*** tosky has joined #refstack | 07:26 | |
*** tommylikehu is now known as tommylikehu2 | 08:02 | |
*** tommylikehu2 is now known as tommylikehu | 08:03 | |
*** tommylikehu is now known as tommylikehu_afk | 08:04 | |
*** tommylikehu_afk is now known as tommylikehu | 08:09 | |
openstackgerrit | Arx Cruz proposed openstack/python-tempestconf master: Removing dependence from SERVICE_VERSIONS https://review.openstack.org/587816 | 09:19 |
---|---|---|
*** rmart04 has joined #refstack | 12:14 | |
openstackgerrit | Chandan Kumar proposed openstack/python-tempestconf master: add jobs for python-tempestconf only https://review.openstack.org/581084 | 12:17 |
openstackgerrit | Chandan Kumar proposed openstack/python-tempestconf master: Document CLI options https://review.openstack.org/580981 | 12:44 |
arxcruz | tosky: chandankumar are you guys seeing something that I'm not here http://logs.openstack.org/16/587816/8/check/python-tempestconf-tempest-packstack-admin/6880e04/job-output.txt.gz ? | 13:42 |
arxcruz | it's constantly failing on the same test | 13:42 |
arxcruz | i check the previous jobs from other patches, the tempest.conf generated is basically the same... | 13:42 |
tosky | arxcruz: the relevant difference is the list of extensions set into volume-feature-enabled, which may show a problem in the patch, but not sure how it could be related to that test | 13:48 |
arxcruz | tosky: yeah, i 'm trying to figure out that also, because the logic remains the same... :/ | 13:48 |
arxcruz | and yeah, shouldn't interfere on this test in particular | 13:48 |
tosky | but let's see if fixing it would fix the other issue | 13:49 |
fungi | hogepodge: did you have any thoughts on what to do about the refstack team since we talked about it a month ago? as far as continuing to be ptl for the stein cycle to oversee wind-down of the team, folding the refstack service and client into the interop wg, what happens to python-tempestconf, et cetera? or would you like help getting those discussions rolling sooner? | 14:06 |
fungi | your friendly neighborhood tc liaison is here to assist however you need ;) | 14:07 |
*** trevormc has joined #refstack | 14:13 | |
trevormc | hey not urgent but i was just curious if the python-tempestconf team has got a chance to review the story https://storyboard.openstack.org/#!/project/912 and if they decided on what to do next regarding the certificate error. | 14:15 |
trevormc | sorry wrong link https://storyboard.openstack.org/#!/story/2003024 | 14:18 |
openstackgerrit | Arx Cruz proposed openstack/python-tempestconf master: Removing dependence from SERVICE_VERSIONS https://review.openstack.org/587816 | 14:23 |
arxcruz | well, found the problem with volume api_extensions | 14:26 |
tosky | trevormc: we have one person less for vacation, and few other deadlines | 14:26 |
arxcruz | trevormc: sorry, martin was taking a look at this, however, he's on vacation this week | 14:26 |
arxcruz | trevormc: but it's in our radar :) | 14:26 |
trevormc | ok no problem, thanks! I'll check back in a week or two :) | 14:27 |
arxcruz | trevormc: as a workaround, you can disable the ssl certification check on urllib | 14:27 |
arxcruz | trevormc: i did something similar on tripleo, because for some reason, tempest on docker complain about ssl certificate being false: https://review.openstack.org/#/c/580384/ | 14:28 |
trevormc | ok, i've been wanting to try this tool out and report back to my team on the results, so ill apply a work around in the meantime. | 14:28 |
arxcruz | trevormc: there are other _CA_BUNDLE variables if CURL_CA_BUNDLE doesn't work | 14:29 |
arxcruz | let me check | 14:29 |
arxcruz | REQUESTS_CA_BUNDLE | 14:30 |
arxcruz | and the other if memory doesn't fails is URLLIB_CA_BUNDLE | 14:30 |
arxcruz | yeah, i think it's only curl and requests ca_bundle | 14:32 |
tosky | I may have a quick patch in few minutes | 14:40 |
openstackgerrit | Luigi Toscano proposed openstack/python-tempestconf master: horizon: don't error out for certificate issues https://review.openstack.org/588593 | 15:11 |
tosky | trevormc: can you please test ^^ ? | 15:11 |
trevormc | tosky: yeah im on it, thanks. | 15:12 |
arxcruz | tosky: so, certificate error = has_horizon set to false? | 15:13 |
tosky | arxcruz: for now, yes | 15:14 |
tosky | arxcruz: it only affects the few tests from the Horizon tempest plugin, if I'm not mistaken | 15:14 |
arxcruz | tosky: then i'll ask you to add a TODO | 15:14 |
tosky | it should not block the process | 15:14 |
tosky | arxcruz: I can add another task on that story | 15:14 |
arxcruz | ok | 15:15 |
tosky | arxcruz: but looking the story, it seems a real issue, or there is no way to get the name of the dashboard | 15:15 |
*** trevormc_ has joined #refstack | 15:16 | |
tosky | arxcruz: the current assumption is "hostname of the dashboard == hostname from the endpoints", which is not the case on the story | 15:16 |
tosky | and I have no idea how you would discover that | 15:16 |
tosky | you probably need to override that information | 15:16 |
tosky | added a task, though | 15:18 |
*** trevormc has quit IRC | 15:18 | |
arxcruz | tosky: isn't the cert_file option ? | 15:20 |
tosky | arxcruz: where? | 15:21 |
arxcruz | tosky: in tempest there's a cert_file option to point to an specific certification file | 15:21 |
arxcruz | i tried without success to fix the ssl issue with that though | 15:22 |
tosky | arxcruz: but the problem is the mismatch between the certificate name and the host name that the code assume to be correct | 15:22 |
tosky | the code tries to use the hostname specified in the endpoints, if I'm not mistaken | 15:22 |
tosky | the name of the horizon host is totally unrelated to that | 15:22 |
arxcruz | tosky: that might not be a problem in all in one | 15:23 |
tosky | at least if the examples given in the story are correct | 15:23 |
tosky | arxcruz: so the exception won't happen | 15:23 |
arxcruz | it seems the certificate is created with the name of the compute machine not the name of dashboard machine | 15:23 |
arxcruz | won't happen, but horizon is enabled, and it's being set to false | 15:23 |
tosky | where is the certificate created with the name of the compute node? Not in the story | 15:24 |
tosky | it's a different case | 15:24 |
arxcruz | yes | 15:31 |
tosky | without the patch, instead of setting horizon as false, it would error out and block the processing, so at least it is better | 15:37 |
*** trevormc_ has quit IRC | 15:38 | |
tosky | arxcruz: ok, the "Document CLI options" patch failed exactly on python-tempestconf-tempest-packstack-admin | 15:58 |
tosky | same error, SSH timeout | 15:59 |
tosky | now I'm puzzled | 15:59 |
tosky | and it was passing until yesterday (of course) | 15:59 |
tosky | well, two days ago | 16:00 |
-openstackstatus- NOTICE: The infra team is renaming projects in Gerrit. There will be a short ~10 minute Gerrit downtime in a few minutes as a result. | 16:04 | |
*** rmart04 has left #refstack | 16:26 | |
*** openstackgerrit has quit IRC | 16:49 | |
*** openstackgerrit has joined #refstack | 17:03 | |
openstackgerrit | Luigi Toscano proposed openstack/python-tempestconf master: Use the common cover job instead of a custom one https://review.openstack.org/588622 | 17:03 |
*** trevormc has joined #refstack | 17:05 | |
trevormc | sorry looks like I got disconnected, I found another bug in python-tempestconf. This one should be a simple fix https://storyboard.openstack.org/#!/story/2003307 | 17:09 |
tosky | thanks a lot again for the detailed analysis | 17:10 |
tosky | the fact that keystone v3 was enabled by default for a long while didn't help :) | 17:14 |
tosky | do you have a keystone v2-only cloud? | 17:14 |
tosky | may I ask also which version of OpenStack? tempestconf is tested from Ocata | 17:15 |
openstackgerrit | Luigi Toscano proposed openstack/python-tempestconf master: Fix: make sure that keystone_v3_support is a boolean https://review.openstack.org/588624 | 17:21 |
-openstackstatus- NOTICE: Project renames and review.openstack.org downtime are complete without any major issue. | 17:30 | |
trevormc | tosky, we test on the kilo version of openstack :( sorry for the delay. We have a v2 and v3 enabled but we only expose the v2 endpoint. | 17:33 |
trevormc | the setup is kind of weird, I will ask specifically how identitiy is setup. I can set OS_AUTH_URL to our v3 endpoint and use the openstack client to list the roles and the debug logs show the v3 endpoint being used but when I run endpoint list it is not listed. :/ | 17:37 |
tosky | oh | 17:54 |
tosky | you may have some issues because, as I mentioned, we only planned the compatibility with Ocata and newer versions | 17:55 |
tosky | so if it works for you, it will show that the some API are really stable :) | 17:55 |
*** openstackgerrit has quit IRC | 18:19 | |
*** trevormc has quit IRC | 20:45 | |
*** tosky has quit IRC | 23:04 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!