| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-02-14 | |||
| 15:06:44 | bauzas | mriedem: dunno, but I can ask | |
| 15:06:55 | bauzas | mriedem: IIRC, we have a different resource | |
| 15:06:56 | mtreinish | I just need to remember what the right one to use is | |
| 15:07:26 | bauzas | mriedem: like https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux_OpenStack_Platform | |
| 15:12:35 | mriedem | melwitt: dansmith: ever seen this in a cells v1 job? http://logs.openstack.org/57/433257/2/check/gate-tempest-dsvm-cells-ubuntu-xenial/1732de2/logs/screen-n-cell-child.txt.gz#_2017-02-13_22_55_18_192 | |
| 15:13:27 | dansmith | mriedem: hmm, no | |
| 15:14:03 | mriedem | could be latent | |
| 15:14:38 | mtreinish | jaypipes, sdague: https://github.com/mtreinish/stestr/commit/9869bfe48ed87d9abf3f3397c46a9d7f8dc60031 give that a shot | |
| 15:15:23 | mriedem | dansmith: actually looks like a regression | |
| 15:15:25 | sdague | mtreinish: that's going to mean that everyone's dbm dies the first time they try to run it again on python2 right? | |
| 15:15:32 | mriedem | http://logstash.openstack.org/#dashboard/file/logstash.json?query=message%3A%5C%22Couldn't%20communicate%20with%20cell%20'child'%5C%22&from=7d | |
| 15:15:46 | mriedem | mtreinish: did something chagne in test_stamp_pattern? ^ | |
| 15:15:48 | sdague | mtreinish: probably also just need an exception around the open and just rebuild the db | |
| 15:15:53 | mriedem | i don't think tempest runs against cells v1 | |
| 15:16:12 | mtreinish | sdague: well for most people it won't matter because stestr uses a seperate repository dir | |
| 15:16:15 | mriedem | https://github.com/openstack/tempest/commit/96c36cffbe115226402600c321c2de701b281f7b | |
| 15:16:16 | mriedem | ha | |
| 15:16:21 | mtreinish | but yeah I can add an exception | |
| 15:16:43 | mriedem | mtreinish: i guess i can abandon this 2 year quest now https://review.openstack.org/#/c/218355/ | |
| 15:16:55 | mtreinish | mriedem: wow, how did that slip by without either of noticing | |
| 15:16:55 | mriedem | dansmith: yeah so nova gate is blocked, i'll put up a skip patch | |
| 15:17:18 | mtreinish | sdague: unless you'd like to push a pull request for that :) | |
| 15:17:34 | dansmith | mriedem: it's 100% fail on that cells thing? | |
| 15:17:41 | mriedem | yes | |
| 15:17:43 | mriedem | it uses security groups | |
| 15:18:24 | mtreinish | mriedem: does the test actually work now? | |
| 15:18:26 | dansmith | sorry, I'm too distracted.. are you saying tempest is now running a secgroup test against a cells deploy that can't handle it? | |
| 15:18:52 | mriedem | dansmith: cells v1 yes | |
| 15:18:56 | dansmith | yeah, okay | |
| 15:19:02 | mriedem | anyway, i've got the fix | |
| 15:19:08 | mriedem | mtreinish: i guess? | |
| 15:19:18 | mriedem | mtreinish: https://review.openstack.org/#/c/431800/ | |
| 15:20:01 | mriedem | maybe killing liberty is what did the trick | |
| 15:20:44 | mriedem | maybe i should start a massive ML thread about how tempest just goes and breaks nova's gate because it doesn't test against cells v1 :) | |
| 15:20:53 | mriedem | same with devstack | |
| 15:20:56 | mriedem | crazy QA guys | |
| 15:21:00 | sdague | mriedem: and nova-net | |
| 15:21:16 | mriedem | i can't remember the last nova-net regression | |
| 15:21:36 | sdague | so, I thought there was a feature flag around security groups | |
| 15:21:43 | mriedem | sdague: that never got it | |
| 15:21:45 | mriedem | *in | |
| 15:21:48 | sdague | oh | |
| 15:22:28 | mriedem | https://review.openstack.org/#/c/225199/ | |
| 15:22:42 | mriedem | nearly a year old, might as well abandon | |
| 15:22:49 | mriedem | ccarmack isn't going to be working on that | |
| 15:26:00 | bauzas | mriedem: planning to cut RC2 by Thurs ? | |
| 15:26:12 | mriedem | bauzas: yes | |
| 15:26:27 | bauzas | okay | |
| 15:26:33 | mriedem | sdague: dansmith: https://review.openstack.org/433707 | |
| 15:26:40 | bauzas | mriedem: I have https://review.openstack.org/#/c/427200/ still WIP | |
| 15:26:54 | bauzas | mriedem: do you think it's still fine for trying to merge it ? | |
| 15:26:59 | bauzas | and then cherry-picking it ? | |
| 15:27:01 | mriedem | bauzas: it's getting a bit late | |
| 15:27:06 | bauzas | yup | |
| 15:27:11 | bauzas | hence my point | |
| 15:27:12 | mriedem | i'd rather just backport to stable/ocata | |
| 15:27:18 | mriedem | after we release ocata | |
| 15:27:26 | bauzas | for a point release ? fine by me | |
| 15:27:47 | sdague | mriedem: +2 | |
| 15:27:48 | bauzas | okay, I'll look after other bugs then | |
| 15:28:22 | mriedem | fudge, this means we'll have to backport this skip patch to ocata too | |
| 15:28:26 | mriedem | and newton | |
| 15:28:27 | mriedem | and mitaka | |
| 15:28:52 | mriedem | and i've got scrum to run to, so bbiab | |
| 16:02:10 | jaypipes | mtreinish: did something happen to -- --failing for ostestr? I'm now getting an error if I run tox -epy35 -- --failing... | |
| 16:02:41 | jaypipes | mtreinish: soemthing about --failing not being a regex. | |
| 16:05:17 | mtreinish | jaypipes: hmm, ostestr should passthrough --failing straight to testr in the subprocess since it's not a known arg | |
| 16:05:34 | mtreinish | we haven't changed anything related to that lately | |
| 16:05:40 | jaypipes | ya... that's what it used to do I think :) | |
| 16:06:21 | mriedem | i gave up on it | |
| 16:06:21 | mriedem | i haven't gotten --failing to work for a long time | |
| 16:06:51 | mtreinish | there might be an ostestr bug related to it | |
| 16:07:03 | mtreinish | I haven't tried it recently | |
| 16:07:08 | mtreinish | jaypipes: can you pastebin the stacktrace? | |
| 16:10:52 | jaypipes | mtreinish: yeah, gimme a bit. running entire functional test suite :) | |
| 16:15:23 | dansmith | mriedem: I use --failing every day | |
| 16:18:39 | mtreinish | dansmith: with ostestr or just testr straight? | |
| 16:19:00 | dansmith | mtreinish: whatever I get if I do "tox -ething -- --failing" | |
| 16:19:44 | mtreinish | dansmith: if it's py27 or anything that doesn't have an whitelist or blacklist file its probably just testr | |
| 16:19:51 | dansmith | okay | |
| 16:19:58 | dansmith | I do it on functional too | |
| 16:21:30 | mtreinish | dansmith: yeah it looks like ostestr is only used on the py35 jobs | |
| 16:21:34 | dansmith | ack | |
| 16:22:42 | mriedem | lyarwood: i'm enjoying working on backports of https://review.openstack.org/#/q/topic:bug/1643017+branch:stable/newton to stable/mitaka for a customer | |
| 16:22:51 | mriedem | lyarwood: i assume you are doing or have done the same | |
| 16:24:04 | macsz | gcb: ping | |
| 16:26:42 | vgadiraj | question - is there an enable_service command? Receiving an error when running a reproduce.sh file that reads "No hosts found to map to cell, exiting". This is for a multi-node gate job. Trying to see if adding "enable_service n-cell" has any positive effect | |
| 16:29:59 | jaypipes | mtreinish: paste showing tox -efunctional -- --failing working and tox -epy35 -- --failing not working: http://paste.openstack.org/show/598861/ | |
| 16:30:36 | jaypipes | mtreinish: must be some difference between pretty_tox.sh and pretty_tox3.sh | |
| 16:31:38 | jaypipes | mtreinish: indeed there is quite a big difference! :)] | |
| 16:32:07 | mtreinish | jaypipes: yeah pretty_tox3.sh is forcing a -r there: https://github.com/openstack/nova/blob/master/tools/pretty_tox3.sh | |
| 16:32:35 | mtreinish | it assumes any args when it's called are going to be a regex | |
| 16:32:42 | mtreinish | which doesn't work if you pass --failing or another flag | |
| 16:46:56 | mtreinish | mriedem: do you know if novnc works on py35? | |
| 16:47:03 | mriedem | no idea | |
| 16:47:12 | mtreinish | we're trying to fix the novnc test on tempest when running under p35 and we're hitting: http://logs.openstack.org/49/429949/8/check/gate-tempest-dsvm-py35-ubuntu-xenial-nv/30158f2/logs/screen-n-novnc.txt.gz#_2017-02-14_14_44_15_302 | |
| 16:47:22 | mtreinish | https://review.openstack.org/#/c/429949/ is the tempest change | |
| 16:55:41 | sdague | mriedem: which job was supposed to fail? | |
| 16:55:46 | sdague | because the nfs one seems to have worked | |
| 16:55:50 | mriedem | sdague: nfs and lvm multibackend | |