<mei_>
/run/ignition/dev_aliases: cannot open `/run/ignition/dev_aliases' (No such file or directory)
<mei_>
createLuks: created device alias for "/dev/sdb": "/run/ignition/dev_aliases/dev/sdb" -> "/dev/sdb" (????)
<mei_>
okay i figured out that ignitation check if the device is luks before creating luks (dunno if that's a safe check to prevent overwrite), but it seems wrong that is saved as an error, it should output a cool message instead
ravanelli has quit [Remote host closed the connection]
<guesswhat[m]>
hey, user systemd units are still not supported out of the box via ingition, right? thinking if is even possible to create dependecny between root systemd and users systemd unit..
Turnikov has joined #fedora-coreos
ravanelli has joined #fedora-coreos
ravanelli has quit [Remote host closed the connection]
<travier[m]>
It's not possible to create dependencies between user & system systemd units
<travier[m]>
(not Fedora CoreOS specific, it's like that in systemd
<guesswhat[m]>
thanks. And with User= stanza in root systemd unit? Do I need to enable linger for that non root user? I know this is also out of scope question :)
<guesswhat[m]>
Hmm, even with loginctl enable-linger 1000 systemd will die to Main process exited, code=exited, status=125/n/a
apiaseck has joined #fedora-coreos
ravanelli has joined #fedora-coreos
Turnikov has quit [Ping timeout: 255 seconds]
Turnikov has joined #fedora-coreos
Turnikov has quit [Ping timeout: 246 seconds]
<guesswhat[m]>
Another question would be, isnt enable-linger blocking zincati updates?
nalind has joined #fedora-coreos
ravanelli has quit [Remote host closed the connection]
ravanelli has joined #fedora-coreos
jpn has quit [Ping timeout: 268 seconds]
dwalsh has quit [Ping timeout: 252 seconds]
jpn has joined #fedora-coreos
dwalsh has joined #fedora-coreos
saschagrunert has quit [Remote host closed the connection]
jlebon has joined #fedora-coreos
mheon has joined #fedora-coreos
bgilbert has joined #fedora-coreos
bgilbert has quit [Remote host closed the connection]
bgilbert has joined #fedora-coreos
vgoyal has joined #fedora-coreos
dwalsh has quit [Ping timeout: 252 seconds]
gursewak has joined #fedora-coreos
jcajka has joined #fedora-coreos
paragan has quit [Quit: Leaving]
plarsen has joined #fedora-coreos
<queeup[m]>
Guys I am missing "/dev/video{10,11,12}" on coreos aarch64 (rpi4). I need them for v4l2 hardware decode. How can I have that? Do I need something to install?
<jlebon>
dustymabe: yeah the conditional networking stuff was designed for first boot only
<jlebon>
something got lost i think in the move from ignition-dracut to f-c-c
<ravanelli>
jlebon: dustymabe let's discuss the cosa-ci-lib changes here? Or do you prefer to schedule a time for it?
bgilbert has joined #fedora-coreos
<jlebon>
good either way :)
<ravanelli>
jlebon: I think make sense now, remove all the `skipTest` in the cosa-ci-lib and run all tests by default. If some test doesn't work, we can add in the denylist. The issue is with the old cosa versions.
<ravanelli>
maybe we can keep a depreciate call and new
<ravanelli>
and decide if we want to backport it in cosa
jpn has quit [Ping timeout: 246 seconds]
<queeup[m]>
dustymabe: I think I am missing `bcm2835_codec` module.
<jlebon>
ravanelli: right, the `kolaTestIso` gets way simpler. it can just call `kola testiso` once and let it run all the tests. for old cosa, i'd like to backport if possible but it depends how hard it'd be. if we don't, we'd have to maintain the two paths in coreos-ci-lib and would still have the denylist problem until they go EOL
<ravanelli>
jlebon: +1
Turnikov has quit [Ping timeout: 260 seconds]
<ravanelli>
jlebon: should we at least split the 4k tests or other test, to create a separate log file?
Turnikov has joined #fedora-coreos
<jlebon>
ravanelli: this is a good topic; are the logs split out into separate dirs per test with your PR?
jpn has joined #fedora-coreos
<ravanelli>
jlebon: no
<ravanelli>
jlebon: However, I can try to split it in the PR
<ravanelli>
so we wouldn't need to work on that in the lib
<dustymabe>
jlebon: back - sorry was lunching
<dustymabe>
ravanelli: i'm down to do a video session if you think that would be useful
<ravanelli>
dustymabe: I think chat here is ok ;)
<dustymabe>
WFM
<dustymabe>
jlebon: one thing - we do run some testiso tests in parallel today.. if we have a single testiso call we lose that (because the testiso code itself runs them serially). This is OK. Just something to call out.
<ravanelli>
dustymabe: good point
<jlebon>
dustymabe: good point. ravanelli: how are the log files from a run laid out right now?
<jlebon>
does one stomp on the other?
<ravanelli>
I will need to run that again to see
<dustymabe>
honestly I think the right move is to rework the testiso code eventually to support running tests in parallel, so even if we lose parallel execution in the pipeline now, I don't see it as that much of a loss
<jlebon>
we'll get "for free" when merging into kola proper too if we don't do it earlier
<jlebon>
ravanelli: i think it's really important that the logs are available and well-separated
mei has quit [Remote host closed the connection]
mei has joined #fedora-coreos
jpn has quit [Ping timeout: 248 seconds]
jaimelm has quit [*.net *.split]
mnaser_ has quit [*.net *.split]
pej0703 has quit [*.net *.split]
mnaser_ has joined #fedora-coreos
jaimelm has joined #fedora-coreos
pej0703 has joined #fedora-coreos
Turnikov has quit [Ping timeout: 260 seconds]
jpn has joined #fedora-coreos
<dustymabe>
bgilbert: jlebon: one thing we probably should do at least is make sure the 68-azure-sriov-nm-unmanaged.rules udev rule gets into the initramfs ?
<bgilbert>
seems like it
<bgilbert>
it'd be nice if there was a better way to track these cloud-specific udev rules
<bgilbert>
feels like we've had a multi-year game of whack-a-mole
<dustymabe>
I think it started out as part of igition itself. so yeah, you'd be right I think
jpn has quit [Ping timeout: 255 seconds]
<jlebon>
cool with committing to that, but would be good to sanity-check how all this stuff is wired up again to make sure it can all work fine on subsequent boots
<jlebon>
obviously, we just found one gap at least, but there may be others
<bgilbert>
yeah. upstream Afterburn has its hostname service enabled on a lot of platforms, and none of them are running on subsequent boots, which I thought it did
<bgilbert>
I'm digging a bit to see whether we can just throw zbus at it now. the issue predates NM+dbus in the initrd
baaash[m] has joined #fedora-coreos
<ravanelli>
dustymabe: To be honestly all the ppc64le people I knew that could help are no longer at IBM. But reach the others now, we may need a BZ that I can tag some people that still around
<ravanelli>
in the Kernel team
<dustymabe>
ravanelli: i'm hoping the root cause is easy enough to determine without having to include experts, but if they are needed we'll try to find someone
<dustymabe>
we still need to actually dig in to see if we can really determine what is failing (beyond the initial investigation from jlebon)
<ravanelli>
dustymabe: what is the multipath version?
<bgilbert>
dustymabe: hmm, I'm not immediately finding a way to change the Azure hostname
<bgilbert>
dustymabe: do we want to record some action items?
<ravanelli>
dustymabe: I also found an issue with 0.8.7-7 version for RHEL9.
<ravanelli>
dustymabe: /me looks the logs
<dustymabe>
ravanelli: rawhide has device-mapper-multipath-0.9.3-1.fc38 and testing-devel has device-mapper-multipath-0.9.0-4.fc37
<dustymabe>
bgilbert: i'm trying to find the best way to stick the udev rule in the initramfs
<dustymabe>
that will solve the immediate problem
<bgilbert>
dustymabe: sure, I mean wrt the broader questions
<dustymabe>
the followups are:
<dustymabe>
1. do we actually want to run networking in the initramfs on every boot for Azure.
<dustymabe>
2. how to solve the 30s delay for azure/afterburn
<dustymabe>
for 1. I think the answer is yes based on our discussion above
<bgilbert>
2 is already reported, I'm more thinking about bugs we need to file
<bgilbert>
there's also 3. do we want to enable hostname setting on every boot
<bgilbert>
(maybe only on some platforms)
<bgilbert>
re 1, we have an inconsistency right now. I don't think there's a need for it unless we enable hostname setting on every boot, and I haven't proven that we need to do that on Azure
<bgilbert>
but we're currently doing 1 without actually making use of it
<dustymabe>
what about boot check in?
<bgilbert>
runs in the initrd on firstboot on RHCOS, and in all other cases runs in the real root
<dustymabe>
oh - ok - so boot check in is needed, just not needed in the initramfs
<bgilbert>
RHCOS only needs it in initrd on firstboot because it's a hack for the OCP installer
<bgilbert>
yeah
<bgilbert>
so maybe I file a "consider" Afterburn bug to set hostnames on subsequent, and for now we disable Azure networking on subsequent?
<bgilbert>
assuming adequate Azure CI, which, idk
<dustymabe>
WFM - jlebon?
* jlebon
reads scrollback
<jlebon>
bgilbert, dustymabe: if we table the Azure hostname setting on subsequent boot bit for now, is there any reason to have coreos-enable-network run on subsequent boots?
<bgilbert>
jlebon: I believe no
<jlebon>
ok, so is the conclusion we're leading to to restrict it to first boot for now?
<bgilbert>
(note: I have not researched the full PR/Git history)
<bgilbert>
yeah
<jlebon>
ok, SGTM
<jlebon>
i think that's a safer path. and i'd be interested if anything breaks from that that we didn't know relied on this
<bgilbert>
+1
<jlebon>
dustymabe, ravanelli: which issue are we talking about there?
<jlebon>
dustymabe, ravanelli: ok, let me know if you want me to help dig into it. i think marmijo[m] was also looking at it.
<dustymabe>
jlebon: bgilbert: ok. so i'll look into disabling coreos-enable-network on first boot (I assume we need a different implementation than in https://github.com/coreos/fedora-coreos-config/pull/2176) and also putting the udev rules in the initramfs (since they should be there anyway)
<bgilbert>
dustymabe: +1
<jlebon>
+1
<jlebon>
s/first boot/subsequent boot/
<dustymabe>
jlebon: right
* dustymabe
heading out for the day
<dustymabe>
see you tomorrow
<jlebon>
see ya!
nalind has quit [Quit: bye for now]
ravanelli has quit [Remote host closed the connection]
jpn has joined #fedora-coreos
jpn has quit [Ping timeout: 260 seconds]
daMaestro has joined #fedora-coreos
ravanelli has joined #fedora-coreos
zpytela_ has quit [Ping timeout: 268 seconds]
zpytela has joined #fedora-coreos
ravanelli has quit [Remote host closed the connection]
ravanelli has joined #fedora-coreos
ravanelli has quit [Remote host closed the connection]