| openstackgerrit | Merged openstack-infra/shade master: Allow a user to submit start and end time as strings https://review.openstack.org/467257 | 00:52 |
|---|---|---|
| openstackgerrit | OpenStack Proposal Bot proposed openstack-infra/shade master: Updated from global requirements https://review.openstack.org/467428 | 01:02 |
| openstackgerrit | Merged openstack/os-client-config master: Revert "Use interface not endpoint_type for keystoneclient" https://review.openstack.org/462144 | 01:02 |
| *** gouthamr has quit IRC | 02:33 | |
| *** gkadam has joined #openstack-shade | 03:20 | |
| *** slaweq has joined #openstack-shade | 03:50 | |
| *** slaweq has quit IRC | 03:55 | |
| *** ChanServ changes topic to "#openstack-shade" | 04:08 | |
| -openstackstatus- NOTICE: Sufficient free space has been reclaimed that jobs are passing again; any POST_FAILURE results can now be rechecked. | 04:09 | |
| openstackgerrit | Rajat Sharma proposed openstack-infra/shade master: Replace assertRaisesRegexp with assertRaisesRegex https://review.openstack.org/467486 | 05:08 |
| *** ioggstream has joined #openstack-shade | 06:51 | |
| *** ioggstream has quit IRC | 07:08 | |
| *** jroll has quit IRC | 07:12 | |
| *** ioggstream has joined #openstack-shade | 07:22 | |
| *** jroll has joined #openstack-shade | 10:25 | |
| *** ioggstream has quit IRC | 11:24 | |
| *** ioggstream has joined #openstack-shade | 11:52 | |
| *** noshankus has quit IRC | 12:05 | |
| *** noshankus has joined #openstack-shade | 12:05 | |
| openstackgerrit | Merged openstack-infra/shade master: Pick most recent rather than first fixed address https://review.openstack.org/467385 | 12:52 |
| *** gouthamr has joined #openstack-shade | 13:50 | |
| Shrews | mordred: https://github.com/ansible/ansible/pull/24881 needs a second shipit to merge, if it looks good to you | 13:55 |
| Shrews | or rcarrillocruz ^^^ | 13:55 |
| *** gkadam has quit IRC | 13:56 | |
| Shrews | mordred: also, bug in image endpoint discovery? https://github.com/ansible/ansible/issues/24786 | 13:58 |
| Shrews | mordred: last thing, i promise... these tempest modules being proposed feel a bit out of place (https://github.com/ansible/ansible/pull/24173 , https://github.com/ansible/ansible/pull/24153). How do you feel about them? | 14:03 |
| mordred | Shrews: image endpoint discovery bug should be fixed in the latest shade that just got cut | 14:20 |
| mordred | Shrews: although- I think we should put in a try/except around image discovery so that we always fall back to whatever's in the catalog | 14:23 |
| Shrews | mordred: that method still exists in the current shade, so shouldn't the KeyError problem still exist? | 14:28 |
| mordred | Shrews: yup | 14:37 |
| Shrews | then i remain confused | 14:43 |
| mordred | Shrews: sorry - we fixed one thing related to image discovery exceptions - but then I realized this is a different issue | 14:44 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Use catalog endpoint on any errors in image version discovery https://review.openstack.org/467644 | 14:44 |
| mordred | Shrews: then I stopped talkined to you so I could write that ^^ | 14:44 |
| Shrews | oh | 14:44 |
| Shrews | by all means, stop talking to me to fix more things! | 14:44 |
| mordred | Shrews: we ALMOST got all the way to the logic of "don't screw the user if image discovery doesn't work" | 14:44 |
| mordred | Shrews: for the tempest modules - I chatted with that guya little while ago and told him they seemed like a fine idea ... there's another module he was working on for making a tempest config that actually needed shade (and needed new methods on shade) | 14:45 |
| mordred | becuase it was talking to the api to find out what extensions and whatnot were enabled then writing a config | 14:46 |
| Shrews | mordred: i left a comment on the tempest things. i just don't think they belong, but if you and others think they do, so be it | 14:46 |
| mordred | that said - I do still think that "os_tempest_results" should just be "subunit" and should be elsewhere | 14:47 |
| Shrews | jlk: btw, i invoked your name on PR 24173 above | 14:47 |
| Shrews | i mean, i have NEVER run tempest before and have no interest in doing so, so I'm not going to test or fix bugs for the os_tempest_* modules | 14:49 |
| Shrews | and this person is adding them for a final project for school, so it's likely he won't either | 14:50 |
| * Shrews returns to non-ansible things | 14:51 | |
| *** pabelanger has quit IRC | 14:52 | |
| *** pabelanger has joined #openstack-shade | 14:52 | |
| *** fbaca has quit IRC | 14:58 | |
| *** slaweq has joined #openstack-shade | 16:49 | |
| *** slaweq has quit IRC | 16:53 | |
| *** slaweq has joined #openstack-shade | 16:53 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Log cloud name on Connection retry issues https://review.openstack.org/467686 | 16:59 |
| mordred | Shrews: btw - the added non-voting py3 functional jobs actually show some py3 errors we're not trapping in unit tests | 17:02 |
| mordred | Shrews: (shade side, obvs) | 17:02 |
| Shrews | neat | 17:02 |
| *** rcarrillocruz has quit IRC | 17:12 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Add time reporting to Connection Retry message https://review.openstack.org/467689 | 17:24 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Fix python3 issues in functional tests https://review.openstack.org/467693 | 17:35 |
| *** rcarrillocruz has joined #openstack-shade | 17:47 | |
| *** rcarrillocruz has quit IRC | 18:27 | |
| *** rcarrillocruz has joined #openstack-shade | 18:36 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Fix python3 issues in functional tests https://review.openstack.org/467693 | 19:01 |
| *** slaweq has quit IRC | 19:31 | |
| *** slaweq has joined #openstack-shade | 19:31 | |
| *** slaweq has quit IRC | 19:40 | |
| *** openstackgerrit has quit IRC | 19:48 | |
| *** rcarrillocruz has quit IRC | 20:27 | |
| *** slaweq has joined #openstack-shade | 20:28 | |
| slaweq | mordred: hello | 20:35 |
| *** rcarrillocruz has joined #openstack-shade | 20:35 | |
| slaweq | mordred: I'm checking designateclient now and it looks that it's still using v1 API for domains and recordsets | 20:35 |
| morgan | mordred: can i say, the whole "admin vs non-admin" API in keystone v2 is terrrrrrrrrrible | 20:36 |
| slaweq | so I think that I will do the same in first step and will only add calls using API v1 (and remove API discovery calls) | 20:36 |
| mordred | morgan: very very bad | 20:36 |
| slaweq | I will also add TODOs with info about discovery to add | 20:37 |
| mordred | slaweq: yes - I think this is the right choice - then adding v2 and discovery later ... | 20:37 |
| mordred | yes - exactly | 20:37 |
| slaweq | and I will add this discovery in second step | 20:37 |
| mordred | \o/ | 20:37 |
| slaweq | ok, thx :) | 20:37 |
| morgan | mordred: hmmm... trying to figure out how to ensure we're acting on admin API in shade when de-client-i-fying | 20:37 |
| mordred | btw - we cut a new shade release today - so nodepool is now running with the neutron REST calls intead of neutronclient | 20:37 |
| slaweq | :) | 20:37 |
| morgan | mordred: is it safe to assume we *always* act on admin api with shade in v2.0? | 20:38 |
| mordred | morgan: yes - we have no end-user api calls (I don't think) in the 2.0 stack | 20:38 |
| morgan | ok cool so i can do just self._keystoneclient.get()... | 20:38 |
| mordred | morgan: also - I *think* we added something in the latest occ to help with this | 20:38 |
| morgan | well, we'll see how it goes :P | 20:38 |
| mordred | morgan: :) | 20:39 |
| mordred | morgan: also - jamielennox said something a few weeks ago about being able to specify interface in endpoint_filters in an individual call | 20:40 |
| mordred | morgan: but I think that assumes making a call on session not on a mounted adapter | 20:40 |
| morgan | mordred: yeah | 20:40 |
| mordred | morgan: but worth considering maybe if you get in a bind | 20:40 |
| morgan | mordred: i'd like to not use self._identity_client, but that can be a future cleanup | 20:42 |
| morgan | oh look... yeah that doesn't work | 20:42 |
| morgan | bah | 20:42 |
| morgan | at all | 20:42 |
| mordred | morgan: I may be reading you backwards ... you want to not-use the adapter? or you're saying step-one workaround is using keystoneclient's get call to switch to the rest calls, and step 2 is swift from keystoneclient.get to ksa.adapter.get ? | 20:43 |
| * mordred is foggy on things sometimes | 20:44 | |
| morgan | eventually i want to use KSA directly | 20:44 |
| morgan | not "identity_client" which is a ksc object | 20:44 |
| morgan | even though we call ".get" on it | 20:44 |
| morgan | thats all | 20:44 |
| morgan | so yeah ksc.get to ksa.adapter.get | 20:44 |
| morgan | but starting with self._identity_client | 20:45 |
| mordred | gotcha. so - fwiw, self._identity_client is the adapter and self.keystone_client is the ksa object | 20:47 |
| mordred | and yes - I think you have described an excellent future state | 20:47 |
| morgan | mordred: oh man. we aren't doing adapter mounting at all here. | 20:56 |
| morgan | ksc does so much magic | 20:56 |
| morgan | ............. | 20:56 |
| mordred | morgan: all of the client libs do an enormous amount of magic, it turns out - and almost all of it just makes things harder | 20:56 |
| morgan | i mean, is it safe to just do: | 20:56 |
| morgan | https://www.irccloud.com/pastebin/vPT5f4yc/ | 20:57 |
| morgan | yes the join is explicitly needed in this case | 20:58 |
| morgan | it seems like it fall back to v2.0 by default | 20:58 |
| *** openstackgerrit has joined #openstack-shade | 20:58 | |
| openstackgerrit | Merged openstack-infra/shade master: Fix python3 issues in functional tests https://review.openstack.org/467693 | 20:58 |
| morgan | if that is safe i need to muck with tests as well. | 20:58 |
| mordred | morgan: it's probably safe - however, go take a peek at what we're doing in _image_client - where we do the api_version hack thing once when we make the adapter | 20:59 |
| morgan | ah | 20:59 |
| morgan | ok | 21:00 |
| morgan | *grump* stupid api versions being weird. | 21:00 |
| mordred | all of these hack should go away when we get version discovery finished in ksa - but we need to get the specs landed first - then I need to write a keystone spec for that ... so shade-level hacks will be here for at least a few weeks | 21:00 |
| mordred | morgan: zomg tell me about it | 21:00 |
| morgan | you mean | 21:00 |
| morgan | https://www.irccloud.com/pastebin/4RksW9n7/ | 21:00 |
| morgan | do something like that... | 21:01 |
| morgan | ? | 21:01 |
| mordred | nope- one sec | 21:01 |
| mordred | morgan: https://github.com/openstack-infra/shade/blob/master/shade/openstackcloud.py#L434 | 21:01 |
| morgan | oh i see | 21:02 |
| mordred | which gets called in https://github.com/openstack-infra/shade/blob/master/shade/openstackcloud.py#L499 | 21:02 |
| mordred | its SO EVIL | 21:02 |
| morgan | i mean... that seems like it will break the magic in ksc | 21:02 |
| morgan | unless we replace all ksc calls at once. | 21:02 |
| morgan | so non-magic calls will suddenly be pointed at the wrong thing? | 21:03 |
| mordred | identity_client isn't ksc | 21:03 |
| morgan | oh it isn't? | 21:03 |
| mordred | self.keystone_client is the ksc client | 21:03 |
| morgan | ooooooh | 21:03 |
| morgan | ok | 21:03 |
| mordred | self._identity_client is the adapter | 21:03 |
| mordred | morgan: so there's a bit of logic you can see in the image version discovery - where we try to honor the version the user tells us - but if we can't we fall back to whatever we can find and then just set the internal config variable to 1 or 2 | 21:04 |
| mordred | the reason is that shade knows how to do both - and absentspecific instructions from the user should prefer the latest version - but sometimes the user wants to tell us that versoin is broken (which happens) | 21:05 |
| mordred | oh - actually ... | 21:05 |
| mordred | morgan: https://review.openstack.org/#/c/467644/ I just wrote that this morning based on a bug report | 21:05 |
| mordred | might want to look at that too | 21:05 |
| morgan | oh wonderful... | 21:27 |
| morgan | keystone doesn't have a "CURRENT" status | 21:27 |
| morgan | all status is "STABLE" | 21:27 |
| mordred | morgan: yup | 21:36 |
| morgan | yeah this is notfun(tm) | 21:37 |
| mordred | morgan: stable and deprecated iirc | 21:37 |
| mordred | morgan: the entire enchilada is here: https://review.openstack.org/#/c/459405/18/guidelines/consuming-version-discovery.rst | 21:38 |
| morgan | now i need to figure out a couple things... but it's mostly in tests now... being wonky due to the next thing | 21:38 |
| morgan | https://www.irccloud.com/pastebin/UsVj0v82/ | 21:38 |
| mordred | covers the keystone stable case and all the rest of it | 21:38 |
| mordred | (and can use review by smart people - especially smart people who are working on version discovery :) ) | 21:38 |
| mordred | morgan: yup- that looks about right | 21:39 |
| morgan | haha... oi i picked a rough one | 21:40 |
| morgan | i should have gone with like update_user :P | 21:40 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade master: First Pass: De-client-ify list_users https://review.openstack.org/467784 | 21:40 |
| morgan | WIP, but i can't lose progress this way | 21:41 |
| mordred | morgan: well - nice thing about picking a rough one first is you get all the plumbing done | 21:41 |
| morgan | i have a RFE for git-review *and* zuul, 1) --WIP, which auto sets WIP mode, or similar when submitting, and in zuul, don't execute if in WIP *except* with a specific comment posted. though order of operations is a bit rough here because how notifies from gerrit work | 21:42 |
| morgan | there should be a "don't run tests unless i explicitly ask" mode for code pushes to gerrit... | 21:42 |
| * morgan tries to figure out how to get there. | 21:42 | |
| morgan | workflow-wise (not implementation wise) | 21:42 |
| mordred | morgan: yup - that's actually been a long-standing request - we kind of had it WAY back in the day when we had a gerrit WIP plugin | 21:47 |
| mordred | but since we upgraded to 2.6 (maybe 4 years ago) the WIP status in our gerrit is just an openstack local config setting so we've been avoiding adding the flag to git-review since there are other git-review users | 21:48 |
| mordred | BUT ... | 21:48 |
| mordred | we're getting closer to actually getting WIP landed upstream | 21:48 |
| mordred | (if you think convincing openstack people about features is hard ...) | 21:48 |
| mordred | and as soon as that's a thing with an interface that other gerrit users can have too - I believe we'll all do a parade and dance in the streets with excitement - cause it'll mean a --wip flag to git-review should be totes happymaking | 21:49 |
| mordred | (this is one of those VERY long dependency chain hells) | 21:49 |
| slaweq | mordred: one more thing about designate zones | 22:02 |
| slaweq | itlooks that it's working only for v2 API | 22:02 |
| slaweq | so I will do it with v2 | 22:02 |
| mordred | wow. hahahaha | 22:03 |
| slaweq | I hope tomorrow I will push this patch to review | 22:03 |
| mordred | slaweq: so we have half a v1 and half a v2 support currently? | 22:03 |
| slaweq | zones are not available for v1 api (there are domains in v1 only) | 22:03 |
| slaweq | in designateclient it's in v2 directory also: https://github.com/openstack/python-designateclient/tree/master/designateclient/v2 | 22:04 |
| slaweq | but quick look on designate client and I think that recordsets are using v2 also | 22:05 |
| mordred | cool - well, supporting v2 is more important than v1 anyway - so if we're doing v2 \o/ | 22:05 |
| slaweq | so we have only v2 used because it's only things from designate in shade currently (or I missed something) | 22:05 |
| slaweq | so should be fine with v2 only | 22:05 |
| slaweq | I will push patch and You will see it | 22:05 |
| mordred | ya. totally agree. if someone shows up who is dying without v1 support, we can talk about it then | 22:05 |
| mordred | excellent | 22:05 |
| mordred | thanks! | 22:06 |
| slaweq | but tomorrow - I need to go sleep now :) It's after midnight for me now | 22:06 |
| slaweq | bye | 22:06 |
| slaweq | and thx for help | 22:06 |
| *** slaweq has quit IRC | 22:08 | |
| *** ioggstream has quit IRC | 22:11 | |
| mordred | thank you! | 22:11 |
| *** gouthamr has quit IRC | 22:32 | |
| *** squid has joined #openstack-shade | 22:32 | |
| *** squid is now known as calebb | 22:32 | |
| *** gouthamr has joined #openstack-shade | 23:04 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack-infra/shade master: Updated from global requirements https://review.openstack.org/467428 | 23:11 |
| openstackgerrit | Morgan Fainberg proposed openstack-infra/shade master: First Pass: De-client-ify list_users https://review.openstack.org/467784 | 23:57 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!