<travier[m]>
we should probably have a full example for AWS in the docs
<guesswhat>
It can work, but I am looking for something more local-ish..
<guesswhat>
Well, I am using ct provider for terraform, doing some templates at deploy time
crobinso has joined #fedora-coreos
nalind has joined #fedora-coreos
<dustymabe>
yes, you can compress pieces of your Ignition config
<dustymabe>
butane should opportunistically do this for you
guesswhat has quit [Quit: Connection closed]
<nemric>
Yes you can gzip some contents, but not with S3 "The type of compression used on the contents (null or gzip). Compression cannot be used with S3."
plarsen has joined #fedora-coreos
nalind has quit [Ping timeout: 246 seconds]
mheon has joined #fedora-coreos
mheon has quit [Quit: WeeChat 3.4]
mheon has joined #fedora-coreos
<dustymabe>
travier[m]: lucab: jlebon: anybody able to run the meeting today? I have to go to kid pickup in the middle of the meeting
<travier[m]>
sorry, I might not be able to be there either
<jlebon>
i can do it
nalind has joined #fedora-coreos
tormath1 has joined #fedora-coreos
<tormath1>
Hello - for your information in latest Flatcar Alpha release we started to use the experimental Afterburn network args feature
<tormath1>
and it should be soon covered by a Mantle test
<travier[m]>
tormath1: Cool. Do you have a link to the changes?
<dustymabe>
jlebon: since you volunteered (thanks). I also created the first version of the F37 changes review ticket. If we could go over that today that would be nice.. The goal is to try to start reviewing the changes earlier to identify any issues early enough that we could participate in the discussion about the change.
<jlebon>
dustymabe: ack. feel free to tag anything else to bring up
<jlebon>
dustymabe: i guess if we want to get in earlier, we should be looking at them pre-approval
saqali has quit [Ping timeout: 255 seconds]
jpn has quit [Ping timeout: 244 seconds]
saqali has joined #fedora-coreos
mnguyen_ has joined #fedora-coreos
saqali_ has joined #fedora-coreos
saqali has quit [Ping timeout: 240 seconds]
mnguyen has quit [Ping timeout: 260 seconds]
paragan has quit [Quit: Leaving]
jpn has joined #fedora-coreos
<jlebon>
all, please tag anything you'd like to bring up in the community mtg with the meeting label
<jlebon>
i wonder if we should just have video meetings every time
<dustymabe>
ehh. we could increase the frequency
<dustymabe>
but I like the IRC meetings too
<dustymabe>
each style has pros/cons
<jlebon>
i do too, some aspects. but at the same time, it feels like we could go over so much more in video
<jlebon>
yeah definitely. alternating could be interesting
<dustymabe>
video meetings require more preparation/planning - they also require more work to "capture" the results
<dustymabe>
but they are more productive if it's a detailed topic
<dustymabe>
IRC meetings are very passive (not a lot of prep) and they also allow people to multi-task effectively (more people are able to come to the meeting)
<dustymabe>
I was gone for 25 minutes of our meeting today and I'm all caught up after 5 minutes of reading
* dustymabe
lunch
<jlebon>
maybe we could call for it when we know we have something larger to work through
* jlebon
also lunches
<bgilbert>
dustymabe: +1
tormath1 has quit [Quit: leaving]
ksinny has quit [Remote host closed the connection]
ravanelli has quit [Remote host closed the connection]
<bgilbert>
nemric guesswhat: dustymabe was talking about inline compression for data URLs, not remote S3 resources
<dustymabe>
yeah. I mean if you can do s3 then do you really care about compression? s3 capacity is pretty large :)
<bgilbert>
right
<bgilbert>
guesswhat: we intentionally don't support config compression. it might look like a good solution for userdata size limits, but it tends to suddenly fail when your config gets larger :-)
<bgilbert>
guesswhat: remote resources are the right solution
<dustymabe>
bgilbert: you *can* compress the userdata directly though can't you?
* dustymabe
remembers we did something like that recently for kola
<dustymabe>
that last comment was a question I had
<dustymabe>
"Ideally the job would run inside the openshift/os repo, but GitHub Actions isn't available there."
<bgilbert>
yeah :-(
<dustymabe>
Is there benefit to having it updating all the time versus just user triggered when needed?
jpn has quit [Quit: Lost terminal]
<dustymabe>
I could see value in getting CI feedback sooner
<dustymabe>
but.. that is only beneficial if someone is looking at the failures
<bgilbert>
reducing friction. whenever there are unvendored changes, there's a PR that we can go look at to see what's outstanding
<dustymabe>
and completely wasteful otherwise
<bgilbert>
if it can't merge for some reason, we can mark it draft
<bgilbert>
and add comments giving the state of things
<bgilbert>
I'm hoping this encourages us to frequently merge smaller updates, rather than letting things pile up
<dustymabe>
yeah. I just don't want yet another process that's running daily and no one looking at the results. in general this sounds like a good idea, though
<bgilbert>
yeah, it's a hypothesis. we can revisit if it isn't working well
<dustymabe>
stamped!
<bgilbert>
I think it's okay if we sometimes ignore failures, though. it's more that we're continually ready to merge, not that that we're continually trying to
<walters>
like I just click build in jenkins and override `COREOS_ASSEMBLER_IMAGE` and also force? any thoughts/objections?
<dustymabe>
theoretically that should be fine unless it somehow gets our built artifacts into some bad state (i.e. upgrade tests fail or builds.json gets out of whack)
<dustymabe>
would probably be better to hack in a different pipeline, but I realize that's cumbersome
<dustymabe>
cc jlebon
<walters>
hmm, mocking up the response from robosignatory in `cmd-sign` might not be too hard
<jlebon>
yeah, should be fine. just to make sure on dustymabe's point, the ostree container change is backwards compatible, right?
<jlebon>
i.e. non-continuous cosa can manipulate container from continuous cosa just fine
<jlebon>
container built by* continuous cosa
<dustymabe>
oh.. signing.. yeah we need the prod repo for that (there is a way to set it up in stage, but aintnobodygottimeforthat