| *** absubram has quit IRC | 00:02 | |
| *** cknight has quit IRC | 00:25 | |
| *** david-lyle_ is now known as david-lyle | 00:35 | |
| *** ameade has quit IRC | 00:57 | |
| *** ameade has joined #openstack-manila | 00:57 | |
| openstackgerrit | Victoria Martinez de la Cruz proposed openstack/manila-ui: Moves OPENSTACK_MANILA_SETTINGS TO local_settings.d/ https://review.openstack.org/397926 | 01:12 |
|---|---|---|
| openstackgerrit | zhongjun proposed openstack/manila-specs: Add spec for enable IPv6 in manila https://review.openstack.org/362786 | 01:18 |
| openstackgerrit | zhongjun proposed openstack/manila-specs: Add spec for enable IPv6 in manila https://review.openstack.org/362786 | 01:31 |
| *** tuanluong has joined #openstack-manila | 02:06 | |
| openstackgerrit | zhongjun proposed openstack/manila-specs: Add detail API to quota-set API collection https://review.openstack.org/390052 | 02:37 |
| *** kaisers_ has joined #openstack-manila | 02:44 | |
| *** kaisers_ has quit IRC | 02:49 | |
| openstackgerrit | Yingzhe Zeng proposed openstack/manila: Change huawei driver to only pass needed info while replica creating https://review.openstack.org/398717 | 03:05 |
| *** kaisers_ has joined #openstack-manila | 03:31 | |
| *** kaisers_ has quit IRC | 03:35 | |
| openstackgerrit | zhongjun proposed openstack/manila-specs: Add detail API to quota-set API collection https://review.openstack.org/390052 | 04:38 |
| *** lpetrut has joined #openstack-manila | 04:59 | |
| *** cdelatte has quit IRC | 05:08 | |
| *** wiebalck has joined #openstack-manila | 05:18 | |
| *** wiebalck has joined #openstack-manila | 05:19 | |
| *** digvijay2016 has joined #openstack-manila | 05:19 | |
| *** vponomaryov has quit IRC | 05:33 | |
| *** kaisers_ has joined #openstack-manila | 05:47 | |
| *** kaisers_ has quit IRC | 05:51 | |
| *** nkrinner_afk is now known as nkrinner | 06:00 | |
| *** lpetrut has quit IRC | 06:02 | |
| *** wiebalck has quit IRC | 06:10 | |
| *** belmoreira has joined #openstack-manila | 06:25 | |
| *** dsariel has joined #openstack-manila | 06:38 | |
| *** pcaruana has joined #openstack-manila | 07:18 | |
| openstackgerrit | Pony Chou proposed openstack/manila: Add QNAP Manila Driver https://review.openstack.org/394703 | 07:26 |
| openstackgerrit | Pony Chou proposed openstack/manila: Add QNAP Manila Driver https://review.openstack.org/394703 | 07:32 |
| *** kaisers_ has joined #openstack-manila | 07:48 | |
| *** akapil has joined #openstack-manila | 07:49 | |
| *** kaisers_ has quit IRC | 07:52 | |
| *** akapil has quit IRC | 07:53 | |
| *** nkrinner is now known as nkrinner_trainin | 07:56 | |
| *** lpetrut has joined #openstack-manila | 08:00 | |
| *** jprovazn has joined #openstack-manila | 08:02 | |
| *** nherciu has joined #openstack-manila | 08:09 | |
| *** kaisers_ has joined #openstack-manila | 08:23 | |
| *** akapil has joined #openstack-manila | 08:25 | |
| *** akapil has quit IRC | 08:29 | |
| *** gouthamr has joined #openstack-manila | 08:45 | |
| openstackgerrit | Merged openstack/manila: devref/driver_requirements: add cephfs protocol https://review.openstack.org/398564 | 08:53 |
| *** igajsin has joined #openstack-manila | 08:57 | |
| *** cknight has joined #openstack-manila | 09:03 | |
| *** cknight1 has joined #openstack-manila | 09:04 | |
| *** akapil has joined #openstack-manila | 09:05 | |
| *** gouthamr has quit IRC | 09:05 | |
| *** akapil has quit IRC | 09:07 | |
| *** cknight has quit IRC | 09:07 | |
| *** akapil has joined #openstack-manila | 09:11 | |
| *** TommyLikeHu has quit IRC | 09:12 | |
| *** TommyLikeHu has joined #openstack-manila | 09:13 | |
| *** akapil has quit IRC | 09:14 | |
| TommyLikeHu | hey any core member can take a look at this spec? https://review.openstack.org/#/c/390052/ much thanks~ | 09:14 |
| *** gouthamr has joined #openstack-manila | 09:15 | |
| *** ociuhandu has joined #openstack-manila | 09:21 | |
| *** akapil has joined #openstack-manila | 09:25 | |
| openstackgerrit | Xiaoyang Zhang proposed openstack/manila: Fix extend operation of shrinked share in generic driver https://review.openstack.org/393589 | 09:31 |
| *** rraja has joined #openstack-manila | 09:37 | |
| *** dsariel has quit IRC | 09:44 | |
| *** openstackgerrit has quit IRC | 09:48 | |
| *** openstackgerrit has joined #openstack-manila | 09:49 | |
| *** cknight1 has quit IRC | 09:49 | |
| *** akapil has quit IRC | 09:50 | |
| arnewiebalck_ | rraja: Good morning! Horizon logs for the dirty auth meta issue are here: http://paste.openstack.org/show/589553/ | 09:57 |
| arnewiebalck_ | rraja: IIRC, the user tried to create two rules which both ended in error, he deleted them, then he successfully recreated one. | 09:59 |
| openstackgerrit | Xiaoyang Zhang proposed openstack/manila: Fix share-service VM restart problem https://review.openstack.org/393594 | 10:03 |
| *** akapil has joined #openstack-manila | 10:04 | |
| openstackgerrit | Xiaoyang Zhang proposed openstack/manila: Fix share-service VM restart problem https://review.openstack.org/393594 | 10:08 |
| *** akapil has quit IRC | 10:08 | |
| rraja | arnewiebalck_: hey! i'm checking the logs. thanks! | 10:14 |
| *** jcsp has quit IRC | 10:16 | |
| rraja | arnewiebalck_: it's not clear to me from the horizon logs how the dirty auth meta issue was hit. can you share the corresponding errors that might have been recorded in m-shr or m-api logs when the user's access rules ended in error state? | 10:19 |
| *** vponomaryov has joined #openstack-manila | 10:19 | |
| arnewiebalck_ | rraja: I think these are the relevant lines: http://paste.openstack.org/show/589555/ | 10:24 |
| arnewiebalck_ | rraja: from m-shr | 10:25 |
| *** akapil has joined #openstack-manila | 10:26 | |
| *** akapil has quit IRC | 10:26 | |
| *** akapil has joined #openstack-manila | 10:27 | |
| *** dsariel has joined #openstack-manila | 10:33 | |
| *** tpsilva has joined #openstack-manila | 10:36 | |
| *** ganso has joined #openstack-manila | 10:36 | |
| rraja | arnewiebalck_: i'm wondering if the problematic code in ceph_volum_client is the same as is for the issue http://tracker.ceph.com/issues/17800 reported by toabctl. | 10:36 |
| rraja | arnewiebalck_: can you share the request/sent json which triggered this traceback? | 10:37 |
| arnewiebalck_ | rraja: As I didn’t have debugging enabled at the time, I don’t think I have that. | 10:41 |
| arnewiebalck_ | rraja: This would be in the client’s (i.e. the m-shr side’s) Ceph logs, right? | 10:42 |
| *** ociuhandu has quit IRC | 10:43 | |
| arnewiebalck_ | rraja: It may very well be that we manoeuvered ourselves into this situation as we touched the caps in order to work around some CephFS issues with Manila created shares. | 10:47 |
| rraja | arnewiebalck_: i'm not sure. but checking the ceph client logs is a good idea. | 10:47 |
| arnewiebalck_ | rraja: The client logs are basically empty. | 10:47 |
| rraja | arnewiebalck_: okay. | 10:48 |
| rraja | arnewiebalck_: so you're saying that it's possible that the mon caps of the auth IDs created by manila/volume client were manipulated outside manila/volumeclient? | 10:50 |
| arnewiebalck_ | rraja: Absolutely, yes. The problem would probably not have occurred if I didn’t share the office with our Ceph admin :) | 10:51 |
| *** gouthamr has quit IRC | 10:53 | |
| rraja | arnewiebalck_: oh okay. :) but it'll be interesting to know why this error was hit, as basically what the volumeclient does is fetch the existing mon caps of auth ID and reapplies it. so unless the MON caps of the auth ID was unset/not set as was the case in the tracker issue I already pointed out this error shouldn't have occurred? anyway I'll try figuring out how this error could've happened. | 10:56 |
| arnewiebalck_ | rraja: this is the bug we’re trying work around: http://tracker.ceph.com/issues/17858 | 10:58 |
| arnewiebalck_ | rraja: Yes, you can assume that the caps have been manipulated (if you don’t find any other explanation). | 11:02 |
| rraja | arnewiebalck_: thanks for sharing the issue that you're trying to work around. | 11:05 |
| rraja | arnewiebalck_: okay. | 11:05 |
| *** nkrinner_trainin has quit IRC | 11:05 | |
| *** khamtamtun has joined #openstack-manila | 11:06 | |
| *** nkrinner_trainin has joined #openstack-manila | 11:07 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 11:11 |
| *** akapil has quit IRC | 11:11 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 11:12 |
| *** khamtamtun has quit IRC | 11:12 | |
| *** ociuhandu has joined #openstack-manila | 11:14 | |
| *** kamtamtun has joined #openstack-manila | 11:14 | |
| *** ianychoi has quit IRC | 11:14 | |
| *** andreaf has quit IRC | 11:26 | |
| *** andreaf has joined #openstack-manila | 11:26 | |
| *** gouthamr has joined #openstack-manila | 11:32 | |
| *** andreaf has quit IRC | 11:36 | |
| *** andreaf has joined #openstack-manila | 11:36 | |
| *** akapil has joined #openstack-manila | 11:38 | |
| *** andreaf has quit IRC | 11:39 | |
| *** akapil has quit IRC | 11:42 | |
| *** andreaf has joined #openstack-manila | 11:43 | |
| *** kamtamtun has quit IRC | 11:45 | |
| *** gouthamr has quit IRC | 11:47 | |
| *** gouthamr has joined #openstack-manila | 11:51 | |
| *** kamtamtun has joined #openstack-manila | 11:52 | |
| *** kamtamtun has quit IRC | 11:53 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila-specs: Remove duplicated imports of specs into index https://review.openstack.org/398965 | 11:58 |
| ganso | vponomaryov: ^ when I first read it, seemed like a spec to remove duplicated specs lol | 12:00 |
| *** zhugaoxiao has quit IRC | 12:01 | |
| mkoderer__ | dmellado: if you have problems with dailing in we move over to hangout | 12:01 |
| *** zhugaoxiao has joined #openstack-manila | 12:01 | |
| dmellado | mkoderer__: I should be there, i got forwarded to a page saying that the organizer would soon admit me | 12:02 |
| *** mkoderer_ has joined #openstack-manila | 12:02 | |
| tbarron | mkoderer__: dmellado i think i'm in by phone | 12:02 |
| mkoderer__ | dmellado: mhhhh | 12:02 |
| *** akapil has joined #openstack-manila | 12:02 | |
| dmellado | let me check the number, I'll just do phone | 12:03 |
| *** tuanluong has quit IRC | 12:04 | |
| *** kamtamtun has joined #openstack-manila | 12:04 | |
| *** kamtamtun has quit IRC | 12:05 | |
| *** dustins has joined #openstack-manila | 12:05 | |
| *** digvijay2016 has quit IRC | 12:06 | |
| *** akapil has quit IRC | 12:07 | |
| mkoderer__ | https://etherpad.openstack.org/p/manila_no_tempest_deps | 12:10 |
| *** kaisers__ has joined #openstack-manila | 12:18 | |
| *** akapil has joined #openstack-manila | 12:19 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila-specs: Add py27 tox env to tox config to unblock CI https://review.openstack.org/398973 | 12:19 |
| *** kaisers_ has quit IRC | 12:20 | |
| *** mkoderer_ has quit IRC | 12:21 | |
| *** mkoderer_ has joined #openstack-manila | 12:21 | |
| *** akapil has quit IRC | 12:23 | |
| *** akapil has joined #openstack-manila | 12:37 | |
| *** catintheroof has joined #openstack-manila | 12:46 | |
| *** akapil has quit IRC | 12:49 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 12:51 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 12:51 |
| *** gouthamr has quit IRC | 12:52 | |
| *** chlong has joined #openstack-manila | 13:00 | |
| *** akapil has joined #openstack-manila | 13:09 | |
| *** akapil has quit IRC | 13:10 | |
| *** akapil has joined #openstack-manila | 13:11 | |
| vponomaryov | ganso, tbarron: manila-specs CI fix -> https://review.openstack.org/#/c/398973/ | 13:15 |
| *** akapil has quit IRC | 13:15 | |
| tbarron | vponomaryov: thanks, after that merges we can do the other one (that removes the duplicatses) | 13:19 |
| tbarron | ganso: ^^^^ you need it too | 13:19 |
| *** akapil has joined #openstack-manila | 13:21 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 13:23 |
| ganso | vponomaryov, tbarron: Thanks! | 13:23 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Data Jobs table https://review.openstack.org/392262 | 13:27 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Data Jobs table https://review.openstack.org/392262 | 13:28 |
| ganso | vponomaryov, tbarron: off to lunch, brb | 13:29 |
| tbarron | ganso: enjoy your early? lunch | 13:29 |
| *** gouthamr has joined #openstack-manila | 13:30 | |
| tbarron | vponomaryov: do you need to put 'commands = ' under [testenv] in https://review.openstack.org/#/c/398973/ ? | 13:31 |
| gouthamr | vponomaryov: ah crap | 13:32 |
| vponomaryov | why we have this py27 job at all? | 13:34 |
| vponomaryov | when we do not have tests | 13:34 |
| gouthamr | yes | 13:34 |
| gouthamr | we should probably get rid of it | 13:34 |
| gouthamr | nova has a couple of tests | 13:35 |
| vponomaryov | cinder has just 1 - https://github.com/openstack/cinder-specs/blob/master/tests/test_titles.py | 13:35 |
| gouthamr | ah, interesting... seems like they want to impose a format.. | 13:36 |
| vponomaryov | haha, after first test run CI bug got immunity against fix )) | 13:38 |
| vponomaryov | evolves | 13:38 |
| vponomaryov | ^_^ | 13:38 |
| *** porrua has joined #openstack-manila | 13:42 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila-specs: Add py27 tox env to tox config to unblock CI https://review.openstack.org/398973 | 13:42 |
| tbarron | vponomaryov: ^^ :) | 13:43 |
| TommyLikeHu | vponomaryov: hey could you take a look this spec, again? <vponomaryov> | 13:45 |
| TommyLikeHu | https://review.openstack.org/#/c/390052/ | 13:45 |
| gouthamr | vponomaryov tbarron: heh if i did that on my internal CI, the QA engineer would come after my soul #dummyCIJobs | 13:45 |
| vponomaryov | gouthamr: and will why you do no test that does not exist? | 13:46 |
| vponomaryov | s/will/will ask/ ? | 13:46 |
| vponomaryov | s/no/not/ | 13:46 |
| gouthamr | vponomaryov: nah.. nodes are expensive, we pay money, the cloud is someone else's computer kinda arguments. | 13:47 |
| *** akapil has quit IRC | 13:47 | |
| gouthamr | not saying 'openstack i n f r a' folks would be okay with this. | 13:48 |
| *** akapil has joined #openstack-manila | 13:48 | |
| vponomaryov | gouthamr: we are ok if they disable it | 13:48 |
| *** eharney has quit IRC | 13:48 | |
| vponomaryov | gouthamr: what is the problem? | 13:48 |
| gouthamr | vponomaryov: can't we disable it? | 13:48 |
| vponomaryov | gouthamr: we can | 13:48 |
| vponomaryov | gouthamr: but, I guess, better to add some test, for example, port Cinder's | 13:48 |
| vponomaryov | gouthamr: till that moment we can have that placeholder | 13:49 |
| vponomaryov | gouthamr, tbarron: it passed ) | 13:49 |
| vponomaryov | it is right time to either +2 it or -2 it | 13:49 |
| *** akapil has quit IRC | 13:49 | |
| tbarron | gouthamr: my irc alerts me on 'm a n i l a' :) | 13:50 |
| vponomaryov | tbarron, right, I summoned you ) | 13:50 |
| gouthamr | tbarron: sheesh | 13:50 |
| tbarron | we can always remove the test later ... | 13:52 |
| tbarron | if we decide there's no useful content to put in that slot | 13:52 |
| vponomaryov | tbarron: "remove test"? we don't have test | 13:52 |
| tbarron | vponomaryov: yeah, bad wording. disable the job. | 13:53 |
| vponomaryov | but FIRST, we should add laucnhpad manila-specs project ) | 13:54 |
| vponomaryov | to be able to file bugs | 13:55 |
| vponomaryov | about all that kind of stuff | 13:55 |
| tbarron | vponomaryov: delegate that to bswartz | 13:55 |
| *** ianychoi has joined #openstack-manila | 14:10 | |
| *** eharney has joined #openstack-manila | 14:12 | |
| *** akapil has joined #openstack-manila | 14:18 | |
| bswartz | what? | 14:19 |
| bswartz | bugs for specs? | 14:20 |
| bswartz | just fix them | 14:20 |
| vponomaryov | bswartz: to fix them we should file them | 14:21 |
| vponomaryov | bswartz: and track it | 14:21 |
| bswartz | why not just push change to gerrit? | 14:21 |
| vponomaryov | now we did so, yes | 14:21 |
| *** akapil has quit IRC | 14:23 | |
| *** wiebalck has joined #openstack-manila | 14:24 | |
| wiebalck | Is there a reason why shrinking shares is not permitted? | 14:26 |
| ganso | wiebalck: why do you mean by not permitted? | 14:27 |
| bswartz | wiebalck: shrinking shares is allowed, but only if driver supports it | 14:27 |
| wiebalck | when trying to reduce the size of a CephFS share, at least the UI says the new size must be larger than the current one | 14:28 |
| bswartz | wiebalck: maybe the UI doesn't know how to call the shrink API? I'd need to check | 14:30 |
| *** timcl has joined #openstack-manila | 14:31 | |
| ganso | wiebalck: new size must be larger? that's the extend feature... or a bug in shrink functionality | 14:32 |
| wiebalck | ganso: correct, it’s extend | 14:33 |
| ganso | wiebalck: ok so you would like to extend or shrink? I am confused | 14:33 |
| wiebalck | ganso: sorry for the confusion, the attempt was to extend to a smaller size (wasn’t aware that the cli has a ’shrink’ subcommand) | 14:34 |
| rraja | wiebalck: :) | 14:34 |
| gouthamr | weibalck: not a subcommand, a whole different 'action' API | 14:34 |
| gouthamr | sry, wiebalck - i before e. :) | 14:35 |
| rraja | gouthamr: :) | 14:35 |
| wiebalck | will check UI, but smells like PEBKAC … | 14:37 |
| openstackgerrit | Daniel Mellado proposed openstack/manila: [Tempest] Port remote_client into Manila https://review.openstack.org/396067 | 14:37 |
| vponomaryov | wiebalck: googled, PEBKAC and laughed a lot )) "Problem Exists Between Keyboard And Chair" ^_^ | 14:39 |
| wiebalck | vponomaryov: :) | 14:40 |
| openstackgerrit | Merged openstack/manila-specs: Remove duplicated imports of specs into index https://review.openstack.org/398965 | 14:42 |
| *** cdelatte has joined #openstack-manila | 14:44 | |
| vponomaryov | wiebalck: and really was surprised that you tried to "extend" a share to "smaller" size. It sounds to me as "curly and bald head" | 14:45 |
| ganso | vponomaryov, wiebalck: yes lol, TIL that one as well, I feel like using that acronym everyday now | 14:45 |
| wiebalck | From what I see, I can shrink my CephFS based share via the cli, but not via the UI. The UI only offers extend. | 14:49 |
| gouthamr | i think that's a nice interaction... as a user, i just want to resize the share. | 14:49 |
| wiebalck | gouthamr: I agree. | 14:50 |
| gouthamr | manila-ui can have a resize option that's smart enough to do the right thing.. | 14:50 |
| *** tommylikehu_ has joined #openstack-manila | 14:51 | |
| vponomaryov | gouthamr: we have a volunteer for new shiny UI feature! | 14:51 |
| wiebalck | At minimum, not offering “shrink” should be fixed, no? | 14:51 |
| gouthamr | wiebalck: +1 | 14:51 |
| gouthamr | vponomaryov: shiny bald | 14:52 |
| vponomaryov | gouthamr: i don't mind if manila features are implemented by someone being shiny bald )) | 14:54 |
| gouthamr | wiebalck: some share drivers may not allow shrinking shares... so, it feels like more of a reason to just have 'resize' and hide the API interaction from the user and spit out the error message if unable to shrink.. however, if a user tries to shrink a share that can't really be shrunk, the state transitions to 'shrinking_error' | 14:54 |
| gouthamr | vponomaryov: oh not me then | 14:54 |
| vponomaryov | why? ) | 14:54 |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila-specs: Add spec for mountable snapshots https://review.openstack.org/321213 | 14:54 |
| vponomaryov | gouthamr: you can save money on shampoo! ) | 14:55 |
| tbarron | gouthamr is bragging about his great hair again | 14:55 |
| * tbarron has not actually heard gouthamr do that ... | 14:55 | |
| gouthamr | :D yeah! | 14:55 |
| gouthamr | bswartz: can we make shrink a capability? | 14:55 |
| gouthamr | :P | 14:56 |
| openstackgerrit | Goutham Pacha Ravi proposed openstack/manila-specs: Fix and improve Access Rules https://review.openstack.org/399049 | 14:57 |
| gouthamr | ^ WIP | 14:57 |
| gouthamr | #tardy | 14:57 |
| ganso | gouthamr: so the admin can create a "shrinkable" share type? | 14:58 |
| bswartz | gouthamr: it's worth considering | 14:58 |
| *** xyang_ has joined #openstack-manila | 14:59 | |
| bswartz | gouthamr: extend/shrink were added in the bad old days when we were just copying ideas from cinder so we didn't think about that | 14:59 |
| bswartz | despite that, it's possible that shrink is okay the way it is | 15:00 |
| gouthamr | ganso bswartz: it would be easier error reporting... because it is an optional feature.. otherwise, the interaction is broken if we do add it to manila-ui, user may shrink an unshrinkable share and has to call the admin to reset state | 15:00 |
| bswartz | I need to consider it | 15:00 |
| *** zengyingzhe has joined #openstack-manila | 15:05 | |
| wiebalck | I created https://bugs.launchpad.net/manila/+bug/1642627 | 15:05 |
| openstack | Launchpad bug 1642627 in Manila "Option to "shrink" a share is missing on the UI " [Undecided,New] | 15:05 |
| openstackgerrit | Merged openstack/puppet-manila: Prepare 10.0.0 release https://review.openstack.org/397424 | 15:07 |
| gouthamr | wiebalck: thanks | 15:07 |
| *** alyson_ has joined #openstack-manila | 15:16 | |
| *** akapil has joined #openstack-manila | 15:25 | |
| openstackgerrit | Daniel Mellado proposed openstack/manila: [Tempest] Port remote_client into Manila https://review.openstack.org/396067 | 15:30 |
| *** absubram has joined #openstack-manila | 15:34 | |
| *** cdelatte has quit IRC | 15:35 | |
| *** absubram_ has joined #openstack-manila | 15:35 | |
| *** akapil has quit IRC | 15:38 | |
| *** absubram has quit IRC | 15:38 | |
| *** absubram_ is now known as absubram | 15:38 | |
| *** timcl1 has joined #openstack-manila | 15:39 | |
| *** akapil has joined #openstack-manila | 15:41 | |
| *** timcl has quit IRC | 15:42 | |
| *** akapil has quit IRC | 15:45 | |
| *** akapil_ has joined #openstack-manila | 15:45 | |
| bswartz | tpsilva, ganso: ping | 16:01 |
| bswartz | also tbarron | 16:01 |
| tpsilva | bswartz: pong | 16:01 |
| * ganso moves over to this channel | 16:01 | |
| bswartz | https://review.openstack.org/#/c/321213/ | 16:01 |
| markstur | o/ | 16:01 |
| *** zengyingzhe has quit IRC | 16:01 | |
| bswartz | In order to support this feature, a new extra spec called mount_snapshot_support will be added | 16:02 |
| ganso | bswartz: LN 52 | 16:02 |
| ganso | bswartz: ya | 16:02 |
| bswartz | no capability mentioned in the driver impact section | 16:02 |
| bswartz | I'll post more specific feedback | 16:02 |
| *** absubram has quit IRC | 16:03 | |
| bswartz | The new capability is important and definitely impacts drivers | 16:03 |
| tpsilva | it will look for the implemented methods and will add the capability, just like revert_to_snapshot or the current snapshot_support | 16:03 |
| tpsilva | is that what you want to be on the spec? | 16:03 |
| bswartz | markstur: did you have something about mountable snapshots? | 16:03 |
| ganso | bswartz: humm I thought that stating that a new "extra_spec" is being added, it implies that it is a new capability for drivers to implement and report support | 16:03 |
| tbarron | ganso: it implies, but we should *say* | 16:04 |
| bswartz | markstur: or just friendly waving? | 16:04 |
| markstur | I was trying to show that I'm on time | 16:04 |
| openstackgerrit | Marc Koderer proposed openstack/manila: Make share size configurable https://review.openstack.org/399116 | 16:04 |
| markstur | mountable snapshots should be good to code (the spec is missing a little on capability, but not a big road-block) | 16:04 |
| bswartz | tpsilva: it's something that simply has to be called out in driver impact | 16:05 |
| tpsilva | bswartz: ok, that's a simple edit, I can add that | 16:05 |
| *** nkrinner_trainin is now known as nkrinner | 16:05 | |
| * tbarron suggests a new review-focus deadline on high prio specs, perhaps December 1. | 16:06 | |
| tbarron | ganso: bswartz I'm not suggesting moving the cutoff date closer. | 16:06 |
| tbarron | ganso: bswartz just a date where high prio specs that have review focus get attention from all cores | 16:07 |
| bswartz | tbarron: it's a good point | 16:07 |
| tbarron | ganso: bswartz don't know if it's the best idea, just brainstormiinng about how to avoid last-minute crunch | 16:07 |
| openstackgerrit | zhongjun proposed openstack/manila-specs: Add detail API to quota-set API collection https://review.openstack.org/390052 | 16:07 |
| bswartz | tbarron: more serious concern is when the specs should be done by | 16:07 |
| bswartz | tbarron: personally I've been dedicating myself to reviewing the low priority stuff and not spending any time on my own spec | 16:08 |
| tbarron | i hate having to +2 when a spec really needs some fixing (more than typos and spelling) and we really think the basic idea is sound and we should do the work in this release | 16:08 |
| TommyLikeHu | bswartz: :( | 16:08 |
| * tbarron spent some time on bswartz's spec :) | 16:09 | |
| bswartz | thanks | 16:09 |
| tpsilva | bswartz: on question | 16:09 |
| bswartz | okay I'm going to push 3 changes to designate specs high priority | 16:09 |
| tpsilva | one* | 16:09 |
| *** pcaruana has quit IRC | 16:10 | |
| tpsilva | bswartz: you pointed that the drivers should report the capability, rather than automatically looking for the implemented methods like it's currently being done with snapshot_support | 16:10 |
| tpsilva | bswartz: are you against automatically reporting it as true or false depending on the methods? | 16:10 |
| bswartz | tpsilva: if we can make the capability automatically report that that maybe be better | 16:11 |
| gouthamr | tpsilva: we don't need the automatic check.. you can default to False and if drivers support it with their implementation, they will override it | 16:11 |
| bswartz | I like simple solutions though | 16:12 |
| gouthamr | tpsilva: look at replication_type or thin_provisioning | 16:12 |
| bswartz | we need to consider a case when the driver supports mountable snapshots on some pools but not all pools | 16:12 |
| gouthamr | ^ yeah, so more sense not to do any automatic setting | 16:13 |
| bswartz | just letting the driver set the boolean seems pretty idiot-proof to me | 16:13 |
| tpsilva | gouthamr, bswartz: ok... I just proposed to be automatically because that's how revert_to_snapshot is being implemented | 16:13 |
| ganso | gouthamr: revert-to-snapshot is already doing that even though there is no existing code to automatically detect https://review.openstack.org/#/c/340502/20/manila/share/driver.py@819 | 16:14 |
| gouthamr | :( i thought we only did that for create_share_from_snapshot support | 16:14 |
| gouthamr | -1 | 16:14 |
| bswartz | tpsilva: following the model for revert is a good idea, but they will both suffer from the same problem | 16:14 |
| bswartz | I'm not sure if any driver every would need to support the capability on some pools and not others | 16:14 |
| bswartz | but if they did then this design would cause problems | 16:15 |
| tpsilva | I agree | 16:15 |
| bswartz | maybe we need to propose a change to cknight's spec | 16:15 |
| tpsilva | we should report manually | 16:15 |
| tpsilva | yes | 16:15 |
| tpsilva | +1 | 16:15 |
| *** akapil_ has quit IRC | 16:15 | |
| markstur | +1 | 16:15 |
| gouthamr | +1 | 16:15 |
| ganso | bswartz: doesn't the scheduler already accept it to be assigned to that backed even if the share type extra spec is False but the capability is True? | 16:16 |
| tpsilva | I'll quickly fix the issues that you pointed out and I would appreciate more reviews :D | 16:16 |
| markstur | I think auto was useful when we were defaulting a feature to true and wanted to automatically weed out the ones that didn't have it | 16:16 |
| bswartz | ganso: no! | 16:16 |
| bswartz | ganso you're thinking of the case when the extra spec is set to None not False | 16:16 |
| ganso | bswartz: oh yes, the "dont care" extra specs | 16:16 |
| markstur | For default false the drivers can manually opt-in when they implement it and document support, etc | 16:17 |
| bswartz | if the extra spec is set to None then the scheduler won't filter on it, but end users will still think the value is false | 16:17 |
| *** falavinha has quit IRC | 16:17 | |
| vponomaryov | bswartz: in case auto detection is not suitable driver just should redefine it | 16:17 |
| markstur | vponomaryov: but that means a driver maintainer needs to do it. Best part about auto is dealing with those that did nothing | 16:18 |
| vponomaryov | bswartz: so it will have correct False value if not implemented and when implemented - driver wil ldefine it | 16:18 |
| bswartz | vponomaryov: as long as the code to do that isn't too fancy I', fine with it | 16:18 |
| gouthamr | i think we're in agreement here: for both revert_to_snapshot and mount_snapshot use cases, drivers *have* to implement additional logic - this isn't the same thing as snapshot_support becoming optional and un-overloading create_share_from_snapshot_support | 16:18 |
| ganso | markstur: but there is no existing code to take advantage of the auto, the interface to be implemented does not exist | 16:18 |
| gouthamr | where we implemented these changes expecting to maintain backwards compatibility | 16:18 |
| tpsilva | gouthamr: +1 | 16:19 |
| *** timcl has joined #openstack-manila | 16:19 | |
| markstur | So we probably all agree here but could keep arguing anyway now that there is no # endmeeting | 16:19 |
| ganso | markstur: lol | 16:19 |
| gouthamr | #endmeeting | 16:20 |
| gouthamr | :P | 16:20 |
| markstur | ganso: lol | 16:20 |
| * markstur put a space after # just to make sure I didn't end manila | 16:20 | |
| TommyLikeHu | lol | 16:21 |
| bswartz | this room is like a 24/7 meeting | 16:21 |
| markstur | nice try gouthamr | 16:21 |
| *** timcl1 has quit IRC | 16:21 | |
| markstur | It was | 16:21 |
| bswartz | you can check out but you can never leave | 16:21 |
| markstur | gouthamr: ended it | 16:21 |
| tbarron | gouthamr has a flight to catch | 16:21 |
| gouthamr | bingo | 16:21 |
| markstur | Welcome to Hotel California | 16:21 |
| tpsilva | it's all over | 16:21 |
| dustins | bswartz: Oh god I'm in 20 meetings at once O_o | 16:21 |
| tpsilva | nice working with you | 16:22 |
| bswartz | gouthamr: is Insight done then? | 16:22 |
| markstur | dustins: or are you in no meetings | 16:22 |
| dustins | markstur: I don't even know any more :) | 16:22 |
| gouthamr | bswartz: got done today - flying home tomorrow | 16:22 |
| *** tommylikehu_ has quit IRC | 16:22 | |
| bswartz | gouthamr: you mean flying to India right? | 16:23 |
| gouthamr | home=india | 16:23 |
| gouthamr | yeah :P | 16:23 |
| dustins | gouthamr: Safe travels, man! | 16:23 |
| bswartz | okay have a nice flight gouthamr | 16:23 |
| gouthamr | thanks dustins bswartz :) | 16:23 |
| ganso | gouthamr: have a nice flight! | 16:23 |
| ganso | bswartz: btw, I am confused if it has been agreed that the data jobs table spec will be withdrawn | 16:23 |
| gouthamr | thank you :) | 16:23 |
| *** gouthamr has quit IRC | 16:23 | |
| bswartz | ganso: well if you withdraw it then we can spend time focusing on the other low priority specs | 16:24 |
| bswartz | it's your decision | 16:24 |
| ganso | bswartz, tbarron, markstur: and btw I am glad that I got some feedback on taskflow, still would like to know it better to justify not going with it | 16:24 |
| bswartz | you could just git mv from ocata to pike to make the intention clear | 16:24 |
| vponomaryov | ganso: look at peace of code in Cinder | 16:25 |
| bswartz | and remove it from review focus etherpad | 16:25 |
| bswartz | s/peace/piece/ | 16:25 |
| ganso | vponomaryov: "peace of code" is good | 16:25 |
| ganso | vponomaryov: lol :P | 16:25 |
| vponomaryov | >.< | 16:25 |
| vponomaryov | sorry ) | 16:25 |
| bswartz | ganso: in addition to looking at how it works in cinder, talk to the developers who maintain that code | 16:26 |
| bswartz | the code in cinder is so hard to understand that even though the community is in favor of removing it, nobody has volunteered to do that work | 16:27 |
| tbarron | ganso: i'm interested in jobs table and in progressing investigation/design of it during ocata even if we don't target implementation till pike | 16:27 |
| ganso | tbarron: thanks tbarron :) | 16:28 |
| tbarron | bswartz: i'd like to understand if taskflow works at the wrong granularity for job restarts or if we just never mapped things right in cinder. I agree that developers tend to shy away from it but am not totally convinced that it couldn't be used more effectively. | 16:30 |
| tbarron | bswartz: mostly I just think we really need to be sure we aren't doing NIH and have good reasoning in the spec. | 16:30 |
| tbarron | bswartz: ganso it really seems to me that tracking long-running jobs, restart after failure, etc. are not manila-specific issues. | 16:31 |
| tbarron | bswartz: ganso maybe we solve them in manila with manila data service and then generalize | 16:31 |
| *** nkrinner is now known as nkrinner_afk | 16:32 | |
| *** dsariel has quit IRC | 16:32 | |
| tbarron | bswartz: ganso but I'd like us to keep the possibility of the general case in mind | 16:32 |
| bswartz | tbarron: in Cinder there are a few problems | 16:32 |
| ganso | tbarron: they are not, but that logic I would like to integrate with the approach that will be taken for HA A-A | 16:32 |
| bswartz | one is that taskflow was used for a few workflows but not nearly all of them | 16:32 |
| bswartz | two is that it doesn't actually persist its state in the DB so many of the purported benefits are not realized | 16:32 |
| tbarron | ganso: yes, we will need to think it through for HA A-A for sure | 16:32 |
| ganso | tbarron: so, I am proposing a phased implementation, and the parts that you and toabctl highlighted are in fact not going to be resolved in that first phase | 16:33 |
| tbarron | bswartz: agree, not sure if that's inherent or just cinder implementation | 16:33 |
| bswartz | and three is that the code is so byzantine that understanding it is comparable to learning a new programming language | 16:33 |
| bswartz | it's the last one that makes me oppose it | 16:33 |
| tbarron | ganso: phasing is fine, but first phase can't paint us in a corner for second phase. Not saying it does, just explaining my wanting to be somewhat deliberate. | 16:34 |
| bswartz | I like idiot proof code | 16:34 |
| tbarron | bbiab | 16:34 |
| *** akapil has joined #openstack-manila | 16:35 | |
| *** belmoreira has quit IRC | 16:35 | |
| ganso | tbarron: yes, that is another concern I have, If I should just wait for this mechanism to exist prior to adding the jobs table because it will be incomplete and needs to be compatible with something that does not exist yet | 16:35 |
| markstur | ganso: I used taskflow for a multi-step retype in the HP driver (hpe3par_common.py?) | 16:36 |
| ganso | tbarron: on the other hand, what is proposed fixes existing problems... like, there is nothing tracking migration at all, so user's experience using migration may not be ideal | 16:36 |
| markstur | ganso: What it did for me was allow retype to work in phases and taskflow framework took care of how to rollback earlier phases if a later one failed | 16:36 |
| *** akapil has quit IRC | 16:37 | |
| markstur | It's not idiot proof, but helped with rollback | 16:37 |
| ganso | markstur: that sounds like a mechanism for HA A-A as well, but I wouldn't want to use that if manila is not onboard and does not want that mechanism for m-shr HA A-A | 16:37 |
| markstur | ganso: I've heard after I was told I should use it that some folks wanted to get rid of it | 16:38 |
| ganso | markstur: yea, if we have a feature tied to it, it makes it even more complicated to remove | 16:39 |
| markstur | ganso: I think more specifically many just don't like how it was used in cinder create_volume(?) and maybe it isn't maintained(?) | 16:39 |
| markstur | ganso: So I'd be reluctant to use it. | 16:40 |
| *** gcb has quit IRC | 16:40 | |
| ganso | markstur: ok I think I will take a step back and redesign the spec, maybe remove stuff proposed so I have a smaller risk of adding something that will not be compatible with our future HA A-A mechanism | 16:40 |
| toabctl | ganso, maybe you can ask jharlow in the oslo channel | 16:40 |
| toabctl | about the taskflow state | 16:41 |
| markstur | ganso: ...but it could still be worth a look if it does what you'd end up writing anyway | 16:41 |
| *** gcb has joined #openstack-manila | 16:41 | |
| ganso | toabctl: thanks, I will join that channel and try to talk to him | 16:42 |
| toabctl | I just want to avoid the NIH syndrom | 16:42 |
| ganso | markstur: even if it does, it would not be implemented in the first phase anyway | 16:42 |
| markstur | It fails 2 of Ben's criteria: 1) idiot-proof 2) not fancy | 16:42 |
| *** lpetrut has quit IRC | 16:44 | |
| *** akapil has joined #openstack-manila | 16:45 | |
| *** rraja has quit IRC | 16:51 | |
| openstackgerrit | Ben Swartzlander proposed openstack/manila-specs: Mark Eliminate Race Conditions high priority https://review.openstack.org/399135 | 16:57 |
| *** wiebalck has quit IRC | 16:58 | |
| *** xyang_ has quit IRC | 16:59 | |
| openstackgerrit | Ben Swartzlander proposed openstack/manila-specs: Mark Fix and improve Access Rules high priority https://review.openstack.org/399138 | 17:02 |
| openstackgerrit | Ben Swartzlander proposed openstack/manila-specs: Mark Enable IPv6 high priority https://review.openstack.org/399140 | 17:06 |
| *** lpetrut has joined #openstack-manila | 17:12 | |
| *** akapil_ has joined #openstack-manila | 17:12 | |
| *** akapil has quit IRC | 17:12 | |
| ganso | markstur, bswartz, toabctl: waiting for feedback in https://review.openstack.org/392291/ ... seems ready so far | 17:14 |
| *** xyang_ has joined #openstack-manila | 17:19 | |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila-specs: Add spec for mountable snapshots https://review.openstack.org/321213 | 17:19 |
| *** xyang_ has quit IRC | 17:21 | |
| *** lpetrut has quit IRC | 17:26 | |
| *** jcsp has joined #openstack-manila | 17:34 | |
| *** akapil_ has quit IRC | 17:36 | |
| *** akapil has joined #openstack-manila | 17:36 | |
| *** jcsp has quit IRC | 17:40 | |
| *** jcsp has joined #openstack-manila | 17:40 | |
| ganso | bswartz: ping | 17:44 |
| *** akapil has quit IRC | 17:55 | |
| openstackgerrit | Mauricio Lima proposed openstack/manila-specs: Spec for openstack client support https://review.openstack.org/395775 | 18:02 |
| *** xyang_ has joined #openstack-manila | 18:03 | |
| *** xyang_ has quit IRC | 18:05 | |
| bswartz | ganso: pong | 18:05 |
| ganso | bswartz: I would like to discuss an aspect of the data jobs spec | 18:05 |
| bswartz | k | 18:06 |
| ganso | bswartz: I remember at the design summit session that some folks cringed when I mentioned that we would store the job details as string JSON in the database | 18:06 |
| bswartz | yeah I was one of them | 18:06 |
| ganso | bswartz: so I came up with an alternative that has its advantages and disadvantages | 18:06 |
| ganso | bswartz: so if you are familiar with the ShareServerBackendDetails table, it is exactly the same | 18:06 |
| * markstur doesn't cringe | 18:07 | |
| ganso | bswartz: it is a separate table "JobDetails" where each key-value pair is a row | 18:07 |
| *** akapil has joined #openstack-manila | 18:07 | |
| bswartz | yeah that's the alternative I would have suggested | 18:07 |
| bswartz | it doesn't allow for infinitely nested structures like json does, but it scales better | 18:07 |
| ganso | bswartz: the disadvantage is that, for the amount of jobs we may have, it will probably not scale so well to seek the information in the JobDetails table every time the Jobs table is queried | 18:08 |
| markstur | ganso: index the table | 18:08 |
| bswartz | ganso: I don't see why it would be any worse than other cases where keys/values are stored in a separate table | 18:08 |
| ganso | markstur: it will be indexed, yes | 18:08 |
| bswartz | think of access rules, type keys, driver private share data | 18:09 |
| ganso | bswartz: it is not really worse than those, pretty much equivalent | 18:09 |
| bswartz | indexed lookups generally perform very well on SQL databases | 18:09 |
| ganso | bswartz: but seems to me the previous approach would scale better if the data is limited and contained in the same table | 18:09 |
| bswartz | the first part of that IF is the crux of the problem | 18:10 |
| bswartz | if the data is limited, then every solution works well | 18:10 |
| bswartz | the problem is that we (apparently) don't want to limit the data | 18:11 |
| *** akapil has quit IRC | 18:11 | |
| ganso | bswartz: it seems like we never do limit the data hehe | 18:12 |
| bswartz | yeah | 18:12 |
| ganso | bswartz: never do *want | 18:12 |
| ganso | bswartz: if we think about 1 million rows, the JSON alternative scales better | 18:12 |
| ganso | bswartz: ... with the JSON being limited to like 255 chars | 18:13 |
| bswartz | so if we don't know what fields the jobs will required ahead of time, or we don't know how wide those fields would be, then we're stuck with either a potentially huge JSON blob or possibly a large number of rows in some join table | 18:13 |
| bswartz | I don't think it's practical to put a size limit on JSON blobs -- it introduces a huge new set of error conditions and ugly workarounds | 18:14 |
| bswartz | unless the limit is so high that nobody will ever hit it in practice | 18:15 |
| ganso | bswartz: well, I will just state in the spec that the JobDetails table is preferred approach then... | 18:15 |
| bswartz | thanks | 18:15 |
| bswartz | some examples will help reviewers | 18:15 |
| *** xyang_ has joined #openstack-manila | 18:16 | |
| ganso | bswartz: for the preferred approach it is detailed | 18:16 |
| ganso | bswartz: thanks | 18:16 |
| *** xyang_ has quit IRC | 18:18 | |
| *** xyang_ has joined #openstack-manila | 18:21 | |
| markstur | ganso: minor thing in migration spec and then hopefully done | 18:23 |
| *** xyang_ has quit IRC | 18:23 | |
| bswartz | ganso, tbarron: you put +2 on 2/3 of the high priority spec changes -- did you opposed the 3rd one? | 18:24 |
| ganso | bswartz: I need to re-read the spec, haven't had time yet | 18:24 |
| ganso | markstur: thanks | 18:24 |
| tbarron | bswartz: i didn't intentionally oppose adding any of the high prio specs that we all talked about, which review looks like I'm opposing? | 18:25 |
| bswartz | tbarron: I pushed 3 changes in a row ^^^ and you +2 two of them | 18:26 |
| bswartz | didn't know if you just missed the 3rd or intentionally didn't +2 it | 18:26 |
| tbarron | bswartz: i think i just missed one then, looking ... | 18:26 |
| *** xyang_ has joined #openstack-manila | 18:26 | |
| ganso | bswartz: re-reading the spec I will be taking a more conscious decision regarding the risks, because hi-pri means "we want that feature in ocata" | 18:26 |
| markstur | I +2*3, but honestly that IPv6 one is the situation where sure it's nice to have the core work merged, but when only 1 driver implements a new feature it kind of looks bad | 18:28 |
| tbarron | bswartz: ganso I am in favor of that being hi prio; i see it as not high risk but would benefit from more review; also not a big diversion of resources - we'll need to review but huawei folks will do the implementation | 18:28 |
| tbarron | ganso: do you think it's risky? | 18:29 |
| tbarron | ganso: my sense is that the spec is almost ready. Just mostly nits and need to nail down the capabilities/extra-spec a bit. | 18:29 |
| bswartz | markstur: It's important to decouple the support for a feature in manila from support in drivers | 18:29 |
| markstur | So there's that fence between better to let it sit in gerrit ready for drivers to play with and merge day 1 vs merge w/ lousy support | 18:29 |
| markstur | I'm sitting on that fence, but this time I guess I lean to agree with merging would be best | 18:30 |
| bswartz | markstur: this is the approach we've taken with other major features | 18:30 |
| markstur | Yes. Single driver features aren't great. Problem with the other approach though... | 18:31 |
| bswartz | when adding a new features we require support in a small handful of drivers to validate the design, and to enable automated gate testing, but we allow all other driver to implement the feature and their leisure | 18:31 |
| markstur | is if we delay merge until day1 of next release. Would drivers just delay yet another release... | 18:31 |
| markstur | So OK | 18:31 |
| tbarron | markstur: we should require a first-party driver implementation at least | 18:31 |
| markstur | bswartz: Totally agree that we HAVE TO allow drivers time to implement. | 18:31 |
| markstur | It's the only 1 driver did it situation that looks bad. | 18:32 |
| markstur | but like I said above. Could be that the alternative is worse. So let's go. | 18:32 |
| bswartz | IPv6 *should* be easy enough to add to most drivers that I doubt we'll end up in that boat | 18:32 |
| ganso | tbarron: I need to catch up with the latest patches in the IPv6 spec, the last one I had read I was against and saw high-risk | 18:32 |
| markstur | bswartz: What's the deadline for hi-pri feature to merge? | 18:32 |
| tbarron | ganso: ack | 18:32 |
| ganso | tbarron: in today's meeting someone said it has been decoupled, so I need to revisit that | 18:33 |
| bswartz | yes, in this case the alternative is that Manila remains in the 90s with it's IPv4-only support | 18:33 |
| bswartz | markstur: same as any feature -- the feature freeze | 18:33 |
| tbarron | ganso: so it's being hi-prio doesn't mean it has to be approved. You have till ocata-2 to -2 :) | 18:33 |
| * markstur mind wanders back to the 90s | 18:33 | |
| markstur | bswartz: So IPv6 merges late in the release. Most drivers won't jump on it until next release. Unfortunate situation vs early merge. | 18:34 |
| tbarron | making it hi-prio means we *want* to have it but is not a guarantee | 18:34 |
| markstur | Anywho. Let's see if we can get it soon and right. The spec needs work, but sure. It is 2016. | 18:35 |
| markstur | By the way!!! Anyone have a good design for an extra spec that says <all> ('4','6') | 18:36 |
| *** lpetrut has joined #openstack-manila | 18:36 | |
| tbarron | markstur: we have to accomodate the non-IP backends too with the new extra-specs/capabilities ... | 18:37 |
| markstur | We need a user-friendly way of saying '4&6' or 'both' but it should be listy | 18:37 |
| tbarron | markstur: that's the area of the spec that needs the most thought I think | 18:37 |
| bswartz | tbarron: what? | 18:37 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Data Jobs table https://review.openstack.org/392262 | 18:37 |
| * bswartz hopes tbarron was making a joke that he missed | 18:38 | |
| tbarron | bswartz: well right now every driver is saying I can do v4, i can do v6, or i can do both, and it's defaulting to v4. | 18:38 |
| tbarron | markstur: do I have that right? | 18:38 |
| bswartz | you mean the spec says that or that's reality? | 18:38 |
| markstur | tbarron: So. An IPv4, an IPv6, and a non-IP walk into a bar... | 18:38 |
| markstur | ^ that's how you start a joke | 18:39 |
| tbarron | bswartz: i think that's what the spec says atm, but the capabilites/extra spec stuff was added to the spec quite late. | 18:39 |
| bswartz | If someone proposes a fibre channel driver for manila I'm going to start throwing chairs | 18:39 |
| tbarron | that's why i didn't want to rush approval +2 by ocata-1 even though I think this is totally implementable | 18:39 |
| tbarron | bswartz: no, i'm talking about exports and access lists, think native cephfs, hdfs, native glusterfs | 18:40 |
| markstur | Agree that the spec isn't right, but fixing it could be punted to the code. I think. | 18:40 |
| bswartz | tbarron: oh are you referring to non-IP *access rules*? | 18:40 |
| tbarron | markstur: if we make it high-prio spec we have time to fix the spec | 18:40 |
| markstur | tbarron: Didn't get the "non-IP" comment. | 18:40 |
| markstur | We did make it hi-pri. So we're good. | 18:41 |
| tbarron | bswartz: isn't the capability advertising what kind of exports one can support? | 18:41 |
| markstur | But I don't see a great way of doing a spec that says I want "both" | 18:41 |
| markstur | reluctant to use "both" because of some future IPvX | 18:41 |
| bswartz | tbarron: access rules != export locations | 18:41 |
| bswartz | export locations are now and always will be IP address-based | 18:42 |
| markstur | We can have driver report ('4','6') as we already support both capabilities like dhss (True, False) | 18:42 |
| bswartz | access rules can be IP, user, cert, etc | 18:42 |
| tbarron | bswartz: true that | 18:42 |
| markstur | But there isn't a good way to do extra-spec that says I want ip_version = '4' and ip_version = '6' | 18:42 |
| bswartz | markstur: yes there is | 18:42 |
| markstur | unless you are allowed to repeat a spec like that. Which would be OK I guess, but not if it is used as a key | 18:42 |
| *** ociuhandu has quit IRC | 18:43 | |
| bswartz | you create 2 extra specs ipv4_support=True, ipv6_support=True | 18:43 |
| markstur | bswartz: good. What is it. Let's get it in the spec | 18:43 |
| markstur | OK. | 18:43 |
| markstur | I didn't get a chance to test it | 18:43 |
| markstur | Seems like I've accessed that as a keyed dict | 18:43 |
| bswartz | I provided that feedback in the ML thread | 18:43 |
| markstur | Oops. I didn't read your suggest correctly | 18:43 |
| markstur | I see what you mean | 18:43 |
| bswartz | maybe it didn't get reflected in the spec yet | 18:43 |
| markstur | Cool. I didn't review all the reviews last night. | 18:44 |
| markstur | We should be able to settle this with the hi-pri deadline | 18:44 |
| tbarron | bswartz: yeah, your point is sound, but the spec needs work on the extra_spec b/c it says "Add new optional extra_spec 'ip_version for share type that indicate[sic] the access ip version". Sounds like access rules though export locations are likely intended ?? | 18:45 |
| bswartz | tbarron: yes | 18:45 |
| tbarron | bswartz: ok, we're on the same page. I let the language muddle me up. | 18:45 |
| bswartz | I think our friends a huawei are probably in bed by now but we should keep helping straighten the spec out | 18:46 |
| markstur | Yeah. I thought at first that export_locations was being dropped from the feature, but then I read on and let it go. | 18:46 |
| bswartz | if someone wants to clean up the language maybe ask them if they'd mind someone else pushing an update to the spec | 18:46 |
| markstur | I feel bad I wanted to post my feedback last night so TommyLikeHu could work on it, but this morning I found that the publish button I thought I hit before I had to run off did not get hit :( | 18:47 |
| bswartz | markstur: ouch I hate that | 18:47 |
| markstur | Glad we have extended w/ the hi-pri. We'll work this one out | 18:47 |
| markstur | bswartz: Especially when I wake up mid-manila meeting open my laptop and see the review right there waiting for me to press the button | 18:48 |
| bswartz | have you ever made the mistake where you posted a bunch a feedback against the wrong changeset, then when you pressed publish it *still* didn't publish? | 18:48 |
| markstur | bswartz: Don't think so. I have of course found stuff in drafts way after I thought I was being ignored | 18:48 |
| tbarron | I put a comment in the spec about export_locations vs export rules. TommyLikeHu will pick that up first thing tomorrow and we'll see a new draft I'm sure. | 18:49 |
| *** timcl1 has joined #openstack-manila | 18:53 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila-specs: Add spec for Share Migration Ocata improvements https://review.openstack.org/392291 | 18:54 |
| ganso | toabctl, tbarron: if you could please take a look at the updated data jobs spec, I hope I have addressed the concerns, please let me know if there are any more outstanding concerns | 18:56 |
| *** timcl has quit IRC | 18:56 | |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila-specs: Add spec for mountable snapshots https://review.openstack.org/321213 | 18:57 |
| openstackgerrit | Tiago Pasqualini da Silva proposed openstack/manila-specs: Add spec for mountable snapshots https://review.openstack.org/321213 | 18:58 |
| *** timcl1 has left #openstack-manila | 19:09 | |
| tbarron | ganso: I made a couple more remarks in the review, will look again fresh tomorrow but have to do some other stuff the rest of today. | 19:10 |
| ganso | tbarron: ok! thanks Tom! | 19:14 |
| tbarron | ganso: yw, good stuff! | 19:14 |
| *** xyang_ has quit IRC | 19:28 | |
| *** xyang_ has joined #openstack-manila | 19:28 | |
| *** nherciu_ has joined #openstack-manila | 19:33 | |
| *** xyang_ has quit IRC | 19:35 | |
| *** xyang_ has joined #openstack-manila | 19:35 | |
| *** wiebalck has joined #openstack-manila | 19:37 | |
| *** nherciu has quit IRC | 19:37 | |
| *** porrua has quit IRC | 19:38 | |
| *** openstackgerrit has quit IRC | 19:48 | |
| *** openstackgerrit has joined #openstack-manila | 19:49 | |
| *** porrua has joined #openstack-manila | 19:56 | |
| *** wiebalck has quit IRC | 20:08 | |
| *** wiebalck has joined #openstack-manila | 20:09 | |
| waj334 | How does that Auto-Negotiation feature in the Manila API work exactly? | 20:12 |
| *** xyang_ has quit IRC | 20:20 | |
| *** xyang_ has joined #openstack-manila | 20:21 | |
| *** wiebalck has quit IRC | 20:25 | |
| waj334 | Nevermind. Found the blueprint | 20:27 |
| waj334 | again | 20:27 |
| openstackgerrit | Marc Koderer proposed openstack/manila: [Tempest] Make share size configurable in scenario tests https://review.openstack.org/399116 | 20:27 |
| bswartz | waj334: are you talking about the microversion negotiation? | 20:27 |
| *** wiebalck has joined #openstack-manila | 20:39 | |
| *** timcl has joined #openstack-manila | 20:43 | |
| *** timcl has left #openstack-manila | 20:44 | |
| waj334 | bswartz: Yes. I'm trying to propose doing something similar for Cinder | 20:49 |
| bswartz | waj334: the way it works is that the client always sends requests with the highest microversion it supports -- and if it receives a version error it retries with the highest mutually supported version (if there is one) or fails (if there's no mutually supported version) | 20:50 |
| bswartz | servers are expected to support several releases worth of old microversions to support old clients, and client are expected to support several releases worth of old microversions to support old servers | 20:52 |
| *** xyang_ has quit IRC | 21:01 | |
| *** xyang_ has joined #openstack-manila | 21:01 | |
| *** wiebalck has quit IRC | 21:02 | |
| *** xyang_ has quit IRC | 21:05 | |
| *** xyang_ has joined #openstack-manila | 21:05 | |
| *** alyson_ has quit IRC | 21:07 | |
| *** dustins has quit IRC | 21:13 | |
| *** akapil has joined #openstack-manila | 21:23 | |
| *** akapil has quit IRC | 21:27 | |
| *** sticker has joined #openstack-manila | 21:38 | |
| *** porrua has quit IRC | 21:44 | |
| *** xyang_ has quit IRC | 21:55 | |
| *** xyang_ has joined #openstack-manila | 21:57 | |
| *** xyang_ has quit IRC | 22:00 | |
| *** eharney has quit IRC | 22:12 | |
| *** jcsp has quit IRC | 22:16 | |
| *** wiebalck has joined #openstack-manila | 22:17 | |
| *** wiebalck has quit IRC | 22:19 | |
| *** xyang_ has joined #openstack-manila | 22:22 | |
| *** xyang_ has quit IRC | 22:25 | |
| *** xyang_ has joined #openstack-manila | 22:26 | |
| *** lpetrut has quit IRC | 22:27 | |
| *** jprovazn has quit IRC | 22:31 | |
| *** nherciu_ has quit IRC | 22:34 | |
| *** ganso has quit IRC | 22:36 | |
| *** tpsilva has quit IRC | 23:05 | |
| *** catintheroof has quit IRC | 23:29 | |
| *** catintheroof has joined #openstack-manila | 23:30 | |
| *** catintheroof has quit IRC | 23:35 | |
| *** sticker has quit IRC | 23:42 | |
| *** sticker has joined #openstack-manila | 23:43 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!