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
Kreaper[m] has left #freedesktop [User left]
scrumplex has joined #freedesktop
scrumplex_ has quit [Ping timeout: 480 seconds]
Xe has quit [Ping timeout: 480 seconds]
Xe has joined #freedesktop
<zmike>
getting some incredibly nostalgic mails from historic pipelines tonight
<alanc>
I'm at least getting mail from new gitlab MR's now
eluc1[m] has joined #freedesktop
eluc1[m] has left #freedesktop [User left]
tlwoerner has joined #freedesktop
alanc has quit [Remote host closed the connection]
alanc has joined #freedesktop
tlwoerner_ has quit [Ping timeout: 480 seconds]
eluks has quit [Remote host closed the connection]
eluks has joined #freedesktop
Kayden has quit [Quit: reboot]
swatish2 has joined #freedesktop
Kayden has joined #freedesktop
Kayden has quit [Quit: back to xe]
jsa1 has joined #freedesktop
sghuge has quit [Remote host closed the connection]
sghuge has joined #freedesktop
swatish2 has quit [Ping timeout: 480 seconds]
ximion has quit [Remote host closed the connection]
AbleBacon has quit [Read error: Connection reset by peer]
tzimmermann has joined #freedesktop
swatish2 has joined #freedesktop
<hakzsam>
is it safe to create new MRs and merge now? or should I wait more?
<mupuf>
I would hold on merging mesa MRs, but if you are OK with losing comments in your MR go ahead I would say
<hakzsam>
okay, I will wait, no probleù
<hakzsam>
thanks for all the work :)
jsa1 has quit [Ping timeout: 480 seconds]
jsa1 has joined #freedesktop
<bilboed>
bentiss: set all gstreamer projects to 3month pipeline auto expire
<bentiss>
bilboed: thanks!
<bilboed>
"I have set..."
Kayden has joined #freedesktop
<bilboed>
how's the s3 mega-sync-from-hell coming along ?
<bentiss>
globally on pause, still figuring out a plan
<bentiss>
but I managed to get the amount of mesa data yesterday: 24.182 TB -> and it's relatively easy to prune because there is a "folder" per day
<bilboed>
24TB :D
<bentiss>
so my current plan is to manually purge everything older than 1 year, which should make S3 on euinix smaller and faster
<bilboed>
👍️
mripard has joined #freedesktop
<eric_engestrom>
bentiss: should we still wait with releases as well? (mesa in my case)
<eric_engestrom>
if we have to go back to a backup, I can re-push the tag, nothing will be lost (it will be the same git object, same hash), and the tarballs are on another server (kemper iirc?)
swatish2 has quit [Ping timeout: 480 seconds]
<eric_engestrom>
(ah no, the tarballs are uploaded to annarchy, but the point remains that it's on another server, so not affected by a potential gitlab rollback)
<eric_engestrom>
so I think it's safe, right?
swatish2 has joined #freedesktop
jsa1 has quit [Ping timeout: 480 seconds]
<bentiss>
eric_engestrom: yeah, should be fine. We had a good day yesterday, so I don't think we'll need to rollback
<eric_engestrom>
glad to hear \o/
<eric_engestrom>
I'll make the release this afternoon (CET) then
<eric_engestrom>
thanks for everything!
jsa1 has joined #freedesktop
Dragoon has joined #freedesktop
Dragoon is now known as Guest11786
DragoonAethis has quit [Ping timeout: 480 seconds]
Guest11786 has quit []
DragoonAethis has joined #freedesktop
krei-se- has quit []
Kayden has quit [Quit: reboot]
krei-se has joined #freedesktop
Kayden has joined #freedesktop
swatish2 has quit [Ping timeout: 480 seconds]
DragoonAethis has quit [Remote host closed the connection]
DragoonAethis has joined #freedesktop
swatish2 has joined #freedesktop
DragoonAethis has quit [Ping timeout: 480 seconds]
DragoonAethis has joined #freedesktop
DragoonAethis has quit [Remote host closed the connection]
DragoonAethis has joined #freedesktop
<jani>
can we push stuff to git repos, or should they still be considered read-only?
<jani>
git repos in gitlab, that is
guludo has joined #freedesktop
<slomo>
just labeled a new issue as spam and the spam bot is alive and well :)
<bentiss>
jani: should be fine (see above)
<jani>
bentiss: thanks!
<Xe>
27TB is pretty good, i'd have expected more lol
swatish21 has joined #freedesktop
jsa1 has quit [Ping timeout: 480 seconds]
swatish2 has quit [Ping timeout: 480 seconds]
jsa1 has joined #freedesktop
swatish21 has quit [Ping timeout: 480 seconds]
MathieuBridon[m] has joined #freedesktop
<MathieuBridon[m]>
hi all, I'm having some problem with gitlab.freedesktop.org: I pushed a new version of my website (gitlab pages) and the CI didn't run for the "pages" task, it has apparently been skipped 😕
<MathieuBridon[m]>
is that related to the unfinished migration to the nex datacenter? anything I can do to get the pages updated?
<MathieuBridon[m]>
a user reported a bug because they had installed incorrectly my software, so I pushed a new version of the pages explaining how to install it properly and was hopinr the page to be updated, but I guess I'll just have to wait 🤷
<MathieuBridon[m]>
any idea what's missing or how people can help getting the migration along? 😊
<svuorela>
MathieuBridon[m]: sleep(48h)
<vyivel>
what's missing in on the status tracker (linked above, updated manually), can't really help
<vyivel>
s/in/is/
<bentiss>
MathieuBridon[m]: not sure why the jobwas skipped. The runners are not impacting how the jobs are scheduled by gitlab
<bentiss>
MathieuBridon[m]: you should try restating a pipeline on main
bnieuwenhuizen has quit [Quit: Bye]
<bentiss>
restarting
bnieuwenhuizen has joined #freedesktop
<MathieuBridon[m]>
Ford_Prefect: the pipeline in question had the "test" job which passed 😕
<bentiss>
MathieuBridon[m]: yes, but you restarted it after gitlab marked the next job as skipped
<MathieuBridon[m]>
how do I restart it now, then? 😕
<bentiss>
and I guess there might be an upstream bug there where the job hasn't restarted or it could be that the gitlab instance is too busy cleaning up the old pipelines
<bentiss>
MathieuBridon[m]: in the menu on the left Build->Pipelines then "new Pipeline" in the top right
<bentiss>
select main and start
guludo has quit [Quit: WeeChat 4.5.2]
<MathieuBridon[m]>
ah, so start a new one rather than restart the existing one, I was searching for the latter and couldn"t find any way to do that ^^
guludo has joined #freedesktop
<bentiss>
we've got a latency of 25s for the sidekiq jobs to run, so expect some delay between creation of pipeline and processing of the jobs
<MathieuBridon[m]>
sure, no problem, I'm not in a hurry
<MathieuBridon[m]>
I just want it to run again so the install page can be correct 😉
<MathieuBridon[m]>
it's hard to tell users how to install my software if the install docs are wrong 😛
<MathieuBridon[m]>
the pages job is now pending, so it should all work I suppose 😊
<bentiss>
yep :)
<MathieuBridon[m]>
it does, except I made a typo 🤦
<MathieuBridon[m]>
stupid software that can't fix obvious typos, I thought this was all about AI, software ought to be interlligent 🤪
<vyivel>
in its intelligence it has decided to give you an opportunity to learn from mistakes :)
ximion has joined #freedesktop
<MathieuBridon[m]>
Except for another typo it's now all good, the doc is now fixed, thank you 😊
jsa1 has quit [Ping timeout: 480 seconds]
<bentiss>
\o/
<bentiss>
daniels: BTW, I'm working on making fleeting working today... just in case you decided to work on that too :)
<bentiss>
(should have probably told you that a few hours ago)
jsa1 has joined #freedesktop
<daniels>
bentiss: heh yeah, I saw you provisioned the server and went back to hacking on some other stuff I need to have done this week :\
<bentiss>
daniels: heh :)
<eric_engestrom>
has whatever change needed to s3cp been figured out already?
<bentiss>
eric_engestrom: yes, but I'm not sure we need s3cp instead
<bentiss>
eric_engestrom: also in the sense that we kind of need to tweak a little bit the deployment because it accepts too much uploads ATM
<eric_engestrom>
I don't quite understand ^^' are you saying "instead of fixing s3cp we need to replace it with a better solution?"
<bentiss>
so much simpler than having to replicate the S3 protocol from python
<bentiss>
though I haven't tried with big files
<eric_engestrom>
I see
<eric_engestrom>
I'm mostly asking because mesa CI is currently broken because it can't upload build artifacts
<eric_engestrom>
so we should replace the s3cp command we have in there with that curl command, instead of waiting for an update that fixes s3cp, right?
ximion has quit [Remote host closed the connection]
<DragoonAethis>
Just beware, it's abusing concurrency *A LOT*
<DragoonAethis>
not sure if it helps your current use case though
<bentiss>
DragoonAethis: the thing is s3-proxy is not talking S3 anymore on the client side, you have to use multipart-upload with a PUT request
<DragoonAethis>
bentiss: ah, okay
jsa1 has quit [Ping timeout: 480 seconds]
<eric_engestrom>
bentiss: I tested it, it does the double filename thing :/
<eric_engestrom>
if I understand correctly, that's something to fix on your side, right?
<bentiss>
yeah
<bentiss>
eric_engestrom: basically we need to remove the filename from the upload url, add the filename in the headers, and then in the policy I check for that
<eric_engestrom>
ack
<bentiss>
problem is, if people fake the filename, I have no way to know
<bentiss>
because I don't have it in the request s3-proxy makes (thus the extra header)
<eric_engestrom>
"fake the filename", you mean if the header doesn't match the file uploaded? that sounds like a user error, not something to care about on the server side?
<bentiss>
*or* if anyone knows about go, we can probably send a PR
<eric_engestrom>
if you know the codebase and can point me to the function, I can take a stab at it
<bentiss>
well, problem is some buckets allow a strict filename (git-cache), so you could upload movie.mp4 but add x-fdo-file=awesome-project.tar.gz, and OPA will validate the "filename", but S3-proxy will upload movie.mp4
<eric_engestrom>
yeah, that was v1, I have a couple of local changes, but that's basically it
<eric_engestrom>
(mainly dropping `-v`)
<bentiss>
maybe if you just drop the "results.tar.zst" from the URL that will be enough for most of the things that don't require filename validation
<bentiss>
and if we can tweak s3-proxy, we can also drop the extra header "x-fdo-file"
<bentiss>
actually the only problem is for git-cache, the easiest solution is to drop the requirement
<bentiss>
we can always see if someone is doing bad things...
<eric_engestrom>
ack, I'll drop the filename from the url, and keep it only in the header
<eric_engestrom>
I have to go though, so -> tomorrow
<bentiss>
sure, have a good one!
<bentiss>
eric_engestrom: for tomorrow: but I think we should drop the name in the extra custom header too, so just keeping the --form and URL without the filename would be fine
jsa1 has joined #freedesktop
tzimmermann has quit [Quit: Leaving]
<eric_engestrom>
daniels, bentiss: the https://mesa.freedesktop.org/archive/ redirection seems to have been dropped, which means the release tarballs can't be fetched anymore :(
* bentiss
looks
<eric_engestrom>
I just remembered we also have https://archive.mesa3d.org/, which does work, so I'll tell people to use that in the meantime
<eric_engestrom>
I see, thanks daniels, I'll put that in place tomorrow :)
<bentiss>
oh, if we can drop those from the main config that would be awesome ;)
<bentiss>
daniels: so far, it would seem that we need to have the cloud runners in separate hcloud projects
<bentiss>
I *think* this is fine with the credits, but we should try ;)
<bentiss>
(separate projects because the x86 and arms are fighting each other with the autoscaler put in place by the fleeting plugin, which is always called "group=hetzner/fsn1/hetzner-docker-autoscaler"
<bentiss>
hmm... I can try putting them in fsn and nbg
<bentiss>
my bad... name: "Name of the fleeting plugin instance group. The created instance names will be prefixed using this name. " I guess I should give them different names
Guest11809 has quit [Ping timeout: 480 seconds]
ximion has quit [Remote host closed the connection]
todi has quit []
TrinitronX has quit [Ping timeout: 480 seconds]
todi has joined #freedesktop
TrinitronX has joined #freedesktop
jsa1 has joined #freedesktop
AbleBacon has joined #freedesktop
<bentiss>
daniels: it works! though I need to have one corrdinator per class of devices