16:31:13 #startmeeting fedora_coreos_meeting 16:31:13 Meeting started Wed Mar 31 16:31:13 2021 UTC. 16:31:13 This meeting is logged and archived in a public location. 16:31:13 The chair is bgilbert. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:31:13 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:31:13 The meeting name has been set to 'fedora_coreos_meeting' 16:31:16 #topic roll call 16:31:23 .hi 16:31:25 cyberpear: cyberpear 'James Cassell' 16:31:25 .hello siosm 16:31:28 travier: siosm 'TimothΓ©e Ravier' 16:31:52 .hello jaimelm 16:31:53 PanGoat: jaimelm 'Jaime Magiera' 16:31:56 .hello2 16:31:58 jlebon: jlebon 'None' 16:32:12 .hello2 16:32:13 dustymabe: dustymabe 'Dusty Mabe' 16:32:37 #chair dustymabe bgilbert 16:32:41 #chair cyberpear travier PanGoat jlebon dustymabe 16:32:41 Current chairs: PanGoat bgilbert cyberpear dustymabe jlebon travier 16:32:41 * sumantro is here 16:32:47 #chair sumantro 16:32:47 Current chairs: PanGoat bgilbert cyberpear dustymabe jlebon sumantro travier 16:32:48 .hello2 16:32:49 πŸ‘‹ sumantro 16:32:49 slowrie: slowrie 'Stephen Lowrie' 16:32:55 dustymabe, hey!! 16:33:05 #chair slowrie 16:33:05 Current chairs: PanGoat bgilbert cyberpear dustymabe jlebon slowrie sumantro travier 16:33:13 .hello jasonbrooks 16:33:14 jbrooks: jasonbrooks 'Jason Brooks' 16:33:41 #chair jbrooks 16:33:41 Current chairs: PanGoat bgilbert cyberpear dustymabe jbrooks jlebon slowrie sumantro travier 16:35:03 Since I don't have privs to label, I'll just mention that I have an item regarding https://github.com/coreos/fedora-coreos-docs/issues/257 16:35:15 .hello2 16:35:15 walters: walters 'Colin Walters' 16:35:17 bgilbert: so am I driving this? 16:35:24 #chair walters 16:35:24 Current chairs: PanGoat bgilbert cyberpear dustymabe jbrooks jlebon slowrie sumantro travier walters 16:36:01 * jlebon assumes yes 16:36:05 #topic Action items from last meeting 16:36:13 πŸ˜… 16:36:14 jlebon: yep 16:36:23 so last week we had the joint FCOS/podman meeting led by travier 16:36:36 there's a bunch of things that came out of that, but we don't have to go over them here 16:36:52 thanks a lot travier for managing that, it was very productive i think! 16:36:57 I don't think we have any major decision to take right now 16:36:58 recording posted from last week? 16:37:16 travier: one thing that came out was migration to cgroups v2 16:37:16 #link https://www.youtube.com/watch?v=9xOEroULZ5w 16:37:23 πŸŽ‰ 16:37:24 did we manage to land that in the f34 rebase? 16:37:53 right, worth highlighting that no decisions were taken because we want to do it through the community mtg 16:37:57 dustymabe: that's the next topic :) 16:38:06 ahh k 16:38:17 anything else to add on this subject otherwise? 16:38:38 I still have on my todo list to summarize the update & size requests into issues 16:38:48 but this is not major 16:38:56 travier++ 16:38:58 #link https://github.com/coreos/fedora-coreos-tracker/issues/768 16:39:22 #info the FCOS/podman joint meeting went well 16:39:43 #action travier to summarize outcome in https://github.com/coreos/fedora-coreos-tracker/issues/768 16:39:51 ok, let's move on 16:40:11 #topic cgroups v2 strategy 16:40:14 #link https://github.com/coreos/fedora-coreos-tracker/issues/292 16:40:26 so `next` is now on f34 16:40:38 it has new moby-engine, which supports cgroupsv2 16:41:05 so we can now move next to default to cgroupsv2 16:41:23 this is what https://github.com/coreos/fedora-coreos-config/pull/910 does 16:41:52 i'm proposing doing this for the next release 16:42:11 (the next release of next that is) 16:42:15 We also need to update the docs for this change. 16:42:31 we'll need to publicize the change as well 16:42:34 +1 16:43:03 to be clear, the strategy is that upgrading nodes will stay on v1, and only new nodes will default on v2 16:43:10 anybody think of reasons to not do this? 16:43:26 we'll want docs for upgrading nodes which want to move to v2, and new nodes which want to move to v1 for whatever reason 16:43:32 ^^ 16:43:33 dustymabe, publishing the change sounds good 16:43:51 #link https://github.com/coreos/fedora-coreos-docs/issues/267 16:43:55 just filed that one 16:44:58 travier: let's try to do that this week and send an email linking to it, so there's enough advance notice for the release in two weeks 16:45:11 Will do 16:45:42 what list do you send info on ? 16:46:08 bgilbert: how do you feel about timing? i know you like more lead time usually :) 16:46:11 copperi_: coreos-status and coreos-devel 16:46:15 https://lists.fedoraproject.org/archives/list/coreos-status@lists.fedoraproject.org/ 16:46:27 https://lists.fedoraproject.org/archives/list/coreos@lists.fedoraproject.org/ 16:46:45 Thanks 16:47:09 jlebon: eh, I don't think we need a lot of lead time to change _next_ in a way that doesn't affect upgrades 16:47:28 +1 agreed 16:47:30 jlebon: from my perspective I think we're basically free to do whatever we want on next; the stream itself disappears/reappears so it's unlikely that anyone is expecting consistent behavior on it if they're running nodes there 16:47:32 but a couple months for stable would be nice 16:48:17 slowrie: to clarify, it doesn't disappear, it's just that it matches testing most of the time 16:48:24 * dustymabe wonders if it would be useful to have a console-login-helper-message that tells you if you are on cgroups v1 or v2 16:48:29 dustymabe: +1 16:48:43 dustymabe: +1 16:48:59 dustymabe: +1 16:49:00 :0 16:49:02 jlebon: did that change; I roughly recall that at one point we just didn't publish new images for it and existing nodes pulled updates from testing 16:49:05 oops meant :) 16:50:06 dustymabe: maybe just if you're on v1 ? 16:50:33 jlebon: +1 16:50:41 yeah, I don't mind it for both for now, but I see why that would be useful 16:51:22 slowrie: nahh, we've always rebuilt it and published its own artifacts 16:51:59 i can see it being helpful for upgrading nodes, but seems odd to default to emitting a message like "hey, you're using a default!" otherwise 16:52:52 we can bikeshed this in tickets maybe :) 16:52:58 yep 16:53:08 ok, anything else on this? 16:53:55 cool cool 16:53:59 ok next topic 16:54:07 #topic Consider alternating video/IRC meeting for FCOS group meetings 16:54:13 #link https://github.com/coreos/fedora-coreos-tracker/issues/779 16:54:23 travier: want to discuss this? 16:54:53 Sure 16:55:21 Last week video meeting went rather well so this is a suggestion to move some meeting to video chat 16:55:42 Probably not all meetings but we can start every other week or monthly or some other frequency 16:55:46 wdyt? 16:56:08 a video meeting is much higher-bandwidth for the participants and much worse for the non-participants 16:56:17 an hour-long recording is basically inaccessible 16:56:35 agree 16:56:40 if we'd like to try this, we should make sure to post minutes, preferably with timestamps for each topic 16:56:41 The IRC meetings have a lot of benefits, but so do video. Different tradeoffs. I'd prefer to keep most meetings like they currently are 16:57:01 maybe quarterly, sort of a recap and see people meeting. 16:57:09 maybe once a month or bi-monthly video meetings with targetted goals? 16:57:22 once a month sounds good 16:57:28 +1 to trying once/month 16:57:31 * cyberpear listens to the recording on 2x speed :P 16:57:40 +1 1/month 16:58:08 I'd personally lean towards having it be more of a special setup where if we have things that need the extra bandwidth we go through that process but I can live with whatever 16:58:10 irc meetings are easier to read but video can bring other benefits 1/month 16:58:32 slowrie: +1 that's where i lean too i think 16:58:36 +1 16:58:38 slowrie: +1 - I was just thinking if we didn't have a healthy set of topics for video meeting then it might be a bit lean 16:58:59 best to record any video meeting IMO (Workstation WG don't record theirs, which is annoying) 16:59:08 there are also different accessibility limitations. IRC is obscure by modern standards, and conversely, some folks may not want/be able to participate in a live meeting 16:59:37 You can't play back IRC while working on something else. 16:59:50 you can skim the chat log though 16:59:52 well, you can with text-to-speech, but it would be weird :) 16:59:59 right, but that's active 17:00:06 not passie 17:00:08 passive* 17:00:17 chat logs have highlights to make this relatively fast 17:00:24 all the #info and stuff 17:00:33 Having be "when it's needed" is interesting but needs to be planned in advance 17:01:00 which makes it harder :/ 17:01:06 travier: Yes, I'd assume it'd be done when something is marked as a meeting topic and it's known it's particularly contentious / needs a lot of additional bandwidth for the discussion 17:01:26 Sometimes the good stuff is not what's noted as a bullet point but what went in between. Being new here, I'll defer. 17:01:27 yeah, a good candidate would be if we invite another team to discuss something 17:01:31 but I agree that having a video meeting for nothing is not great 17:02:04 Or a special invite to FCOS users to meet the working group folks. 17:03:03 one option is to just fill time with "bug scrub" like activities if we have nothing else 17:03:20 we certainly don't do enough of that today 17:03:23 those can be effective. We do that occasionally with OKD 17:03:38 what we could do is schedule them as needed, but call it out in the previous week's IRC meeting 17:03:55 +1 jlebon 17:03:58 i'm supportive of trying out new things :) 17:04:34 so e.g. i can say "i'd like to discuss cgroupsv2 in a video meeting next week" and if there's agreement, it's scheduled 17:04:36 the bug scrub and maybe doc scrub would be great. 17:04:39 It is good to try 17:05:26 e.g. travier has been making lots of doc changes the past few days. It would be helpful to have a video meeting to go over what he, or the group, envisions. 17:06:07 totally cool with just trying it out once a month and see how it goes 17:07:11 πŸ‘ 17:07:15 once a month doesn't sound onerous 17:07:19 travier: you want to try it out? 17:07:23 first meeting of month (just to make it easy?) 17:07:34 though, that would be next week :) 17:07:35 sure 17:07:39 +1 17:08:16 +1 17:08:17 #proposal Setup a monthly video meeting for the Fedora CoreOS group 17:08:21 first meeting of the month SGTM 17:08:32 +1 17:08:34 yes 17:08:35 ack 17:08:56 #proposed Setup a monthly video meeting for the Fedora CoreOS group, planned for the first meeting of the month 17:09:23 travier: we should shoot an email to the list about this too 17:09:36 sure 17:10:05 +1 to latest 17:11:21 #agreed Setup a monthly video meeting for the Fedora CoreOS group, planned for the first meeting of the month 17:11:27 (based on previous =1) 17:11:30 +1 17:11:51 cool, thanks travier! 17:11:56 let's move on to the next topic 17:11:56 We can reconsider if that does not turn out well 17:12:26 this is it for tracker issues, but PanGoat you had something? 17:12:52 yeah... I have some basic ign and text for NFS mounts. I'm unclear where to put them. 17:13:05 Should all of these types of requests get put in the tutorials? 17:13:13 ok, let's hold that for a second 17:13:39 (and discuss it in open floor) 17:13:43 sure. 17:13:52 are there any other issues anyone wants to discuss before we move to open floor? 17:14:56 ok doesn't look like it :) 17:14:59 #topic Open Floor 17:15:17 PanGoat: right ok, so you're asking where to put the NFS mount example? 17:15:31 right, and similar requests that come in. 17:15:37 "how do I...?" 17:16:06 I would put it under User Guides -> System Configuration 17:16:36 i.e. like: https://docs.fedoraproject.org/en-US/fedora-coreos/managing-files/ 17:16:53 i think if it fits well within an existing category, i'd put it there. for this one, maybe https://docs.fedoraproject.org/en-US/fedora-coreos/storage/ ? 17:17:00 PanGoat: if you wanted to have a FAQ entry "How do I mount NFS?" and then point to the user guide 17:17:17 that was my thought. being new around here, not sure about criteria for what shoudl be a tutorial vs. just an entry in sys config 17:17:40 Yeah, storage makes sense. 17:17:44 I'd prefer not to add more faq entries pointing to permanent docs 17:17:51 jlebon: that could work too.. along with a FAQ entry that links to the subsection in storage 17:17:56 travier: oh? 17:18:02 travier: what's your thinking 17:18:15 FAQ entries should be more about specific question rather than doc examples 17:18:29 (to me) 17:18:43 otherwise we will end up with all examples as faq entries 17:18:51 and our faq is already very large 17:18:57 i think a major issue related to this is that AFAICT there's no search function 17:19:05 i was thinking if the user couldn't find the information otherwise, they go to FAQ and then the FAQ points to the right location in the docs 17:19:14 jlebon +1 17:19:24 jlebon, right - search would help 17:19:57 I'd argue that the last tutorial about system user units could be converted into a standalone page or merged into a new system page 17:20:03 +1 to minimizing FAQ size 17:20:07 travier: the answer to the question in the FAQ would just be See [docs](https://docs.fedoraproject.org/en-US/fedora-coreos/storage/#NFSMounting) 17:20:32 travier: I actually wrote it to be short and sweet, but put it in the wrong place because I didn't realize each one was strung together into a thread. 17:20:34 dustymabe: sure, but does that help? 17:20:48 we should ask the docs folks if there's a way to add a built-in search tool 17:20:48 I think so, when you're clicking around and can't find what you're looking for 17:21:09 I'm happy to move the user units to a separate page. 17:21:17 go to FAQ - Ctrl-F - search NFS - Find FAW entry with link 17:21:17 PanGoat: no worries, I updated it to make it look like the other but now that we also have an NFS example, maybe we simply need a system (system and user) page for examples 17:21:20 FAQ* 17:21:26 dustymabe: IMO it'd be better to fix the underlying problem 17:21:29 travier: right, right 17:21:44 systemd* page 17:21:45 bgilbert: probably - I don't have any solutions there though :) 17:22:18 yeah, systemd page would be good and maybe reference from the storage page 17:22:38 I would prefer the FAQ to focus on history/general issues/specific FCOS choices and keep all examples in docs pages 17:22:44 as a user, i would look in the storage place for NFS 17:22:56 jlebon: good point 17:23:30 search + metadata. If only. 17:24:02 a site:https://docs.fedoraproject.org/en-US/fedora-coreos/ xyz google search helps 17:24:24 so let's just put it there for now and dig into the discoverability issue separately? 17:24:34 good plan 17:25:07 PanGoat: would you be able to to create a new ticket about the latter? 17:25:18 yep 17:25:32 awesome, thanks! 17:25:53 #action PanGoat to file an issue about FCOS docs discoverability 17:26:10 anything else? 17:26:23 we're renaming FCCT to Butane 17:26:40 bgilbert: +1 17:26:42 πŸŽ‰ 17:26:48 nice 17:26:57 "FCCT" is unwieldy and the tool applies to more than just Fedora CoreOS 17:27:10 the configs will now be "Butane configs" 17:27:25 the change should be backward-compatible, aside from the command name 17:27:27 re. extension, do we actually want to recommend a custom one vs just .yaml ? 17:27:37 jlebon: was going to use .bu, yeah 17:27:49 it matters for editors and website highlighters 17:28:00 "Hey, do you have the latest version of C4H10 installed?" 17:28:10 (bgilbert: to be clear, i'm arguing for not recommending a custom one) 17:28:13 πŸ˜„ 17:28:15 PanGoat: saves one character :-P 17:28:26 :) 17:28:48 yeah, I almost wonder if our ignition examples should use .json 17:28:53 jlebon: the config doesn't actually say it's a Butane config, so unless you know to recognize variant/version from context 17:29:06 it's not obvious that some random YAML is a Butane config 17:29:07 example.bu.yaml 17:29:14 example.ign.json 17:29:16 I'm in favor of things saying what they are? 17:29:32 bgilbert: +1 17:29:48 good point 17:30:15 jlebon: thoughts? 17:31:06 bgilbert: meh, not strongly opposed i guess. it's slightly more friction for the editor experience 17:31:17 it's true 17:31:19 I have mixed feelings. We will have to get our extensions added everywhere but this is mostly a bu=yaml and ign=json change 17:31:27 it's something i've thought about multiple times for ignition too when hacking locally 17:31:42 dustymabe's .bu.yaml and .ign.json is a nice compromise 17:32:10 right, users can do whatever they want. but we could change our examples to match ^^ 17:33:05 I'm okayish with the double extension 17:33:21 it's not a code change in any event. jlebon, maybe want to file a ticket for discussion? 17:33:48 ack sure 17:33:52 πŸ‘ 17:33:58 while we're bikeshedding 17:34:18 dustymabe: better be quick, we're over time :) 17:34:19 I'm sure .but (vs .bu) was considered and discarded for probably obvious reasons? 17:34:25 it's a change to get into other projects but a rather simple one. Doing editorconfig and some big others should be enough 17:34:29 I do like the 3 letters and the symetry with .ign, though 17:34:40 at least it's not .butt 17:34:45 :D 17:34:50 dustymabe: in my head, yes. I didn't bother writing that process into a ticket :-) 17:35:35 EOM 17:35:54 and PanGoat, I did actually consider c4h10 O:-) 17:36:34 ha 17:36:43 heh, that'd be painful to type 17:36:55 ok, going to close this meeting in 45s unless there's anything else 17:37:00 +1 17:37:19 thanks jlebon, thanks everyone 17:37:40 #endmeeting