| htaccess | thanks for response re my bug, i'll test new release when its available, for the moment i just pinned to version 1.20 in our installer | 00:21 |
|---|---|---|
| *** slaweq has joined #openstack-shade | 00:35 | |
| *** slaweq has quit IRC | 00:39 | |
| *** yolanda has joined #openstack-shade | 00:46 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Cleanup volumes in functional tests in parallel https://review.openstack.org/473969 | 01:10 |
| *** yolanda has quit IRC | 01:16 | |
| mordred | htaccess: awesome. and sorry for the bug | 01:29 |
| *** slaweq has joined #openstack-shade | 01:35 | |
| *** slaweq has quit IRC | 01:40 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Cleanup volumes in functional tests in parallel https://review.openstack.org/473969 | 01:43 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Convert delete server mocks to requests_mock https://review.openstack.org/473839 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Properly expand server dicts after rebuild and update https://review.openstack.org/473972 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Convert get_server_by_id https://review.openstack.org/473838 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Add boot from volume unit tests https://review.openstack.org/473616 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Migrate non-list server interactions to REST https://review.openstack.org/473971 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Don't fetch extra_specs in functional tests https://review.openstack.org/473583 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Update tests for server calls that aren't list https://review.openstack.org/473970 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Convert create_server mocks to request_mock https://review.openstack.org/473554 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: RESTify create_server https://review.openstack.org/473555 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Convert delete server calls to REST https://review.openstack.org/473840 | 02:15 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Break early from volume cleanup loop https://review.openstack.org/474027 | 02:35 |
| *** slaweq has joined #openstack-shade | 03:37 | |
| *** slaweq has quit IRC | 03:41 | |
| openstackgerrit | Merged openstack-infra/shade master: Cleanup volumes in functional tests in parallel https://review.openstack.org/473969 | 03:51 |
| *** gouthamr_ has joined #openstack-shade | 04:07 | |
| *** gkadam has joined #openstack-shade | 04:08 | |
| *** gouthamr has quit IRC | 04:09 | |
| *** slaweq has joined #openstack-shade | 04:38 | |
| *** slaweq has quit IRC | 04:42 | |
| *** slaweq has joined #openstack-shade | 05:12 | |
| *** slaweq has quit IRC | 05:17 | |
| *** gouthamr_ has quit IRC | 05:29 | |
| openstackgerrit | Rajat Sharma proposed openstack-infra/shade master: Replace six.iteritems() with .items() https://review.openstack.org/474071 | 06:18 |
| *** slaweq has joined #openstack-shade | 06:19 | |
| *** slaweq has quit IRC | 06:23 | |
| openstackgerrit | Deepak Mourya proposed openstack-infra/shade master: Remove py34 and pypy in tox https://review.openstack.org/474074 | 06:26 |
| *** yfried has joined #openstack-shade | 06:27 | |
| *** htaccess has quit IRC | 06:58 | |
| *** htaccess has joined #openstack-shade | 07:10 | |
| *** slaweq has joined #openstack-shade | 07:19 | |
| *** slaweq has quit IRC | 07:26 | |
| *** yfried has quit IRC | 07:38 | |
| *** yolanda_ has joined #openstack-shade | 08:09 | |
| *** gkadam is now known as gkadam-afk | 09:20 | |
| *** lastmikoi has joined #openstack-shade | 09:21 | |
| *** cdent has joined #openstack-shade | 09:22 | |
| *** slaweq has joined #openstack-shade | 09:23 | |
| lastmikoi | Hello everyone. I've asked a question in #openstack-infra and was redirected here since it was specifically about shade. Here is the question: I am unable to use shade 1.21.0 (through ansible) with my company's openstack-based provider because their neutron endpoint does not have a trailing slash. This causes an issue with urllib.parse.urljoin as it transforms 'https://domain.tld/neutron' into 'https://domain.tld/v2.0' (https:// | 09:24 |
| lastmikoi | github.com/openstack-infra/shade/blob/1.21.0/shade/openstackcloud.py#L513). | 09:24 |
| lastmikoi | https://wiki.openstack.org/wiki/VersionDiscovery states that href elements are read to be folder elements, but I don't know if this document applies to endpoint definition. Is this an actual bug or is my provider's keystone 'misconfigured' ? | 09:24 |
| *** slaweq has quit IRC | 09:28 | |
| *** yolanda_ has quit IRC | 09:55 | |
| *** slaweq has joined #openstack-shade | 10:24 | |
| *** slaweq has quit IRC | 10:28 | |
| *** ioggstream has joined #openstack-shade | 11:06 | |
| *** ioggstream has quit IRC | 11:20 | |
| mordred | lastmikoi: hrm. well, even if your provider's keystone is misconfigured - it's still a bug | 11:23 |
| mordred | I'll see if I can't get a patch up for you | 11:23 |
| *** slaweq has joined #openstack-shade | 11:24 | |
| lastmikoi | mordred: sounds awesome, thanks a lot ! Shall I file a bugreport in the storyboard then ? | 11:25 |
| *** slaweq has quit IRC | 11:29 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Fix urljoin for neutron endpoint https://review.openstack.org/474162 | 11:30 |
| mordred | lastmikoi: nah. that should fix you ^^ | 11:30 |
| mordred | rods: oh! check out the bug in pagination that my concurrent-delete patch has uncovered | 11:33 |
| mordred | rods: http://logs.openstack.org/16/473616/2/gate/gate-shade-functional-legacy/67233ea/console.html#_2017-06-14_03_27_00_284234 | 11:33 |
| lastmikoi | mordred: indeed it works as expected when applying the patch locally, looking forward to seeing this bugfix released then. thanks again! :) | 11:34 |
| mordred | lastmikoi: sure thing! I need to make a release today for another bugfix as well - so it should be soon (assuming I can get this other gate bug sorted) | 11:36 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Ignore 404 errors while fetching paginated volumes https://review.openstack.org/474165 | 11:41 |
| mordred | rods: ^^ I'm _sure_ that's not the 100% correct answer | 11:42 |
| *** gkadam-afk is now known as gkadam | 11:45 | |
| *** cdent has quit IRC | 11:53 | |
| mordred | samueldmq: on projectupdate - I think we can treat the enabled=True behavior as a bug | 11:55 |
| lastmikoi | is shade aimed to be backwards compatible with older (in this case pre-mitaka) openstack deployments ? | 12:08 |
| mordred | lastmikoi: yes it is | 12:09 |
| mordred | lastmikoi: our expectation is to support all openstack clouds an end-user might be using | 12:09 |
| lastmikoi | oh nevermind, the issue I'm having has already been fixed in the master branch (https://github.com/openstack-infra/shade/commit/755890ba413a67806ceb40a9d413c95cdd233f95) | 12:13 |
| mordred | lastmikoi: yah - I'm going to release that today | 12:14 |
| mordred | so as soon as the urljoin patch above lands | 12:16 |
| lastmikoi | sounds good, I'll go forward and use a development snapshot of shade until then | 12:17 |
| samueldmq | mordred: exactly, I plan to report and fix in a followup | 12:21 |
| mordred | samueldmq: awesome | 12:21 |
| *** slaweq has joined #openstack-shade | 12:25 | |
| *** slaweq has quit IRC | 12:30 | |
| openstackgerrit | Merged openstack-infra/shade master: Remove py34 and pypy in tox https://review.openstack.org/474074 | 12:34 |
| rods | mordred there seems to be something wrong with pagination in this output http://logs.openstack.org/16/473616/2/gate/gate-shade-functional-legacy/67233ea/console.html#_2017-06-14_03_27_00_284234, mind if I take a look? | 12:36 |
| rods | s/take/give/g | 12:37 |
| mordred | rods: please do! | 12:37 |
| mordred | rods: I pushed up a workaroud patch: https://review.openstack.org/474165 | 12:37 |
| mordred | rods: the issue at hand (which that test will now highlight) is that there were 8 volumes. each of them had a delete call sent to the server | 12:38 |
| mordred | then, we're now doing a "while True: list_volumes()" and looking for the volumes we deleted in the list output and waiting until list doesn't have any of them | 12:39 |
| mordred | rods: so it's _completely_ reasonable for one of the volumes to go away which could make the pagination call not work anymore | 12:40 |
| mordred | rods: I'm wondering if, maybe we should put a try/except further up in list_volumes so that if any of the calls 404, we try the whole thing again - and have a limited number of retries | 12:40 |
| rods | mordred ya, I was about to suggest that :) | 12:41 |
| mordred | since if there are 8 volumes, the pagination length is 6, if volume 7 goes away between the initial call and the next call, volume 8 will be missing from the final list | 12:41 |
| mordred | rods: ++ | 12:41 |
| mordred | rods: then I agree with you! | 12:41 |
| *** gouthamr has joined #openstack-shade | 12:44 | |
| openstackgerrit | Merged openstack-infra/shade master: Replace six.iteritems() with .items() https://review.openstack.org/474071 | 12:47 |
| openstackgerrit | Merged openstack-infra/shade master: Add boot from volume unit tests https://review.openstack.org/473616 | 13:14 |
| *** gkadam has quit IRC | 13:17 | |
| *** openstackgerrit has quit IRC | 13:18 | |
| *** noshankus has joined #openstack-shade | 13:24 | |
| *** slaweq has joined #openstack-shade | 13:26 | |
| *** openstackgerrit has joined #openstack-shade | 13:28 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Do pagination test in its own project https://review.openstack.org/474200 | 13:28 |
| *** slaweq has quit IRC | 13:31 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Fix urljoin for neutron endpoint https://review.openstack.org/474162 | 13:38 |
| mordred | frickler, lastmikoi: ^^ update to that patch | 13:38 |
| mordred | lastmikoi: frickler made a good review point about the previous patch - do you have a moment to verify that the updated version works for you too? | 13:38 |
| lastmikoi | my provider doesn't present the versioned endpoint but I'll try it out | 13:40 |
| lastmikoi | mordred: no regression to report, the updated version works :) | 13:49 |
| mordred | lastmikoi: yay! thanks | 14:21 |
| *** slaweq has joined #openstack-shade | 14:27 | |
| *** slaweq has quit IRC | 14:40 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Do pagination test in its own project https://review.openstack.org/474200 | 14:45 |
| *** cdent has joined #openstack-shade | 15:21 | |
| *** yolanda_ has joined #openstack-shade | 15:23 | |
| *** ioggstream has joined #openstack-shade | 15:40 | |
| rods | mordred want me to work on that try/except for list_volumes we talked about before? | 16:34 |
| *** yolanda_ has quit IRC | 16:40 | |
| *** yolanda_ has joined #openstack-shade | 16:40 | |
| mordred | rods: yah - if you don't mind, I think it'll be better | 16:42 |
| mordred | rods: I also got fancy with the cleanup: https://review.openstack.org/#/c/474200 :) | 16:43 |
| *** yolanda_ has quit IRC | 16:43 | |
| rods | mordred yah, saw that. Pretty cool :) | 16:43 |
| *** yolanda_ has joined #openstack-shade | 16:43 | |
| *** cdent has quit IRC | 17:17 | |
| openstackgerrit | Merged openstack-infra/shade master: Don't remove top-container element for sec group REST API calls https://review.openstack.org/473770 | 17:58 |
| *** yolanda__ has joined #openstack-shade | 18:09 | |
| *** yolanda_ has quit IRC | 18:10 | |
| *** ioggstream has quit IRC | 18:20 | |
| samueldmq | mordred: re https://review.openstack.org/#/c/473581/3/shade/openstackcloud.py | 18:23 |
| samueldmq | mordred: domain_id will go away if it is None (in the _normalize._split_filters logic) | 18:24 |
| samueldmq | so the test needs to be modified anyway | 18:25 |
| mordred | samueldmq: I wonder why it wasn't being passed by keystoneclient before | 18:27 |
| samueldmq | mordred: maybe that's related to the fact you have different domains for different things there in ksc? | 18:28 |
| samueldmq | project_domain, user_domain, ... | 18:29 |
| samueldmq | nvm, that;s list_proejct | 18:29 |
| samueldmq | not the role assignments thing | 18:29 |
| samueldmq | mordred: did the unit tests use keystoneclient? | 18:30 |
| mordred | samueldmq: yah - we were passing down domain to keystoneclient.v3.projects.list before | 18:30 |
| mordred | samueldmq: yah | 18:30 |
| mordred | and it passed domain_id=base.getid(domain), inside | 18:30 |
| mordred | so I'd expect the REST keystoneclient generated to be the same since we haven't changed that logic | 18:31 |
| samueldmq | yes, but looks to be a bug there then? | 18:31 |
| samueldmq | since the url should contain ?domain_id since the beginning | 18:31 |
| samueldmq | that's the whole point of providing a domain_id anyways | 18:31 |
| mordred | yah - I think it was a bug and the current thing *looks* right | 18:33 |
| mordred | but I can't point my finger to what the bug actually was and why ksc wasn't passing domain_id in this case :) | 18:33 |
| samueldmq | mordred: I looked into the ksc code and it looks fine, should be passing domain_id down :( | 18:35 |
| samueldmq | mordred: maybe we are in the other case in our tests | 18:35 |
| samueldmq | and we should write a test that does not provide a domain_id | 18:36 |
| samueldmq | nah, that's just inverting the issue | 18:36 |
| samueldmq | the thing is that moving from ksc to rest is inconsistent | 18:36 |
| mordred | samueldmq: yah - I went ahead and landed that one - let's keep our eyes on this though | 18:36 |
| samueldmq | mordred: yes, our code seems to be fine | 18:37 |
| samueldmq | btw, I wanted to take another look at _split_filters | 18:37 |
| mordred | ++ | 18:37 |
| samueldmq | overall seems like we combine the filters, them split them back in the same thing we had before | 18:37 |
| samueldmq | tldr: seems unnecessary processing ? | 18:37 |
| samueldmq | I can try a patch | 18:39 |
| samueldmq | that is always easier to explain | 18:39 |
| mordred | sounds good to me | 18:47 |
| samueldmq | mordred: in the unit tests, do we "run" v2.0 or v3? | 18:55 |
| openstackgerrit | Merged openstack-infra/shade master: Fix urljoin for neutron endpoint https://review.openstack.org/474162 | 18:58 |
| samueldmq | nvm, domain_id may come inside filters, not always in the domain_id param :( | 19:10 |
| openstackgerrit | Merged openstack-infra/shade master: De-client-ify Project List https://review.openstack.org/473581 | 19:16 |
| openstackgerrit | Merged openstack-infra/shade master: De-client-ify Project Delete https://review.openstack.org/473628 | 19:17 |
| mordred | samueldmq: it depends per-unit test | 19:22 |
| mordred | samueldmq: you can add self.use_keystone_v2() to the top of a unittest and it'll set things up for that test to use v2 | 19:23 |
| openstackgerrit | Merged openstack-infra/shade master: De-client-ify Project Create https://review.openstack.org/473609 | 19:23 |
| openstackgerrit | Merged openstack-infra/shade master: De-client-ify Project Update https://review.openstack.org/473629 | 19:23 |
| samueldmq | mordred: kk I am debugging a bit | 19:24 |
| samueldmq | seeing how this https://github.com/openstack-infra/shade/blob/master/shade/operatorcloud.py#L1663-L1664 relates to the change I made | 19:25 |
| samueldmq | that data is passed down to list_projects, and that data contains 'domain' not 'domain_id' as my patch expects, I am bit confused right now | 19:25 |
| samueldmq | nah that's fine, I do filters['domain_id | 19:26 |
| samueldmq | filters['domain_id'] = data['domain'] | 19:26 |
| * samueldmq needs vacation :) | 19:26 | |
| samueldmq | mordred: aha I found it!!! | 19:31 |
| samueldmq | our code was doing this https://github.com/openstack-infra/shade/blob/master/shade/operatorcloud.py#L1663-L1664 | 19:31 |
| samueldmq | even before my patches, I didnt touch that | 19:31 |
| samueldmq | that sets 'domain_id' into filtesr | 19:32 |
| mordred | samueldmq: AHHHH | 19:32 |
| samueldmq | in list_projects the _split_filters thing was expecting 'domain' | 19:32 |
| mordred | nod. so I thnk the change to domain_id caused this to also attach | 19:32 |
| samueldmq | so our old code was not passing down domain_id to ksc, but filtering it in the shade filter logic afterwards | 19:32 |
| mordred | yah | 19:32 |
| samueldmq | phew, I was not crazy | 19:32 |
| mordred | PHEW. I like understanding | 19:33 |
| mordred | yes | 19:33 |
| samueldmq | ++ we dont need to keep that in mind anymore, it wont break :) | 19:33 |
| samueldmq | brb | 19:33 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Do pagination test in its own project https://review.openstack.org/474200 | 19:45 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Skip pagination test for now https://review.openstack.org/474329 | 19:45 |
| mordred | rods: ^^ I'm tossing in a skipTest on the pagination functional test for now ... we're seeing too many timeouts from cinder with it currently | 19:45 |
| mordred | rods: I think the pagination code itself works fine atm - so we can iterate on the test without stress | 19:46 |
| rods | mordred yah, makes sense. I should be able to work on it in a few | 19:47 |
| mordred | rods: I have a hunch this is going to be an annoying one - the most recent issues were cinder failing to create volumes in other tests | 19:48 |
| rods | mordred do you have any log url handy? | 19:50 |
| mordred | yah - one sec | 19:51 |
| mordred | rods: http://logs.openstack.org/00/474200/2/check/gate-shade-functional/33b27bd/console.html | 19:52 |
| mordred | is an example | 19:52 |
| rods | mordred ok, thx | 19:52 |
| mordred | it hits status='error' here: http://logs.openstack.org/00/474200/2/check/gate-shade-functional/33b27bd/console.html#_2017-06-14_18_43_27_500878 | 19:52 |
| *** ioggstream has joined #openstack-shade | 20:02 | |
| *** ioggstream has quit IRC | 20:22 | |
| mordred | rods: wow: http://logs.openstack.org/00/474200/3/check/gate-shade-functional-py35/278808a/ | 20:36 |
| mordred | rods: it's NOT pagination. just sometimes we're not succeeding in creating volumes | 20:36 |
| mordred | rods: maybe the volumes in their own project is also too much | 20:37 |
| rods | ya, wondering if 8 volumes are too much and cinder is having hard time | 20:38 |
| *** gouthamr has quit IRC | 20:45 | |
| mordred | rods: oh. hahaha. I'm stupid | 20:45 |
| mordred | rods: that chanage un-skips the pagination test. I was starting to worry skipping the test wasn't heling | 20:45 |
| mordred | but I clearly need new eyeballs | 20:46 |
| rods | :) | 20:50 |
| rods | mordred ah, I can reproduce the timeout for test_list_volumes_pagination locally | 20:55 |
| rods | digging | 20:55 |
| mordred | rods: excellent! | 20:55 |
| *** gouthamr has joined #openstack-shade | 21:12 | |
| openstackgerrit | Merged openstack-infra/shade master: Skip pagination test for now https://review.openstack.org/474329 | 21:18 |
| rods | mordred interesting if I run the test like this `python -m testtools.run discover --load-list ../my-list` it always pass running it like this `tox -e functional -- 'shade.tests.functional.test_volume.TestVolume.test_list_volumes_pagination'` fails | 21:45 |
| mordred | rods: that's FANTASTIC | 21:52 |
| *** cdent has joined #openstack-shade | 21:57 | |
| *** yolanda__ has quit IRC | 22:07 | |
| *** yolanda__ has joined #openstack-shade | 22:15 | |
| *** yolanda__ is now known as yolanda | 22:15 | |
| *** yolanda has quit IRC | 22:23 | |
| rods | mordred I hay have found something about this log you pasted http://logs.openstack.org/00/474200/3/check/gate-shade-functional-py35/278808a/logs/testr_results.html.gz | 22:39 |
| rods | http://logs.openstack.org/00/474200/3/check/gate-shade-functional-py35/278808a/logs/screen-c-sch.txt.gz#_Jun_14_20_24_42_053439 | 22:39 |
| *** cdent has quit IRC | 22:44 | |
| openstackgerrit | Merged openstack-infra/shade master: Convert create_server mocks to request_mock https://review.openstack.org/473554 | 22:45 |
| openstackgerrit | Merged openstack-infra/shade master: Don't fetch extra_specs in functional tests https://review.openstack.org/473583 | 22:45 |
| mordred | rods: aha! | 22:48 |
| mordred | rods: so we really do need to wait for those volumes to be destroyed - otherwise we run out of space | 22:48 |
| rods | yeah | 22:49 |
| rods | mordred but I'm seeing a weird issue with with test_list_volumes_pagination http://paste.openstack.org/show/612601/ | 22:52 |
| rods | test is timing out after 60 secs | 22:52 |
| rods | even tough timeout is supposed to be 180 https://github.com/openstack-infra/shade/blob/master/shade/tests/functional/test_volume.py#L101 | 22:53 |
| mordred | rods: ah - well, we have a 60-second timeout defined in the base test: https://github.com/openstack-infra/shade/blob/master/shade/tests/base.py#L40 | 22:54 |
| mordred | rods: maybe we just need to add a TIMEOUT_SCALING_FACTOR to that test | 22:54 |
| mordred | rods: we've got one in test_volume_backup.py already | 22:55 |
| openstackgerrit | Merged openstack-infra/shade master: RESTify create_server https://review.openstack.org/473555 | 22:56 |
| openstackgerrit | Merged openstack-infra/shade master: Convert get_server_by_id https://review.openstack.org/473838 | 22:59 |
| openstackgerrit | Merged openstack-infra/shade master: Convert delete server mocks to requests_mock https://review.openstack.org/473839 | 22:59 |
| openstackgerrit | Merged openstack-infra/shade master: Convert delete server calls to REST https://review.openstack.org/473840 | 23:00 |
| openstackgerrit | Merged openstack-infra/shade master: Update tests for server calls that aren't list https://review.openstack.org/473970 | 23:00 |
| rods | mordred yup, TIMEOUT_SCALING_FACTOR does it | 23:01 |
| mordred | rods: WOOT | 23:01 |
| mordred | rods: maybe we don't even need that complex cleanup bit | 23:02 |
| rods | mordred I think this logic should be enough https://github.com/openstack-infra/shade/blob/master/shade/tests/functional/test_volume.py#L98-L108 | 23:05 |
| rods | I guess the problem was that since that test was timing out, another test was started but volume were still there waiting to be deleted | 23:06 |
| rods | *volues | 23:06 |
| openstackgerrit | Monty Taylor proposed openstack-infra/shade master: Add some release notes we forgot to add https://review.openstack.org/474385 | 23:08 |
| mordred | rods: sweet | 23:09 |
| openstackgerrit | Rosario Di Somma proposed openstack-infra/shade master: Increase timeout for volume tests https://review.openstack.org/474388 | 23:24 |
| rods | mordred ^^ that should do it, let's see what happens | 23:24 |
| mordred | \o/ | 23:26 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!