Friday, 2017-12-15

*** lennyb has joined #openstack-cinder00:02
*** ntpttr_laptop has quit IRC00:05
*** masber has joined #openstack-cinder00:12
*** wanghao has joined #openstack-cinder00:30
*** wanghao has quit IRC00:32
*** crushil_ has joined #openstack-cinder00:35
*** markvoelker has quit IRC00:36
*** xyang has joined #openstack-cinder00:37
*** xyang has quit IRC00:44
*** AlexeyAbashkin has joined #openstack-cinder00:45
*** wanghao has joined #openstack-cinder00:46
*** AlexeyAbashkin has quit IRC00:50
*** psachin has joined #openstack-cinder00:54
*** yangyapeng has joined #openstack-cinder01:10
*** crushil_ has quit IRC01:18
*** r-daneel has quit IRC01:26
*** wanghao_ has joined #openstack-cinder01:27
*** wanghao has quit IRC01:31
*** oomichi_afk is now known as oomichi01:36
*** linkmark has quit IRC01:40
*** dalgaaf has quit IRC01:43
*** dalgaaf has joined #openstack-cinder01:44
*** sapd has joined #openstack-cinder01:45
*** sapd has quit IRC01:55
*** sapd has joined #openstack-cinder01:57
*** sticker has quit IRC01:58
*** korean101 has joined #openstack-cinder02:05
korean101HI guys02:05
korean101cinder-volume can possible Active-Active HA? (in Newton Release + Ceph Backend)02:07
tommylikehuhey korean101  I guess no till now:)02:09
*** masber has quit IRC02:10
korean101tommylikehu: guess? hmmm. thanks02:10
korean101somebody tells me that in ceph backend can A/A HA02:10
*** mriedem has quit IRC02:11
*** stvnoyes has quit IRC02:12
*** liverpooler has joined #openstack-cinder02:12
korean101HA A-A: Add cluster configuration option to allow grouping hosts that share the same backend configurations and should work in Active-Active fashion. (
korean101what's this??02:13
*** psachin has quit IRC02:16
*** wanghao_ has quit IRC02:17
*** wanghao has joined #openstack-cinder02:18
*** salv-orlando has joined #openstack-cinder02:28
*** stakeda has quit IRC02:28
*** stakeda has joined #openstack-cinder02:29
*** salv-orl_ has quit IRC02:31
*** zengyingzhe has quit IRC02:32
*** Apoorva has quit IRC02:34
*** markvoelker has joined #openstack-cinder02:37
*** markstur has quit IRC02:38
*** harlowja has joined #openstack-cinder02:41
*** harlowja has quit IRC02:42
*** harlowja has joined #openstack-cinder02:43
*** namnh has joined #openstack-cinder02:45
tommylikehukorean101:  that's part of the works we need to support HA02:45
korean101tommylikehu: yes02:47
korean101tommylikehu: you think not yet supprorted A/A in HA?02:48
korean101tommylikehu: even Ceph RBD backend?02:48
korean101(in newton releases)02:48
tommylikehukorean101:  yea, I think so02:50
*** psachin has joined #openstack-cinder02:50
*** moshele has quit IRC02:54
*** imran_ansari has joined #openstack-cinder02:54
*** crushil_ has joined #openstack-cinder03:05
*** markvoelker has quit IRC03:11
*** harlowja has quit IRC03:15
*** jappleii__ has joined #openstack-cinder03:17
*** jappleii__ has quit IRC03:19
*** jappleii__ has joined #openstack-cinder03:19
*** threestrands_ has quit IRC03:20
*** yangyapeng has quit IRC03:24
*** abhishekk has joined #openstack-cinder03:30
korean101tommylikehu: many thanks!!!03:31
*** markstur has joined #openstack-cinder03:34
*** masber has joined #openstack-cinder03:38
*** ganso has quit IRC03:45
openstackgerritNeha Alhat proposed openstack/cinder master: V2/V3 jsonschema validation: snapshots
*** armax has quit IRC03:47
*** r-daneel has joined #openstack-cinder03:52
*** liverpooler has quit IRC03:54
*** jappleii__ has quit IRC03:56
*** armax has joined #openstack-cinder04:05
*** r-daneel has quit IRC04:07
*** markvoelker has joined #openstack-cinder04:08
*** jmlowe_ has joined #openstack-cinder04:12
*** jmlowe has quit IRC04:14
*** rmcall has joined #openstack-cinder04:15
*** armax has quit IRC04:17
*** vivsoni_ has quit IRC04:24
*** adisky__ has joined #openstack-cinder04:27
*** chhavi has joined #openstack-cinder04:29
*** s-shiono has joined #openstack-cinder04:32
*** vivsoni has joined #openstack-cinder04:33
*** harlowja has joined #openstack-cinder04:36
*** markvoelker has quit IRC04:42
*** harlowja has quit IRC04:45
*** pgadiya has joined #openstack-cinder04:48
*** zhurong has joined #openstack-cinder04:56
*** kwathore__ has quit IRC05:10
*** kwathore has quit IRC05:10
*** markstur has quit IRC05:15
*** kwathore has joined #openstack-cinder05:18
*** kwathore__ has joined #openstack-cinder05:18
*** harlowja has joined #openstack-cinder05:21
*** harlowja has quit IRC05:22
*** harlowja has joined #openstack-cinder05:24
*** harlowja has quit IRC05:29
kwathore@Team, Some one facing issue in attachment of lvm volume in Master setup, after attachment volume status become reserve and no specific entry found in /dev/disk/by-path05:32
kwathoreis there any activity going around in attachment of a volume?05:32
*** harlowja has joined #openstack-cinder05:38
*** markvoelker has joined #openstack-cinder05:39
openstackgerritMerged openstack/python-cinderclient master: Backup create is not available from 3.0 to 3.42
*** sapd_ has joined #openstack-cinder06:01
*** sapd has quit IRC06:01
*** sage__ has quit IRC06:08
*** sage__ has joined #openstack-cinder06:10
*** markvoelker has quit IRC06:11
*** links has joined #openstack-cinder06:18
*** wanghao_ has joined #openstack-cinder06:25
*** wanghao has quit IRC06:26
*** lpetrut has joined #openstack-cinder06:31
*** harlowja has quit IRC06:32
openstackgerritBrin Zhang proposed openstack/cinder master: add enough notification for QoS
*** rcernin has quit IRC06:47
openstackgerritNiraj Singh proposed openstack/cinder master: Add service_token for cinder-nova interaction
openstackgerritNiraj Singh proposed openstack/cinder master: Add service_token for cinder-glance interaction
*** markvoelker has joined #openstack-cinder07:09
openstackgerritChhavi Agarwal proposed openstack/cinder master: Run backup-restore operations on native thread
openstackgerritPooja Jadhav proposed openstack/cinder master: V3 jsonschema validation: Clusters
*** moshele has joined #openstack-cinder07:32
*** lpetrut has quit IRC07:35
openstackgerritqtlu proposed openstack/cinder master: [WIP] RBD: get manageable volumes
*** moshele has quit IRC07:42
*** markvoelker has quit IRC07:42
*** lpetrut has joined #openstack-cinder07:43
*** imran_ansari has quit IRC07:44
*** imran_ansari has joined #openstack-cinder07:44
openstackgerritImran Ansari proposed openstack/cinder master: 3PAR - Fix temp snapshot that remains post online copy
openstackgerritImran Ansari proposed openstack/cinder master: 3PAR - Fix temp snapshot that remains post online copy
*** armaan has joined #openstack-cinder07:54
*** zhurong has quit IRC07:54
*** AlexeyAbashkin has joined #openstack-cinder07:59
*** lpetrut has quit IRC08:05
*** crushil_ has quit IRC08:10
*** lhx_ has joined #openstack-cinder08:20
*** amoralej|off is now known as amoralej08:21
*** armaan has quit IRC08:23
*** armaan_ has joined #openstack-cinder08:23
*** armaan_ has quit IRC08:24
*** armaan has joined #openstack-cinder08:24
*** s-shiono has quit IRC08:34
*** wanghao_ has quit IRC08:34
*** wanghao has joined #openstack-cinder08:34
*** wanghao_ has joined #openstack-cinder08:37
*** markvoelker has joined #openstack-cinder08:39
*** wanghao has quit IRC08:40
*** imran_ansari has quit IRC08:41
*** wanghao_ has quit IRC08:41
*** wanghao has joined #openstack-cinder08:41
*** wanghao_ has joined #openstack-cinder08:43
*** wanghao has quit IRC08:46
*** salv-orl_ has joined #openstack-cinder08:55
*** salv-orlando has quit IRC08:59
*** armaan has quit IRC09:04
*** armaan has joined #openstack-cinder09:05
*** imran_ansari has joined #openstack-cinder09:07
*** stakeda has quit IRC09:08
*** markvoelker has quit IRC09:13
*** lhx_ has quit IRC09:25
*** lhx_ has joined #openstack-cinder09:25
*** armaan has quit IRC09:28
*** armaan has joined #openstack-cinder09:28
*** wanghao_ has quit IRC09:30
*** lhx__ has joined #openstack-cinder09:32
*** lhx_ has quit IRC09:33
*** prateek_ has quit IRC09:40
*** e0ne has joined #openstack-cinder09:43
*** armaan_ has joined #openstack-cinder09:52
*** armaan has quit IRC09:52
swamireddyquick Q: volume_clear=0 is this option wipes old data and overwrite with '0'09:56
swamireddyfor all types of volumes like lvm and ceph...09:56
*** prateek has joined #openstack-cinder09:57
e0neswamireddy: it's only for lvm and maybe  for nfs09:58
*** lpetrut has joined #openstack-cinder09:58
*** pgadiya has quit IRC09:59
*** mvk has quit IRC09:59
swamireddye0ne: it won't work with ceph volumes? is there way for ceph for the data wipe purpose?09:59
*** pgadiya has joined #openstack-cinder10:04
*** armaan_ has quit IRC10:06
*** armaan has joined #openstack-cinder10:07
*** hoangcx has quit IRC10:10
*** markvoelker has joined #openstack-cinder10:10
e0neswamireddy: you don't need it for ceph10:12
e0neswamireddy: ceph does it by itself10:12
imran_ansarie0ne: Regarding - here I'm spawning a worker thread, that is responsible for tracking asynchronous task(s) running on the array10:13
e0neimran_ansari: you have to refactor your code or tests (e.g. make tests more granular)10:13
e0neusing sleeps in unit tests is a very bad practice10:14
imran_ansarie0ne: I agree with your comment... However, please me explain what I'm trying to achieve in the unit test and then we can discuss if there is any other approach10:15
*** lhx__ has quit IRC10:15
imran_ansariPlease allow me to explain*10:15
imran_ansariThanks :)10:15
*** lhx__ has joined #openstack-cinder10:16
imran_ansariMy design of task-tracker is based on observer pattern10:16
imran_ansariWherein task tracker tracks status of asynchronous tasks running on array for the regitered listeners10:17
imran_ansariThe task tracker then allows the listeners to take some action for different status of task. As of now I've one listener that is responsible for clean up of a temporary snapshot post task-completion10:19
imran_ansariWe have other use cases as well that would be using this mechanism10:19
imran_ansariNow, in the unit test part, I'm mocking the client API for the array - so returning different task status is being mocked10:20
imran_ansariI'm returning two status as part of getTask.side_effect - 1. ACTIVE, 2. DONE10:21
imran_ansariSince getTask is invoked by Task Tracker on a worker thread, my main thread needs to wait until it finishes - there is a race condition here10:23
e0neseems that you test Task Tracker and driver feature in one test10:23
imran_ansariThere is a separate test as well for task tracker10:24
e0neif you split them into the several tests, you can mock task tracker in driver tests and implement new tests for task tracker10:24
imran_ansariAlso, the worker thread needs to keep running all the time - it doesn't exit until someone from outside does that10:24
imran_ansariSo the main thread cannot call thread.join() until it stops the worker thread from running10:25
imran_ansariNow the issue is when does the main thread stop the worker thread?10:25
imran_ansariHow would the main thread know that the worker thread invoked the function to delete temp snapshot?10:26
imran_ansariThis I'm achieving by checking how many times getTask got invoked by task tracker10:26
e0nedo we real need threads in unit tests?10:27
e0nemocks should be enopgh, I guess10:27
imran_ansariUnit test may not need it... But I've a funtionality that needs to delete tmp snapshot. How do I test that via unit test?10:28
*** mvk has joined #openstack-cinder10:28
imran_ansariAnd this functionality happens to run on a separate thread10:28
imran_ansariSo to continute the explanation - mock_tt.getTask.call_count allows the main thread to know that task tracker has executed it10:29
imran_ansariSince I've configured getTask.side_effect with two values, as soon as the count becomes 2 the main thread knows that delete snapshot logic is either about to be called or already called10:29
e0necan you just mock threads and checks that all methods were invoked with needed params?10:30
*** namnh has quit IRC10:31
imran_ansariThe thing is what I'm looking at is the action that must be eventually invoked on listener - was it invoked or not is what I thought I must check10:31
imran_ansariI could even mock task tracker... but then some of functionality would remain untested10:33
e0neit means that code should be refactored a bit to be more friendly for tests10:34
*** odyssey4me has joined #openstack-cinder10:34
odyssey4memorning folks - I'm seeing inconsistent information in the docs for upgrades and need clarification before pushing up a patch10:34
odyssey4mein it has the note: Assumed service upgrade order is cinder-api, cinder-scheduler, cinder-volume and finally cinder-backup.10:35
odyssey4meThen in it says that cinder-api should go first.10:35
odyssey4meSo which is it?10:35
odyssey4melol, my brain needs more coffee apparently10:36
odyssey4meah, ok - I figured out where the issue was - it's in
imran_ansarie0ne: The design i think is straight forward based on observer pattern... I think the complexity is there due to a separate thread10:36
*** abhishekk has quit IRC10:37
imran_ansariWhich required introduction to the sleep call as without it the scheduler doesn't allow worker thread to run :)10:38
e0neyou don't need running threads10:38
e0neyou need check that threads start calls were invoked10:39
*** daidv has quit IRC10:39
*** daidv has joined #openstack-cinder10:39
imran_ansariRight, but as I said earlier, is there a way to also check if the registered listener got notified of the task state or not and if the required action was performed by the listener?10:40
imran_ansarie0ne: I recorded the time the relevant unit tests take to complete10:41
imran_ansariThere are just three of them:10:42
imran_ansaritest_create_cloned_volume [0.114273s]10:42
imran_ansaritest_clone_volume_with_vvs [0.120239s]10:42
imran_ansaritest_task_tracker_delete_tmp_snapshot [0.115865s]10:42
imran_ansariDo you think these execution times are acceptable?10:43
*** markvoelker has quit IRC10:44
e0neexecution time looks good, but I care not only about it but for code too10:45
*** lhx_ has joined #openstack-cinder10:48
*** lhx__ has quit IRC10:49
imran_ansarie0ne: I agree. I would request you to have a look at the code, in case you haven't had a chance to do so, and please suggest whatever is best10:49
imran_ansariBTW, we have a precedence of using sleep in unit test :)10:50
*** daidv has quit IRC10:50
*** daidv has joined #openstack-cinder10:51
e0neimran_ansari: could you please point me on such test? I'll fix it:)10:51 -> test_create_delete_volume_with_encrypted_volume_type10:51
e0nethanks. we have to fix it10:52
*** abhishekk has joined #openstack-cinder10:53
imran_ansariAlso please suggest how we can test listener in conjunction with task tracker?10:54
e0neI have to go deeper into your code to provide such advice. unfortunately, I have no time for it today:(10:57
e0nebut I'll be happy to review new patch if any10:58
e0neif community decides to approve this patch even with sleeps on weekly meeting or mailing list - I'll remove my -210:58
imran_ansariI've pushed one today with Jay's comments addressed. But it has sleep :(10:58
*** daidv has quit IRC10:58
imran_ansariIt would be helpful though if u could have a look at the task tracker implementation and provide your comments. If you are ok I can run you through the code quickly10:59
imran_ansariBased on your availability some time early next week if not today11:00
e0neI can't promise anything but I will try11:00
imran_ansariThanks Ivan for your time! I appreciate! :)11:01
*** abhishekk has quit IRC11:05
*** abhishekk has joined #openstack-cinder11:09
*** abhishekk has quit IRC11:13
*** wanghao has joined #openstack-cinder11:23
openstackgerritLucian Petrut proposed openstack/os-brick master: Windows: fix connectors 'disconnect_volume' signature
openstackgerritLucian Petrut proposed openstack/os-brick master: Windows SMBFS: allow mounting vhd/x images
*** wanghao has quit IRC11:27
*** wanghao has joined #openstack-cinder11:27
*** sapd_ has quit IRC11:28
*** sapd_ has joined #openstack-cinder11:28
*** markvoelker has joined #openstack-cinder11:41
openstackgerritHelen Walsh proposed openstack/cinder master: VMAX doc - Queens feature documentation
*** wanghao has quit IRC11:42
*** wanghao has joined #openstack-cinder11:42
*** moshele has joined #openstack-cinder11:48
*** wanghao has quit IRC11:57
*** imran_ansari has quit IRC11:57
*** chrisyang_0660 has quit IRC12:04
*** erlon has joined #openstack-cinder12:05
*** markvoelker has quit IRC12:13
openstackgerritJeremy Zhang proposed openstack/cinder master: RBD: get manageable volumes
*** dave-mccowan has joined #openstack-cinder12:19
*** mvenesio has joined #openstack-cinder12:21
*** zhaochao has quit IRC12:21
*** dave-mcc_ has joined #openstack-cinder12:22
*** dave-mccowan has quit IRC12:24
pooja_jadhavtommylikehu: Hi12:26
*** catintheroof has joined #openstack-cinder12:31
*** lhx_ has quit IRC12:50
*** abhitechie has joined #openstack-cinder12:54
*** markvoelker has joined #openstack-cinder12:56
*** markvoelker has quit IRC12:58
*** abhitechie has quit IRC13:13
openstackgerritKushal Wathore proposed openstack/cinder master: HPE 3PAR - Implement Tiramisu feature on 3PAR
tommylikehupooja_jadhav:  hi13:18
*** lhx_ has joined #openstack-cinder13:20
*** _alastor__ has quit IRC13:22
*** _alastor__ has joined #openstack-cinder13:22
*** melwitt has quit IRC13:23
*** melwitt has joined #openstack-cinder13:25
*** melwitt is now known as Guest7554713:25
*** pgadiya has quit IRC13:29
*** liverpooler has joined #openstack-cinder13:36
*** takedakn has joined #openstack-cinder13:48
*** mriedem has joined #openstack-cinder13:51
*** amoralej is now known as amoralej|lunch13:53
smcginnisodyssey4me: A bug has been opened on that and a patch put up to fix it.13:58
*** liverpooler has quit IRC13:59
*** liverpooler has joined #openstack-cinder13:59
odyssey4mesmcginnis to fix that review?13:59
smcginnisodyssey4me: Sorry, I just saw you pointing out the inconsistency earlier. Is there a duplicate patch?14:00
smcginnisodyssey4me: OK, just seeing you commented on that now. Just catching up. :)14:01
smcginnisodyssey4me: Yeah, looks like that needs some more work.14:02
odyssey4mesmcginnis well, the review says the docs are inconsistent - but I don't see the inconsistency14:03
smcginnisodyssey4me: There was another part that has a different order pointed out in the bug. So looks like we have the order in a few different parts, but they don't agree.14:03
*** psachin has quit IRC14:07
*** takedakn has quit IRC14:08
*** stvnoyes has joined #openstack-cinder14:23
*** links has quit IRC14:31
*** markvoelker has joined #openstack-cinder14:31
*** moshele has quit IRC14:33
e0nesmcginnis, jungleboyj: hi. should we add release note while removing deprecated config options?14:40
smcginnise0ne: Yeah, usually we've been putting notice of the removal under the "upgrades" section.14:41
*** shaner has quit IRC14:41
e0nesmcginnis: :(. it will be harder than I expected. now, I have to add release notes14:42
e0nesmcginnis: I'm going to remove ~5 deprecated config options now14:43
smcginnise0ne: It doesn't need to be much. Just something noting that they are actually removed so deployments hopefully see it and know they need to change their config if they missed them being marked deprecated.14:44
e0nesmcginnis: sure14:44
*** amoralej|lunch is now known as amoralej14:45
*** openstackgerrit has quit IRC14:48
*** rcernin has joined #openstack-cinder14:57
*** xyang1 has joined #openstack-cinder14:58
*** xyang1 has quit IRC14:59
*** xyang1 has joined #openstack-cinder14:59
*** xyang has joined #openstack-cinder15:00
*** xyang has quit IRC15:01
*** xyang has joined #openstack-cinder15:01
*** felipemonteiro_ has joined #openstack-cinder15:02
*** crushil_ has joined #openstack-cinder15:02
*** openstackgerrit has joined #openstack-cinder15:03
openstackgerritIvan Kolodyazhny proposed openstack/cinder master: Remove deprecated 'nas_ip' config option
openstackgerritIvan Kolodyazhny proposed openstack/cinder master: Remove deprecated HP 3PAR config options
openstackgerritIvan Kolodyazhny proposed openstack/cinder master: Remove deprecated 'netapp_eseries_host_type' config option
openstackgerritIvan Kolodyazhny proposed openstack/cinder master: Remove deprecated 'pybasedir' config option
*** armaan has quit IRC15:05
*** jmlowe_ has quit IRC15:05
*** xyang1 has quit IRC15:05
*** felipemonteiro__ has joined #openstack-cinder15:06
*** Apoorva has joined #openstack-cinder15:06
*** felipemonteiro_ has quit IRC15:09
*** harlowja has joined #openstack-cinder15:11
*** shaner has joined #openstack-cinder15:12
*** harlowja has quit IRC15:13
openstackgerritKushal Wathore proposed openstack/cinder master: HPE 3PAR - Implement Tiramisu feature on 3PAR
openstackgerritJeremy Zhang proposed openstack/cinder master: RBD: get manageable volumes
*** dustins has joined #openstack-cinder15:25
swamireddye0ne: Thanks...for ceph volumes, data wiping not needed...can you please let me know, how its doing @ ceph?15:28
*** Apoorva has quit IRC15:28
e0neswamireddy: I don't know details on it15:28
swamireddye0ne: ok,,NOP...I will dig more on this..Thank you15:29
*** markvoelker has quit IRC15:29
*** armaan has joined #openstack-cinder15:30
*** armax has joined #openstack-cinder15:31
*** rcernin has quit IRC15:41
*** abhitechie has joined #openstack-cinder15:42
*** vivsoni__ has joined #openstack-cinder15:44
vivsoni__smcginnis: Hi15:46
vivsoni__mriedem: has uploaded a new patchset on fixing the tempest failure (passing empty connector to terminate_connection)15:47
*** markstur has joined #openstack-cinder15:47
*** markstur has quit IRC15:47
*** markstur has joined #openstack-cinder15:48
vivsoni__smcginnis & Team, please review
*** mvk has quit IRC16:05
*** e0ne has quit IRC16:05
vivsoni__smcginnis: Great ... Thanks for your precious time :)16:09
vivsoni__smcginnis: Thanks for review16:10
adrianofrHey guys. Could you please this review when you have some time?16:12
smcginnisvivsoni__: No problem. Just been tied up with other things lately, so didn't get a chance to get back to that after the last update. ;)16:12
vivsoni__smcginnis: just to re-confirm it .. do we need to handle this in our drivers code ?16:14
smcginnisvivsoni__: Yes16:14
jungleboyjsmcginnis:  It is in the driver?16:15
smcginnis _______________________________16:15
smcginnis< The driver needs to be fixed! >16:15
smcginnis -------------------------------16:15
smcginnis        \   ^__^16:15
smcginnis         \  (oo)\_______16:15
smcginnis            (__)\       )\/\16:15
smcginnis                ||----w |16:15
smcginnis                ||     ||16:15
SwansonThe hell?16:18
*** markvoelker has joined #openstack-cinder16:18
*** Guest75547 is now known as melwitt16:18
erlonSwanson: exactly, all drivers will burn there16:18
smcginnisWe discussed this in channel and in a meeting a few months back and realized we all need to handle this situation.16:19
erlonsmcginnis: lost that contex, you mean about the new attachment API?16:23
*** crushil_ has quit IRC16:24
*** yangyapeng has joined #openstack-cinder16:24
*** crushil_ has joined #openstack-cinder16:24
Swansonerlon, I think this "passing empty connector to terminate_connection"16:26
*** abhitechie has quit IRC16:27
ildikovsmcginnis: with this particular terminate_connection issue, it's not the drivers, it's the microversion concept or just simply the inability to efficiently handle API changes...16:27
ildikovsmcginnis: my head is blowing up every single day that we're introducing painful workarounds everywhere to avoid bumping the Nova API microversion16:28
*** vivsoni__ has quit IRC16:28
*** vivsoni__ has joined #openstack-cinder16:29
ildikovsmcginnis: and just to the margin it's for a functionality that everyone wants to burn on hell fires and I'm not sure how many people rely on that particular volume state that wold change in this case16:29
smcginnisildikov: We needed to handle terminate_connection without a connector being passed in for the case of force detach. That was before multiattach.16:29
smcginnisildikov: But with multiattach, it appears we needed to handle that in the code to *not* pass that down to the driver if it was not a force detach situation.16:30
ildikovsmcginnis: well, we're reverting and working around also shelve which I think hit some of these connector issues16:30
*** vivsoni__ has quit IRC16:30
smcginnisildikov: Yeah, I think it was the shelve/unshelve scenario that uncovered this.16:30
*** vivsoni__ has joined #openstack-cinder16:30
ildikovsmcginnis: so with the new flow we shouldn't put the volume into 'in-use'16:30
smcginnisildikov: I haven't been following that part too close. I think nova does want it that way.16:31
ildikovsmcginnis: it should be in 'reserved' state, which means that deleting the attachment wouldn't go that far to having to deal with an empty connector as it would be deleted in the volume/api.py16:31
smcginnisildikov: I think that's how jgriffith had it, but then that was all reverted because someone somewhere maybe possibily on a Tuesday might have expected the old behavior.16:32
*** vivsoni__ has quit IRC16:33
*** itlinux has joined #openstack-cinder16:34
*** vivsoni__ has joined #openstack-cinder16:34
ildikovsmcginnis: if it's now all because of that one particular workaround we started to avoid a Nova API mv bump I'm still strongly against this and think we should bump that version rather and make it sane16:36
ildikovsmcginnis: even if mriedem will invest in voodoo dolls after this comment of mine16:36
mriedemi guess i'm not goign to get my results written up today16:39
mriedemildikov: the microversion bump only changes behavior for that version forward16:39
mriedemit doesn't fix anything for all the versions before that16:39
mriedemthe canonical way for a client to wait for a volume to be attached to a server is waiting for the volume status to be in-use16:39
mriedemregardless of the instance status16:40
mriedemthat's what novaclient does, that's what tempest does, etc16:40
mriedemif you change that to never go to in-use but reserved b/c of shelved offloaded, you potentially break a bunch of client code16:40
ildikovin a shelved offloaded case I would argue with what attaching actually means16:40
mriedemildikov: at this point it doesn't matter16:40
mriedemthat ship has sailed16:40
mriedemi was just trying to defend compute API stability questions in the 1-year release cycle thread last night16:41
ildikovit still blows my mind as I only saw half of the consequences before and even at that time this whole thing blew my mind16:41
ildikovon my todo list to read it16:41
*** AlexeyAbashkin has quit IRC16:41
jungleboyjmriedem: and smcginnis  Just to make sure I don't do something stupid ...  We do want this merged.  Right?16:44
vivsoni__jungleboyj: :-)16:44
jungleboyjvivsoni__:  Making sure you aren't leading me to hell.16:45
smcginnisjungleboyj: Yeah, that's almost besides the point to the overall issue IMO.16:45
smcginnisI'm only aware of force being a case where the drivers need to handle not having that.16:45
smcginnisSo I think it's fine to only send it to the drivers when the force flag is set.16:46
jungleboyjsmcginnis:  Ok, that was what I thought but wanted to tripple check.16:46
mriedemjungleboyj: smcginnis: probably need to talk with erlon on the bottom patch16:47
mriedemerlon: see my reply?16:47
mriedemdep order doesn't matter here, they are both separate bugs that impact different CI systems in different ways16:47
mriedemif you reverse the dep order, the other one will fail some CIs16:47
smcginnisBottom patch?16:47
mriedemthey have to both be merged16:47
erlonmriedem: not yet, Ill look to it16:47
mriedem fixes FC backends16:48
mriedemwhere wwpns in the connector is a list16:48
mriedem fixes 3HPAR or whatever backends that don't handle a None connector passed to terminate_connection16:48
mriedemi'm not going to squash those changes16:48
erlonmriedem: how do I know that the first patch will pass if the second is merged? for me that look a dependency16:49
erlonmriedem: furthermore, if all will merge anyways doesnt not change for you if they are chained, but we are sure that they all work together16:50
erlonmriedem: Im not suggesting a squash, just a dependency so all are tested together16:52
erlonmriedem: anyway merged, lets see how that goes16:53
erlonjungleboyj: wanna join the merge party :D16:57
jungleboyjerlon:  Will do when I can take a look.16:57
erlonjungleboyj: thanks!16:58
*** sapd_ has quit IRC17:06
*** lhx_ has quit IRC17:06
*** dustins has quit IRC17:06
*** ntpttr_laptop has joined #openstack-cinder17:09
*** amoralej is now known as amoralej|off17:11
openstackgerritAseel Awwad proposed openstack/cinder master: Check if already managed before manage snapshot
mnaseris there a gate issue in cinder or was this just a spurious timeout?
openstackgerritAseel Awwad proposed openstack/cinder master: Check if already managed before manage snapshot
smcginnismnaser: Spurious I think.17:18
*** crushil_ has quit IRC17:23
*** mvk has joined #openstack-cinder17:25
openstackgerritLucian Petrut proposed openstack/cinder master: [WIP - UT needed] SMBFS: allow snapshot ro attach
openstackgerritLucian Petrut proposed openstack/cinder master: SMBFS: fix detecting if a volume is in-use
openstackgerritLucian Petrut proposed openstack/cinder master: Add Windows volume backup support
*** lpetrut has quit IRC17:38
jungleboyjerlon:  Done.17:43
erlonjungleboyj: Thanks!17:44
jungleboyjerlon:  Welcome.17:46
*** crushil_ has joined #openstack-cinder17:50
*** ntpttr__ has joined #openstack-cinder17:55
openstackgerritSean McGinnis proposed openstack/cinder master: Correct documented service upgrade order
*** ntpttr_ has quit IRC17:56
*** rmcall has quit IRC18:00
*** imacdonn_ is now known as imacdonn18:01
*** aarefiev has quit IRC18:04
*** catintheroof has quit IRC18:07
*** catinthe_ has joined #openstack-cinder18:07
*** chhavi has quit IRC18:09
*** mvenesio has quit IRC18:11
*** jose-phillips has quit IRC18:15
*** jose-phillips has joined #openstack-cinder18:15
*** AlexeyAbashkin has joined #openstack-cinder18:16
*** AlexeyAbashkin has quit IRC18:21
*** AlexeyAbashkin has joined #openstack-cinder18:22
*** AlexeyAbashkin has quit IRC18:26
*** thiagolib has quit IRC18:29
*** nicolasbock has joined #openstack-cinder18:29
*** dustins has joined #openstack-cinder18:35
*** imacdonn has quit IRC18:35
*** imacdonn has joined #openstack-cinder18:36
*** rmcall has joined #openstack-cinder18:36
*** armaan has quit IRC18:38
*** lpetrut has joined #openstack-cinder18:39
openstackgerritMerged openstack/cinder master: NEC driver: delete an unused configuration parameter.
openstackgerritMerged openstack/cinder master: Store host connector in volume_attachment.connector column
openstackgerritMerged openstack/cinder master: Don't call driver.terminate_connection if there is no connector
*** mriedem has quit IRC18:51
*** dalgaaf has quit IRC18:51
*** ntpttr_laptop has quit IRC18:52
*** lpetrut has quit IRC18:53
*** e0ne has joined #openstack-cinder19:00
*** dalgaaf has joined #openstack-cinder19:02
*** ntpttr_laptop has joined #openstack-cinder19:07
*** crushil_ has quit IRC19:08
openstackgerritSean McGinnis proposed openstack/cinder master: Add thin provisioning package to install guide
*** crushil_ has joined #openstack-cinder19:09
openstackgerritMerged openstack/cinder master: NetApp ONTAP: Copy offload bugfix
*** catinthe_ has quit IRC19:21
*** e0ne has quit IRC19:22
*** vivsoni__ has quit IRC19:31
*** moshele has joined #openstack-cinder19:32
*** ntpttr_laptop has quit IRC19:32
*** moshele has quit IRC19:43
*** moshele has joined #openstack-cinder19:45
*** dalgaaf has quit IRC19:57
*** dalgaaf has joined #openstack-cinder20:08
*** moshele has quit IRC20:22
*** armaan has joined #openstack-cinder20:27
*** dustins has quit IRC20:31
imacdonnTrying to figure out if mriedem's fix will address the Oracle ZFSSA blow-ups ... maybe I'm not reading it right, but I think it won't20:37
imacdonnseems that terminate_connection() is getting a connector that's not None, but it has no 'initiator'20:37
*** e0ne has joined #openstack-cinder20:40
*** e0ne has quit IRC20:42
Swansonimacdonn, Yeah. So if you need that to whack a connection then you have an issue.20:47
imacdonnI'm trying to figure out if it's valid to call an iSCSI volume driver with a connector that has no initiator20:47
imacdonndoesn't seem valid to me .... ???20:48
*** openstackgerrit has quit IRC20:48
imacdonnsmcginnis, jungleboyj: open to comments on above ;)20:48
Swansonimacdonn, That was my take but I believe that take was over ridden.20:48
imacdonnI can probably just exist as a noop (or something?) if there's no initiator in the connector.... but don't think I should have to, and not really sure what the right action would be (do nothing, or detach the volume from all initiators??)20:50
imacdonnjust exit* (as in return)20:50
smcginnisimacdonn: Do you know how it is getting it with no initiator?20:51
imacdonnsmcginnis: I know that it's getting passed a connector that is not None (because it specifically checks for that), but it then tries to get connector['initiator'] and barfs a KeyError20:52
smcginnisimacdonn: But under what circumstance is that happenign?20:53
imacdonnseems to be related to AttachVolumeShelveTestJSON20:54
imacdonnthis is CI issue, if that wasn't obvious20:54
SwansonI think it should do that with Force detach, too, right?20:54
smcginnisimacdonn: OK, if it's related to shelve then mriedem's patch is not going to help with that.20:55
smcginnisForce detach won't have a connector at all, so that will still be sent through to the driver.20:55
smcginnisAnd the driver needs to handle that and delete any attachements.20:55
imacdonnright, and I wrote code to specifically handle that20:55
smcginnisBut in this case the check will see that there is a connector and pass it through.20:55
imacdonnit assumes that if connector is None, it's expected to disconnect from any and all initiators20:56
smcginnisSo there's probably something on the nova side with shelve/unshelve that mucks with that, but probably a better question for mriedem.20:56
imacdonnright. What's his typical schedule?20:57
smcginnisUsually around now, but appears to have taken off.20:58
smcginnisHe's central US time.20:58
imacdonnOK. I'll go grab lunch and hope he pops back in then .. thanks ;)20:58
*** itlinux has quit IRC20:59
*** sdague has quit IRC21:09
*** openstackgerrit has joined #openstack-cinder21:10
openstackgerritJohn Griffith proposed openstack/cinder master: Check for migrated UUID in SolidFire delete
*** jmlowe has joined #openstack-cinder21:11
*** armaan has quit IRC21:14
*** armaan has joined #openstack-cinder21:15
*** rmcall has quit IRC21:16
*** jmlowe has quit IRC21:17
*** openstackgerrit has quit IRC21:18
*** openstackgerrit has joined #openstack-cinder21:23
openstackgerritOpenStack Proposal Bot proposed openstack/cinder master: Updated from global requirements
*** dustins has joined #openstack-cinder21:28
imacdonnsmcginnis: where did we end up with the test_snapshot_create_volume_description_non_ascii_code issue ... where it seemed that something was assuming that the description for a snapshot would be copied from the parent volume ....21:34
*** rcernin has joined #openstack-cinder21:36
*** ntpttr_laptop has joined #openstack-cinder21:36
*** itlinux has joined #openstack-cinder21:40
smcginnisimacdonn: Looks like that has merged:
imacdonnsmcginnis: hmm, I'm still seeing failures (apparently) related to it21:42
imacdonnTraceback (most recent call last):21:42
imacdonn  File "/opt/stack/cinder/cinder/tests/tempest/api/volume/", line 70, in test_snapshot_create_volume_description_non_ascii_code21:42
imacdonn    self.assertEqual(description, snapshot_info['description'])21:42
imacdonn  File "/usr/local/lib/python2.7/dist-packages/testtools/", line 411, in assertEqual21:42
imacdonn    self.assertThat(observed, matcher, message)21:42
imacdonn  File "/usr/local/lib/python2.7/dist-packages/testtools/", line 498, in assertThat21:42
imacdonn    raise mismatch_error21:42
imacdonntesttools.matchers._impl.MismatchError: u'\u05e7\u05d9\u05d9\u05e4\u05e9' != None21:42
imacdonndoes that mean that the submitters haven't updated their branches ?21:43
openstackgerritMarc proposed openstack/cinder master: Solving permission errors due to directory ownership on NFS
imacdonnnot sure where all the bits come from21:43
smcginnisimacdonn: That's possible. They may just need to rebase on current master.21:44
*** jmlowe has joined #openstack-cinder21:44
imacdonnyeah, they're not all failing on that ... most are just failing on the shelve thing .. thanks21:46
smcginnisimacdonn: Cool, that was easy. :)21:47
*** openstackgerrit has quit IRC21:48
*** jmlowe has quit IRC21:49
*** erlon has quit IRC21:50
*** openstackgerrit has joined #openstack-cinder21:50
openstackgerritSean McGinnis proposed openstack/cinder master: Fixes creation of mirrored volumes due to wrong type
openstackgerritOpenStack Proposal Bot proposed openstack/os-brick master: Updated from global requirements
*** itlinux has quit IRC22:01
*** crushil_ has left #openstack-cinder22:05
*** mriedem has joined #openstack-cinder22:07
*** dave-mcc_ has quit IRC22:10
*** nicolasbock has quit IRC22:11
*** dustins has quit IRC22:23
*** adrianofr has quit IRC22:28
*** armaan has quit IRC22:28
*** catintheroof has joined #openstack-cinder22:29
*** stvnoyes has quit IRC22:30
imacdonnmriedem: ping22:34
imacdonnmriedem: see discussion while you were out, about terminate_connection() getting passed a connector that is not None, but has no 'initiator' (it's an empty dict)22:36
*** AlexeyAbashkin has joined #openstack-cinder22:42
*** AlexeyAbashkin has quit IRC22:47
*** felipemonteiro__ has quit IRC23:35
*** Swimingly has quit IRC23:40
*** Swimingly has joined #openstack-cinder23:41
*** Swimingly has joined #openstack-cinder23:41
*** mriedem has quit IRC23:46
*** adisky__ has quit IRC23:56

Generated by 2.15.3 by Marius Gedminas - find it at!