<fifofonix>
fyi I've stopped/disabled zincati on my testing fleet because the CIFS SELinux related issues still in fedora38 would be problematic. only mentioning it because i'm aggressive about updates generally and it is rare I push them off.
<dustymabe>
fifofonix: I appreciate that. I added an update to the BZ
<fifofonix>
dustymabe: investigating another odd f38 issue. i think what i'm seeing is that once a node has updated to f38 it is unable to do another zincati update to a further f38 release. i'm suspecting that crypto settings of mirror site are not sufficient.
<fifofonix>
dustymabe: thought it was corp proxy related or due to FIPS or NEXT crypto settings but I've just eliminated both of those scenarios with an AWS-based DEFAULT setting node.
<dustymabe>
fifofonix: hmm. if that were true then people wouldn't have been able to update all along the `next` releases
<fifofonix>
commissioning new nodes obviously works fine. and the update of an existing next node works fine to f38. but i wonder whether anyone would have noticed they haven't been receiving additional next updates.
<dustymabe>
fifofonix: are you on `aarch64` by chance?
<dustymabe>
fifofonix: if you can repro the update issue I'd be very interested (we have a rollout of testing coming up here in the next 30 minutes that we could stop if there is a problem we need to investigate)
<dustymabe>
cc jlebon
<dustymabe>
fifofonix: any chance you have any logs?
<fifofonix>
I enabled -vvv on zincati but nothing more than timeout error shown with regular settings.
<fifofonix>
Apr 18 14:11:52 d-pdm-mini-0-1 zincati[2053]: [INFO zincati::update_agent::actor] target release '38.20230414.1.0' selected, proceeding to stage it
<fifofonix>
Apr 18 14:12:22 d-pdm-mini-0-1 zincati[2053]: error: While pulling 7453dda78c55cdbad93d949093d539fc53f163a39eeb0017452fc8c3be50a209: No valid mirrors were found in mirrorlist 'https://ostr>
<fifofonix>
(sorry for pasting here like that - expediency)
<fifofonix>
And if I do verbose curl to that I see TLS1.2
<dustymabe>
fifofonix: you're seeing this in AWS?
<fifofonix>
yes.
<fifofonix>
(i mean i initially had it on premises but I think i've reproduced on AWS with no crypto mods)
<fifofonix>
the last time i saw this exact behaviour was when originally migrating from CoreOS to FCOS. i had tried to use NEXT crypto but could not get upgrades to work so fell back to FIPS at the time.
<dustymabe>
seems like the release notes never run
bgilbert has joined #fedora-coreos
<darknao>
I notice that too, something it doesn't load correctly, but if you click on any buttons (stable/testing/next for instance) it loads fine
<darknao>
sometime*
<dustymabe>
darknao: ok I see that it does sometimes load
<dustymabe>
maybe the problem is that it's trying to load all the release notes (i.e. a lot of data)
<dustymabe>
i think our old page only loaded 5 (but it would be nice to have an option to "load more")
<dustymabe>
darknao: can the old page(s) still be accessed somewhere?
<darknao>
it only loads a few release notes at start, then load more as you scroll
<darknao>
I think the issue is just it's trying to display something that is not loaded yet, then crash
<dustymabe>
fifofonix: i just launched in AWS (vanilla 38.20230408.1.1 node with almost no configuration) and it seems to be talking to zincati fine https://paste.centos.org/view/a406c3b6
<darknao>
as far as I can tell, the old getfedora site is not accessible anymore
<darknao>
but I'll see if I can get this fixed in priority
<dustymabe>
darknao: there is one more feature that we'd like to restore to the release notes page (it would be much easier to show you if the old site was still accessible)
<fifofonix>
dustymabe: apologies but i think this is my issue after all. i just this second got a node working. thought i had removed all proxy config but I had left rpm-ostreed.service proxy-configured. still don't understand this on my side but i'm going to investigate more.
<dustymabe>
we used to summarize the changed packages and also allow you to expand the full list of packages in a particular build/release
finsight[m] has quit [Quit: You have been kicked for being idle]
paragan has quit [Quit: Leaving]
samuelbernardo has quit [Quit: ZNC 1.8.2 - https://znc.in]
<jlebon>
dustymabe: to answer your question, yes I think I have seen it on rawhide (is that where you saw it too). looks scary so we should dig into it at some point. could you file an issue when you have a chance?
<jlebon>
saw it too?*
c4rt0 has quit [Ping timeout: 240 seconds]
samuelbernardo has joined #fedora-coreos
<dustymabe>
jlebon: I was doing a local build and `cosa run` and saw it
<baude>
bgilbert, we will keep looking, but this just does not exist
<bgilbert>
:-(
<baude>
anyways, your head didn't blow up; that's enough for me rn
<bgilbert>
heh
<baude>
bgilbert, ok, file writing now works ... first try, how abouts that
<baude>
i even put debug stuff in there expecting to debug failures ... lol
<bgilbert>
nice!
<baude>
i had to export a func in libhvee so perhaps by tomorrow I will have the PR updated
<bgilbert>
+1
<baude>
i.e ;libhvee first; and then cut a release
jdoss_ is now known as jdoss
<jdoss>
Is there a way to make rpm-ostree fetch new refs? I pushed an update to my container layer and the registry and podman pull are pulling the new ref but rpm-ostree is barking about error: Old and new refs are equal
<jdoss>
I got it to fetch the new refs by switching to a different tag and the back to my stable tag. Weird.