ChanServ changed the topic of #freedesktop to: https://www.freedesktop.org infrastructure and online services || for questions about freedesktop.org projects, please see each project's contact || for discussions about specifications, please use https://gitlab.freedesktop.org/xdg or xdg@lists.freedesktop.org
alanc has quit [Remote host closed the connection]
<xorgy>
Just wanted to congratulate all involved in the successful migration... the new GitLab is way faster (for now).
scrumplex has joined #freedesktop
scrumplex_ has quit [Ping timeout: 480 seconds]
dcunit3d has quit [Remote host closed the connection]
tlwoerner has joined #freedesktop
zerozero1 has joined #freedesktop
zerozero2 has joined #freedesktop
zerozero3 has joined #freedesktop
zerozerozero has quit [Ping timeout: 480 seconds]
zerozero1 has quit [Ping timeout: 480 seconds]
zerozero2 has quit [Ping timeout: 480 seconds]
ximion has quit [Remote host closed the connection]
swatish2 has joined #freedesktop
zerozerozero has joined #freedesktop
zerozero3 has quit [Ping timeout: 480 seconds]
sima has joined #freedesktop
sghuge has quit [Remote host closed the connection]
sghuge has joined #freedesktop
cascardo_ has joined #freedesktop
cascardo has quit [Ping timeout: 480 seconds]
zerozerozero has quit [Quit: WeeChat 4.5.2]
zerozerozero has joined #freedesktop
jsa1 has joined #freedesktop
tzimmermann has joined #freedesktop
swatish21 has joined #freedesktop
swatish2 has quit [Ping timeout: 480 seconds]
AbleBacon has quit [Read error: Connection reset by peer]
zerozerozero has quit [Ping timeout: 480 seconds]
<jadahl>
Currently getting "ERROR: The git server, Gitaly, is not available at this time. Please contact your administrator." when doing 'git fetch'
<jadahl>
nm, seems to have been temporary
swatish21 has quit [Ping timeout: 480 seconds]
<bentiss>
jadahl: yeah, sorry, I have been restarting them a couple of time as I'm setting up the backups
<bentiss>
should be good now, I seem to have found the correct parameters so I can do incremental backups on gitaly
<jadahl>
alright, thanks!
swatish2 has joined #freedesktop
ximion has joined #freedesktop
ximion has quit [Remote host closed the connection]
<bentiss>
daniels: good morning. For the arm runners, do we need kvm as well?
<bentiss>
(basically, if we do, that means we need to request one RX220, for $244 a month)
<bentiss>
or have it hosted at PSU with the HW arm promised
<eric_engestrom>
anyone with experience writing nginx rules?
<eric_engestrom>
dj-death: the mesa ci is not working yet (looks like there's one last bug in the proxy used by the rpi farm and all lava farms, I'm working on it)
<eric_engestrom>
so I would advise against merging anything
MrBonkers has quit []
<eric_engestrom>
most likely, anyone trying would just get a ci pipeline that's all red and it wouldn't get merged anyway, which is why I haven't specifically said anything (and also I keep hoping to find the fix Very Soon™)
<dj-death>
vyivel: apparently most of the builders are failing
<vyivel>
right yeah just checked the maintenance tracker
<vyivel>
sorry
jsa2 has joined #freedesktop
jsa1 has quit [Ping timeout: 480 seconds]
<bentiss>
dj-death: which jobs?
<eric_engestrom>
I assume dj-death means the mesa ci build jobs, which are failing right now because they try to upload the build to s3 using the broken s3cp
<bentiss>
k, cause we're having also some jobs trying to access the kvm module when they are not using the tag kvm, which kind of annoys me a little :)
<eric_engestrom>
bentiss: do you know nginx a bit? I just realized that while `nginx -t` does parse the config files, when I restart nginx it's not even reading them, I'm really lost
<bentiss>
eric_engestrom: can't really help you there. I'm fixing the couple of nginx files we have here, but everytime I'm paging things out
<bentiss>
plus the k8s deployment is slightly different from plain nginx
<eric_engestrom>
yeah, it might not translate to being able to help me :/
<bentiss>
FWIW, once you fixed the pipelines, I've another request for mesa MR: do not create 400 useless jobs for users when they never uses them, but rely on parent child pipeline instead
<bentiss>
the bulk of the db is full of jobs declaration that are never used. If we could dynamically create them, that would actually reduce the db size
MrBonkers has joined #freedesktop
<bentiss>
the idea is that if it's marge or schedule, or anything that needs to have a non manual pipeline the pipeline is created normally, but for all "manual" actions, this ould be a single job that triggers a child pipeline with a special variable that forces the creation of the pipeline
<bentiss>
I'll try to look further later, but for now, lunch time -> bbl
<eric_engestrom>
I'll just write everything here, hopefully someone will have an idea:
<eric_engestrom>
when I modify /etc/nginx/snippets/uri-caching.conf, which in included by /etc/nginx/sites-enabled/fdo-cache, which is included by /etc/nginx/nginx.conf, and I run `nginx -t` to test the config, it tells me if there's a parse error, but when I run `nginx -s reload` or `systemctl restart nginx`, it ignores whatever I typed in this file, I can delete everything and the server still behaves as if all the config was there
<eric_engestrom>
bentiss: re- only creating the jobs the user wants, I get the sysadmin need, but for mesa ci admin this is very non-trivial to write, and would require all users to use our script for the jobs to simply exist, when users are already complaining that they can't just click "run" on the web ui, so they'll push back hard if we actually delete the jobs from the web ui :/
<eric_engestrom>
I'll create an issue trying to gather feedback on the idea though, so see if maybe the people who complained in the past have changed their mind
<MrCooper>
eric_engestrom: it should be at most one additional step to trigger the child pipeline
<eric_engestrom>
yes, but it's a step that requires running the script as gitlab's web ui does not offer this kind of feature
<nirbheek_>
eric_engestrom: did you fix it? This answer suggests that using `set` to assign it to a variable would do the trick: https://serverfault.com/a/923274
<svuorela>
a3g/win 26
<svuorela>
woops
<MrCooper>
eric_engestrom: per our comments, I'm afraid you read too much into the request :)
<eric_engestrom>
bentiss, MrCooper: ah indeed, I misunderstood; I replied on the issue, but basically: yes, it should be trivial, I'll do that next
<eric_engestrom>
nirbheek_: no, not fixed yet, and I can't use the double-`set` trick because the variable only exists on that line because it's the regex match, but maybe there's an nginx way to do this anyway that I don't know (I know very little about it)
<eric_engestrom>
right now I've been trying to get it to evaluate a lua script in the right moment/with the right variable, to use `ngx.escape_uri()`, but I'm failing
zerozerozero has joined #freedesktop
zerozerozero has quit [Quit: WeeChat 4.5.2]
zerozerozero has joined #freedesktop
haaninjo has joined #freedesktop
<Hazematman>
eric_engestrom: I don't know much about nginx either but is that line just grabbing the whole url and then adding it to the end? Could you try using $uri or $request_uri instead of the captured $1?
<eric_engestrom>
Hazematman: unfortunately that doesn't work, these two vars contain the original uri, not the one in the `location:` header
<eric_engestrom>
I'm tempted to disable the farm with the buggy proxy so that we can re-enable the rest of the ci... ie. the ci-tron farms 😅
<eric_engestrom>
(and all the builds)
<eric_engestrom>
ie. disable the lava farms, and the baremetal farms (I just realised rpi is not the last baremetal farm as I thought)
<eric_engestrom>
daniels, sergi: thoughts? disabling farms is really something I want to never do if it can be avoided, but until someone comes up with a fix, it's either that or no ci for anyone
<eric_engestrom>
actually, is it possible to drop the proxy and have the duts talk to the internet directly? it will be a bandwidth cost, but it might be better than the two alternatives above
* eric_engestrom
tries bypassing the proxies
pzanoni_ has left #freedesktop [WeeChat 4.4.3]
pzanoni has joined #freedesktop
<daniels>
eric_engestrom: please don't bypass the proxies, no
<daniels>
I can try to look at nginx later tonight if you like
<eric_engestrom>
yes please
<eric_engestrom>
and ack on "no to bypass"
<daniels>
and itmt if you want to bring only ci-tron back, go for it
<daniels>
it's not just a bandwidth cost per se, but if you have hundreds of machines all pulling the entire rootfs for every single job, it just doesn't complete in time
<eric_engestrom>
ack, I'll disable all the other farms then :(
<eric_engestrom>
ah true, I forgot about that pesky "time" concept xD
dbrouwer has joined #freedesktop
zerozero1 has quit [Quit: WeeChat 4.5.2]
zerozerozero has joined #freedesktop
zerozerozero has quit []
swatish2 has joined #freedesktop
jsa1 has quit [Ping timeout: 480 seconds]
<bentiss>
daniels: "but if you have hundreds of machines all pulling the entire rootfs for every single job" -> maybe we can use fastly there??? We will see how this behaves with the registry, but that's probably something they are good at (local caching near the requester)
swatish2 has quit [Ping timeout: 480 seconds]
<sergi>
eric_engestrom: I agree to disable the farms with the problem. We can allow others to integrate while we work on the problem.
alanc has joined #freedesktop
nephyrin has quit [Quit: ... besides, it was hot]
AbleBacon has joined #freedesktop
tzimmermann has quit [Quit: Leaving]
<eric_engestrom>
bentiss: if your big post jsut now (thanks for it btw!), you dropped the `0` at the end of the mesa MR, which will no doubt be confusing to anyone following that link ^^
<bentiss>
eric_engestrom: fixed
ity1 has quit []
ity has joined #freedesktop
<bentiss>
and thanks for reporting this
<eric_engestrom>
thanks!
jsa1 has joined #freedesktop
Kayden has quit [Ping timeout: 480 seconds]
ximion has joined #freedesktop
zerozerozero has joined #freedesktop
Kayden has joined #freedesktop
dcunit3d has joined #freedesktop
jsa1 has quit [Ping timeout: 480 seconds]
zerozero1 has joined #freedesktop
zerozero1 has quit [Quit: WeeChat 4.5.2]
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozerozero has quit [Ping timeout: 480 seconds]
zerozerozero has joined #freedesktop
jsa1 has joined #freedesktop
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozero1 has joined #freedesktop
zerozero1 has quit []
zerozerozero has quit [Ping timeout: 480 seconds]
zerozerozero has joined #freedesktop
JanC has quit [Ping timeout: 480 seconds]
sima has quit [Ping timeout: 480 seconds]
broski[m] has joined #freedesktop
imre has quit [Ping timeout: 480 seconds]
jsa1 has quit [Ping timeout: 480 seconds]
<f_>
Hey congrats on the migration, happy it went mostly ok ^^
Kayden has quit [Remote host closed the connection]