16:30:56 <lucab> #startmeeting fedora_coreos_meeting
16:30:56 <zodbot> Meeting started Wed May 27 16:30:56 2020 UTC.
16:30:56 <zodbot> This meeting is logged and archived in a public location.
16:30:56 <zodbot> The chair is lucab. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:30:56 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
16:30:56 <zodbot> The meeting name has been set to 'fedora_coreos_meeting'
16:31:01 <cyberpear> .hello2
16:31:05 <zodbot> cyberpear: cyberpear 'James Cassell' <fedoraproject@cyberpear.com>
16:31:13 <slowrie> .hello2
16:31:14 <zodbot> slowrie: slowrie 'Stephen Lowrie' <slowrie@redhat.com>
16:31:16 <jlebon> .hello2
16:31:17 <bgilbert> #topic roll call
16:31:18 <zodbot> jlebon: jlebon 'None' <jonathan@jlebon.com>
16:31:23 <bgilbert> oh yeah, circular dependency
16:31:24 <bgilbert> .hello2
16:31:25 <zodbot> bgilbert: bgilbert 'Benjamin Gilbert' <bgilbert@backtick.net>
16:31:30 <lucab> #topic roll call
16:31:31 <coremodule> .hello2
16:31:33 <zodbot> coremodule: coremodule 'Geoffrey Marr' <gmarr@redhat.com>
16:31:34 <lorbus> .hello2
16:31:36 <zodbot> lorbus: lorbus 'Christian Glombek' <cglombek@redhat.com>
16:31:37 <kparal> .hello2
16:31:39 <zodbot> kparal: kparal 'Kamil Páral' <kparal@redhat.com>
16:31:42 <jdoss> .hello2
16:31:43 <zodbot> jdoss: jdoss 'Joe Doss' <joe@solidadmin.com>
16:32:14 <lucab> #chair bgilbert jdoss kparal lorbus cyberpear slowrie jlebon
16:32:14 <zodbot> Current chairs: bgilbert cyberpear jdoss jlebon kparal lorbus lucab slowrie
16:32:31 <skunkerk> .hello sohank2602
16:32:34 <zodbot> skunkerk: sohank2602 'Sohan Kunkerkar' <skunkerk@redhat.com>
16:32:46 <lucab> #chair coremodule skunkerk
16:32:46 <zodbot> Current chairs: bgilbert coremodule cyberpear jdoss jlebon kparal lorbus lucab skunkerk slowrie
16:33:28 <sumantro> .hello sumantrom
16:33:29 <zodbot> sumantro: sumantrom 'Sumantro Mukherjee' <sumukher@redhat.com>
16:33:44 <lucab> #chari sumantro
16:34:14 <lucab> * #chair sumantro
16:34:26 <lucab> #chair sumantro
16:34:26 <zodbot> Current chairs: bgilbert coremodule cyberpear jdoss jlebon kparal lorbus lucab skunkerk slowrie sumantro
16:34:36 <slowrie> lucab: third times the charm
16:35:07 <lucab> slowrie: eh, editing message from Matrix does not work well in IRC ;)
16:35:20 <lucab> anyway, let's start
16:35:31 <lucab> #topic Action items from last meeting
16:36:10 <lucab> 1. lorbus to try out OKD on our `next` stream so we can work out any kinks before switching `stable` to f32
16:36:30 <lucab> 2. cyberpear to take a look to see if we can change the libicu dep in samba-client-libs to a recommends (#452)
16:36:45 <lucab> 3. bgilbert to PR c-i and f-c-c SOP for new signing key
16:37:14 <lucab> for 1. I've seen Vadim's comment at https://github.com/coreos/fedora-coreos-tracker/issues/372#issuecomment-632174381
16:37:41 <bgilbert> #info bgilbert PR'd signing key to c-i, dustymabe PR'd to f-c-c
16:38:14 <lucab> for 2. I've seen some movement at https://github.com/coreos/fedora-coreos-tracker/issues/452
16:38:50 <lucab> so I think nothing too strange on F32, and for unicode data still going on?
16:39:55 <jlebon> yeah, not sure if it's worth pursuing that one further
16:40:13 <jlebon> (that one = item 2)
16:41:04 <lorbus> item 1 is still WIP.
16:41:11 <lucab> I'm not too concerned. Yes it's big, but it's a library and it can be dropped freely in the future (hopefully)
16:41:47 <jlebon> yeah, true
16:41:57 <lucab> lorbus: ack, I think we have the F32 switch as one of the next topics
16:42:09 <lucab> so let's just go to that
16:42:27 <lucab> #topic F32 rebase
16:42:35 <lucab> #link https://github.com/coreos/fedora-coreos-tracker/issues/372
16:43:28 <lucab> jlebon: I think you were pointing to https://github.com/coreos/fedora-coreos-config/pull/394#issuecomment-634776712
16:43:59 <jlebon> right yeah. accordign to the schedule we agreed to last week, the next testing release should be f32: https://github.com/coreos/fedora-coreos-tracker/issues/372#issuecomment-631586457
16:45:19 <jlebon> bug-wise, there's https://github.com/coreos/fedora-coreos-tracker/issues/496 which is worrisome. i'm looking into this now, though i don't think it will be made worse by the f32 move
16:46:42 <lucab> that's a regression on our side anyway, it can't get more broken
16:46:49 <lucab> (famous last words)
16:47:27 <jlebon> there's also https://github.com/coreos/fedora-coreos-tracker/issues/488, which has a backport workaround already in next-devel, so that should be ok too
16:47:46 <jlebon> apart from those, AFAIK we're in a good position
16:48:19 <lucab> jlebon: so the ostree in F32 is fixed too right?
16:48:27 <jlebon> yup, indeed
16:49:03 <lucab> I think we can merge #394 once CI is green today/tomorrow
16:49:09 <jlebon> unless anyone has any concerns with the next testing release being f32, i'd say let's merge https://github.com/coreos/fedora-coreos-config/pull/394 on green
16:49:18 <jlebon> lucab: heh :)
16:49:29 <lucab> are there any other bits to tweak for a fedora-major bump?
16:49:31 <slowrie> +1
16:49:40 <bgilbert> +1
16:49:52 * kparal notes this topic is related to https://github.com/coreos/fedora-coreos-tracker/issues/491
16:49:53 <lorbus> +1
16:50:06 <jlebon> lucab: there's the releng key signing thing we'll need to take care of
16:50:33 <lucab> kparal: you are right, I should have tackled that first
16:51:15 <lucab> jlebon: can you reference that in #372 so that we don't forget to do it
16:51:19 <jlebon> i guess we can move to that topic?
16:51:41 <lucab> yes, let's just say not technical blocker to rebase testing to F32
16:51:51 <jlebon> lucab: yup
16:51:58 <lucab> #info no technical blocker to rebase testing to F32
16:52:17 <jlebon> (by "that topic" i meant test day, to be clear :) )
16:52:20 <lucab> #topic Fedora Test day for our `next` stream (Fedora 32)
16:52:28 <lucab> #link https://github.com/coreos/fedora-coreos-tracker/issues/491
16:53:05 <lucab> kparal: here you go
16:53:15 <kparal> OK, hi all, we talked to Dusty and offered to help you run a FCOS 32 test day. It might be a good PR for you, and not only a way to find some bugs, but also identify documentation or user experience issues. There is a document to do some brainstorming about possible test cases here: https://hackmd.io/lX41BH4TSlqDGqgwr_xWVg?edit
16:53:36 <kparal> and sumantro set up a preliminary test day results page here: http://testdays.fedorainfracloud.org/events/84
16:53:56 <kparal> it links to draft testcases (or documentation directly, when we haven't managed to create testcases yet)
16:54:01 <lucab> kparal: do we have to pick a day for this or is it already set?
16:54:09 <jlebon> kparal: that's great
16:54:17 <kparal> you can pick any date you want, sumantro might have some suggestions
16:54:24 <sumantro> lucab, we need to pick a date
16:54:25 <kparal> it depends how much in a hurry you are
16:55:05 <lucab> I think we'd like to switch testing to F32 for the next release
16:55:09 <coremodule> and any suggestions on specific things to test/testcases are very welcome
16:55:10 <kparal> the testcases are not finished yet, and they will need help from your side, because you're the ones who know fcos stuff, we don't know the technical details that much :)
16:55:15 <lucab> which would be next week
16:56:07 <kparal> testing can be switched to next before or after this test day, that doesn't really matter I think (unless you want to detect some bugs before switching it to testing)
16:56:15 <kparal> but it should happen before you switch it to stable :)
16:56:32 <jlebon> hmm, maybe we should do it between f32 hitting testing and stable
16:57:09 <coremodule> as kparal said, we are not coreos experts, so your expertise on what should be tested and help developing test cases is much appreciated
16:57:10 <jlebon> having it in testing for test day would make it easier for participants
16:57:35 <lucab> I don't know which target date Dusty had in mind
16:57:40 <kparal> I already reported on issue related to this test day, so you might expect more "ux" feedback like this, which is I think beneficial: https://github.com/coreos/fedora-coreos-tracker/issues/493
16:57:42 <kparal> *one
16:58:15 <lucab> but I think we can more or less target the second half of next week, after testing release
16:58:52 <jlebon> right yeah. or even early in the week of the 8th
16:58:54 <bgilbert> kparal: +1, we need more reports of that kind
16:59:22 <jlebon> heh, we should probably fix that for test day
16:59:36 <sumantro> lucab, does 12th sound good to you all?
17:00:01 <kparal> the testcases are in fedora wiki, anyone logged in can edit them (I believe), please help us shape them up. Or you can put the steps into the hackmd document and we'll convert it to wiki. Just have in mind your target audience, it should either reference documentation or list concrete steps so that even beginners can try them (unless the testcase is for advanced/expert audience only)
17:00:43 <lucab> sumantro: I think as jlebon suggested, 8th or 9th would be better
17:01:02 <sumantro> awesome then 8th sounds great
17:01:24 <lucab> sumantro: because I guess the stable switch would be around the 16th
17:01:51 <lucab> #info tentatively scheduled test day on June 8th
17:02:40 <lucab> anything else on this topic?
17:03:03 <lucab> I guess a general request for "come up with more manual test scenarios"
17:03:30 <kparal> I'll also include a test case for some exploratory testing (do what you want) and documentation review
17:03:47 <kparal> so it doesn't need to strictly be just manual testing in the narrow sense
17:04:13 <kparal> but yes, we need ideas in the hackmd document and implementation of those ideas in wiki or again in the hackmd document
17:04:41 <lucab> ack
17:04:47 <kparal> thanks, that's all from me :)
17:04:53 <jlebon> maybe we can split the testcases among us to fill out?
17:05:03 <jlebon> anyway, we can discuss that after :)
17:05:10 <lucab> kparal: thanks!
17:05:41 <lucab> last one is from my side
17:06:19 <lucab> #topic network interface name
17:06:30 <lucab> #link https://github.com/coreos/fedora-coreos-tracker/issues/484
17:06:53 <lucab> (I should probably retitle this ticket with something better)
17:07:09 <lucab> I was digging into this yesterday/today
17:07:35 <lucab> I think we made a mistake some time ago and we didn't realize for quite a bit
17:08:03 <lucab> we assumed that `.link` units are used by systemd-networkd
17:08:24 <lucab> and we purged both sides from FCOS images
17:09:04 <lucab> it turns out both code and docs agree that they are used by udev, to (among other thing) rename interfaces in a persistent way
17:09:23 <lucab> and that's my investigation so far
17:09:51 <lucab> fixing it for new images can be done by just removing a -config line
17:10:21 <lucab> I think there is consensus that it may break static config on bare-metal on auto-updates
17:10:47 <lucab> so we should at least try brainstorming a bit more
17:11:57 <lucab> my gut feeling is that 1) we want to eventually use persistent names for new nodes 2) we care a bit about trying not to break everything if we can 3) we are not in a super-hurry to solve this (i.e. not next release)
17:12:03 <lucab> s/next/upcoming/
17:12:33 <lorbus> I'd like to argue with 3
17:13:15 <lorbus> OKD is planning to go GA in 1-2mo. I'd like to solve this before that happens
17:13:39 <lucab> 3) assumes that we can keep old nodes on old names, and new nodes on new names
17:13:54 <cyberpear> agreed, would be best to solve in time for OKD GA, or otherwise, patch it OKD side if necessary
17:14:32 <lucab> does OKD itself care about what's the name of an interface?
17:15:14 <lorbus> not really, but I'd rather not find out about any implications of this after GA
17:15:56 <lucab> fair
17:16:13 <jlebon> it makes it harder for OKD users to write network configs i suppose
17:16:25 <lucab> so the ballpark ETA for this is, after F32 but before OKD GA
17:16:49 <jlebon> lucab: i think we had a suggestion that should work for the majority of existing nodes, right?
17:17:54 <lucab> jlebon: I still need to sleep on it a bit more, but yes some kind migration script (writing a file) and a barrier to force node through that
17:18:47 <jlebon> hmm, actually, i think just adding a net.ifnames=0 karg in the migration script would cover it, no?
17:19:07 <cyberpear> ^ so we may end up w/ a barrier for the F32 upgrade after all
17:19:09 <lucab> jlebon: kargs are a bazooka, but yes
17:19:14 <jlebon> we can chat about it in the other channel if you prefer
17:19:48 <lucab> cyberpear: subtly different, we place barrier whenever we need, not explicitly on major bumps
17:20:00 <jlebon> we did something similar for cgroupsv1/2: https://github.com/coreos/fedora-coreos-config/pull/238
17:20:16 <cyberpear> right; just happens to be adjacent to a major bump here
17:20:31 <lucab> cyberpear: if they happens to be the same release, it's fine, but a coincidence
17:20:42 <lucab> jlebon: yes, I'll keep adding stuff to the ticket
17:20:58 <jlebon> +1
17:21:30 <lucab> lorbus: when is OKD tentative GA? (roughly)
17:21:52 <lorbus> lucab: Mid July
17:22:22 <lucab> ack
17:23:26 <lucab> #info barrier for netnames pinning script should better be in place for mid-July for OKD GA
17:23:43 <lucab> ok, I think that's all
17:23:51 <lucab> #topic Open Floor
17:24:09 <cyberpear> wrt https://github.com/coreos/fedora-coreos-tracker/issues/452 and the extra 32MiB added: supposedly https://fedoraproject.org/wiki/Changes/RemoveExcessiveLinking should have helped, but that was an F30 change, so apparently didn't actually help
17:24:27 <cyberpear> ^ just a comment, nothing further from me on it
17:24:48 <lucab> lorbus: isn't minimization also hitting this?
17:25:36 <cyberpear> once/if they look at SSSD, or other packages that are impacted, but I don't know SSSD has been a target of Minimization specifically yet
17:25:43 <lorbus> hm I'm unaware of this
17:26:29 <jlebon> cyberpear: when i briefly looked at it, it did seem like libsmbclient was actually using the library and not just linking unnecessarily
17:26:30 <lucab> perhaps it's worthy bouncing it there just as a FYI
17:27:10 <jlebon> +1
17:27:29 <lorbus> +1
17:27:55 <cyberpear> minimization: https://pagure.io/minimization/issues https://docs.fedoraproject.org/en-US/minimization/
17:27:58 <lucab> unless anything else, I'll close in a few seconds
17:28:16 <lorbus> thanks all!
17:28:23 <cyberpear> thanks!
17:28:35 <lucab> #endmeeting