2023-06-03T14:28:05  <acidsys> moin cboltz, when you have time to review my SR for LimeSurvey .. it's a bit bigger, would appreciate feedback
2023-06-03T17:02:32  *** teepee_ is now known as teepee
2023-06-03T17:17:38  <acidsys> updating/rebooting meet.o.o .. seems no conferences are active atm
2023-06-03T18:27:34  <cboltz> acidsys: for sr 1090646, I have two nitpicks:
2023-06-03T18:27:41  <cboltz> - find . -type f \( -name '.*' -o -name '*keep' -o -name '*.htaccess' \) -delete
2023-06-03T18:28:08  <cboltz> are there really *.htaccess (and not just .htaccess) files around?
2023-06-03T18:28:25  <cboltz> - ReadWritePaths=/etc/limesurvey in limesurvey-php-fpm-systemd
2023-06-03T18:28:38  <cboltz> does limesurvey write to its own config file?
2023-06-03T18:29:09  <cboltz> besides that, I didn't notice an obvious error
2023-06-03T18:36:43  <acidsys> hi cboltz, thanks for the input. I will remove the asterisk. unfortunately it does need to write to all of its configuration files, even adding an existing configuration without using the installer
2023-06-03T18:37:11  <acidsys> I did start to patch out a lot of "isWritable" clauses but gave up
2023-06-03T18:37:35  <cboltz> if the asterix is superfluous, then the whole .htaccess is superfluous since the find command already checks for .*
2023-06-03T18:38:06  <cboltz> for the writeable config - it's not nice, but this sounds like a case where reality wins ;-)
2023-06-03T18:43:33  <acidsys> good point .. and yes :-) sent an updated SR
2023-06-03T18:52:26  <cboltz> https://gitlab.infra.opensuse.org/infra/salt/-/jobs/12917 just failed with "no space left on device", can you please have a look at gitlab-runner2?
2023-06-03T18:53:18  <acidsys> my favourite
2023-06-03T18:55:33  <acidsys> # docker image prune
2023-06-03T18:55:37  <acidsys> < long pause >
2023-06-03T18:55:41  <acidsys> Total reclaimed space: 22.5GB
2023-06-03T18:55:48  <cboltz> oh, nice ;-)
2023-06-03T18:56:13  <acidsys> I think I am expecting too much from GitLab to clean up after itself?
2023-06-03T18:56:18  <acidsys> maybe in the $$$ edition
2023-06-03T18:57:50  <cboltz> maybe in the cron.daily edition? ;-)
2023-06-03T19:01:23  <acidsys> heh
2023-06-03T19:19:18  <cboltz> in !642, can you please temporarily add some invalid nginx and sudo config? 54 seconds for the nginx test look too good to be true...
2023-06-03T19:33:45  <acidsys> cboltz: interesting, it complains about missing "virtual.None" SLS, should grains.virtual not be "container" or similar, but even if, we wouldn't have a file for that
2023-06-03T19:34:09  <acidsys> also nice that it exits as 0 regardless
2023-06-03T19:34:29  <acidsys> I think I will add --retcode-passthrough in a later MR
2023-06-03T19:35:23  <acidsys> ah, now I get it, we are overriding the virtual grain in prepare_test_Env
2023-06-03T19:38:53  <cboltz> indeed, failing with $? == 0 is "nice"
2023-06-03T19:40:02  <acidsys> also nice that this is the second case today a redirect to /dev/null fooled me ... ;-)
2023-06-03T19:41:49  <cboltz> I'm a bit surprised why virtual fails now - unless I overlook something in your MR (especially in prepare_test_env.sh), it was empty before, and still is
2023-06-03T19:42:07  <cboltz> and obviously the test now doesn't like that it's empty
2023-06-03T19:44:56  <acidsys> That puzzles me as well - for now I try to just remove the two empty lines from the grains file .. but nothing in that regard should have changed
2023-06-03T19:45:01  <acidsys> are you sure the tests worked before?
2023-06-03T19:45:40  <cboltz> yes, if you look at the pipeline of another MR or production, you'll see a warning about a deprecated syntax for ssl
2023-06-03T19:46:25  <cboltz> for example https://gitlab.infra.opensuse.org/infra/salt/-/jobs/12958 (latest production)
2023-06-03T19:47:40  <cboltz> I guess setting   virtual: virtual   might be a better idea than dropping it ;-)
2023-06-03T19:47:49  <acidsys> interesting
2023-06-03T19:48:34  <acidsys> I'm not a big fan of overriding the built-in grains, I guess the idea is to simulate a kvm or physical machine inside the test environment, but for the nginx test it should not be needed
2023-06-03T19:48:58  <acidsys> theoretically inside a docker container the default grain should be "container" or something similar
2023-06-03T19:49:00  <cboltz> (actually I wonder if the two pillar/virtual/* files are even worth a separate directory, or if they could become a {% if ...} somewhere - but that would be a separate MR)
2023-06-03T19:49:17  <acidsys> I would very much like to drop these as well
2023-06-03T19:49:37  <cboltz> looking at the log, we seem to have  virtual: None   aka undefined
2023-06-03T19:50:21  <acidsys> that's from the printf "virtual:\n" ... in prepare_test_env, no?
2023-06-03T19:50:40  <cboltz> yes, probably
2023-06-03T19:51:03  <cboltz> feel free to do a small MR to get rid of pillar/virtual/ and then rebase 642
2023-06-03T19:52:12  <acidsys> just tried inside Podman, grains.virtual resolves to "physical" .. I think that's a bug
2023-06-03T19:52:17  <acidsys> alright
2023-06-03T19:52:50  <acidsys> can I just drop the ntp tinker override? we mostly use chronyd now anyways
2023-06-03T20:02:17  <cboltz> good question - I'm afraid my ntp knownledge is too small to answer this (but mostly using chrony sounds like a good reason to just drop this)
2023-06-03T20:02:20  <acidsys> fun, now the tests fail with zypper not being able to add o:i repo, New repository or package signing key received
2023-06-03T20:02:27  <acidsys> New repository or package signing key received:
2023-06-03T20:02:29  <acidsys> https://gitlab.infra.opensuse.org/infra/salt/-/jobs/12969
2023-06-03T20:03:12  <acidsys> cboltz: fair, I moved it for now, figured it's better replace the whole ntp pillar with chronyd separately at some point
2023-06-03T20:03:49  <cboltz> ok
2023-06-03T20:04:06  <cboltz> hmm, that zypper error is interesting[tm]
2023-06-03T20:04:38  <acidsys> it's again a thing where I could "fix" it by importing the key but .. why did it work before
2023-06-03T20:05:02  <cboltz> did it, or did it just   ignore_error; exit 0   ?
2023-06-03T20:05:24  <acidsys> that's more likely
2023-06-03T20:05:32  <cboltz> I mean, after all the errors, I see a green   PASSED
2023-06-03T20:05:51  <acidsys> I should just quickly add the /dev/null back and have you verify the success based on the runtime
2023-06-03T20:06:22  <cboltz> indeed ;-)  (+ the ssl deprecation note)
2023-06-03T20:07:29  <acidsys> in my case the include error I introduced
2023-06-03T20:25:49  <cboltz> right, that one failed as expected, and I also see the expected deprecation note for ssl
2023-06-03T20:26:28  <cboltz> so if you un-break pagure again, the MR should be ready for merge
2023-06-03T20:47:09  <acidsys> seems good
2023-06-03T20:55:41  <cboltz> indeed :-)
2023-06-03T20:55:48  * cboltz dared to click the merge button
2023-06-03T21:01:26  * acidsys hopes for the best
2023-06-03T21:34:56  *** teepee_ is now known as teepee
2023-06-03T21:44:57  <cboltz> well, it breaks my test setup because it blindly expects /srv/pillar/id/ (which I "fixed" with a symlink for now)
2023-06-03T21:46:24  <cboltz> and the changed include order in pillar/top.sls means that a role now wins over salt_cluster/* - but I won't complain about that since it means i can now override everything with roles/cboltz, and don't need to do it in salt_cluster/geeko anymore
2023-06-03T21:55:57  <acidsys> in production it worked as expected
2023-06-03T21:56:26  <acidsys> one could add a file.exists clause but I think /srv/pillar/ is the most common path for people to store the pillar root in so I think you're a bit of a special case there
2023-06-03T21:56:48  <cboltz> yeah, I know that my test setup is a bit special ;-)
2023-06-03T21:58:07  <acidsys> speaking of /srv/{salt,pillar}, the -ng reactor seems to work quite nicely, I think we can soon drop the GitFS
2023-06-03T21:58:10  <cboltz> (/srv/pillar/ is common for production use, I'm not that sure when it comes to development setups - but anyway, needing a symlink is not that bad)
2023-06-03T21:58:36  <cboltz> nice :-)
2023-06-03T22:31:43  <cboltz> for 638,   chmod +x bin/test_haproxy.sh   would be a good idea ;-)