| *** threestrands has joined #openstack-manila | 00:09 | |
| *** andrebeltrami has quit IRC | 03:08 | |
| *** carloss has quit IRC | 03:20 | |
| openstackgerrit | inspurericzhang proposed openstack/manila-tempest-plugin master: Add py38 package metadata https://review.opendev.org/731411 | 03:41 |
|---|---|---|
| *** threestrands has quit IRC | 03:49 | |
| *** tosky has joined #openstack-manila | 07:44 | |
| *** tkajinam has quit IRC | 08:37 | |
| openstackgerrit | Liron Kuchlani proposed openstack/manila-tempest-plugin master: Preparation for support scenario tests on CephFS Native https://review.opendev.org/725487 | 08:57 |
| openstackgerrit | Liron Kuchlani proposed openstack/manila-tempest-plugin master: Preparation for support scenario tests on CephFS Native https://review.opendev.org/725487 | 09:41 |
| *** tosky_ has joined #openstack-manila | 09:57 | |
| *** tosky is now known as Guest48633 | 09:57 | |
| *** tosky_ is now known as tosky | 09:57 | |
| openstackgerrit | Maurice Escher proposed openstack/manila master: add share server update to manila-manage share update_host https://review.opendev.org/731486 | 10:09 |
| openstackgerrit | Liron Kuchlani proposed openstack/manila-tempest-plugin master: Preparation for support scenario tests on CephFS Native https://review.opendev.org/725487 | 11:41 |
| openstackgerrit | Maurice Escher proposed openstack/manila master: add manila-manage service cleanup https://review.opendev.org/731504 | 11:48 |
| *** carloss has joined #openstack-manila | 11:59 | |
| *** eharney has joined #openstack-manila | 13:03 | |
| vkmc | can I get some eyes on https://review.opendev.org/#/c/726514/? | 14:22 |
| vkmc | easy one | 14:22 |
| vkmc | and this one https://review.opendev.org/#/c/725680/ | 14:23 |
| vkmc | also easy one | 14:24 |
| openstackgerrit | Victoria Martinez de la Cruz proposed openstack/python-manilaclient master: Add initial documentation for OSC https://review.opendev.org/705834 | 14:37 |
| openstackgerrit | Victoria Martinez de la Cruz proposed openstack/python-manilaclient master: Improve OSC test coverage https://review.opendev.org/710830 | 14:38 |
| *** maaritamm has joined #openstack-manila | 14:52 | |
| *** andrebeltrami has joined #openstack-manila | 14:57 | |
| *** vhari has joined #openstack-manila | 14:59 | |
| *** felipe_rodrigues has joined #openstack-manila | 14:59 | |
| carloss | sure, vkmc | 15:00 |
| vkmc | :D | 15:02 |
| *** jmlowe has quit IRC | 15:34 | |
| *** jmlowe has joined #openstack-manila | 15:37 | |
| gouthamr | ganso: looks like cinder addressed this a while ago: https://review.opendev.org/#/c/405578/ | 16:02 |
| gouthamr | ganso: routing volume extensions via the scheduler i mean | 16:02 |
| ganso | gouthamr: hmmm so it could make sense for us too to go through the scheduler | 16:03 |
| gouthamr | ganso: then again, they also made their volume types immutable: https://review.opendev.org/#/c/440680/ | 16:04 |
| gouthamr | in-use volume types* | 16:04 |
| ganso | gouthamr: hmmm that's debatable. Has its pros and cons, and that's a far greater change at the moment | 16:05 |
| gouthamr | ganso: agreed, routing extensions through the scheduler still make senses - we should have a doc on what mutability of share types means | 16:05 |
| ganso | greater in the sense of user impact | 16:05 |
| ganso | s/greater/bigger for disambiguity | 16:06 |
| gouthamr | :) | 16:06 |
| ganso | gouthamr: I believe we may have code in place to handle the situation of not having enough space to do an extend operation. I remember this as well from a previous discussion. Changing to go through the scheduler now would render that stale | 16:08 |
| ganso | but it could be a better solution perhaps | 16:09 |
| gouthamr | ganso: oh? i didn't see it while reviewing haixin's proposed change | 16:09 |
| ganso | gouthamr: I believe it is handled in the manager/driver | 16:09 |
| gouthamr | ganso: yep, will check again | 16:09 |
| gouthamr | ganso: ack | 16:09 |
| gouthamr | dviroel lseki: o/ wouldn't it help to have a user message if share server deletion fails? without leaking the abstraction, the message could be as generic as "De-allocation of share network ports did not succeed. Contact your administrator to determine the cause and rectify." | 16:11 |
| dviroel | gouthamr: i think that make sense, yes. Is port allocation the only thing that will affect users? | 16:13 |
| dviroel | regarding resources that they can see | 16:13 |
| gouthamr | dviroel: yeah pretty much, we allow cleaning up shares/networks/groups etc.. | 16:16 |
| gouthamr | dviroel: a similar message could be used while unmanaging a share server, because its a very similar situation | 16:18 |
| carthaca | gouthamr ganso: what you think: will/should the extend through scheduler modification still keep the old feature to skip the scheduler? e.g. via a force-without-scheduler option or by having the possibility to define the set of scheduler filters to be used for this action? | 16:18 |
| ganso | carthaca: what is the use case for maintaining the old approach? | 16:19 |
| lseki | gouthamr: yes, sounds more helpful than just raising an exception | 16:19 |
| carthaca | ganso: we keep a high reserved_percentage to stop new provisioning early, but existing shares should be allowed to grow (and snapshots ideally, too) | 16:21 |
| ganso | carthaca: I think that the impact of that depends on the implementation | 16:24 |
| ganso | carthaca: in my mind, the extend operation wouldn't go through the regular "scheduling routines" | 16:24 |
| gouthamr | carthaca: i see.. | 16:25 |
| gouthamr | ganso: in the current approach haixin's taking, we'll be going through all the scheduler filters | 16:25 |
| gouthamr | same thing cinder does, i think | 16:25 |
| ganso | carthaca: the host is predefined, it would just check the available space | 16:25 |
| ganso | gouthamr: yea, cinder does that as well "host_passes_filter()" | 16:26 |
| gouthamr | ganso: just checking available space would be hard, we'd need to do the overprovisioning, headroom calculations as well, for one? | 16:27 |
| ganso | I don't personally think it makes a lot of sense. However, in either way the reserved_percentage would come into play | 16:27 |
| dviroel | in that case, would be good to have a way of avoiding scheduler, like we do for create from snapshot | 16:28 |
| gouthamr | yes, and prevent users from extending their existing shares | 16:28 |
| ganso | gouthamr: what I'm trying to say is that we don't need to run filters to decide a "host", like the weighers etc | 16:28 |
| gouthamr | ganso: i agree, it's like you've done for qualifying a migration destination host - just run it through the filters | 16:29 |
| ganso | gouthamr: yes | 16:29 |
| gouthamr | dviroel carthaca: this is a good concern to take to haixin's patch: https://review.opendev.org/#/c/700393/ | 16:29 |
| gouthamr | let's continue the discussion there, this will def be one of the top concerns to address | 16:31 |
| dviroel | gouthamr: ack | 16:32 |
| *** maaritamm has quit IRC | 16:32 | |
| *** jv_ has quit IRC | 16:38 | |
| *** jv_ has joined #openstack-manila | 16:41 | |
| *** jv_ has quit IRC | 16:53 | |
| *** jv_ has joined #openstack-manila | 16:54 | |
| *** maaritamm has joined #openstack-manila | 18:46 | |
| openstackgerrit | Maari Tamm proposed openstack/manila master: Add user message when share shrinking fails at the driver https://review.opendev.org/726514 | 19:55 |
| *** maaritamm has quit IRC | 19:59 | |
| *** felipe_rodrigues has quit IRC | 20:18 | |
| gouthamr | dviroel carloss: friends, one of you willing to pull the trigger here? https://review.opendev.org/#/c/726291/ :) | 21:25 |
| carloss | gouthamr will take a look :) | 21:25 |
| carloss | s/will/I will | 21:26 |
| gouthamr | carloss++ | 21:26 |
| *** threestrands has joined #openstack-manila | 21:45 | |
| *** tkajinam has joined #openstack-manila | 22:54 | |
| *** tosky has quit IRC | 23:20 | |
| *** andrebeltrami has quit IRC | 23:26 | |
| openstackgerrit | Merged openstack/manila master: CIFS extension failing because of volume in use https://review.opendev.org/498445 | 23:28 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!