Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2017-01-12
14:04:24 huanxuan I cannot see any information about it
14:05:01 stevemar huanxuan: that's fine, i was hoping there was information about in the volume, but i guess not
14:05:15 stevemar huanxuan: i'm just concerned about mike's last comments
14:05:38 stevemar he really doesn't like the --remote option, sounds like he would have liked manage/unmanage commands
14:06:18 dtroyer I really don't like manage/unmanage, but don't understant the choice of —remote either
14:06:20 huanxuan stevemar: yes
14:06:42 stevemar dtroyer: did we release those commands yet?
14:06:42 dtroyer is 'set —manage | —unmanage' a thing here?
14:07:04 dtroyer stevemar: no idea…probably, this goes back to before the previous release or two
14:07:47 dtroyer if we make a change in direction, deprecating —remote is no big deal
14:08:17 huanxuan yes, we can make a change if necessary
14:08:21 stevemar grr, went into 3.6.0 and 3.5.0
14:08:28 dtroyer my concern with manage/unamanage is that it is fairly meaningless without already knowing exactly what cinder defines it to mean, not something that can ge used elsewhere sicne no other API that I know of has this concenpt
14:08:31 stevemar thats true
14:08:51 huanxuan yeah
14:09:00 dtroyer those really are db-only operations
14:09:18 RuiChen sorry, I think make OSC version match openstack version make sense for downstream team, it's easier to pick right osc version for stable openstack , so what's your advise? dtroyer
14:09:18 dtroyer if I wanted to be a jerk, I'd say they belong in cinder-manage :)
14:09:50 dtroyer RuiChen: I'd like to finish the cinder thing first…
14:09:59 stevemar we can always deprecate the option or add a new manage command (the two options for creating can co-exist)
14:10:13 stevemar dtroyer: i think we're done
14:10:28 RuiChen sure, here you go, sorry
14:10:29 dtroyer right. this is a case where I do think we need to step back a bit and make sure we build the right UI
14:10:39 dtroyer ok, thanks guys
14:11:03 huanxuan RuiChen: thanks
14:11:10 dtroyer RuiChen: the current OSC will always have features that an older cloud can not use, but that does not make it unusable on older clouds
14:11:53 stevemar if someone comes up with a nice UI, they can build it. the two options can co-exist. just like git checkout -b new_branch; and git branch -d <new_branch>
14:11:55 dtroyer if that idea is a problem for your downstream, then it will always be a problem during every release
14:12:12 huanxuan stevemar: sounds good
14:13:02 huanxuan stevemar: so we are going to add manange and unmanage commands now, right?
14:13:12 stevemar huanxuan: not now, whenever :P
14:13:19 dtroyer the cases where matching stable OSC to a cloud release are usually due to distributions that can not co-install multiple python bits from different releases due to dependency conflicts
14:13:58 huanxuan stevemar: OK, get it! :)
14:14:53 dtroyer RuiChen: so what specifically is blocking your use of the current OSc releases?
14:16:52 RuiChen our user confused about why some osc commands raise error in stable openstack, they don't know openstack too much
14:17:21 RuiChen i means latest osc run against old version openstack cloud
14:18:11 dtroyer are those due to differences in the API or due to deployment choices?
14:18:15 RuiChen and the dependency conflicts is other reason too
14:19:24 RuiChen for example, that SSL bug, we deploy all openstack services in SSL
14:19:42 dtroyer and I agree that is one that probably should be backported
14:20:27 dtroyer can you prepare that one for a backport?
14:20:42 RuiChen thank you, I will
14:21:12 RuiChen I saw the backport patch is abandoned, so ask these
14:22:19 dtroyer fortunately, stable releases will not be stuck in our upcoming freeze so there is not a time pressure from our side for that
14:23:19 RuiChen \o/, I will prepare the patch
14:23:55 RuiChen just make sure, will we release a stable osc version for it?
14:24:21 dtroyer yes
14:24:48 RuiChen thank you :)
14:32:29 stevemar RuiChen: thanks for your help with osc :)
14:33:05 dtroyer huanxuan: that image test is still racy, failed py35 but passed py27
14:33:33 huanxuan dtroyer: yes, I saw it
14:33:49 dtroyer I'm having a hard time reproducing it locally, can you?
14:34:20 huanxuan dtroyer: I am working on it
14:35:07 dtroyer I am wondering if we need to reset the mock inside the for loop
14:35:27 dtroyer well, reset is at the bottom, I mean re-initialize it
14:37:22 huanxuan dtroyer: re-initialize?
14:38:00 dtroyer move lines 855-857 below 859
14:40:13 huanxuan dtroyer: Oh! get your point!
14:43:13 dtroyer stevemar, huanxuan: https://review.openstack.org/#/c/419461/ looks ready to go
14:43:50 openstackgerrit Huanxuan Ao proposed openstack/python-openstackclient: Fix gate broken by auth_ref in image unit test https://review.openstack.org/419445
14:44:12 huanxuan dtroyer: +A
15:06:03 openstackgerrit Huanxuan Ao proposed openstack/python-openstackclient: Fix gate broken by auth_ref in image unit test https://review.openstack.org/419445
15:08:20 stevemar dtroyer: i dont understand the if statements in the tests
15:08:28 reedip ralonsoh : hehe, gate did it :D
15:11:35 dtroyer stevemar: Where? are you talking about test_image_set_membership_option()?
15:11:52 stevemar dtroyer: commented
15:11:55 stevemar https://review.openstack.org/#/c/419461/1
15:14:00 openstackgerrit Steve Martinelli proposed openstack/python-openstackclient: Fix gate broken by auth_ref in image unit test https://review.openstack.org/419445
15:14:34 dtroyer ok… ugh, more parallel testing side effects maybe?
15:14:37 stevemar dtroyer: rebased https://review.openstack.org/#/c/419445/4 on top of the qos fix
15:15:03 stevemar dtroyer: yeah, i was wondering if we're just opening ourselves up to more racey tests
15:16:28 dtroyer totally. am I off base for just wanting unrolled multiple tests that don't try to share too much?
15:17:13 dtroyer my gut says the for loop in the image test is part of our pain
15:17:18 dtroyer too
15:18:14 dtroyer I'm going to do an unroll of the qos tests and see how bad it looks
15:18:44 dtroyer although not gettgin them to fail locally makes it hard to prove it is worth the effort
15:20:29 openstackgerrit Reedip proposed openstack/python-openstackclient: Fix quota show output https://review.openstack.org/419067
15:20:33 stevemar hehe
15:21:27 dtroyer so I'm going to go find it, but I had to do a lot of deepcopy() in the past to avoid this sort of thing with shared mock return values, I think we've forgotten that
15:25:39 huanxuan dtroyer: failed again :(
15:40:16 dtroyer so I'm trying a simple unrolling of that loop and making it into 3 tests
15:43:16 openstackgerrit Dean Troyer proposed openstack/python-openstackclient: Fix image member unit tests occasional failures https://review.openstack.org/419528
15:43:25 dtroyer let's see what that does ^^^
15:45:58 huanxuan dtroyer: good idea, thank you!
15:47:07 huanxuan I am also trying in other ways
15:48:00 etoews cdent elmiko edleafe i have a conflicting meeting today and won't be able to attend the api wg meeting
15:49:05 etoews edleafe: you hoser
15:49:06 edleafe :)
15:49:17 openstackgerrit Huanxuan Ao proposed openstack/python-openstackclient: Fix gate broken by auth_ref in image unit test https://review.openstack.org/419445
15:49:38 openstackgerrit Ed Leafe proposed openstack/api-wg: Add guidelines for boolean names https://review.openstack.org/411529
15:52:22 elmiko edleafe: ack
15:52:36 elmiko er etoews lol
15:52:59 edleafe elmiko: that's ok. etoews and I are practically indistinguishable
15:53:45 elmiko hehe =)
15:55:59 mordred edleafe: when is the api wg meeting?
15:56:14 edleafe mordred: 5 minutes
15:56:28 edleafe mordred: in #openstack-meeting-3
15:57:00 mordred edleafe: neat
16:02:13 openstackgerrit Merged openstack/python-openstackclient: Fix Network QoS rule CLI attrs parsing https://review.openstack.org/419461
16:07:49 waj334 dtroyer: What's the verdict on this patch? https://review.openstack.org/#/c/412614/

Earlier   Later