Earlier  
Later
New Search
Posted Nick Remark
#openstack-sdks - 2018-02-28
16:39:28 openstackgerrit Matthew Treinish proposed openstack/python-openstackclient master: Cleanup error messages on failure https://review.openstack.org/548656
16:42:32 openstackgerrit Matthew Treinish proposed openstack/python-openstackclient master: Cleanup tox.ini for functional tests https://review.openstack.org/548657
16:45:12 mtreinish dtroyer: ^^^ still reading things, but that first patch might help
16:45:23 mtreinish I'm having a hard time reading the giant blob of text
16:45:57 dtroyer I think the entry point failures is the key, that they are there at all, not which ones specifically
16:46:39 dtroyer I've narrowed it down to working fine locally without https://review.openstack.org/#/c/538478/ and failing with it, and that commit is the only difference in the venv
16:46:59 dtroyer the dependencies that changed in there do not change between test runs
16:47:08 dtroyer (I am removing the venv each tme so it is always refreshed)
16:48:32 dtroyer that cleanup is certainly worth doing no matter what, thanks
16:49:31 mtreinish yeah, parsing the giant blob in the logs I see the entrypoint failure now
16:49:39 mtreinish and I agree that's likely the issue
16:50:41 mtreinish I just looked at the pip freeze logs from the run and it seems like there are some weird issues there: http://logs.openstack.org/78/538478/5/check/osc-functional-devstack-tips/356323d/tox/functional-5.log
17:11:51 openstackgerrit Merged openstack/osc-lib master: Add project lookup utils https://review.openstack.org/522861
17:27:24 openstackgerrit OpenStack Proposal Bot proposed openstack/cliff master: Updated from global requirements https://review.openstack.org/548672
17:39:11 tobixen dtroyer: thanks, clean.sh did the trick. Just verified that I managed getting some tests failing with a fresh devstack, and getting all tests passing by disabling tempest and doing a clean.sh
19:57:52 frickler mordred: this one looks fun, too ;) https://launchpad.net/bugs/1751822
19:57:54 openstack Launchpad bug 1751822 in cliff "EXCEPTION of type 'AttributeError' when using 'help' from openstack-cli" [Undecided,In progress] - Assigned to Corey Bryant (corey.bryant)
20:12:44 openstackgerrit Corey Bryant proposed openstack/cliff master: Align parsed() call with cmd2 versions >= 0.7.3 https://review.openstack.org/548625
20:16:08 openstackgerrit Corey Bryant proposed openstack/cliff master: Align parsed() call with cmd2 versions >= 0.7.3 https://review.openstack.org/548625
22:16:27 openstackgerrit Tobias Brox proposed openstack-infra/shade master: Simple fix making the TestNetwork functional test cleanup more robust and a bit faster, by passing network id instead of network name. Task: 6594 Story: 2001617 https://review.openstack.org/548714
22:18:16 openstackgerrit Tobias Brox proposed openstack-infra/shade master: Functional test robustness and performance fix. In TestNetwork._cleanup_networks, pass network id instead of network name. Task: 6594 Story: 2001617 https://review.openstack.org/548714
22:35:19 openstackgerrit Tobias Brox proposed openstack-infra/shade master: Add subnetpool_name_or_id parameter to create_subnet https://review.openstack.org/548722
#openstack-sdks - 2018-03-01
02:12:30 mordred mtreinish: yah - those warnings are a known issue with python installed from git repos on zuul nodes - pip parses git origin urls to find the string to use to print the version of the package - but there is no origin url on the repos there - it just produces ugly output though
02:22:57 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
03:31:34 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
03:40:12 openstackgerrit Monty Taylor proposed openstack-infra/shade master: Functional test robustness and performance fix https://review.openstack.org/548714
03:40:26 openstackgerrit Monty Taylor proposed openstack-infra/shade master: Add subnetpool_name_or_id parameter to create_subnet https://review.openstack.org/548722
03:43:33 mnaser mordred: i see i'm not the only one who's unable to sleep :p
03:43:39 mordred mnaser: :)
05:07:27 openstackgerrit Merged openstack-infra/shade master: Functional test robustness and performance fix https://review.openstack.org/548714
05:20:46 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
05:23:35 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
05:40:28 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
07:25:52 openstackgerrit OpenStack Proposal Bot proposed openstack/python-openstackclient master: Imported Translations from Zanata https://review.openstack.org/548850
09:45:38 openstackgerrit Tobias Brox proposed openstack-infra/shade master: Fix proper domain support in user role grants https://review.openstack.org/548887
10:49:26 openstackgerrit Dean Troyer proposed openstack/osc-lib master: Revert "Updated from global requirements" https://review.openstack.org/548908
15:49:03 openstackgerrit Dongcan Ye proposed openstack/python-openstacksdk master: DNM: osc functional check https://review.openstack.org/548977
18:05:31 frickler mordred: shouldn't https://review.openstack.org/548702 be on master initially?
20:32:04 slaweq mordred: hi
20:32:37 slaweq mordred: I have a question about https://review.openstack.org/#/c/548737/ - why this revert of revert is proposed?
20:33:03 slaweq I see that this patch wasn't even merged: https://review.openstack.org/#/q/1ef3ce7ac0ab2598a494e85b8575b0114f562296
20:33:50 slaweq mordred: sorry if I missed something but I don't understand now if there is any issue with this preferred FIP or not :)
20:39:14 frickler slaweq: iiuc your original patch introduced a regression. the revert is amended with a test for that regression. the revert of the revert is being edited so that it hopefully will fix both the original issue and the regression
20:39:55 slaweq frickler: but rever wasn't merged so why "revert of revert"?
20:39:57 slaweq :)
20:41:12 frickler slaweq: mordred is preparing two steps at once, so that when both are ready and can be merged in one batch, things wills work fine
20:41:30 slaweq frickler: ok, thx :)
20:41:34 frickler without the first revert, it would not be so easy to show the test working
20:41:47 frickler see https://storyboard.openstack.org/#!/story/2001619 for a description of the regression
20:43:01 slaweq ok, I see now
20:43:23 slaweq so both are necessary and we have to merge it on proper order, right?
21:14:49 frickler slaweq: yes. might instead do two new patches, one adding the tests and one fixing the regression. but that's a matter of taste I'd say
21:15:29 slaweq frickler: yes, now I at least undersand why there are two patches :)
21:15:37 slaweq thx a lot for help with understanding that
22:17:05 openstackgerrit Tobias Brox proposed openstack-infra/shade master: Fix proper domain support in user role grants https://review.openstack.org/548887
#openstack-sdks - 2018-03-02
06:22:27 openstackgerrit Robin Cernin proposed openstack/python-openstackclient master: Adds missing Project information in server list https://review.openstack.org/548733
07:40:12 openstackgerrit Monty Taylor proposed openstack-infra/shade master: Fix private_v4 selection related to floating ip matching https://review.openstack.org/549116
07:46:11 mordred slaweq, frickler: I've found the actual issue ^^ so I think we'll be in a position to just land a fix ratherthan a revert and a revert ...
07:47:18 mordred slaweq, frickler: I'm not sure I had an *actual* plan with the twopatches - as much as ansible inventory was broken for some people and that seemed like a thing that should be fixed quickly - but luckily the real problem isn't too bad
07:47:39 mordred and also yes ... should fix it on master too :)
07:52:11 openstackgerrit Monty Taylor proposed openstack-infra/shade master: Fix private_v4 selection related to floating ip matching https://review.openstack.org/549116
07:52:55 mordred slaweq, frickler: ^^ and that should fix the 2 tests that the fix broke (yay for nova network!)
07:54:01 slaweq mordred: thx, +2 already :)
07:54:25 openstackgerrit Monty Taylor proposed openstack/python-openstacksdk master: Fix private_v4 selection related to floating ip matching https://review.openstack.org/549117
07:54:26 slaweq please remember to cherry-pick it to openstacksdk also as it is there as well
07:54:35 mordred slaweq: done! :)
07:54:47 mordred slaweq: I will be happy when we no longer need to cherry-pick between the two
07:54:59 slaweq me too but for now we have to
07:56:49 mordred mtreinish, dtroyer: I guess the osc-functional-devstack-tips failure is what you were looking at yesterday - any luck finding the issue yet (other than narrowing down to the entrypoint issue) or should I start digging?
08:05:28 mordred mtreinish, dtroyer: also - it occurs to me that we have an assymetrical gate going there and that perhaps we should add osc-functional-devstack-tips to keystoneauth, os-client-config and cliff
08:06:03 mordred although actually I'll probably write a patch today to go ahead and remove the os-client-config depend from osc and just use the sdk version that's already there
09:23:37 dtroyer mordred: I've narroed it down to the last global req review in osc-lib. I can't explain why yet, doing it locally I get the same things installed with and without it, but I haven't had a contiguous block to look at this yet
10:13:26 openstackgerrit melissaml proposed openstack/python-openstackclient master: Update links in README https://review.openstack.org/549159
13:24:20 mordred dtroyer: kk. same here - I put in an auto hold with zuul - cause I can't make it fail locally
13:27:23 mordred dtroyer: oh - you know what - I haven't tried a full from-scratch install using constraints yet
13:28:59 mordred dtroyer: I've got a node held from a failed job - do you want me to put an ssh key for you on it?
13:29:44 dtroyer mordred: yes…
13:30:50 mordred dtroyer: zuul@ubuntu-xenial-ovh-gra1-0002764775:~/src/git.openstack.org/openstack/python-openstackclient$ .tox/functional/bin/openstack help
13:30:58 mordred is definitely unhapy in a way it's not for me loocally :)
13:31:28 dtroyer I've duplicated it locally, or found another way of achieving the same result anyway
13:31:36 mordred dtroyer: did you get this:
13:31:40 mordred ContextualVersionConflict: (openstacksdk 0.11.1.dev64 (/home/zuul/src/git.openstack.org/openstack/python-openstacksdk), Requirement.parse('openstacksdk>=0.11.2'), set(['osc-lib']))
13:32:22 dtroyer I do not see that locally, but I'm just running the functional-tips against a pre-built devstack on a differnet node
13:34:31 mordred nod. so - I got that from running openstack --debug -v help ... so I'm very confused by why soomething is getting 0.11.1.dev64 on master
13:35:55 dtroyer that narrows things down, my local failure is with 0.11.2 in requirements.txt. with 0.9.xx there it works. and of course that shouldn't matter at all since we're installing from source. but I don't see any transitive deps change either
13:37:16 mordred dtroyer: so - I think this points to something being slightly off with tags, pbr, branches and tox-siblings
13:38:11 mordred dtroyer: I'm 99% sure this will go away if we cut a release on the master branch - but I think digging in to why it's broken it worthwhile - since this is the first release we've cut with tox-siblings in play
13:38:38 mordred dtroyer: which is to say - I think it's a gate issue and not an actual bug in the wild
13:40:05 dtroyer so I have mixed feelings there :) then I am misunderstanding something for my local problem
13:43:17 mordred dtroyer: oh - it's possible there is a second problem that is being masked by the tag issue
14:03:48 mordred dtroyer: if I add a tag to the sdk repo it fixes the tox functional job
14:05:40 mordred dtroyer: however - there's another fun issue I'm going ot see if I can come up with a slution for ... which is that the functional-tips tox env in osc might not be producing the same results as what's happening in the gate due to the logic in tox-siblings ... I'm going to see if I can extract that logic in a consumable fashion so that we can have functional-tips be a completely accurate local
14:05:42 mordred reproduction
14:07:42 dtroyer mordred: that is probably why I'm having similar-but-different symptoms
14:08:05 mordred yah. I don't like local reproduction cases being similar but different
14:08:41 dtroyer although I have also narrowed it to sdk 0.11.1+
14:09:08 mordred dtroyer: is yours still an entrypoints loading error? or are you seeing a different error altogether?
14:10:07 dtroyer it is the same entrypoint errors
14:10:08 mordred dtroyer: (and if you did something like 'git tag 0.11.5' in the openstacksdk dir, does your error go away?)
14:10:18 dtroyer I'll try that

Earlier   Later