17:59:24 #startmeeting 17:59:24 Meeting started Tue Oct 13 17:59:24 2020 UTC. The chair is paddatrapper. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:59:24 Useful Commands: #action #agreed #help #info #idea #link #topic. 17:59:30 wow a wouter 17:59:41 agenda: http://deb.li/oNCD 17:59:46 #topic roll call 17:59:51 o/ 17:59:52 * highvoltage rolls 17:59:56 * wouter calls 17:59:59 0/ 18:00:07 \o 18:01:01 Please add anything you wish to talk about to the agenda - just created one for today 18:01:19 #topic DC20 Videos 18:02:08 anyone know what the status of the review issues is? 18:02:15 * pollo hasn't done any video review since last week 18:02:19 I reviewed 2 of them but I kind of thing we're going about it the wrong way 18:02:25 still haven't looked at them 18:02:34 wouter: Can you help out with the issues here? 18:02:42 highvoltage: why? 18:02:46 (I think instead of having core video team members reviewing it we should do recruiting for that instead or at least try) 18:03:08 from my PoV the problem isn't the reviewing, it's fixing the issues 18:03:11 Sure, we just need people to do that, whoever they are 18:03:17 and volunteers won't help unless their work is useful 18:03:25 tumbleweed: I haven't looked at the list of issues, and it's in topic (which is currently hidden) 18:03:39 #link http://deb.li/3I0Ms 18:03:48 checking 18:04:19 from what I've seen on the sheet most of the videos reviewed so far seemed fine? how would it be not useful to review more of those? 18:04:31 it would be useful 18:04:47 wouter: if spinning up dc20 sreview to fix those issues is too much work, please say so 18:05:08 We can always fix them in a nonlinear video editor using the raw files 18:05:57 pollo: it's not *that* much work, but I'll be on holiday in two days, so I won't have time until I'm back in two weeks 18:06:19 pollo: if it can wait until then, I'll hack up a perl script to conver the db dump so it can be imported in vittoria 18:06:30 if not, a nonlinear editor might be a better idea 18:06:34 what needs to be converted? 18:06:40 I think it would be less work overall to use a nonlinear video editor then 18:06:57 There are not tons of problems atm 18:07:09 +1 18:07:12 tumbleweed: the primary keys start again from zero, so you have to bump them up to whatever vittoria has available, while keeping track of the foreign keys 18:07:23 wouter: can you give us a set of params for encoding? 18:07:27 it's not *that* hard, but it needs to be done 18:07:38 can't you just blow away the DB and start a new one? 18:07:46 then those PKs are available 18:07:59 tumbleweed: I... suppose I could, but that feels wrong ;-) 18:08:10 seems like the right way to approach restoring backups 18:08:17 hacking the backup to restore into a dirty DB feels wrong to me 18:08:18 yeah, I guess we can do that 18:08:34 then only the filenames may need to be adjusted, but that should be fine 18:08:45 is the config there? 18:08:58 if I want to do this, is everything I need somewhere I can find on vittoria? 18:09:10 in the tarball that I copied from the review host 18:09:16 I'll see if I can find the time tomorrow 18:09:30 if it's too much work, I'll figure out what needs doing and will let you know 18:09:40 does that work? 18:09:44 that works 18:09:51 after all, I meant to talk to you 2 weeks ago, and haven't 18:09:56 :) 18:10:09 well, I meant to be in the past X meetings, and haven't, so don't stress 18:10:27 we do want to get this done, of course... 18:10:32 sure 18:10:34 #action wouter to try restore the sreview backup to vittoria, and let tumbleweed know what needs doing if unable to get the time 18:10:37 esp before the minidebconf 18:10:45 right 18:10:48 I attended Arch Conference 2020 over the weekedn 18:11:07 highvoltage: can that be in AOB? 18:11:16 tumbleweed: I take it no progress on YouTube and PeerTube uploads? 18:11:19 they have promised and committed to have all the videos released by the time the COVID-19 epidemic has ended, so hopefully we can at least do better than that :) 18:11:28 lol :) 18:11:32 I'm sure we can 18:11:39 paddatrapper: yeah. I'll poke at them *right now* 18:11:50 #info no progress on YouTube and PeerTube uploads so far 18:12:06 #action tumbleweed to poke at Youtube and peertube uploads *right now* 18:12:07 :P 18:12:22 #topic Miniconfs - Infra state 18:12:47 debian.ch said they'll set up the groups today 18:13:00 #info debian.ch should set up a video team group today 18:13:14 don't think it's done yet, but can follow-up if we haven't heard back from them by tomorrow 18:13:27 (I know the admin has been traveling maybe he's just still tired) 18:13:27 thanks for that! 18:13:32 The current infra doesn't list monitoring 18:13:35 what are our deadlines here? 18:13:35 Do we need it? 18:13:42 #action highvoltage to follow up if we haven't heard back from them by tomorrow 18:13:42 monitoring is useful, yes 18:13:48 pollo: good point 18:14:04 Should still be under the budget approved though 18:14:07 pollo: we have some wiggle room in the budget if we need another vm for monitoring 18:14:08 we can probably co-host the monitoring with something else 18:14:08 doesn't need to be a large machine 18:14:16 +1 18:14:40 tumbleweed: not if you want to do something like prometheus, but meh 18:14:51 (it's insane how many resources that wants...) 18:15:05 the prometheus vm at dc20 was hardly taking any strain 18:15:06 unless they fixed it since I looked at making it be better 18:15:09 #info initial infra list missed out monitoring 18:15:12 but those machines were beefy 18:15:31 were we monitoring the monitor? 18:15:35 on the scale of debconf infra prometheus shouldn't be a problem at all 18:15:39 tumbleweed: did that prometheus setup come from ansible? 18:15:43 it did 18:15:46 olasd: fair point 18:15:54 so it should be fairly trivial to implement again? 18:15:58 yes 18:16:01 The graphs for dc20 also have been backed uo 18:16:03 definitely 18:16:14 so is there anything left to talk about on monitoring right now? 18:16:36 nope 18:16:42 #topic Miniconfs - Setup 18:16:58 How is this different from Infra state? 18:17:24 we might want to discuss if we want to do things differently from dc20? 18:17:32 for one, I'm not going to do the insane SReview stuff anymore 18:17:39 because terrible 18:17:49 sounds like a very solid plan 18:17:53 :) 18:17:54 you're going to do even more insane SReview stuff? 18:18:08 * pollo is OK with trying to use the stream dumps for the miniconfs 18:18:13 highvoltage: no, I'm going to revert to how SReview was meant to be used 18:18:34 as in, do what we did previously, and don't try to hack in the unmodified prerecordings 18:18:46 #agreed use stream dumps for recordings 18:18:51 for a dozen or so talks I think it's fine taking a simpler approach 18:18:53 pollo: can you expand on that? 18:19:13 you don't want to record on voctomix? 18:19:24 tumbleweed: what wouter said basically, use the voctomix recording for everything 18:19:31 #undo 18:19:31 Removing item from minutes: 18:19:39 #agreed use voctomix output for recordings 18:19:39 Yes, bad terminology 18:19:54 will the streams still be recorded elsewhere as a backup? 18:19:54 OK, +1 18:19:58 yes 18:20:00 ok cool 18:20:12 Then how do we handle pre-recordings collection? 18:20:30 paddatrapper: we can still use the upload functionality in SReview 18:20:34 ah ok 18:20:44 #info we can still use the ipload funcionality in SReview 18:20:50 It's just going to be a separate thing, and it won't be used at all for the released recordings 18:20:51 #undo 18:20:51 Removing item from minutes: 18:20:57 #info we can still use the upload funcionality in SReview 18:21:04 * wouter likes ipload actually 18:21:11 heh 18:21:22 sounds like a traffic test tool or something ;) 18:21:55 Any other changes we would like to implement for the miniconfs? 18:22:33 I guess we'll only be using one stream, but beyond that can't think of anything 18:23:00 #topic Miniconfs - Training 18:23:18 anything more to add on this one? 18:23:27 I don't think we can do much until we have infra? 18:23:33 yeah 18:23:34 I suppose it should be possible to have a more or less working stack by the end of the month? 18:23:54 if the groups happen and someone can create the vms, sure 18:24:00 I mean, enough bits to do some training on 18:24:18 so we'd probably need a call for volunteers too 18:24:40 I think we should wait another week 18:24:40 or we could just train the people from mdco3+mdco4 18:25:02 another week before discussing this further or another week after the end of the month for training? 18:25:02 pollo: so a week after stack up have the call for volunteers? 18:25:24 Another week before sending a call for volunteers 18:25:46 Nothing is up yet, too much incertitudes 18:25:59 incertitudes? 18:26:13 on, uncertainties? sorry 18:26:17 Yes that 18:26:32 paddatrapper: ok I think it would be good to move on 18:26:34 sorry, didn't mean to be pedantic, just didn't understand :) 18:26:55 #topic Any other Business 18:27:28 only that I'll be off for two weeks, but I already said that :) 18:27:35 highvoltage: how was the CCC stack for the arch conf? 18:27:53 leaving on thursday, will be back on the 25th or thereabouts 18:28:13 wouter: enjoy the break 18:28:20 paddatrapper: thanks, I'll need it 18:28:26 first proper time off in over two years 18:28:34 I actually wonder what part of the CCC stack they really used 18:28:37 (going to Kruger) 18:28:57 if it was anything like SotM, mainly the RTMP network and review infra 18:28:59 pollo: jitsi + some vocto stuff, I didn't follow the details on their stack 18:29:03 pollo: it worked ok 18:29:12 (I don't know what they use for live mixing of remote events) 18:29:17 Ah they used vocto! 18:29:19 oh, right, FOSDEM is also trying to figure out online conf 18:29:32 they also streamed to twitch.tv 18:29:36 lots of people followed there 18:29:41 we had a discussion a week or two ago to look at things, but nobody's really started to actually implement anything yet 18:30:04 if anyone is interested in helping out, do speak up 18:30:27 I would really love to but I'm maxed out 18:30:39 potentially, but I'm going to be away for much of December and no idea what Jan looks like yet 18:30:40 happy to join a conversation, but can't really commit to driving anything 18:30:47 well, insert an "and has time" in there then ;-) 18:31:11 that rules out all the channels on this server that starts with #deb 18:31:12 paddatrapper: we discussed that we really want to have the infra up and running in early January, so 18:31:20 highvoltage: I'm sure :) 18:31:48 (at least the prerecording part, for which some insane team mates want to do something OBS-based..) 18:32:29 pre-recording over the internet? 18:32:58 tumbleweed: yes 18:33:03 the goal of prerecording (for me) was that it removes the unreliable network bit from the equation. And allows for mistakes to be corrected 18:33:10 would be nice if we could have that too but I think next month would be too soon 18:33:20 so, local recording seems a must. But hey, if it works... :) 18:33:29 tumbleweed: exactly, and I've been trying to tell them, but they be like "oh you can restart so it's fine lalala" 18:33:38 well they only need good internet connection at the time of the recording 18:33:39 yeah, which is true 18:33:53 well, kind of, but still 18:33:53 if the power trips or their connection dodges out they can restart or continue after restoration 18:34:18 anyway, it all means I've been volunteered to add more API to SReview, so... 18:35:42 #info people interesting in helping with FOSDEM talk to wouter 18:35:59 anything else here? 18:36:12 nothing more from me 18:36:47 nope 18:36:57 (not that I was really there :p) 18:37:02 #topic Next meeting 18:37:21 20 Oct @ 18:00 UTC (next tuesday)? 18:37:48 Wfm 18:37:57 I'll miss that one, but I can do the one on the 27th 18:38:16 (and nothing works for me next week, so no need to reschedule on my behalf) 18:38:32 #agreed 20 October 2020 @ 18:00 UTC 18:38:36 #endmeeting