proposed openstack/cinder: Implement making volume snapshots public  https://review.openstack.org/15937212:51
*** cdelatte has quit IRC12:55
*** cdelatte has joined #openstack-cinder12:56
*** xyang1 has joined #openstack-cinder12:56
*** geguileo has joined #openstack-cinder12:57
*** cdelatte has quit IRC13:01
*** cdelatte has joined #openstack-cinder13:01
*** bkopilov has quit IRC13:04
*** jistr has quit IRC13:04
*** emagana has joined #openstack-cinder13:05
*** jistr has joined #openstack-cinder13:05
*** coolsvap is now known as coolsvap|afk13:06
*** nlevinki has quit IRC13:08
*** tbarron has quit IRC13:08
openstackgerritharsh mishra proposed openstack/cinder: Fix for  inconsistent cinder-services state change  https://review.openstack.org/16010413:09
*** Yogi1 has joined #openstack-cinder13:12
*** markvoelker has joined #openstack-cinder13:13
*** bkopilov has joined #openstack-cinder13:18
*** kaufer has joined #openstack-cinder13:18
*** rushiagr is now known as rushiagr_away13:20
*** nlevinki has joined #openstack-cinder13:21
*** dustins has joined #openstack-cinder13:22
openstackgerritJohn Griffith proposed openstack/python-cinderclient: Move unit tests into test directory  https://review.openstack.org/16223013:23
*** changbl has quit IRC13:26
*** topshare has joined #openstack-cinder13:31
*** tbarron has joined #openstack-cinder13:31
*** jungleboyj has quit IRC13:33
*** tbarron has quit IRC13:33
*** thangp has joined #openstack-cinder13:33
*** tbarron has joined #openstack-cinder13:33
*** topshare has quit IRC13:35
*** anuragpalsule has joined #openstack-cinder13:36
*** zhipeng has joined #openstack-cinder13:41
*** tbarron has quit IRC13:43
*** tbarron has joined #openstack-cinder13:44
*** tbarron has quit IRC13:44
*** amoturi_ has joined #openstack-cinder13:46
*** jcru has joined #openstack-cinder13:47
*** mriedem has joined #openstack-cinder13:48
*** jcru has quit IRC13:48
*** jcru has joined #openstack-cinder13:48
openstackgerritMarkus Zoeller (markus_z) proposed openstack/os-brick: Adjust os-brick to support FCP on System z systems  https://review.openstack.org/16267813:48
amoturi_Hi, Could someone take a look at review https://review.openstack.org/#/c/152676/ . Has been in waiting for a while now. thanks13:49
*** annegentle has quit IRC13:51
*** mriedem has quit IRC13:52
*** mtanino has joined #openstack-cinder13:54
*** mriedem has joined #openstack-cinder13:56
*** lcurtis has joined #openstack-cinder13:57
jgriffithxyang1: DuncanT thougths: https://review.openstack.org/#/c/162532/4//COMMIT_MSG13:57
jgriffithxyang1: DuncanT did something change, or did I miss something?13:57
jgriffithseems like an awful paradigm to me13:57
xyang1jgriffith: I'll take a look, not sure what this is13:57
jgriffithxyang1: heeh... so13:57
*** r-daneel has joined #openstack-cinder13:58
*** bswartz has joined #openstack-cinder13:58
jgriffithxyang1: the commit message and bug states: "when deleting an imported volume, cinder calls unmanage"13:58
jgriffithxyang1: if we're doing that it's "wrong" IMO13:58
jgriffithxyang1: it should in fact delete the volume, not unmanage.  If a user wants to unmanage then call unmanage13:58
xyang1jgriffith: Seems like it should be the other way around13:58
jgriffithxyang1: yeah, that's what I thought too.13:59
jgriffithxyang1: so the patch itself is fine as it really has nothing to do with that13:59
xyang1jgriffith: Could be just wrong description13:59
jgriffithxyang1: but I flagged it for the commit message, and also becasue if we're doing what is described there it's wrong13:59
*** thingee has joined #openstack-cinder13:59
jgriffithxyang1: that's what I'm hoping yes :)13:59
xyang1jgriffith: Let me check what it is doing14:00
*** anuragpalsule has quit IRC14:00
jgriffithxyang1: awesome, thanks14:00
*** timcl has quit IRC14:00
*** tbarron has joined #openstack-cinder14:00
*** rushil has joined #openstack-cinder14:01
*** nshaikh has joined #openstack-cinder14:03
xyang1jgriffith: I added comment there14:05
*** akerr has joined #openstack-cinder14:07
*** timcl has joined #openstack-cinder14:07
DuncanTjgriffith: That looks totally wrong14:07
xyang1jgriffith: I think he just need to say: implement unmanage14:07
DuncanTjgriffith: If we're calling unmanage in delete then we have a problem14:08
*** annegentle has joined #openstack-cinder14:08
jgriffithDuncanT: yeah, that was my point too14:08
*** zhipeng_ has joined #openstack-cinder14:08
DuncanTjgriffith: The stack trace in the bug is not exactly helpful14:09
xyang1DuncanT: jgriffith so when api does unmanage, it calls delete volume in manager.  Then the manager calls driver's unmanage.14:09
xyang1In this case, unmanage was not implemented14:10
*** zhipeng has quit IRC14:10
*** lpetrut has quit IRC14:11
*** annegentle has quit IRC14:11
*** lpetrut has joined #openstack-cinder14:12
*** lpetrut has quit IRC14:12
*** rushiagr_away is now known as rushiagr14:13
DuncanTxyang1: Yeah, got that. It looks like just an unclear/misleading bug report, which is fine14:14
jgriffithxyang1: yeah, it's misleading because it sets the unmanage_only flag14:15
jgriffithDuncanT: +114:15
jgriffithso I'd just like to see the bug description cleared up and the commit message more accurate14:15
jgriffithand then I'm happy to +2/A the patch14:15
jgriffithDuncanT: xyang1 it may seem silly, but I worry that what's there is going to introduce confusion and misunderstanding14:16
xyang1DuncanT: jgriffith I asked him to clarify how delete volume ended up doing unmanage14:16
jgriffithxyang1: great, thanks!14:16
DuncanTjgriffith: +1. I marked the bug as invalid/needs more info14:16
jgriffiththat'll work :)14:16
jgriffithxyang1: DuncanT thanks to both of you for the help14:17
xyang1jgriffith: DuncanT delete imported volume should be no difference from delete regular volume14:17
jgriffithxyang1: correct14:17
DuncanTxyang1: +114:17
jgriffithxyang1: I think the confusion was just because we "use" delete for unmanage14:17
jgriffithwhich is probably a really bad idea14:17
xyang1jgriffith: That is probably it14:18
* DuncanT hasn't done any bug triage for a while14:18
*** BharatK has joined #openstack-cinder14:18
DuncanTProbably time to go spend an hour going through some14:18
jgriffithDuncanT: you do an hour now, and I'll do an hour later today14:18
jgriffithDuncanT: that way we don't overlap and knock some stuff out14:19
DuncanTjgriffith: Excellent, I'll hopefully have taken care of the easy ones by then ;-)14:19
jgriffithDuncanT: hey wait....  I'll go first, then you go tomorrow :)14:19
*** eharney has joined #openstack-cinder14:21
*** topshare has joined #openstack-cinder14:21
*** topshare has quit IRC14:21
*** topshare has joined #openstack-cinder14:22
*** topshare has quit IRC14:22
*** topshare has joined #openstack-cinder14:23
dulekthangp: Hi!14:24
*** Mandell has joined #openstack-cinder14:25
dulekthangp: I'm just curious - why you're manually tracking changes in metadata field (which is DictOfStringsField) in Snapshot object14:25
*** jungleboyj has joined #openstack-cinder14:25
thangpdulek: you have to so that when you save, it gets saved in the db table14:27
thangpdulek: snapshot_metadata is in a separate table14:27
dulekthangp: So versionedobjects doesn't take care of it out of the box?14:28
thangpdulek: nope14:28
*** lpetrut has joined #openstack-cinder14:28
dulekthangp: I thought otherwise from the quick glance into the code. Okay, thanks!14:29
thangpdulek: managing mutliple db tables is up to the implementer of the versionedobjects14:29
*** IanGovett has quit IRC14:30
*** Guest75125 has joined #openstack-cinder14:30
*** emagana has quit IRC14:31
*** zhithuang has joined #openstack-cinder14:32
*** yuriy_n17 has quit IRC14:33
*** timcl1 has joined #openstack-cinder14:33
*** zhithuang is now known as winston-d_14:34
winston-d_dulek: ping14:34
*** timcl has quit IRC14:34
winston-d_dulek: ping14:34
dulekwinston-d_: hi!14:34
winston-d_dulek: hey, want to check with you about the comment related to 'error out'14:35
winston-d_dulek: due to my ignorance about taskflow, i need some education here.14:35
dulekwinston-d_: sure, it would be definitely easier to discuss here :)14:36
*** kaufer has left #openstack-cinder14:36
*** markvoelker has quit IRC14:37
*** markvoelker has joined #openstack-cinder14:37
*** ankit_ag has quit IRC14:39
winston-d_dulek: yes14:39
dulekwinston-d_: So any specific questions or you want me to try to explain it in details?14:40
winston-d_dulek: so by moving the retry out of taskflow, error_out_volume() doesn't work anymore?14:40
*** IanGovett has joined #openstack-cinder14:41
dulekwinston-d_: Point is - volume is errored-out inside the flow in case of NoValidHost.14:41
dulekwinston-d_: So when I have NoValidHost in manager volume is already updated in the DB and notification on the error send.14:41
*** markvoelker has quit IRC14:42
dulekwinston-d_: So I shouldn't retry whole flow, because we don't want multiple notifications.14:42
winston-d_dulek: yes, that sounds reasonable.14:42
dulekwinston-d_: So the correct approach would be to get error_out_volume execution out of TaskFlow.14:42
dulekwinston-d_: But this breaks the point of using TaskFlow.14:43
winston-d_dulek: ah, i see your point.14:43
*** Guest75125 is now known as notsogentle14:43
dulekwinston-d_: Because in the end we start to take out more parts and then whole flow can be moved back to manager14:43
*** notsogentle is now known as agentle_14:43
*** rwsu has joined #openstack-cinder14:44
*** nshaikh has quit IRC14:44
dulekwinston-d_: I'm not TaskFlow evangelist, it has it flaws, but such big change for a simple bug fix sounds too overwhelming14:44
winston-d_dulek: i'd agree with that.14:44
openstackgerritharsh mishra proposed openstack/cinder: Fix for  inconsistent cinder-services state change  https://review.openstack.org/16010414:45
dulekwinston-d_: Your point is correct - we need a generic solution, but simply using jgriffith's decorator for retries will work in cases of operations that are running inside the manager.14:45
dulekwinston-d_: I'm glad we can agree on that. :)14:46
winston-d_dulek: well, my first thought was we don't retry but wait, but then DuncanT suggest using retry until successful or timeout to shorten the delay.14:46
winston-d_dulek: that's probably best we can do, without modification to scheduler.14:47
dulekwinston-d_: Yes. I've tested it and normally just one retry is required (after your change with init_host_with_rpc), so this is nice.14:47
winston-d_dulek: but the thing is, or invalid request, we end up wasting resource doing useless retries.14:48
dulekwinston-d_: True, I agree.14:48
dulekwinston-d_: Do you have a plan how to cleanly implement raising NoValidHostNotComplete? Current way of returning two variables from a method doesn't seem appealing.14:48
jgriffithwinston-d_: +10000014:48
*** markvoelker has joined #openstack-cinder14:49
winston-d_dulek: another idea that I have is, do not even start trying, if schedule hasn't reached the state where it can consider itself has the so-called 'complete' view of all backends.14:50
*** hemnafk is now known as hemna14:50
jgriffithwinston-d_: how would you define that?14:51
winston-d_dulek: that requires some other changes to scheduler though, at least one additional DB query is needed.14:51
dulekwinston-d_: Can you explain it in more details?14:52
winston-d_jgriffith: query DB for all active c-vol services; and check if scheduler itself has the up-to-date stats from all of them, here up-to-date means stats['timestamp'] - now < CONF.service_down_time14:53
jgriffithwinston-d_: hmmm14:53
*** jaypipes has quit IRC14:53
*** e0ne is now known as e0ne_14:54
*** eharney has quit IRC14:54
jgriffithwinston-d_: so that would be a one time query on startup of the scheduler14:54
jgriffithwinston-d_: not terrible; and you'd still easily pick up "new" services from rabbit if/when they come online?14:55
openstackgerritRushil Chugh proposed openstack/cinder: Fixing mount when state_path is configured with a final '/'  https://review.openstack.org/16303214:55
dulekjgriffith: I think this query should be run at each request in this 60-seconds time window at scheduler start14:55
winston-d_dulek: this scheduler-complete-view logic is already in my PoC scheduler change, but only used for return a set of out-of-date c-vols, then scheduler raises NoVolidNotComplete when out-of-date set isn't empty.14:55
winston-d_dulek: no, just one time14:56
winston-d_dulek: it's about scheduler's view, not related to any specific request14:56
dulekwinston-d_: Ah, and then scheduler can update it's view when receiving updates from c-volume14:57
*** anshul has quit IRC14:57
*** e0ne has joined #openstack-cinder14:57
winston-d_jgriffith: unless we plan to do more stuff about scheduler not having 'complete' view situation in normal cinder request processing, we don't care if scheduler has complete view or not, for most of the time.14:58
winston-d_dulek: yes14:58
*** eharney has joined #openstack-cinder14:58
dulekSo it would be your PoC + doing _update_host_state_map on scheduler start14:59
openstackgerritPeter Wang proposed openstack/cinder: Fixes VNX NotImplementedError of unmanage  https://review.openstack.org/16253214:59
openstackgerritPeter Wang proposed openstack/cinder: Fixes VNX NotImplementedError of unmanage  https://review.openstack.org/16253215:00
DuncanTwinston-d_: With your patch to get the updates faster in the normal case, just waiting isn't such a problem, there are not normally any long stalls, so I'm not as bothered as I was15:00
winston-d_dulek: yes.15:00
rhagartyhemna, https://blueprints.launchpad.net/python-cinderclient/+spec/volume-type-description15:01
*** TobiasE has joined #openstack-cinder15:02
dulekwinston-d_, DuncanT, jgriffith: So how should I proceed with the patch now?15:03
winston-d_DuncanT: but we still need to figure a way to be able to dynamically alter 'scheduler_delay', otherwise, the 'just wait' solution will have a bunch of request waiting for 'scheduler_delay'15:03
winston-d_DuncanT: or, the simplest solution is, just wait, but change 'scheduler_deplay' to a much shorter duration, say, 5 sec?15:04
jgriffithdulek: I'm going to defer to winston-d_ and DuncanT as they've been more involved with this than myself as of late15:04
e0nerhagarty: hi. is bp for cinder approved? for patch https://review.openstack.org/#/c/140906/ you provided link for cinderclient bp15:04
*** TobiasE1 has quit IRC15:04
DuncanTwinston-d_: 5 seconds might well not be enough with a large, busy system though15:04
winston-d_DuncanT: that's true.15:04
*** agentle_ has quit IRC15:05
DuncanTwinston-d_: I prefer retrying again now - we aren't contacting a backend in the case of failure, just doing some internal work, so it isn't a cost15:05
dulekDuncanT, winston-d_: We can increase the backoff in retries.15:05
*** yuriy_n17 has joined #openstack-cinder15:05
DuncanTdulek: Optimally we should retry after each status comes in15:06
*** lpetrut has quit IRC15:06
dulekDuncanT: This increases complexity of the patch as we're unable to use utils.retry decorator.15:06
rhagartyeOne: taking a look now. Originally there was one blueprint to handle both vol type description and name change, but we split up the patches - one for description (already merged), and one for name change15:07
DuncanTdulek: Backing off doesn't buy us much, the retry is only an internal datastructure lookup, it really doesn't matter if we retry a thousand times, it costs us nothing since the process can't do anything useful anyway15:07
winston-d_DuncanT: not easy to get notified for that stats coming-in event though.15:07
*** vilobhmm has joined #openstack-cinder15:08
dulekwinston-d_: It's possible to notify TaskFlow's flow from the outside.15:08
dulekwinston-d_: But it would get code more complicated which when TaskFlow is involved I would prefer to avoid15:08
winston-d_DuncanT: yeah, worst case, one bad request (e.g. with wrong scheduler hint) comes in, and scheduler process eats one processor core. :)15:08
winston-d_DuncanT: that was my motivation of scheduler raising new exception to let caller knows if scheduler has all stats it needs.15:11
dulekwinston-d_: This won't fix blocking the thread tough.15:12
*** vilobhmm has left #openstack-cinder15:12
winston-d_dulek: what blocking thread?15:12
winston-d_dulek, DuncanT: if scheduler is able to adjust 'schedule_delay' based on stats it received, i think it's better to just wait and check (and don't start taskflow).15:13
dulekwinston-d_: Maybe I've misunderstood DuncanT, [16:07:04]15:13
rhagartyhemmna, eOne: here is the Cinder blueprint which is approved for kilo - https://blueprints.launchpad.net/cinder/+spec/volume-type-description15:14
dulekwinston-d_:Is it? You mean with your PoC patch?15:15
rhagartyhemna, e0ne: should I add additional blueprint tag in all of the patches?15:17
winston-d_dulek: no, my PoC doesn't do that, but I can change that if that's desired.15:18
openstackgerritBrianna Poulos proposed openstack/cinder: Add project_id to barbican keymgr wrapper  https://review.openstack.org/15587515:19
rhagartyhemna, e0ne: unfortunately, both the cinder and python-cinderclient BPs have the same name15:19
dulekwinston-d_: So host_manager and scheduler_driver would report when all the stats are received?15:19
hemnarhagarty, yes, please add the bp info in the commit message15:19
*** topshare has quit IRC15:20
*** dannywilson has joined #openstack-cinder15:20
e0nei've asked Duncun earlier, now i would ask other cores to review https://review.openstack.org/162488. it's a patch for openstack-dev15:21
*** dannywilson has quit IRC15:21
e0nei mean for devstack15:21
winston-d_dulek: not report, but flip a switch, like changing a flag 'open_to_business' once host manager has got all stats.15:21
*** dannywilson has joined #openstack-cinder15:21
*** boris-42 has quit IRC15:22
*** Mandell has quit IRC15:22
winston-d_dulek: scheduler manager can sleep and check for self.host_manager.open_for_business flag to come true and then start feeding requests to driver.15:22
*** david-lyle_afk is now known as david-lyle15:23
openstackgerritBrianna Poulos proposed openstack/cinder: Add project_id to barbican keymgr wrapper  https://review.openstack.org/15587515:23
hemnae0ne, so the volume type description/name change is all good.  the confusion is that the link in the commit message is pointing to the cinderclient BP15:23
hemnae0ne, the real BP for cinder is all good.15:23
winston-d_well, in that sense, 'schedule_delay' is redundent.15:23
dulekwinston-d_: True.15:23
e0nehemna: thanks for clarification of it! now i can +2/+A for it15:23
dulekwinston-d_: manager doesn't have access to host_manager I think, so this would need to be reported trough the driver also.15:24
hemnae0ne, thanks15:24
dulekwinston-d_: I mean scheduler_driver.15:24
winston-d_dulek: correct15:24
rhagartyhemna, e0ne: thanks!15:24
dulekwinston-d_: This is still doable from what I see.15:25
winston-d_dulek: yes, it is.15:26
e0nerhagarty, hemna: where is link for bp in cinder? i can't find it:(15:27
*** lpetrut has joined #openstack-cinder15:27
*** zhipeng_ has quit IRC15:27
rhagartye0ne: https://blueprints.launchpad.net/cinder/+spec/volume-type-description15:27
dulekwinston-d_: So in that case we're getting rid of retries but we stop sleeping just once we have the stats so long delay shouldn't be a problem.15:28
e0nerhagarty: thanks!15:28
dulekwinston-d_: This seems fine, I'll start reworking the patch.15:28
*** pradipm has joined #openstack-cinder15:28
*** emagana has joined #openstack-cinder15:28
winston-d_dulek: yes15:29
*** nlevinki has quit IRC15:29
winston-d_dulek: please split the change into two parts: 1st one for scheduler change; 2nd one is to address the bug15:29
dulekOkay, I'll try to do that. And of course these should depend on your change.15:30
e0nehemna: i've rebased your patch on my with shadow tables: https://github.com/e0ne/cinder/commit/8ae973852fdf11b7c01f9e268361fa1d319846d4. sqlite downgrade looks a bit ugly :(15:31
hemnae0ne, I'm about to push a new patchset up for it.  covering the comments15:31
winston-d_dulek: yes, the init_host_after_rpc is still very important fix to shorten the time for scheduler to gather all stats.15:32
hemnae0ne, ouch yah, the shadow tables makes a mess of migrations.   :(  I really don't like it.15:32
*** kallebe has quit IRC15:32
e0nehemna: yep:(15:32
*** shyama has quit IRC15:33
e0nei'm thinking how can we move it to oslo.db15:33
*** tsekiyama has joined #openstack-cinder15:34
winston-d_dulek: ok, it's close to midnight, i need to drop off now. looking forward to your new change(s). ttyl.15:35
dulekwinston-d_: Thank you!15:35
*** winston-d_ has quit IRC15:35
*** changbl has joined #openstack-cinder15:35
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder: Add volume multi attach support  https://review.openstack.org/8584715:37
*** delattec has joined #openstack-cinder15:37
*** cdelatte has quit IRC15:39
*** boris-42 has joined #openstack-cinder15:39
*** annashen has joined #openstack-cinder15:40
*** esker has joined #openstack-cinder15:42
*** nlevinki has joined #openstack-cinder15:44
*** annashen has quit IRC15:44
katcois the size specified in a create volume request in GB or GiB?15:45
*** Apoorva has joined #openstack-cinder15:46
*** changbl has quit IRC15:47
e0nekatco: do you mean GB or MB?15:48
e0nekatco: oh.. i've got your question15:50
katcoe0ne: the documentation states GB, but i wanted to double check what the actual implementation is15:51
DuncanTkatco: 1024 * 1024 * 1024 bytes15:57
katcoDuncanT: so it looks like that's a GiB15:58
DuncanTkatco: Pass, sorry, I never figured those units out15:58
DuncanTpradipm: Hi15:58
katcoDuncanT: hehe15:58
katcoDuncanT: tyvm for the answer15:59
*** crose has joined #openstack-cinder15:59
*** markvoelker has quit IRC15:59
pradipmDuncanT: I was looking into the 'get_pools', where intent is to add a volume-type parameter to retrieve all the pools matching the extra-specs present in the vol-type15:59
*** markvoelker has joined #openstack-cinder16:00
*** kallebe has joined #openstack-cinder16:00
*** bill_az has joined #openstack-cinder16:01
DuncanTpradipm: Seems useful16:01
*** bkopilov has quit IRC16:02
pradipmDuncanT:I can pass another parameter as voltype(default to None). But we have already one parameter (most likely unused), called filters. Which is most likely not used/supported.16:03
*** changbl has joined #openstack-cinder16:03
*** nlevinki has quit IRC16:03
*** emagana has quit IRC16:04
*** markvoelker has quit IRC16:05
*** rmesta has joined #openstack-cinder16:06
DuncanTpradipm: I don't know that code well enough to comment without spending some time reading it16:06
*** rmesta has left #openstack-cinder16:07
pradipmDuncanT:sure. thanks. sorry to disturb you.16:08
*** emagana_ has joined #openstack-cinder16:08
DuncanTpradipm: No problem. I'll get back to you when I've looked at the code16:09
pradipmDuncanT:I shall ping you after 19th.16:09
*** agentle_ has joined #openstack-cinder16:10
*** delatte has joined #openstack-cinder16:10
*** esker has quit IRC16:12
*** TobiasE has quit IRC16:13
*** delattec has quit IRC16:13
*** changbl has quit IRC16:13
*** tsekiyama has quit IRC16:14
*** emagana_ has quit IRC16:16
*** tsekiyama has joined #openstack-cinder16:17
*** emagana has joined #openstack-cinder16:17
*** ronis has quit IRC16:19
*** vilobhmm1 has joined #openstack-cinder16:20
*** emagana has quit IRC16:21
vilobhmm1thangp : ping16:21
*** thingee has quit IRC16:23
*** leeantho has joined #openstack-cinder16:24
*** hodos has joined #openstack-cinder16:24
openstackgerritSean McGinnis proposed openstack/cinder: Dell FC driver inheritance order causing failures  https://review.openstack.org/16307516:25
smcginnisAny cores ^^16:25
smcginnisCritical bug for us with trivial fix.16:25
openstackgerritharsh mishra proposed openstack/cinder: Fix for  inconsistent cinder-services state change  https://review.openstack.org/16010416:27
hemnasmcginnis, as soon as jenkins +1's I'll push it through16:28
*** stefan_amann has joined #openstack-cinder16:28
smcginnishemna: Thank you sir!16:28
*** e0ne is now known as e0ne_16:29
*** e0ne_ is now known as e0ne16:31
smcginnishemna, jgriffith: Appreciate the help!16:31
smcginnis+1 for third party CI as well. :)16:32
openstackgerritBob Callaway proposed openstack/cinder: Fix LUN misalignment issue with NetApp iSCSI drivers  https://review.openstack.org/16237616:32
*** alecv has quit IRC16:34
*** delattec has joined #openstack-cinder16:35
*** delatte has quit IRC16:38
*** _cjones_ has joined #openstack-cinder16:39
hemnasmcginnis, jdg beat me to it. :P16:39
smcginnishemna: :)16:40
*** jaypipes has joined #openstack-cinder16:40
*** harlowja_away is now known as harlowja_16:44
*** pradipm has quit IRC16:48
*** delatte has joined #openstack-cinder16:51
*** delattec has quit IRC16:51
*** markstur has quit IRC16:52
*** esker has joined #openstack-cinder16:56
*** lpetrut has quit IRC16:57
*** agentle_ has quit IRC16:58
*** timcl has joined #openstack-cinder16:58
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder: Namespace updates for _i18n and imageutils & fileutils  https://review.openstack.org/15795616:58
*** timcl1 has quit IRC16:58
*** markstur has joined #openstack-cinder16:58
*** timcl has quit IRC16:59
*** timcl has joined #openstack-cinder16:59
*** timcl has quit IRC17:00
*** timcl has joined #openstack-cinder17:00
*** timcl has quit IRC17:01
*** jistr has quit IRC17:01
*** timcl has joined #openstack-cinder17:01
*** timcl has quit IRC17:02
*** sbadia has quit IRC17:04
*** _cjones_ has quit IRC17:05
*** changbl has joined #openstack-cinder17:06
*** Mandell has joined #openstack-cinder17:06
*** Longgeek has quit IRC17:06
*** sbadia has joined #openstack-cinder17:07
*** jaypipes has quit IRC17:07
*** markvoelker has joined #openstack-cinder17:09
*** _cjones_ has joined #openstack-cinder17:09
*** annashen has joined #openstack-cinder17:09
*** krtaylor has quit IRC17:14
openstackgerritTom Barron proposed openstack/cinder: Refactor Swift backup driver and introduce chunking driver  https://review.openstack.org/14972517:17
stefan_amannhemna: on the cinder changes for System z: the cert results show two failures because our images are 3GB. Some of the test cases have fixed values for quotas. We're trying to figure out these places. I'm still working on a clean run. But can't promise we will achieve that in the next hours. Is a clean run a requirement to get our changes merged?17:18
hemnayah i would say so.17:19
*** markvoelker has quit IRC17:19
*** markvoelker has joined #openstack-cinder17:19
*** emagana has joined #openstack-cinder17:21
*** changbl has quit IRC17:21
*** markvoelker_ has joined #openstack-cinder17:21
*** patrickeast has joined #openstack-cinder17:22
jungleboyjstefan_amann: That was able to be worked around with the first run.17:23
*** markvoelker has quit IRC17:24
stefan_amannthe first run was on Intel. We are using different images there. The current run is on System z and our images are roughly 3 GB. And some of the test cases don't deal with it17:24
*** changbl has joined #openstack-cinder17:24
jungleboyjDoh!  Ok.17:25
*** krtaylor has joined #openstack-cinder17:27
*** timcl has joined #openstack-cinder17:27
openstackgerritMichal Dulko proposed openstack/cinder: Allow scheduler to receive volume stats when starting service  https://review.openstack.org/15862317:31
openstackgerritMichal Dulko proposed openstack/cinder: Add is_ready method to scheduler driver  https://review.openstack.org/16309817:31
openstackgerritMichal Dulko proposed openstack/cinder: Add waiting for the driver to SchedulerManager  https://review.openstack.org/16309917:31
*** lpetrut has joined #openstack-cinder17:31
*** stefan_amann has quit IRC17:32
*** coolsvap|afk is now known as coolsvap17:32
*** changbl has quit IRC17:33
*** dulek has quit IRC17:34
*** thingee has joined #openstack-cinder17:34
*** _cjones_ has quit IRC17:36
*** _cjones_ has joined #openstack-cinder17:37
*** esker has quit IRC17:40
*** changbl has joined #openstack-cinder17:41
*** e0ne has quit IRC17:42
openstackgerritTom Barron proposed openstack/cinder: Refactor Swift backup driver and introduce chunking driver  https://review.openstack.org/14972517:43
*** timcl has quit IRC17:49
*** rhagarty has quit IRC17:49
*** esker has joined #openstack-cinder17:51
thingeehemna: wrt https://review.openstack.org/#/c/85847/46/cinder/api/contrib/volume_actions.py17:53
thingeehemna: can you pass an instance_uuid or host and it'll just work?17:53
hemnathe current api is just detach w/o any params17:54
hemnabecause it's only allowed to be attached to 1 thing17:54
thingeeright, so take the instance uuid or host in the body17:54
hemnaw/ multiattach it's attachment_id17:54
thingeeinstead of attach id17:54
hemnabecause we don't know if it's a host or instance_uuid17:54
*** resker has joined #openstack-cinder17:54
thingeeI've talked to some of the ops folks here on the current idea and people agree it's a bit cumbersome.17:55
hemnaand the attachment_id is a unique identifier for that specific attachment.17:55
thingee1) look up the instance or host identifier17:55
thingee2) look up all attach ids and match it17:55
thingee3) give the attach id17:55
hemnathe attachment id comes with the attachments list on a volume17:55
hemnayou fetch the volume17:55
hemnait has all of the attachments17:55
hemnafind the one you want to detach, and pass it.17:55
thingeeI think you should just be able to pass what you want that volume to detach or attach from17:55
hemnaI disagree17:56
*** markus_z has quit IRC17:56
hemnathis is how it was designed from the get go17:56
thingeewell I got users here who disagree :)17:56
hemnawe churned on this for a long time.17:56
thingeeit's too bad the input is coming now, but they will ultimately be using it17:56
*** jdurgin has joined #openstack-cinder17:56
hemnathe host, or instance uuid isn't enough17:56
hemnabecause it's not guaranteed to be unique17:56
hemnato find that specific attachment.17:57
thingeewhy? if you have the volume and host/instance, look it up. what else is there going to be?17:57
hemnaonly the attachment id is unique for each individual attachment.17:57
*** esker has quit IRC17:57
*** ronis has joined #openstack-cinder17:57
hemnait can be attached multiple times.17:57
hemnawhich one do you get?17:57
hemnathere are 4 of them17:57
hemnadoesn't work.17:57
thingeeright but a volume is only going to be attach to one specific vm and another specific vm right?17:57
thingeeor host17:58
hemnanot on a vmware clustered system17:58
hemnawhere the volume looks like it's attached to the same thing multiple times.17:58
thingeethere are cases where you want a vm attached to the same exact vm?17:58
hemnathat's the vmware clustering model works.17:58
hemnait looks like it's attached to the same thing multiple times.17:59
*** stefan_amann has joined #openstack-cinder17:59
hemnathis is the only way it will always work, regardless.17:59
*** kallebe has quit IRC17:59
*** rhagarty has joined #openstack-cinder17:59
thingeebut why?17:59
*** victorfeitosa has quit IRC17:59
thingeejust because a company does it mean it's right for openstack.17:59
hemnaI just explained why17:59
thingeeyou said because vmware does it17:59
thingeethat's not a reason why they do it17:59
*** asselin_ has joined #openstack-cinder18:00
hemnawe want to be able to support vmware as the hypervisor for openstack.18:01
hemnaI'm not sure why shooting ourselves in the foot makes sense.18:01
hemnawhen it's a clean/simple api.18:01
thingeeAnyways, speaking for ops here.18:01
hemnawould have been nice if you asked these questions a year ago, when this was discussed then.18:02
thingeeI know this patch has been dragging, but I feel like you're letting your frustration get in the way of wanting to let people give you feedback, even if it is late.18:02
hemnayes, I'm frustrated18:03
*** victorfeitosa has joined #openstack-cinder18:03
hemnabecause it's awesome that folks ignore this work for over a year and now just dump on it.18:03
hemnait's not like this has been a surprise18:03
hemnawe even talked about it at the Paris summit18:03
thingeehave you ever solicited the ops list?18:03
hemnawhere were the reviews and discussions about it ?18:03
hemnawhy is this just happening now?18:03
*** Mandell has quit IRC18:04
hemnathat's not the point.18:04
*** jaypipes has joined #openstack-cinder18:04
thingeewell if you want feedback, the list might be a good place to start.18:04
*** bkopilov has joined #openstack-cinder18:05
*** changbl has quit IRC18:05
*** mathrock has joined #openstack-cinder18:05
hemnaat this point, I'm just ready to abandon the effort all together.  it's pointless.18:05
*** sgotliv has quit IRC18:06
openstackgerritMitsuhiro Tanino proposed openstack/cinder: Implement IET target driver  https://review.openstack.org/15882918:06
thingeeI'm really sorry you're frustrated and I know I wasn't involved with it this release, because it wasn't my particular focus that I told people I would do initially. I wasn't involved with the earlier releases either because initially we just gave up after the nova side wouldn't get merged.18:07
thingeeBut I've been trying to tell you that I want to work with you on this and make sure we merge it first thing when L starts after the version change merges.18:07
*** yuriy_n17 has quit IRC18:08
thingeebut I'll help ask people on feedback with the implementation. I also have interest in helping to test it with the concerns I had in the patch.18:09
*** IanGovett has quit IRC18:10
hemnaand you haven't been involved in this review the entire K release until now, just to prevent this from landing.18:11
hemnathat's fine, it's your job as PTL to say no.18:11
hemnaI get it.18:11
hemnaI'm over it.18:11
*** timcl has joined #openstack-cinder18:12
*** stefan_amann has quit IRC18:12
thingeeI would rather think it's my job to help you be successful. I think making sure you have priority in early L for landing to get sufficient gate testing and help encouraging the nova side is a good solution. But people want to rush this in K because they have no faith in my word on that.18:14
thingeelook at k-3. its been pretty successful in my opinion. I think we can do this for multi-attach too. I'll help do the rebase for you too.18:14
*** IanGovett has joined #openstack-cinder18:15
*** Mandell has joined #openstack-cinder18:15
*** victorfeitosa has quit IRC18:17
*** karimb has quit IRC18:24
*** takedakn has joined #openstack-cinder18:26
*** [1]Thelo has joined #openstack-cinder18:26
*** dulek has joined #openstack-cinder18:26
*** takedakn has quit IRC18:27
*** resker has quit IRC18:27
*** Thelo has quit IRC18:28
*** [1]Thelo is now known as Thelo18:28
*** annashen has quit IRC18:34
*** e0ne has joined #openstack-cinder18:37
*** _cjones_ has quit IRC18:38
*** changbl has joined #openstack-cinder18:44
*** agentle_ has joined #openstack-cinder18:50
*** _cjones_ has joined #openstack-cinder18:54
*** agentle_ has quit IRC18:55
*** agentle_ has joined #openstack-cinder18:55
*** sgotliv has joined #openstack-cinder19:00
*** markvoelker_ has quit IRC19:01
*** emagana has quit IRC19:01
*** Mandell has quit IRC19:01
*** emagana has joined #openstack-cinder19:02
e0nejungleboyj: hi Jay! do we still plan make oslo.middleware landed in Kilo?19:05
*** emagana has quit IRC19:06
*** rushiagr is now known as rushiagr_away19:07
*** Mandell has joined #openstack-cinder19:07
*** jaypipes is now known as jay-afk19:09
*** jay-afk is now known as jaypipes-afk19:10
*** angela-s has joined #openstack-cinder19:10
*** BharatK has left #openstack-cinder19:13
vilobhmm1thangp : ping19:14
vilobhmm1dulek : ping19:14
thangpvilobhmm1: hey19:14
vilobhmm1thangp : was working on quota objects19:15
*** rushiagr_away is now known as rushiagr19:15
dulekvilobhmm1: Hi19:15
vilobhmm1the way def quota_update(context, project_id, resource, limit): or quota_create is defined19:16
vilobhmm1which doesn't take "values" (dict) as an input in the create and save methods for the Quota class the method obj_get_changes19:16
vilobhmm1won't be applicable19:16
vilobhmm1so i will keep it simple19:17
vilobhmm1to accept project_id, resource, limit as input19:17
vilobhmm1or just to be in sync do you prefer changing the quota* api at the db/sqlalchemy layer to the way snapshot/volume/backup have theres19:17
vilobhmm1dulek : hi…please see above que ^^19:18
*** ndipanov has quit IRC19:18
vilobhmm1which doesn't take "values" (dict) as an input ; so in the create and save methods for the Quota class the method19:18
dulekvilobhmm1: Oh, I get it. I would prefer not to change existing DB API.19:19
thangpvilobhmm1: I suggest you look at https://github.com/openstack/nova/blob/master/nova/objects/quotas.py19:20
vilobhmm1sure thangp19:20
*** coolsvap is now known as coolsvap|afk19:20
vilobhmm1thanks thangp, dulek19:20
*** BharatK has joined #openstack-cinder19:20
thangpvilobhmm1: nova uses the same constructs was we do: quota_create(context, project_id, resource, limit)19:21
vilobhmm1thangp : or vice versa :) since we were intially part of nova :P thanks btwn19:22
*** erlon has joined #openstack-cinder19:22
*** Mandell has quit IRC19:23
*** bill_az has quit IRC19:24
openstackgerritMichal Dulko proposed openstack/cinder: Add is_ready method to scheduler driver  https://review.openstack.org/16309819:26
openstackgerritMichal Dulko proposed openstack/cinder: Add waiting for the driver to SchedulerManager  https://review.openstack.org/16309919:26
*** annashen has joined #openstack-cinder19:27
*** stefan_amann has joined #openstack-cinder19:30
*** emagana has joined #openstack-cinder19:31
openstackgerritMichal Dulko proposed openstack/cinder: Add is_ready method to scheduler driver  https://review.openstack.org/16309819:31
openstackgerritMichal Dulko proposed openstack/cinder: Add waiting for the driver to SchedulerManager  https://review.openstack.org/16309919:32
*** mathrock has quit IRC19:32
*** timcl has quit IRC19:35
*** hodos has quit IRC19:36
*** rushiagr is now known as rushiagr_away19:36
*** vishy has joined #openstack-cinder19:38
*** asselin_ has quit IRC19:38
jungleboyje0ne: Would be nice if we could.  Any progress with the change to Grenade?19:40
e0nejungleboyj: it was proposed to not make such changes in Kilo there19:41
jungleboyje0ne: Still hoping to get my changes for config-generator in but that isn't going to happen until tomorrow at the earliest.  Busy with meetings, code reviews, etc.19:41
e0nejungleboyj: but i don't know how could we deprecate middleware from oslo-incubator19:41
jungleboyje0ne: If they won't take the Grenade change, I guess we won't be able to do that until L.19:42
e0nejungleboyj: yes. did you see their comments?19:42
openstackgerritTom Barron proposed openstack/cinder: Refactor Swift backup driver and introduce chunking driver  https://review.openstack.org/14972519:42
jungleboyjNo, let me looking.19:43
*** Yogi11 has joined #openstack-cinder19:43
jungleboyje0ne: Oh, I did see that.  Ok, I guess we will need to take that approach instead.19:44
*** Mandell has joined #openstack-cinder19:44
*** Yogi1 has quit IRC19:47
*** dulek has quit IRC19:56
e0nejungleboyj: ok. what is the correct way to deprecate such things?19:57
*** aix has quit IRC19:57
jungleboyje0ne: Good question.  Guessing we need to Log a deprecation message when oslo.middleware is used.19:58
e0nejungleboyj: but we won't use it in Kilo19:58
e0neand we can't change code in incubator19:58
jungleboyje0ne: Doh.19:58
jungleboyjSorry, I meant openstack.common.middleware .19:59
jungleboyjYeah, we can't change that.19:59
jungleboyjHold on.19:59
jungleboyjtbarron: You there?19:59
tbarronjungleboyj: yup19:59
jungleboyjYou able to fix up a couple of comments on :  https://review.openstack.org/#/c/14972520:00
e0nejungleboyj: it's not such easy as i think:(20:00
*** bswartz has quit IRC20:01
jungleboyje0ne: Going to go chat with my friends in the oslo channel.20:01
openstackgerritTom Barron proposed openstack/cinder: Refactor Swift backup driver and introduce chunking driver  https://review.openstack.org/14972520:15
*** annashen has quit IRC20:15
*** takedakn has joined #openstack-cinder20:16
*** delatte has quit IRC20:19
*** angela-s has quit IRC20:21
*** casusbelli has joined #openstack-cinder20:25
*** akerr has quit IRC20:32
*** jungleboyj has quit IRC20:33
*** annashen has joined #openstack-cinder20:34
openstackgerritVipin Balachandran proposed openstack/cinder: VMware: Fail immediately for images in a container  https://review.openstack.org/16317020:37
*** emagana has quit IRC20:41
*** cdelatte has joined #openstack-cinder20:41
*** boris-42 has quit IRC20:42
*** tshefi has quit IRC20:42
*** tbarron has quit IRC20:44
*** esker has joined #openstack-cinder20:46
*** tbarron has joined #openstack-cinder20:47
*** tbarron1 has joined #openstack-cinder20:47
mtaninoe0ne: Thank you for the review :) https://review.openstack.org/#/c/161036/520:49
stefan_amannhemna: the cinder certification now runs successfully on System z. I had to install fixes for a new bug we opened. The fix reolves the issue that some of the tests implement a fix value for quotas and/or image file sizes. I hope the cinder changes for System z are now ready to be merged. Thanks for reviewing!20:49
*** ronis has quit IRC20:50
*** tbarron has quit IRC20:51
*** changbl has quit IRC20:58
*** vilobhmm1 has quit IRC21:03
*** vilobhmm1 has joined #openstack-cinder21:03
*** jungleboyj has joined #openstack-cinder21:07
*** thingee has quit IRC21:07
hemnastefan_amann, ok let me check :)21:12
stefan_amannthanks much!21:12
hemnaFailed: 021:13
*** angela-s has joined #openstack-cinder21:14
openstackgerritWalter A. Boring IV (hemna) proposed openstack/cinder: Adjust Cinder to support FCP on System z systems  https://review.openstack.org/14925621:14
*** emagana has joined #openstack-cinder21:14
openstackgerritEric Harney proposed openstack/cinder: Tests: Harden fake_notifier asserts  https://review.openstack.org/16319121:18
*** kaisers1 has joined #openstack-cinder21:19
*** boris-42 has joined #openstack-cinder21:22
*** rhagarty_ has joined #openstack-cinder21:26
*** dannywilson has quit IRC21:30
*** stefan_amann has quit IRC21:30
*** stefan_amann has joined #openstack-cinder21:30
*** Yogi11 has quit IRC21:31
*** dustins has quit IRC21:31
*** jcru has quit IRC21:40
*** chlong has joined #openstack-cinder21:40
*** crose has quit IRC21:43
*** mriedem has quit IRC21:44
*** stefan_amann has quit IRC21:45
*** casusbelli has quit IRC21:46
*** dflorea has joined #openstack-cinder21:47
*** vilobhmm1 has quit IRC21:49
*** jcru has joined #openstack-cinder21:49
*** vilobhmm1 has joined #openstack-cinder21:49
*** vilobhmm1 has quit IRC21:49
*** vilobhmm1 has joined #openstack-cinder21:50
*** lpetrut has quit IRC21:50
jungleboyjhemna: They got it!?!21:51
hemnalooks like they fixed it21:51
hemnajust waiting on jenkins21:51
*** thangp has quit IRC21:52
openstackgerritWalter A. Boring IV (hemna) proposed openstack/python-cinderclient: Update to change name for volume type client  https://review.openstack.org/14118721:52
*** jcru has quit IRC21:53
jungleboyjhemna: Thanks for bearing with us while we got everything worked out.21:53
hemnajungleboyj, np21:53
*** nikesh_vedams has quit IRC21:54
jungleboyjhemna: Looks like e0ne pushed up another patch for shadow tables.  https://review.openstack.org/#/c/131182/21:56
jungleboyjThought we were going to do what was discussed at the end of the day yesterday.21:56
e0nejungleboyj, hemna: i just rebased it. I'll set WIP status21:57
jungleboyje0ne: Ok.  Thanks.  Just wanted to make sure I hadn't missed something.21:57
*** vilobhmm1 has quit IRC21:57
*** vilobhmm1 has joined #openstack-cinder21:57
jungleboyjhemna: Looks like I still need to look through your latest multi-attach patch.21:59
*** jaypipes-afk is now known as JAYPIPES21:59
*** JAYPIPES is now known as jaypipes21:59
hemnajungleboyj, ok that'd be great thanks.   though, I think thingee is dead set against it.21:59
jungleboyjhemna: Understood, many others aren't though.22:00
jungleboyjDon't know what to say there.22:00
*** takedakn has quit IRC22:00
hemnayah, it is what it is.22:00
hemnaso, separate topic22:00
hemnaare we -2'd new feature patches now?22:00
*** takedakn has joined #openstack-cinder22:00
hemnasince the "deadline" has passed?22:00
e0nehemna: i'm sorry, didn't have a time to test your patch:(. i looked only on db-related code22:01
jungleboyjNot only are we before the real deadline but we have time to test and stabilize.22:01
jungleboyjhemna: I am not going to do that right now.  Lets talkt about that in tomorrow's meeting.22:02
hemnaok sounds good.22:02
hemnayah I  just wasn't sure what to do about in flight patches now.22:02
*** takedakn has quit IRC22:02
hemnaguess we also hold off on +3 ?22:03
*** coolsvap|afk has quit IRC22:03
jungleboyjWell, I think the System Z one is fine if you are good with it and the Swift recactor.  Those that are in the pipline right now that we ahve been watching.22:03
jungleboyjThe others on the list aren't ready to go and won't be by EOD.22:04
*** dflorea has quit IRC22:05
jungleboyjI need to drop off for a bit to drive home.  Back online shortly.22:05
hemnaok l8s22:05
mtaninohemna: I'm sorry I can't join tomorrow's meeting(you proposed CI topic for target object)22:05
hemnamtanino, it's ok, the topic is more of a general purpose discussion22:05
hemnaabout CI and target objects22:05
mtaninohemna: I got it. instead of me, anish will be join I think :)22:06
*** jungleboyj has quit IRC22:06
hemnaok sounds good22:06
*** chlong has quit IRC22:07
anishyep, I shall represent22:07
*** agentle_ has quit IRC22:07
*** annashen has quit IRC22:20
*** Lee1092 has quit IRC22:20
anishhemna: answered22:20
hemnaso...it's not needed? :P22:21
anishwell it's strictly "code cleanup"22:23
*** bswartz has joined #openstack-cinder22:23
anishbut I do think that many mocks in a test case is not good22:23
hemnawell it depends on what you are trying to test really.22:23
*** agentle_ has joined #openstack-cinder22:25
openstackgerritJulie Gravel proposed openstack/cinder: Validate volume type extra specs  https://review.openstack.org/16148022:31
*** rushil has quit IRC22:36
*** jungleboyj has joined #openstack-cinder22:46
*** agentle_ has quit IRC22:47
*** etta_ has joined #openstack-cinder22:49
*** jcru has joined #openstack-cinder22:51
*** jcru has quit IRC22:51
*** lcurtis has quit IRC22:52
*** eharney has quit IRC22:55
*** tbarron1 has quit IRC23:03
*** _cjones_ has quit IRC23:05
jgriffithanybody around that uses the initiator list method for sec?23:06
jgriffithxyang1: ping23:08
*** hemna is now known as hemnafk23:08
*** esker has quit IRC23:08
*** esker has joined #openstack-cinder23:09
xyang1jgriffith: Hi23:10
jgriffithxyang1: hi ya23:10
*** angela-s has quit IRC23:10
jgriffithxyang1: wondered if I could pick your brain on your use of iniitator mappings23:10
xyang1jgriffith: Sure23:11
jgriffithxyang1: so the idea is like an access group no?23:11
*** erlon has quit IRC23:11
jgriffithxyang1: so you can say "these hosts/initiators" have access and use that instead of chap23:11
jgriffithxyang1: and the auto-add thing just looks at connect info and sets up the access group automagically23:11
xyang1jgriffith: You'll have to add them on array in our case23:12
xyang1jgriffith: It is not automatic23:12
jgriffithahh... ok but other than that the basic principal is the same?23:12
jgriffithso admin setup up the groups on the array23:12
*** annashen has joined #openstack-cinder23:13
xyang1jgriffith: Ya, something like that23:13
jgriffithxyang1: so what i was wondering is..... a number of products have this sort of thing I think23:13
jgriffithLIO, zfs etc23:13
jgriffithand I noticed yours is a nice generic term23:13
jgriffithwas thinking of using it rather than introducing yet another one23:14
*** BharatK has quit IRC23:14
jgriffithbut I just realized it's explicitly in the emc section23:14
xyang1jgriffith: Oh, you are talking about anto zoning23:14
xyang1jgriffith: Np:)23:15
xyang1jgriffith: So this is for your FC driver?23:15
jgriffithxyang1: well, it can work for FC or iSCSI23:15
*** _cjones_ has joined #openstack-cinder23:15
jgriffithso the idea is instead of chap you can add a list of initiator IQN's or WWNN's23:16
xyang1jgriffith: ok23:16
jgriffithwe call it an access group23:16
xyang1jgriffith: So you want to combine them into one23:16
jgriffiththen you can add volumes to that access group and those initiators will be able to "see" the volumes in that group23:16
patrickeastinteresting, we have kind of the same concept but still do chap for the “host” that may have iqn’s/wwn’s on it23:16
jgriffithxyang1: well I was thinking about it... but I'm not sure it matters23:16
jgriffithxyang1: patrickeast yeah, I think there are similar concepts among a few of us23:17
patrickeasti setup our driver to create them automatically though23:17
jgriffithI hate to introduce sf specific conf options23:17
patrickeasti think the juno release they had to be setup beforehand on the array23:17
xyang1jgriffith: So the FC one goes thru lookup service23:17
jgriffithpatrickeast: yeah, I was thinking about doing that as well, but then I thought... maybe someobdy would want this for tenant isolation across compute nodes?23:18
xyang1jgriffith: That is specific to FC though23:18
jgriffithI dunno23:18
jgriffithxyang1: yeah, looking at that code now.... seems like it is a bit different than I thought23:18
jgriffithor at least than what I'm doing23:18
xyang1jgriffith: patrickeast only has iscsi23:18
jgriffithmaybe the config option thing is just a pet peave of mine and isn't a big deal23:18
patrickeastthe biggest reason we went for auto creation was the immediate complaints about adding dozens of iqn’s wwn’s manually23:19
jgriffithit's almost worse I guess to have a bilion options in the base driver :)23:19
patrickeasti’ll be adding FC in L23:19
jgriffithpatrickeast: yeah, I could see the objection to overhead you mention above23:19
xyang1jgriffith: I haven't looked at that in pure driver23:19
jgriffithxyang1: patrickeast ok, maybe I'll just mess with a few things first23:20
xyang1jgriffith: patrickeast will be intreresting to see how pure implement FC23:20
jgriffiththen figure out if there's value in common config options23:20
jgriffithxyang1: patrickeast thanks for the info23:20
xyang1jgriffith: Sure, common config options are good23:20
jgriffithxyang1: yeah... our list is growing :)23:20
xyang1jgriffith: Yes:)23:21
patrickeastxyang1: hehe yea i haven’t really looked into it yet… need to figure out how to do it23:21
patrickeastjgriffith: np23:21
jungleboyjAnyone know why grenade is failing for Kilo right now?23:29
*** Apoorva has quit IRC23:30
*** david-lyle is now known as david-lyle_afk23:33
jgriffithjungleboyj: got a recent fail log handy?23:37
jgriffithjungleboyj: not seeing anything in gerrit the past day or so23:37
jgriffithjungleboyj: guess not23:48
*** chlong has joined #openstack-cinder23:55
openstackgerritVipin Balachandran proposed openstack/cinder: VMware: Fix exception logging  https://review.openstack.org/16323123:58
openstackgerritVilobh Meshram proposed openstack/cinder: Quota Objects  https://review.openstack.org/16323223:59

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!