| *** ianychoi has joined #openstack-manila | 00:02 | |
| *** catintheroof has joined #openstack-manila | 00:08 | |
| *** akapil has quit IRC | 00:22 | |
| *** akapil has joined #openstack-manila | 00:23 | |
| *** mtanino has quit IRC | 00:27 | |
| *** zengyingzhe_ has quit IRC | 00:48 | |
| *** zengyingzhe_ has joined #openstack-manila | 00:48 | |
| *** akapil has quit IRC | 00:50 | |
| *** kaisers_ has joined #openstack-manila | 02:08 | |
| *** kaisers has quit IRC | 02:11 | |
| openstackgerrit | Ha Van Tu proposed openstack/manila: [api-ref] Refactor Manila scheduler stats API https://review.openstack.org/376354 | 02:14 |
|---|---|---|
| *** catintheroof has quit IRC | 02:16 | |
| *** catintheroof has joined #openstack-manila | 02:16 | |
| *** catintheroof has quit IRC | 02:20 | |
| *** zhonghua has quit IRC | 02:49 | |
| *** zhonghua has joined #openstack-manila | 03:00 | |
| *** gcb has joined #openstack-manila | 03:15 | |
| *** catintheroof has joined #openstack-manila | 03:25 | |
| openstackgerrit | zhongshengping proposed openstack/puppet-manila: Fix the annotation https://review.openstack.org/404007 | 03:58 |
| *** tuanluong has joined #openstack-manila | 04:06 | |
| openstackgerrit | Pony Chou proposed openstack/manila: Add QNAP Manila Driver https://review.openstack.org/394703 | 04:10 |
| *** dsariel has quit IRC | 04:11 | |
| *** catinthe_ has joined #openstack-manila | 04:23 | |
| *** catintheroof has quit IRC | 04:26 | |
| *** catinthe_ has quit IRC | 04:26 | |
| *** catintheroof has joined #openstack-manila | 04:27 | |
| *** catintheroof has quit IRC | 04:27 | |
| *** gouthamr has joined #openstack-manila | 05:13 | |
| *** shausy has joined #openstack-manila | 05:35 | |
| *** gouthamr has quit IRC | 05:37 | |
| *** gouthamr has joined #openstack-manila | 05:46 | |
| *** dsariel has joined #openstack-manila | 05:52 | |
| *** gouthamr has quit IRC | 05:56 | |
| openstackgerrit | Pony Chou proposed openstack/manila: Add QNAP Manila Driver https://review.openstack.org/394703 | 05:57 |
| openstackgerrit | Pony Chou proposed openstack/manila: Add QNAP Manila Driver https://review.openstack.org/394703 | 06:07 |
| *** gcb has quit IRC | 06:08 | |
| *** gcb has joined #openstack-manila | 06:09 | |
| *** gcb has quit IRC | 06:21 | |
| *** lpetrut has joined #openstack-manila | 06:31 | |
| *** gcb has joined #openstack-manila | 06:41 | |
| *** nkrinner_afk is now known as nkrinner | 07:19 | |
| *** pcaruana has joined #openstack-manila | 07:21 | |
| *** lpetrut has quit IRC | 07:28 | |
| *** makowals has joined #openstack-manila | 07:28 | |
| openstackgerrit | zhongjun proposed openstack/manila: Add manila-manage db purge command https://review.openstack.org/308212 | 07:33 |
| *** jprovazn has joined #openstack-manila | 07:44 | |
| *** nherciu has joined #openstack-manila | 07:54 | |
| *** akapil has joined #openstack-manila | 08:01 | |
| *** gouthamr has joined #openstack-manila | 08:03 | |
| *** nkrinner has quit IRC | 08:28 | |
| *** nkrinner has joined #openstack-manila | 08:34 | |
| *** nkrinner has quit IRC | 08:38 | |
| *** nkrinner has joined #openstack-manila | 08:51 | |
| xinyan | Hi,everybody,please review https://review.openstack.org/#/c/308212/ ,thanks! | 08:57 |
| *** lpetrut has joined #openstack-manila | 09:19 | |
| *** lpetrut has quit IRC | 09:24 | |
| *** lpetrut has joined #openstack-manila | 09:31 | |
| *** senk has joined #openstack-manila | 09:32 | |
| *** mkoderer has joined #openstack-manila | 09:37 | |
| *** lpetrut has quit IRC | 09:41 | |
| *** lpetrut has joined #openstack-manila | 09:43 | |
| *** lpetrut has quit IRC | 09:47 | |
| *** openstackgerrit has quit IRC | 09:48 | |
| *** makowals_ has joined #openstack-manila | 09:48 | |
| *** openstackgerrit has joined #openstack-manila | 09:49 | |
| *** lpetrut has joined #openstack-manila | 09:49 | |
| *** makowals has quit IRC | 09:51 | |
| openstackgerrit | Marc Koderer proposed openstack/manila: Remove unused function in db api https://review.openstack.org/404118 | 09:54 |
| mkoderer | gouthamr: still different timezone? | 10:03 |
| gouthamr | mkoderer: yep.. IST until 13th Dec | 10:03 |
| mkoderer | gouthamr: nice! | 10:04 |
| gouthamr | :) | 10:04 |
| mkoderer | gouthamr: I improved my race condition patch a bit: https://review.openstack.org/#/c/401132/ | 10:04 |
| mkoderer | more documentation :) | 10:05 |
| gouthamr | mkoderer: sure thing, will check it out. | 10:05 |
| gouthamr | that's always good :P | 10:05 |
| mkoderer | gouthamr: and I uploaded a test case that fails due to an existing race: | 10:06 |
| mkoderer | https://review.openstack.org/#/c/403675/ | 10:06 |
| gouthamr | mkoderer: ah. that has a bug i believe.. we hit that in the gate for different reasons; the workaround has been: turn off automatic share server deletion in the CI. guess bswartz will add that case to his race-conditions spec | 10:09 |
| mkoderer | gouthamr: you mean this: https://review.openstack.org/#/c/324300/ | 10:10 |
| mkoderer | I didn't had time to fix that yet | 10:10 |
| mkoderer | gouthamr: share server deletion can be also trigged by the user | 10:11 |
| gouthamr | mkoderer: yes.. that one. i don't recall that patch though.. | 10:11 |
| gouthamr | mkoderer: yep. user initiated, and we have two ways to cleanup via polling in the share manager | 10:11 |
| mkoderer | gouthamr: yep and the periodic cleanup way is the smarter one | 10:12 |
| mkoderer | gouthamr: but we have a general problems with locks and periodic jobs | 10:12 |
| gouthamr | mkoderer: the case we see in the gate is: share server is in the process of being cleaned up, but a new share has been provisioned on the share server; the two actions race and the share and server end up without being cleaned up | 10:12 |
| mkoderer | gouthamr: ok for the gate we should definitly deactivate automatic share server deletion | 10:13 |
| gouthamr | mkoderer: true. but rewriting some of the state transitions carefully and merging this patch and using it for concurrency control is a good alternative :) https://review.openstack.org/#/c/318336/ | 10:14 |
| *** erlon-airlong has joined #openstack-manila | 10:15 | |
| *** senk has quit IRC | 10:18 | |
| *** rraja has joined #openstack-manila | 10:23 | |
| *** alyson_ has joined #openstack-manila | 10:30 | |
| *** senk has joined #openstack-manila | 10:31 | |
| *** tpsilva has joined #openstack-manila | 10:34 | |
| *** senk has quit IRC | 10:41 | |
| *** ganso has joined #openstack-manila | 10:47 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix share writable in host-assisted migration https://review.openstack.org/375047 | 10:54 |
| *** gouthamr has quit IRC | 10:58 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [DNM] test 5 https://review.openstack.org/404163 | 11:31 |
| *** ociuhandu has joined #openstack-manila | 11:31 | |
| *** lpetrut1 has joined #openstack-manila | 11:39 | |
| *** lpetrut has quit IRC | 11:41 | |
| *** lpetrut1 is now known as lpetrut | 11:41 | |
| openstackgerrit | Alexander Pugachev proposed openstack/python-manilaclient: Configures coverage tool https://review.openstack.org/404167 | 11:44 |
| *** a-pugachev has joined #openstack-manila | 11:48 | |
| *** porrua has joined #openstack-manila | 11:57 | |
| *** JoseMello has joined #openstack-manila | 12:11 | |
| *** tuanluong has quit IRC | 12:14 | |
| *** senk has joined #openstack-manila | 12:24 | |
| *** akapil_ has joined #openstack-manila | 12:26 | |
| *** akapil has quit IRC | 12:29 | |
| *** senk has quit IRC | 12:33 | |
| *** catinthe_ has joined #openstack-manila | 12:37 | |
| *** gouthamr has joined #openstack-manila | 12:42 | |
| *** akapil_ has quit IRC | 12:43 | |
| *** akapil has joined #openstack-manila | 12:44 | |
| *** akapil_ has joined #openstack-manila | 12:50 | |
| *** akapil has quit IRC | 12:50 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [Devstack] Use openstack CLI instead of other clients https://review.openstack.org/402169 | 12:54 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [Devstack] Fix DHSS=False setup for Generic driver https://review.openstack.org/403651 | 12:54 |
| *** gcb has quit IRC | 12:56 | |
| *** senk has joined #openstack-manila | 12:59 | |
| *** senk has quit IRC | 13:02 | |
| openstackgerrit | Mauricio Lima proposed openstack/manila-specs: Spec for openstack client support https://review.openstack.org/395775 | 13:03 |
| gouthamr | rraja: ping | 13:17 |
| gouthamr | ^ question regarding glusterFS driver | 13:17 |
| rraja | gouthamr: yes | 13:18 |
| rraja | csaba: ^^ | 13:18 |
| gouthamr | rraja: hey! in the feature support matrix, i see that gluster doesn't support readonly semantics on shares. is this true? | 13:19 |
| rraja | gouthamr: yes. | 13:20 |
| gouthamr | rraja: okay.. hard back end limitation? | 13:20 |
| rraja | gouthamr: this was raised by you in Barcelona. For GlusterFS/Ganesha driver it's not. For GlusterFS/Gluster-NFS driver and GlusterFS_native driver it is. | 13:22 |
| rraja | csaba: ^^ | 13:22 |
| rraja | gouthamr: I'd raised that RFE in glusterfs community more than a year ago. it's still not implemented. | 13:23 |
| gouthamr | rraja: ah. i do recall this discussion now... needed this clarification for a proposal | 13:23 |
| gouthamr | rraja: good stuff. thanks | 13:23 |
| rraja | what I was thinking is write an email to openstack-dev ML, and let the community decide what to do with the driver. | 13:24 |
| rraja | gouthamr: were you thinking of proposing a grade band for drivers? so a gluster driver which does not meet minimum requirement would not be in the first grade (let's say) instead of being removed from the tree? | 13:25 |
| gouthamr | rraja: hey i wasn't looking to make your job easier ;) | 13:26 |
| rraja | :) what was your plan? | 13:27 |
| rraja | if I may ask | 13:27 |
| gouthamr | rraja: but no, maybe there are cases where drivers can't do readonly semantics - i'm going to re-read my notes from the summit.. there were some action items that are forgotten. :P | 13:27 |
| rraja | gouthamr: i'll be surprised if there were AIs on this topic. i don't recall of any. | 13:28 |
| rraja | anyway, what's community view on drivers that don't meet minimum requirements? | 13:29 |
| gouthamr | rraja: ganso and i are trying to cleanup readonly rule semantics for share instances... https://review.openstack.org/#/c/375047 and https://review.openstack.org/#/c/399049/ if you're interested.. | 13:29 |
| rraja | thanks for the links. i'll take a look. | 13:31 |
| gouthamr | rraja: "Drivers must support read-write and read-only access levels for each supported protocol, either through individual access rules or separate export locations." | 13:32 |
| gouthamr | rraja: bswartz may know better about the driver minimum requirements. They have been requirements for being accepted so far and this document was written in Mitaka | 13:33 |
| openstackgerrit | Merged openstack/manila: [api-ref] Refactor Manila scheduler stats API https://review.openstack.org/376354 | 13:33 |
| rraja | gouthamr: okay. thanks! | 13:36 |
| *** maurice_ has joined #openstack-manila | 13:36 | |
| bswartz | rraja: read-only is something that all drivers must support | 13:51 |
| *** cknight has joined #openstack-manila | 13:52 | |
| *** maurice_ has quit IRC | 13:53 | |
| *** sapcc-bot1 has quit IRC | 13:53 | |
| *** tpatzig_1 has quit IRC | 13:53 | |
| *** sapcc-bot has joined #openstack-manila | 13:53 | |
| *** dgonzalez_ has joined #openstack-manila | 13:53 | |
| *** mkoderer_ has joined #openstack-manila | 13:53 | |
| *** databus23_ has joined #openstack-manila | 13:53 | |
| *** maurice_ has joined #openstack-manila | 13:53 | |
| *** tommy_ has joined #openstack-manila | 13:53 | |
| *** tpatzig_ has joined #openstack-manila | 13:53 | |
| *** tommy_ is now known as Guest23410 | 13:54 | |
| *** dgonzalez_ has quit IRC | 13:55 | |
| *** mkoderer_ has quit IRC | 13:55 | |
| *** databus23_ has quit IRC | 13:55 | |
| *** maurice_ has quit IRC | 13:55 | |
| *** tpatzig_ has quit IRC | 13:55 | |
| *** Guest23410 has quit IRC | 13:55 | |
| *** maurice_ has joined #openstack-manila | 14:04 | |
| *** maurice_ has quit IRC | 14:05 | |
| *** senk has joined #openstack-manila | 14:05 | |
| *** xyang1 has joined #openstack-manila | 14:06 | |
| *** carthaca has joined #openstack-manila | 14:09 | |
| *** sapcc-bot has quit IRC | 14:11 | |
| *** sapcc-bot has joined #openstack-manila | 14:12 | |
| *** mkoderer_ has joined #openstack-manila | 14:12 | |
| *** dgonzalez_ has joined #openstack-manila | 14:12 | |
| *** databus23_ has joined #openstack-manila | 14:12 | |
| *** tpatzig_ has joined #openstack-manila | 14:12 | |
| *** david_1 has joined #openstack-manila | 14:12 | |
| *** sapcc-bot has quit IRC | 14:12 | |
| *** david_1 has quit IRC | 14:12 | |
| *** mkoderer_ has quit IRC | 14:12 | |
| *** tpatzig_ has quit IRC | 14:12 | |
| *** dgonzalez_ has quit IRC | 14:12 | |
| *** databus23_ has quit IRC | 14:12 | |
| *** sapcc-bot has joined #openstack-manila | 14:13 | |
| *** mkoderer_ has joined #openstack-manila | 14:13 | |
| *** carthaca_ has joined #openstack-manila | 14:13 | |
| *** databus23_ has joined #openstack-manila | 14:13 | |
| *** tpatzig_ has joined #openstack-manila | 14:13 | |
| *** dgonzalez_ has joined #openstack-manila | 14:13 | |
| *** david_1 has joined #openstack-manila | 14:13 | |
| *** senk has quit IRC | 14:13 | |
| *** carthaca_ has quit IRC | 14:15 | |
| *** databus23_ has quit IRC | 14:15 | |
| *** tpatzig_ has quit IRC | 14:15 | |
| *** mkoderer_ has quit IRC | 14:15 | |
| *** dgonzalez_ has quit IRC | 14:15 | |
| *** david_1 has quit IRC | 14:15 | |
| *** dustins has joined #openstack-manila | 14:15 | |
| *** carthaca has left #openstack-manila | 14:20 | |
| *** carthaca_ has joined #openstack-manila | 14:23 | |
| *** akapil_ has quit IRC | 14:37 | |
| *** akapil has joined #openstack-manila | 14:37 | |
| *** jprovazn has quit IRC | 14:42 | |
| *** tommylikehu_ has joined #openstack-manila | 14:48 | |
| *** mtanino has joined #openstack-manila | 14:51 | |
| mkoderer | tbarron: let me know when you are available | 14:56 |
| ganso | bswartz: not necessarily read-only access rules though, right? | 14:57 |
| ganso | gouthamr: ping | 14:57 |
| gouthamr | ganso: pong | 14:57 |
| ganso | gouthamr: Hi Goutham! | 14:57 |
| ganso | gouthamr: I am having difficulties adressing your comment about CONF.set_override | 14:57 |
| ganso | gouthamr: because of self.driver = self.mock_class("manila.share.driver.ShareDriver", mock.Mock()) | 14:58 |
| *** surabujin has quit IRC | 14:58 | |
| bswartz | ganso: we discussed whether read-only rules should be an optional or a required thing and nobody had a good reason to make them optional | 14:58 |
| bswartz | so we said required feature | 14:58 |
| ganso | gouthamr: I believe your suggestion does not work in this case without some modifications to the mocks in this test class | 14:58 |
| ganso | bswartz: but we agreed that drivers could have a read-only export location if they don't support read-only access rules | 14:59 |
| bswartz | ganso: what is a read only export location? | 15:05 |
| bswartz | and what driver doesn't support read only access rules? I thought we agreed there would be no drivers that didn't support read only access rules -- what am I missing? | 15:06 |
| ganso | bswartz: it is an export location that all access will be read-only, regardless of access rule mode | 15:06 |
| ganso | bswartz: I don't remember which drivers | 15:07 |
| ganso | bswartz: let me look at the matrix | 15:07 |
| bswartz | I understood that we needed to be able to force-readonly on a per-instance basis, not a per-export location basis | 15:08 |
| ganso | bswartz: gluster, HP 3PAR, and Oracle ZFS | 15:08 |
| bswartz | what would be the point of a share with 1 writable export location and 1 readonly export location? | 15:08 |
| ganso | bswartz: to provide read-only support... we've agreed to this in the past | 15:08 |
| bswartz | right but how would that come into existence? | 15:09 |
| bswartz | users cannot set rules per export location | 15:09 |
| bswartz | they can set them per-share | 15:09 |
| ganso | bswartz: http://docs.openstack.org/developer/manila/devref/driver_requirements.html#access-rules | 15:09 |
| ganso | bswartz: they don't | 15:09 |
| bswartz | we force some instances to be readonly, such as replicas or the source of a migrating share sometimes | 15:09 |
| gouthamr | ganso: sorry, in the middle of a meeting... will take a look at your responses... i think you don't need to read the config option in access.py if you use 'cast_to_readonly' as a parameter to the update_access method. | 15:10 |
| ganso | bswartz: for those backends, when the share is created, I believe there is an additional export location, so if the user does not have RW access to the main share export location, the user can access the RO one | 15:10 |
| bswartz | but I don't see a use case for per-export-location rules | 15:10 |
| bswartz | ah, we discussed that possibility for Generic+CIFS because of samba stupidity | 15:10 |
| bswartz | but I think we ultimately said no | 15:10 |
| *** dsariel has quit IRC | 15:11 | |
| bswartz | there would be no special export locations for readonly access | 15:11 |
| ganso | bswartz: not only that, we had driver maintainers involved in this | 15:11 |
| ganso | bswartz: we are feel like changing, should be boot gluster, HP 3PAR and Oracle ZFS for not supporting RO access rules? | 15:11 |
| bswartz | the only driver I recall being involved was the generic driver | 15:11 |
| ganso | bswartz: *if we feel | 15:12 |
| bswartz | ganso: we should fix those driver ideally | 15:12 |
| bswartz | I think we agreed in BCN that we won't handle non-compliant drivers by booting them anymore | 15:12 |
| bswartz | they will simply fail the tests that check for readonly support and we'll advertise that those drivers are broken and unmaintained (assuming they don't fix them) | 15:13 |
| ganso | gouthamr: that would mean bringing the parameter back to the RPCs as well, and that is not necessary | 15:13 |
| gouthamr | ganso: what about 'writable' and 'nondisruptive' migrations? | 15:14 |
| *** shausy has quit IRC | 15:14 | |
| ganso | gouthamr: update_access is not invoked in those cases, and the API blocks any subsequent call | 15:15 |
| ganso | gouthamr: I explained in one of the comment responses | 15:15 |
| gouthamr | ganso: if this is a short-lived bugfix that would need a refactor in ocata and beyond, i don't think we need to change the RPC.. we would not set the destination to read-only in any case. | 15:16 |
| ganso | gouthamr: exactly | 15:16 |
| gouthamr | ganso: https://review.openstack.org/#/c/375047/6/manila/share/manager.py@749 | 15:17 |
| ganso | gouthamr: but my argument in your spec still holds, unless there is a specific scenario I am missing that requires a column cast_to_readonly to be solved | 15:17 |
| gouthamr | ganso: isn't the source 'migrating'? | 15:17 |
| ganso | gouthamr: dang forgot to publish the comments | 15:17 |
| bswartz | ganso: if we had the state flag in the database we wouldn't need to block access rule changes during migrations | 15:18 |
| ganso | gouthamr: there | 15:18 |
| bswartz | there is no good reason to block access rule changes during a migration aside from implementation convenience | 15:18 |
| ganso | bswartz: we could block RW rules if the instance is migrating, and still wouldn't need a column | 15:19 |
| ganso | bswartz: or always cast | 15:19 |
| bswartz | but why block changes? there could be a use case for modifying access rules during a long migration | 15:19 |
| *** surabujin has joined #openstack-manila | 15:19 | |
| ganso | bswartz: we've discussed last midcycle in the case that I proposed unblocking the API to access rules so the admin could check the files at the destination, and the discussion outcome was that we should always block | 15:20 |
| bswartz | part of the value of the writable migration and nondisruptive migration is that you can use the share while it's migrating | 15:20 |
| *** senk has joined #openstack-manila | 15:20 | |
| ganso | bswartz: yes, in a writable migration the existing access rules will remain writable | 15:21 |
| bswartz | I don't like the idea of the admin "checking the files" because it implies that manila is somehow unreliable -- but I do feel that allowing access rule changes is desirable | 15:21 |
| bswartz | I suspect at the midcycle we were only discussing the use case of an admin checking the files | 15:22 |
| ganso | bswartz: right now we have nowhere to record the API parameters submitted to migrate the share. Also, if anything requires checking the driver capabilities to see if it supports a writable migration will defer to the share manager to do it, so there is no way we could return an API response in the case it is not writable | 15:22 |
| bswartz | the use case of a user adding a new access rule or removing an existing access rule during a long migration seems valid | 15:22 |
| bswartz | ganso: we don't need to store all that stuff -- just 1 bit of data per share instance -- the force_read_only flag | 15:23 |
| bswartz | with that 1 bit of data the share manager can always do the right thing | 15:23 |
| ganso | bswartz: ok, the API layer can check that | 15:24 |
| tbarron | mkoderer: sorry for the delay, what's up? | 15:26 |
| *** chlong has joined #openstack-manila | 15:27 | |
| mkoderer | tbarron: yeah wanted to discuss the race thing | 15:30 |
| mkoderer | tbarron: should I rebase it to you tooz patch and see how this would work? | 15:30 |
| tbarron | mkoderer: my name is on the tooz patch b/c I was instigating and pulled over stuff from cinder, but gouthamr really is the owner of that patch now | 15:31 |
| mkoderer | tbarron: oh didn't know that | 15:31 |
| tbarron | mkoderer: but it seems reasonable, let's see what gouthamr says | 15:32 |
| *** jprovazn has joined #openstack-manila | 15:32 | |
| gouthamr | mkoderer: please do.. it uses a similar "synchronized" decorator, but from the coordinator | 15:32 |
| tbarron | mkoderer: and as you say, that DNM patch that fails with your test is good material for the bswartz races spec | 15:33 |
| mkoderer | bswartz: https://review.openstack.org/#/c/403675/ | 15:33 |
| *** eharney has joined #openstack-manila | 15:34 | |
| mkoderer | tbarron: gouthamr: ok sounds like a plan. I will rebase and push it on top. | 15:34 |
| *** senk has quit IRC | 15:38 | |
| *** JoseMello has quit IRC | 15:43 | |
| *** tommylikehu_ has quit IRC | 15:51 | |
| *** madorn has quit IRC | 15:59 | |
| *** rraja has quit IRC | 16:00 | |
| *** chlong has quit IRC | 16:07 | |
| *** StraubTW has joined #openstack-manila | 16:08 | |
| gouthamr | mkoderer: sure thanks | 16:14 |
| *** eharney has quit IRC | 16:19 | |
| *** chlong has joined #openstack-manila | 16:20 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [DNM] test 5 https://review.openstack.org/404163 | 16:32 |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [DNM] test 6 https://review.openstack.org/404331 | 16:33 |
| *** pcaruana has quit IRC | 16:36 | |
| *** mtani____ has joined #openstack-manila | 16:39 | |
| *** mtanino has quit IRC | 16:40 | |
| *** zhongjun_ has quit IRC | 16:46 | |
| *** zhongjun_ has joined #openstack-manila | 16:49 | |
| *** tommylikehu_ has joined #openstack-manila | 16:52 | |
| *** mtanino has joined #openstack-manila | 16:52 | |
| *** mtani____ has quit IRC | 16:53 | |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix share writable in host-assisted migration https://review.openstack.org/375047 | 16:55 |
| *** eharney has joined #openstack-manila | 16:55 | |
| *** tommylikehu has quit IRC | 16:55 | |
| *** tommylikehu_ has quit IRC | 16:57 | |
| *** nkrinner is now known as nkrinner_afk | 17:04 | |
| *** dustins has quit IRC | 17:09 | |
| *** chlong has quit IRC | 17:10 | |
| markstur | Do we have much consistency with our "manage_existing" implementations? | 17:11 |
| markstur | I think we'd prefer that existing share access is revoked so that manila knows/contols all access -- but I doubt that is true across drivers | 17:12 |
| markstur | I also suspect we differ on rename/move vs leave where is (using private data to track). | 17:12 |
| markstur | That would ideally not matter, but it affects existing access and also affects how host/pool selection works or fails | 17:13 |
| *** xinyan has quit IRC | 17:21 | |
| markstur | vponomaryov: ping | 17:22 |
| *** chlong has joined #openstack-manila | 17:24 | |
| *** makowals_ has quit IRC | 17:25 | |
| markstur | ganso: ping | 17:25 |
| gouthamr | markstur: while managing a share, the driver need not change the export paths, but it really can't maintain existing access rules | 17:26 |
| gouthamr | markstur: the reason being that there's no way to let manila know about them during the manage operation | 17:26 |
| markstur | gouthamr: Right. But I think we've discussed that manila "should" have all the access knowledge. But I doubt it's true. | 17:26 |
| gouthamr | markstur: so, the expectation is that all export policies or access rules are removed when the share is brought in.. | 17:27 |
| markstur | So is it better for a driver to eliminate all existing access. Or better to fail and make the admin remove existing access???? | 17:27 |
| gouthamr | markstur: yes... we wanted to discuss this at the summit, but ran out of time. | 17:27 |
| ganso | markstur: pong | 17:27 |
| markstur | ganso: Hi. Was looking for opinions and manage existing ^ | 17:27 |
| gouthamr | markstur: some drivers do clear access rules | 17:27 |
| *** lpetrut has quit IRC | 17:28 | |
| markstur | So the "some drivers" is a problem. Also the lack of doc requirements. | 17:28 |
| gouthamr | yep | 17:28 |
| openstackgerrit | Victoria Martinez de la Cruz proposed openstack/manila: Decouple Manila UI from Manila Devstack plugin https://review.openstack.org/388855 | 17:28 |
| ganso | markstur: I had considered this situation in one of my drivers | 17:29 |
| ganso | markstur: in case we are adding a rule that already exist, we just log a warning | 17:29 |
| ganso | markstur: it may already exist because it was already there before the share was managed | 17:29 |
| markstur | So your managed shares keep pre-existing access | 17:29 |
| ganso | markstur: yes | 17:29 |
| ganso | markstur: in fact, one of my drivers yes | 17:30 |
| markstur | but it seems we don't quite have agreement on if that is a good thing or a bad thing | 17:30 |
| ganso | markstur: because it works differently | 17:30 |
| markstur | a feature or a bug | 17:30 |
| ganso | markstur: my other driver doesn't, because the first rule that is added after the share is managed overwrites everything else | 17:30 |
| gouthamr | a bug | 17:30 |
| markstur | a beature or a fug | 17:30 |
| markstur | a fug | 17:30 |
| gouthamr | lol | 17:30 |
| ganso | "a warm, stuffy, or smoky atmosphere in a room." | 17:30 |
| ganso | seems more like a fug indeed | 17:31 |
| markstur | gouthamr: I'd say it is a bug when it is someone else's driver and a feature in your own <-- but want to make sure I'm not really insulting you. Just in general. | 17:31 |
| gouthamr | markstur: lol.. no, we had these discussions. if it has to be one or two vendors that do it differently, then why manila :) | 17:32 |
| ganso | I think this is something we can't do consistently across all drivers | 17:33 |
| gouthamr | markstur: it makes no sense to preserve access rules and not let manila know.. because the next time a new rule gets added, manila expects the driver to clear everything anyway | 17:33 |
| markstur | With an admin API that is optional and has the option of saying InvalidShare -- there is room for us to differ. But I don't like that. | 17:33 |
| ganso | I recall drivers that could not list existing rules, so they would not be able to remove them | 17:33 |
| markstur | Whether drivers clear the existing ones at any point would probably vary. Some might be able to preserve them. | 17:34 |
| gouthamr | ganso: hmmm, so how does update_access affect those drivers? | 17:34 |
| vponomaryov | markstur: guys are right, your questions were in agenda for summit, but we didn't reach them | 17:34 |
| markstur | Need more summits | 17:34 |
| ganso | gouthamr: some drivers just refactored the old deny and allow into update_access because they had no way to do it differently | 17:34 |
| vponomaryov | markstur: and yes, we do not have explicit description of that feature | 17:35 |
| markstur | Yeah. The new update_access better suggests that access rules really should be in sync with what manila says | 17:35 |
| vponomaryov | markstur: according to current behavior we allow both ways for both questions | 17:35 |
| markstur | "manage share" could mean just mean to tell it to update its user stories and evaluate it annually | 17:36 |
| gouthamr | markstur: let me ask you this question slightly differently. How is the admin supposed to know if clients will lose access or retain access to the share that he is managing for a given back end? | 17:37 |
| markstur | gouthamr: The admin should know. Or should not do destructive things. | 17:37 |
| vponomaryov | gouthamr: he could create clones and then register them | 17:38 |
| markstur | gouthamr: Otherwise it would actually be good for the driver to fail with an error. | 17:38 |
| markstur | If this was not an admin feature, then we'd really need to avoid interfering with existing access I think. | 17:38 |
| gouthamr | markstur: how? think of a cloud with xyz and abc storage and abc behaves differently from xyz wrt to a manila API | 17:39 |
| markstur | So it depends on how many levels of admins you have. | 17:40 |
| markstur | Sure a cloud admin might not have access to backend details | 17:40 |
| * gouthamr is actually in favor of retaining access - nondisruptive operations ftw | 17:40 | |
| markstur | A backend admin would be able to see/understand the backend specifics | 17:40 |
| ganso | bswartz: this needs to be updated: https://blueprints.launchpad.net/manila/+spec/newton-migration-improvements | 17:40 |
| ganso | gouthamr: +1 | 17:41 |
| markstur | gouthamr: ganso: Does that mean it is OK to fail if there is existing access? | 17:41 |
| vponomaryov | gouthamr: retaining access with registration of rules in manila or not? | 17:42 |
| ganso | markstur: I think there is nothing wrong with allowing the existing access to remain there, at least until the first update_access call | 17:42 |
| markstur | ganso: That is bad | 17:42 |
| gouthamr | markstur: vponomaryov suggested that admins strip out all access and manage and run a script to add clients back (least downtime operation) | 17:43 |
| ganso | markstur: it is bad because we currently do not have consistency | 17:43 |
| markstur | ganso: So it would all look nice and go production and then some little change comes a long and things break | 17:43 |
| ganso | markstur: wouldn't it break anyway? a managed share shouldn't just go into production | 17:44 |
| bswartz | ganso: look good now? | 17:44 |
| gouthamr | vponomaryov: if the driver can discover it, i think it can register it with manila.. but bswartz's argument was that the pre-existing access rules may be more complicated than manila can understand/support | 17:44 |
| ganso | bswartz: thanks! | 17:44 |
| markstur | Registering existing access would be cool. | 17:45 |
| markstur | but inconsistently implemented | 17:45 |
| bswartz | what manila supports access-rule-wise is a subset of what each backend supports | 17:45 |
| gouthamr | markstur: not implemented | 17:45 |
| markstur | There is also a difference in whether the share is moved or left in-place. That certainly affects existing users and existing access. | 17:45 |
| bswartz | it's extremely likely that admins can create shares with rules that manila can't manage on most backends | 17:45 |
| markstur | So manage_existing will call-back to allow-access? | 17:46 |
| bswartz | I would prefer for manage to remove all access, or for manage to fail until the admin has removed all access | 17:47 |
| gouthamr | markstur: or just return access rules in a format that manila understands | 17:47 |
| markstur | gouthamr: Needs a spec | 17:47 |
| bswartz | +1 | 17:47 |
| gouthamr | :) #queens | 17:48 |
| bswartz | one alternative I can think of is a 2-phase manage where we give the admin a chance to add manila access rules to the share before we destroy the existing rules | 17:48 |
| vponomaryov | bswartz: then requires empty list of access rules is more safe approach | 17:48 |
| bswartz | that would be a significant change from the current design however | 17:48 |
| vponomaryov | s/requires/requirements of/ | 17:49 |
| bswartz | vponomaryov: yes but many implementations of manage already exist which don't have this requirement | 17:49 |
| markstur | Soooo. I'm thinking the GPFS driver should just fail with existing access for now. It can evolve from there if we add existing access later. | 17:50 |
| bswartz | and at least one implementation exists which tries to preserve pre-existing access rules even after manila access rule are added (which is incorrect behavior IMO) | 17:50 |
| ganso | ^ this would be aiming for disruptive manage operation | 17:52 |
| markstur | And then we have "unmanage_remove_access_rules" in the CONF just to legitimize inconsistent behavior on unman | 17:52 |
| bswartz | manage share as designed today is disruptive, and some drivers try to workaround that to make it less so | 17:52 |
| *** a-pugachev has quit IRC | 17:52 | |
| bswartz | the case for non-disruptive manage share is a pretty weak one, but there are things we can do to accomodate it in the design if we want | 17:53 |
| ganso | we are not solving the use case where a share may already be in production and being managed | 17:53 |
| gouthamr | or a share that's used in an application, unmanaged from manila and re-managed | 17:54 |
| ganso | bswartz: like a "manila manage --non-disruptive" ? :P | 17:54 |
| gouthamr | ganso: "nondisruptive" | 17:54 |
| bswartz | >_< | 17:54 |
| gouthamr | ganso: also, enforce-nondisruptive, allow-disruptive, allow-dataloss | 17:55 |
| gouthamr | all good options :) | 17:55 |
| markstur | kill-kill-kill | 17:55 |
| ganso | gouthamr: --yes-i-really-mean-it | 17:56 |
| * gouthamr top-secret-ci just posted comments on this change | 17:56 | |
| gouthamr | ganso: that classic 8) | 17:57 |
| markstur | ganso: Be sure to make the prompting require an interactive shell | 17:57 |
| markstur | which hangs for most use cases :( | 17:57 |
| * markstur goes to screen to press 'y' | 17:57 | |
| *** JoseMello has joined #openstack-manila | 18:07 | |
| *** lpetrut has joined #openstack-manila | 18:11 | |
| *** mkoderer has quit IRC | 18:13 | |
| ganso | markstur, xyang1, cknight, bswartz, tbarron, toabctl, gouthamr, vponomaryov: Could you please review https://review.openstack.org/375047/ when you have some time? I intend to propose a backport of that one as soon as it merges | 18:18 |
| *** makowals has joined #openstack-manila | 18:30 | |
| *** akapil_ has joined #openstack-manila | 18:44 | |
| *** akapil has quit IRC | 18:44 | |
| markstur | ganso: Still reviewing but there is a messed up log message (see inline) | 18:51 |
| markstur | ganso: Thanks for the comments. I don't like asking for more inline comments but that code lacks context for reading. | 18:52 |
| ganso | markstur: thanks, I am going to prepare the update, going to wait until you are finished so I can add anything else into the update | 18:52 |
| ganso | markstur: you mean, it is still lacking context? | 18:52 |
| markstur | ganso: I meant it WAS. Your comments help. | 18:53 |
| ganso | markstur: great! :) | 18:53 |
| markstur | ganso: Nothing else from me. | 19:06 |
| ganso | great | 19:07 |
| ganso | markstur: gonna git review | 19:07 |
| openstackgerrit | Rodrigo Barbieri proposed openstack/manila: Fix share writable in host-assisted migration https://review.openstack.org/375047 | 19:07 |
| *** akapil_ has quit IRC | 19:09 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [DNM] test 7 https://review.openstack.org/404394 | 19:19 |
| *** lpetrut has quit IRC | 19:21 | |
| *** dustins has joined #openstack-manila | 19:26 | |
| *** ociuhandu has quit IRC | 19:53 | |
| *** senk has joined #openstack-manila | 20:00 | |
| *** carthaca_1 has joined #openstack-manila | 20:03 | |
| *** carthaca_ has quit IRC | 20:03 | |
| *** ociuhandu has joined #openstack-manila | 20:05 | |
| *** senk has quit IRC | 20:06 | |
| markstur | ganso: I suppose your migration patch didn't cause IBM-CI to fail and NetApp-CI to UNREGISTER -- so +2 | 20:08 |
| ganso | markstur: yea most likely didn't | 20:10 |
| ganso | markstur: thanks! | 20:10 |
| *** dustins has quit IRC | 20:13 | |
| *** lpetrut has joined #openstack-manila | 20:14 | |
| bswartz | unregister? | 20:32 |
| bswartz | is our CI that bad? | 20:32 |
| *** catinthe_ has quit IRC | 20:35 | |
| *** alyson_ has quit IRC | 20:38 | |
| *** tommylikehu_ has joined #openstack-manila | 20:54 | |
| openstackgerrit | Valeriy Ponomaryov proposed openstack/manila: [DNM] test 7 https://review.openstack.org/404394 | 20:54 |
| markstur | manila-cDOT-no-ss NOT_REGISTERED | 20:56 |
| markstur | manila-cDOT-ss NOT_REGISTERED | 20:56 |
| openstackgerrit | Goutham Pacha Ravi proposed openstack/manila-specs: Add a spec to fix and improve Access Rules https://review.openstack.org/399049 | 20:56 |
| gouthamr | markstur: we're having some issues on our end. i'll disable voting until that's resolved. | 20:57 |
| *** tommylikehu_ has quit IRC | 20:58 | |
| * tbarron notes ameade is for hire | 20:58 | |
| * markstur wonders how much ameade charges to disable voting (is that a house call?) | 21:00 | |
| * ameade trips running into the room | 21:02 | |
| *** JoseMello has quit IRC | 21:03 | |
| tbarron | ameade: rofl | 21:06 |
| *** gouthamr has quit IRC | 21:11 | |
| *** porrua has quit IRC | 21:18 | |
| *** catintheroof has joined #openstack-manila | 21:34 | |
| *** catintheroof has quit IRC | 21:38 | |
| *** jprovazn has quit IRC | 21:45 | |
| *** kaisers_ has quit IRC | 21:47 | |
| *** catintheroof has joined #openstack-manila | 21:47 | |
| openstackgerrit | Tom Barron proposed openstack/manila: Fix smb configuration outside ubuntu https://review.openstack.org/404433 | 21:50 |
| openstackgerrit | Tom Barron proposed openstack/manila: Fix devstack smb configuration outside ubuntu https://review.openstack.org/404433 | 21:52 |
| *** cknight has quit IRC | 21:52 | |
| *** tpsilva has quit IRC | 21:55 | |
| *** lpetrut has quit IRC | 22:05 | |
| openstackgerrit | Merged openstack/manila: Fix share writable in host-assisted migration https://review.openstack.org/375047 | 22:09 |
| *** kaisers has joined #openstack-manila | 22:11 | |
| *** chlong has quit IRC | 22:12 | |
| *** ganso has quit IRC | 22:16 | |
| *** kaisers has quit IRC | 22:16 | |
| *** kaisers has joined #openstack-manila | 22:17 | |
| *** eharney has quit IRC | 22:18 | |
| *** xyang1 has quit IRC | 22:50 | |
| *** StraubTW has quit IRC | 23:02 | |
| *** catinthe_ has joined #openstack-manila | 23:03 | |
| openstackgerrit | Tom Barron proposed openstack/manila: Fix devstack smb configuration outside ubuntu https://review.openstack.org/404433 | 23:05 |
| *** panatl has joined #openstack-manila | 23:05 | |
| panatl | ^all I am getting some issue while testing in Devstack | 23:06 |
| panatl | SSHException: Check whether private key or password are correctly set. Error connecting via ssh: not a valid EC private key file | 23:06 |
| panatl | any idea? | 23:06 |
| *** catintheroof has quit IRC | 23:06 | |
| tbarron | panatl: maybe the "EC private key file" part is misleading: https://github.com/paramiko/paramiko/issues/521 but you still didn't setup up your ssh keys right? | 23:16 |
| tbarron | panatl: who is trying to ssh to what? | 23:17 |
| panatl | i have the password less ssh keys in .ssh folder | 23:17 |
| panatl | i am getting this after running manila create --name devstack_share --share-network cf621758-bd07-4f99-9fdb-ee1e1fd5e67c NFS 1 | 23:18 |
| panatl | my local.conf is | 23:19 |
| panatl | [[local|localrc]] | 23:19 |
| panatl | enable_plugin manila https://github.com/openstack/manila | 23:19 |
| *** tommylikehu_ has joined #openstack-manila | 23:20 | |
| tbarron | panatl: I've never done it with that simple two-line local.conf but I see it at https://github.com/openstack/manila/blob/master/devstack/README.rst. Will try it out ... | 23:31 |
| panatl | tbarron: thanks! .. do you have a sample local.conf .. which i can try? | 23:39 |
| tbarron | panatl: here are a couple that work for me: https://github.com/tombarron/manila-local.conf | 23:41 |
| openstackgerrit | Merged openstack/puppet-manila: Fix the annotation https://review.openstack.org/404007 | 23:41 |
| panatl | tbarron: thanks! let me give it a try | 23:42 |
| *** tommylikehu_ has quit IRC | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!