*** liuxie is now known as liushy | 09:06 | |
opendevreview | Albin Vass proposed zuul/zuul-jobs master: DNM: Test command creates behaviour https://review.opendev.org/c/zuul/zuul-jobs/+/929897 | 09:27 |
---|---|---|
kevko | Hi, i have a problem to debug one review ... from logs i am unable to see what's happening because everything looks ok ... can i somehow connect to the nodes from pipeline ? | 09:58 |
tonyb | kevko: we can arrange for that to happen. | 10:00 |
tonyb | we can create an autohold for failing jobs. | 10:01 |
tonyb | I'll be in a position to help in about 45mins | 10:01 |
kevko | cool ..please let me know .. | 10:06 |
kevko | i will be here | 10:06 |
kevko | tonyb: i will continue with semi blind debug :) | 10:06 |
tonyb | kevko: Ok, I'm able to help now | 10:51 |
tonyb | If you can give me a change number and a jobname, I can hold the next node run that matches | 10:51 |
tonyb | kevko: from there I'll need an SSH key and then you'll be able to log in | 10:52 |
kevko | tonyb: I am here | 11:19 |
tonyb | kevko: Okay, let me know when you have that information | 11:20 |
kevko | tonyb: https://paste.openstack.org/show/bBFw4WCqLCm1kIgz6lMh/ | 11:20 |
tonyb | kevko: Thanks | 11:21 |
kevko | tonyb: kolla-ansible-ubuntu-multinode-ipv6 , change 929859 | 11:22 |
tonyb | Gah, I messed it up | 11:24 |
tonyb | let me try again | 11:24 |
tonyb | kevko: Okay, recheck the job | 11:26 |
tonyb | when it fails I can add your SSH key | 11:26 |
kevko | okay thanks | 11:28 |
tonyb | kevko: I can see the job is queued | 11:34 |
kevko | tonyb: also | 11:36 |
tonyb | kevko: yes? | 11:38 |
tonyb | Looks like the autohold triggered | 12:03 |
kevko | yeah ..checking | 12:03 |
tonyb | reach out to any of the infra-roots to have us clear the autohold when you're done with it | 12:04 |
kevko | secondary1 has fd::ff:0 and 3306 running | 12:04 |
kevko | it's vip handled by keepalived | 12:04 |
kevko | mysql -h fd::ff:0 -u root_shard_0 -pwEBwgXgITGYKuw6yTIFKvQztvxSiyESWRDtqBNW6 << this is normally working ...so i really don't understand what is going on ... | 12:05 |
kevko | yet | 12:05 |
kevko | hmm ... investigating .. | 12:06 |
kevko | thanks ..i will let you know | 12:06 |
opendevreview | Stephen Finucane proposed openstack/project-config master: Start including team as release meta field https://review.opendev.org/c/openstack/project-config/+/929914 | 12:44 |
stephenfin | fungi: Do we currently have any mechanism for IRC interaction from zuul jobs, or is it just gerritbot? | 13:05 |
fungi | stephenfin: we don't currently. zuul does have an mqtt reporter though which could probably be easily tied into an irc bot | 13:08 |
clarkb | corvus: I've just approved the zuul launcher change | 16:26 |
clarkb | more of an fyi than anything else. I don't think we have a new server for that yet so it should noop other than the test updates | 16:26 |
corvus | clarkb: ack thx! | 16:27 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 17:34 |
clarkb | infra-root ^ some discussion about unbound and dns and dhcp and ovn reminded me this is something I've been meaning to try to get done for us | 17:34 |
clarkb | I'd be curious to hear if there are opinions on whether or not we think that will be less reliable over all (since theoretically tls adds a new complicated layer to the system) | 17:35 |
clarkb | also that is very likely going to be a base-test update first whihc my current change isn't set up for. I'll go ahead and WIP it. I awnted to make sure there weren't any strong objections before I make a copy of the role and set up base-test and all that | 17:36 |
opendevreview | Merged opendev/system-config master: Run zuul-launcher https://review.opendev.org/c/opendev/system-config/+/924188 | 17:40 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 18:17 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Add libjpeg-dev to bindep for Pillow wheels https://review.opendev.org/c/opendev/base-jobs/+/929962 | 18:17 |
clarkb | the blockdiag is no longer maintained and doesn't work with modern pillow has me wondering if we should just use inkscape to make a static svg of what we need | 18:21 |
clarkb | I'm definitely no artist but I suspect I can manage to do something like what pillow/blockdiag produce with inkscape | 18:22 |
clarkb | I want to say johnsom said that they did something similar to drop the blockdiag dependency | 18:22 |
clarkb | https://docs.opendev.org/opendev/base-jobs/latest/docker-image.html is the page in base-jobs with some diagrams. I think sytem-config and zuul-jobs and zuul have similar things | 18:23 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 18:37 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 18:37 |
clarkb | I'm glad I started poking at this. Will be good to get CI working on base-jobs repo again | 18:37 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 19:19 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 19:19 |
clarkb | this is turning into quite the little fun set of updates. Testing that will be difficult due to needing things to pass... Not sure if we just want to send it since the changes are relatively minor or go through a dance of pinning the node type, updating base-test only, then updating everything and running on noble | 19:20 |
clarkb | or maybe disabling linting temporarily to get base-test tested and then reenabling linting | 19:20 |
opendevreview | Monty Taylor proposed zuul/zuul-jobs master: Validate that node_version is set https://review.opendev.org/c/zuul/zuul-jobs/+/921066 | 20:02 |
opendevreview | Monty Taylor proposed zuul/zuul-jobs master: Remove deprecated javascript jobs https://review.opendev.org/c/zuul/zuul-jobs/+/921067 | 20:02 |
opendevreview | Merged zuul/zuul-jobs master: Modernize ensure-nodejs https://review.opendev.org/c/zuul/zuul-jobs/+/929852 | 20:23 |
opendevreview | Monty Taylor proposed zuul/zuul-jobs master: Add support for installing node via NVM https://review.opendev.org/c/zuul/zuul-jobs/+/921304 | 20:24 |
opendevreview | Merged zuul/zuul-jobs master: Validate that node_version is set https://review.opendev.org/c/zuul/zuul-jobs/+/921066 | 20:31 |
clarkb | I think I decided to minimize the total number of changes so that it is easier to reviwe then we just take a small gamble doing the minimal update | 20:32 |
clarkb | later changes can do bigger linter updates | 20:32 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 20:45 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 20:45 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 20:55 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 20:55 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 21:05 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 21:05 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Fixup CI jobs https://review.opendev.org/c/opendev/base-jobs/+/929962 | 21:19 |
opendevreview | Clark Boylan proposed opendev/base-jobs master: Set up Unbound to forward via DNS over TLS https://review.opendev.org/c/opendev/base-jobs/+/929960 | 21:19 |
clarkb | ok https://review.opendev.org/c/opendev/base-jobs/+/929962 should be ready for review now. I'm open to feedback on the approach and how we want to make sure its tested/safe | 21:37 |
corvus | clarkb: what prompted the change to the upload-git-mirror role usage? | 23:26 |
clarkb | corvus: there was a linter error saying "role" is a required attribute | 23:30 |
corvus | i don't suppose you remember that rule name? | 23:30 |
clarkb | corvus: I'm guessing that name is an alias for role but we're using the role + associated var in a less standard way? | 23:30 |
clarkb | I don't but let me see if I can find it | 23:30 |
corvus | i've been looking through the docs and can't find it | 23:30 |
corvus | it certainly was the standard way to do that when we started; and the fact that we could write things like that was actually part of why we thought migrating from jjb builders to ansible would be intuitive and easy | 23:31 |
clarkb | corvus: https://zuul.opendev.org/t/opendev/build/8e8b45b57079429ea40163a8134f077d/log/job-output.txt#1117 | 23:32 |
corvus | i will find it disappointing if they have decided it needs to be more verbose :( | 23:32 |
clarkb | it points to https://docs.ansible.com/ansible/latest/user_guide/playbooks_reuse_roles.html#roles | 23:32 |
corvus | https://ansible.readthedocs.io/projects/lint/rules/schema/ | 23:33 |
clarkb | corvus: if you grep for 'You can pass other keywords to' I based my edit off of that example | 23:34 |
corvus | i feel like there are two ansibles: the one written by the program authors, and the one written by the linter authors. | 23:40 |
clarkb | to be fair the way we were using roles doesn't show up in that document. However, it clearly works so ya not sure what is going on there | 23:41 |
clarkb | that document == the ansible using roles guide | 23:41 |
corvus | ansible makes it very hard to find old docs, and i don't care enough to look any more; the fact that it worked for years without warnings from the actual program is justification enough anyway :) | 23:42 |
corvus | i'm +2 on the change, just continuing to be aggravated by the whole ordeal | 23:43 |
fungi | i continue to be unsure what the actual benefits are we're getting out of ansible-lint | 23:43 |
clarkb | there are actually a couple of things it catches in this chagne that are good. Primarily the octal or not mode specifications | 23:44 |
corvus | something to talk about other than the weather! | 23:44 |
corvus | i like the octal thing. | 23:45 |
clarkb | it also runs the syntax checker without needing to install all of zuul to get zuul_return and zuul_console | 23:45 |
corvus | ansible should deal with the octal thing though :) | 23:45 |
clarkb | corvus: ++ | 23:45 |
clarkb | the checks for proper truthy strings are good too | 23:45 |
fungi | yeah, having a secondary ansible parser that doesn't require executing it to identify egregious errors makes sense | 23:45 |
clarkb | I've noticed we've done a fair bit of the wrong thing around that whcih could result in unexpected behavior. I even notied we check if defined instead if true or false | 23:46 |
clarkb | and the avlue is "True" | 23:46 |
clarkb | but ya I think the primary thing it gets us is syntax checking with minimal setup (we can just mock the modules for zuul_return and zuul_console) | 23:46 |
clarkb | then also a reminder to quote our mode strings | 23:47 |
clarkb | I think if we wanted to drop linting on say system-config or zuul-jobs where we actually do have quite a bit of robust testing I wouldn't care too much. But base-jobs deals with a lot of test that is difficult to test and the linter is something we should put up with there until we can have more robust functional testing | 23:56 |
clarkb | *deals with a lot of stuff that is difficult to test | 23:57 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!