18:00:50 #startmeeting DebConf Video Team Meeting 18:00:50 Meeting started Tue Nov 17 18:00:50 2020 UTC. The chair is tumbleweed. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:50 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:01:01 #link http://deb.li/oNCD Agenda 18:01:04 #topic Roll Call 18:01:06 i'm here but I might have to leave at any moment 18:01:12 * wouter waves 18:01:40 and we don't have an agenda for today 18:01:44 scribble scribble scribble 18:01:57 that makes it a very short meeting? ;-) 18:02:04 (actually not I would guess) 18:05:04 OK, moving on 18:05:13 just the three of us then? 18:05:14 #topic miniconfs - MDCO2 - SReview 18:05:22 wouter: you've been setting things up? 18:05:30 I just imported the schedule 18:05:37 I haven't seen logos or a .svg file yet 18:05:47 Sorry, just finished supper 18:05:47 I see from last week's meeting that highvoltage was working on that 18:06:07 yeah, it's been available for a couple of weeks now 18:06:08 so I'll ask him; if he doesn't have anything, I'll make something myself 18:06:12 but conference schedules always change :) 18:06:12 oh, okay 18:06:38 my schedule parser can run repeatedly if there is a parseable schedule with unique identifiers 18:06:41 wafer provides that 18:06:54 so I could just have set that up, and that would've worked... ah well 18:07:28 are we running current sreview master? 18:07:30 I'll look at making the gridengine stuff work after the meeting 18:07:32 yes 18:07:40 is bs1770gain gone? :) 18:07:50 eh, I think so, but I'll double check that :) 18:07:55 I don't think we have any encoder nodes for the moment 18:08:05 don't need them, vittoria can be encoder node 18:08:11 it's how we always did miniconfs 18:08:17 (vittoria has 8 virtual CPUS) 18:08:47 looks like bs1770gain is commented out, in an uncommitted change 18:09:03 (which is how we did dc20) 18:09:06 right 18:09:22 I'll keep using it that way then 18:09:36 anything else you need? SVGs are presumably the big thing 18:09:47 highvoltage had an action on them 18:09:48 my long term plan is to make the normalization configurable, and then also plug in an ffmpeg-based one 18:09:56 yeah, just the SVGs I think 18:10:09 when we get them, what's the process to add a test talk to sreview? 18:10:16 (so that we can run it through the pipeline) 18:10:31 direct database calls :) 18:10:40 or you can use the API that I added 18:10:45 just the talks table? 18:11:00 yes, though that doesn't give you a speaker 18:11:07 probably good enough 18:11:09 but that's okay, it won't die if you don't add speakers 18:11:10 yup 18:11:13 tumbleweed: oh yes, thanks for the reminder I'll try to get them tonight 18:11:15 yeah, we can link to a speaker 18:11:24 highvoltage: as soon as you get them done, let's try a test 18:11:27 which is in speakers_talks 18:11:38 yeah, I've done that before 18:11:44 right 18:12:13 I had plans to add an admin interface so you would be able to enter a talk through a form this year, but then we had to do an online debconf so it's on the backburner now :) 18:12:23 :) 18:12:41 moving on, then 18:12:47 oh 18:12:54 at least there's API and I learned about Vue, so it should be doable now 18:12:59 how are we going to get video files to vittoria 18:13:06 do we need DSA to add a public key? 18:13:13 or do we have a private one somewhere we should use? 18:13:23 either that, or we can make vittoria pull 18:13:32 which is probably easier 18:13:49 there's a key in ~sreview/.ssh which *should* be in ansible already 18:14:22 OK, I'll look at that 18:14:35 #action highvoltage to create title slides for SReview 18:14:41 #action tumbleweed to get video to sreview 18:14:46 (afk for a minute, brb) 18:14:49 #action tumbleweed to do an all-up test through sreview 18:15:00 #topic miniconfs - MDCO2 - Static Loops 18:15:05 highvoltage: you've got another hanging action here 18:17:14 I'll carry it over 18:17:26 #action highvoltage to generate statics loops 18:17:36 #topic miniconfs - MDCO2 - Training 18:17:38 Training is done 18:17:48 Docs are updated to be closer to MDCO2 18:17:52 anything more to say here? 18:18:06 (back) 18:18:34 Is the volunteer system up and working? 18:19:13 hi - I couldn't make the volunteer system work yet. is it a blocker? 18:19:33 being a small event, I would suggest just using the wiki or an etherpad 18:19:35 it could be scheduled in a pad instead 18:20:25 Either way, we need to publish something soon 18:20:47 so, we need somebody to create a pad/something and email all the registered volunteers 18:20:54 any volunteers? 18:20:59 I can 18:21:44 #action pollo to create a pad/something and email all the registered volunteers 18:22:04 #topic miniconfs - .br mini 18:22:12 there was an agenda item last weekt o schedule a meeting the the BR team 18:22:17 has anything happened there? 18:22:37 I assume somebody needs to start a framadate / something and let people know 18:22:41 phls: ^^ 18:22:45 tumbleweed: phls came in during our meeting afterall 18:23:16 yes 18:23:28 I take it there are no more details that need to be hammered out, then 18:23:32 he said we didn't need to meet 18:23:34 might still be good to have a a proper meeting though? 18:23:37 oh, nvm then :) 18:24:00 terceiro: do you think there's something still necessary? or are you not involved? 18:24:09 there are actions to carry over here 18:24:18 #action phls (or someone from mcdobr) to provide a slide with sponsor logos + a general logo for the conf 18:25:07 I am involved, but won't be available during at least the Saturday 18:25:46 wouter: I assume we should be setting up sreview for that mini, too 18:25:52 AFAICT things are in place 18:25:54 they have a published schedule 18:26:00 oh, cool 18:26:03 let me do that now then 18:26:11 #action wouter to set up sreview for the .br mini 18:26:29 terceiro: did we train enough .br volunteers last weekend, or are we going to need to organise more training? 18:27:03 tumbleweed: I don't know, I didn't really follow this 18:27:18 I imagine we're going to need more .in training - the time slots we had weren't great for them. And nobody pitched for the one we added at the last minute 18:27:53 terceiro: anyone we should be relaying that question to? or drop it? 18:28:20 tumbleweed: phls and lenharo 18:28:51 I'd be curious to know how pre-recording has been going and where the videos are stored atm 18:29:14 tumbleweed: I *think* they have a call for volunteers happening atm 18:29:18 well they're both here. No need for me to repeat that to them 18:29:29 they should have news on that Soon(TM) 18:29:59 schedule for mdcobr has been imported into SReview 18:30:02 OK, well if the .br team needs anything, let us know 18:30:08 #info mdco.br has been imported into sreview 18:30:16 (but mdco2 is still the default) 18:30:55 #topic mdco.br - infra 18:31:20 I'll be around for some of the .br mini timezone. If you run into problems I'm here 18:31:27 am I wrong in assuming we just use the same infra we used for mdco2, with changed configs? 18:31:36 that's the plan yes 18:31:38 I don't know if anyone will be for the first few sessions of the day, though? 18:32:16 I've offered phls access to the machines, in case he needs them. But he needs to take me up on that offer and provide his ssh key 18:32:21 I can try, I haven't actually looked at the schedule yey 18:32:33 Don't know if .br needs more than that 18:33:20 #link https://mdcobr2020.debian.net/schedule/ mdcobr2020 schedule 18:33:34 the schedule starts at 14:30 local time for me 18:33:35 I mean, it's early but I guess I can take the first slots 18:33:49 it's less early on the sunday, at least 18:33:54 I *should* be able to look at things, but I'm not confident enough that I'll be able to actually fix things if they're broken 18:34:35 let's move on 18:35:00 #info wouter may be around during mdcobr2020 mornings 18:35:07 #info pollo should be around during mdcobr2020 mornings 18:35:19 #info tumbleweed should be around during mdcobr2020 afternoons 18:35:40 #topic video archive documentation 18:35:43 pollo: carry over? 18:35:49 yup 18:36:09 what do you want documented there? 18:36:18 it's just a big file dump that gets rsync'd 18:36:29 migrating https://wiki.debconf.org/wiki/Videoteam/Archive to the debian wiki and updating 18:36:30 and some git annex that nobody uses anymore, I think 18:36:32 it uses git-annex for one 18:36:33 locations, who owns what, how does it get rsynced, etc. 18:36:48 mm, okay 18:37:06 (the git-annex is still used for integrity checking) 18:37:20 and if we got access to those s3 bucket credentials, it could be backed up there 18:37:27 (at least it was when the archive was moved) 18:37:45 okay, fair enough 18:37:45 I have committed dc20 to it 18:39:09 #topic AoB 18:39:46 should I keep doing VP8 encodes? 18:40:03 I don't think so 18:40:04 we currently do an "lq" version of everything, which is scaled down and at lower bitrate 18:40:13 we could do av1 and vp9 18:40:18 (so it's not *just* vp8) 18:40:24 av1 encoding is still *very* slow with ffmpeg 18:40:34 takes multiple seconds to encode just one frame 18:40:34 ah, lq is vp8? 18:40:38 I'd rather wait with that 18:40:38 pollo: yep 18:40:39 pollo: yes 18:40:44 then I'd say yes 18:40:58 why isn't lq also vp9, but with different settings? 18:41:09 historic reasons :) 18:41:24 I think urbec complained once that her old laptop couldn't handle vp9 18:41:41 so I decided to make at least one encode usable on older hardware 18:41:53 I could convert that one to VP9 too, easily 18:42:19 That urbec laptop issue should have been resolved a couple of years ago now 18:42:30 I know, hence "historic" 18:42:33 +1 on vp9 18:42:45 ah, I also have an AoB 18:42:53 currently we just have the instance storage on the voctomix and backend machines 18:43:05 should we add extra volumes for recordings? 18:43:14 how much space is available currently? 18:43:33 I guess I should look at 140G on voctomix1 18:43:38 oops 18:43:44 140G on voctomix1 18:44:14 140G might be a bit tight 18:44:19 only 34G on backend 18:44:24 which is certainly going to be tight 18:44:30 I'd be more comfortable with twice that 18:44:37 definitely need more than 34G, yes :) 18:44:53 so yeah, I'd add some storage if you can make that happen 18:45:05 I think you need to assume 300G for a miniconf 18:45:18 (for all the recordings of that miniconf, obviously) 18:45:35 that's a bit more than what we actually need, but you want some margin 18:46:05 OK, I'll give them both 250G volumes and we can see how it goes after mdco2 18:46:06 minidc hamburg 2019 used 224G 18:46:10 marseille used 158G 18:46:14 vaumarcus 95 18:46:20 they'll cost €10/month each 18:46:28 yeah, that should work 18:46:55 actually, those all used MPEG2 18:47:02 mdco1 used 4.4G, in mpeg4 18:47:14 so we might not need extra storage after all 18:47:33 then again, €10 is not *that* much 18:47:47 maybe do it for now, we can always remove it if we don't end up needing it 18:48:29 dco20 looks like it was < 8G / day / room, every day 18:48:48 so voctomix1 should be fine 18:49:12 it's the backend that's risky. I'll give it 50G or something 18:49:22 sounds good 18:50:00 #topic next meeting 18:50:07 this time next week, as usual. I presume 18:50:11 +1 18:50:13 no need to change 18:50:21 #agreed same time, next week 18:50:26 #endmeeting