18:00:20 #startmeeting 18:00:20 Meeting started Tue Jun 18 18:00:20 2019 UTC. The chair is pollo. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:20 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:00:25 o/ 18:00:26 #topic Roll Call 18:00:36 please say hello if you are here for the meeting 18:00:41 * wouter is here 18:00:42 hello if you are here for the meeting 18:00:45 hello if you are here for the meeting 18:00:47 "hello if you are hear" 18:00:48 (for the first time in months) 18:00:50 hi 18:00:53 * DLange is just lurking 18:00:57 wouter: ack same 18:00:59 hi 18:01:00 o/ 18:01:09 also, have a look at the proposed agenda: http://deb.li/3aA7s 18:01:14 (in the middle of another meeting, but hopefully killiing that soon) 18:01:14 o/ 18:02:10 seems like y'all here 18:02:23 #topic Marseille & Hamburg post-mortem 18:02:41 hurray for two successful events back to back! 18:03:00 we want more of those (events and successful ones) 18:03:11 yeah, I think things went pretty smoothly 18:03:12 cool 18:03:15 at least from my POV 18:03:18 we also found a bunch of problem with our setup. It seems tobi and urbec started a shopping list already 18:03:25 I think two weeks apart is the minimum though 18:03:28 +1 18:03:45 I'll action tobi and urbec to send a mail on ML pointing to a proper shopping list pad 18:03:58 unless someone is taking the stuff from one event to the next 18:04:01 #action tobi & urbec to send a mail on ML pointing to a proper shopping list pad 18:04:03 I should probably add that I used the experience of configuring SReview for the two miniconfs to also document how to update things 18:04:14 great! 18:04:15 there's a README.miniconf (or some such) in ~sreview on vittoria 18:04:28 (can't remember the exact filename, but you can't miss it) 18:04:51 so hopefully that should reduce my bus factor slightly ;) 18:05:14 so do we want to talk about what went wrong that doesn't imply buying more stuff to fix it? 18:05:30 is there anything in that category? ;-) 18:05:44 I was mainly thinking about the ground loop issue 18:05:47 well #1 how did the training sessions go? 18:05:57 oh right 18:06:06 if we have an idea of what caused that, it would be nice to record it here 18:06:39 #info we had a ground loop problem at both the Marseille and the hamburg mini-confs 18:06:44 earth loops are unfortunatly common. mainly caused by differing power at differant ends of a building. 18:06:55 RattusRattus: everything was wired to the same outlet 18:07:18 olasd: including the PA system? 18:07:21 yes 18:07:30 well, the issue happened without the pa system 18:07:35 or that, yes 18:07:36 just the mixer, the camera and the pc 18:07:41 is there perhaps some of our kit that's leaking to ground? 18:07:52 that can also produce hums, IIRC 18:07:55 the probably suspect is the pc 18:07:58 but I'm not the expert here 18:08:08 I also agree the PC is the main suspect 18:08:09 because we tried with another mixer we borrowed, and that didn't fix it 18:08:20 it will almost certainly be the PC (not having a floating PSU) 18:08:28 #info the PC is the main suspect 18:08:34 (also please not that it's nice to have a friendly AV rental company in the same building as the venue) 18:08:39 so how do we fix that? 18:08:40 olasd: can we action you to replace the PSU ? 18:09:01 well, we should try with a different pc first, to confir 18:09:02 m 18:09:05 no replacing PSU will not fix 18:09:12 PC PSUs are NOT issolated 18:09:29 issolation transformer may help 18:09:32 RattusRattus: what would you suggest then? 18:09:37 ^^ 18:09:48 perhaps a UPS could help, too? 18:09:59 (I'm not shipping either of these things) 18:10:15 no we shouldn't 18:10:16 (and as a bonus, it would also protect us against someone tripping over cables) 18:10:25 we didn't have this problem with the Cube PC did we? 18:10:27 olasd: mm, I imagine that might be pretty expensive, yes 18:10:36 RattusRattus: didn't that one use a lot less power? 18:10:48 sorry I'm late 18:10:49 wouter: no idea 18:11:01 olasd: can confirm, but I think it's the same motherboard in a different case 18:11:06 the guts are identical, only the case changed 18:11:13 and PSU? 18:11:18 same? 18:11:22 yes 18:11:23 well, yes 18:11:58 * RattusRattus will think. 18:12:09 #action RattusRattus earth loops and PCs 18:12:10 there's a non-zero chance that the motherboard isn't properly grounded to the case 18:12:17 might have used the wrong screws 18:12:32 IIRC we covered the earth on the PC power plug and that fixed it (Tobi would know for sure) 18:12:43 yeah 18:12:45 covered? 18:12:50 Gaffer 18:12:56 so that it wouldn't make contact anymore? 18:12:59 no, disconnecting the mixer (not the pc) from earth fixed it 18:13:07 right so lifted it. that makes sence 18:13:16 wouter: yes 18:13:22 0V should not be mains earth 18:13:23 but obviously we didn't keep the setup like that 18:13:36 #info there's a non-zero chance that the motherboard isn't properly grounded to the case (wrong screws?) 18:13:44 it should be the same potential and (possibly) common only at 1 point 18:13:47 I don't think we'll get farther than this for now 18:14:08 the other big issue is the audio desync 18:14:09 if someone has something else to add on the confs (that's not about the ground loop), please go ahead 18:14:12 chances are the mother board is 0V and should NOT be connected to the case (mains earth) 18:14:12 ivodd: lifting the pc or the mixing desk would have the same effect (as the cameras have a double isolated power supply with no earth) 18:14:26 olasd: we tried lifting the pc, it didn't help 18:14:36 ivodd: huh. interesting 18:14:42 I'd look at the mixer though ... easier to fix than a PC PSU, these are hard to get circuit separated 18:15:00 even with everything (network, screen, etc) disconnected from the pc 18:15:01 DLange: aparently differant mixer was tried 18:15:04 I don't know where the Gaffer fixed it, I just know it did :) 18:15:16 anyway, yeah, the A/V desync issue still happens 18:15:29 * RattusRattus will take earth loop as a primary task during debconf 18:15:31 and we still don't really understand how 18:15:42 could someone ask CCC VOC if that also happens to them? 18:15:58 A/V desync issue? please describe 18:16:05 olasd: almost forgot, but FOSDEM had issues with that as well, and it *might* actually just be a bug in SReview... sorry :) 18:16:22 can you confirm if the issue exists with the original raw files, too? 18:16:28 it does 18:16:37 okay, *phew*, then it's probably not me :) 18:16:42 it's also noticeable in the voctomix gui 18:16:53 RattusRattus:audio led audio by a significant time after the stream was running for a while 18:16:58 (if people actually listen to that) 18:16:58 okay 18:17:21 restarting the voctomix PC fixed the issue 18:17:21 audio led audio? audio led video perhaps? 18:17:25 yes 18:17:43 RattusRattus: yes, that one... 18:17:52 could someone ask CCC VOC if that also happens to them? I think it would be a good way forward 18:18:15 how much time, and how much lead? 18:18:26 a few secs 18:18:30 ok again this is somthing to test with a camera and voctomix PC. 18:18:32 they've certainly looked into issues like this in the past 18:18:52 tumbleweed: do you know if they found the cause or had a fix? 18:18:53 CarlFK has had test-jigs to demonstrate desync issues in the past, too 18:18:58 maybe https://github.com/voc/voctomix/issues/58 18:18:59 almost certainly voctomix buffer paths 18:19:07 pollo: desync issues have been fixed in the past 18:19:20 or BlackMagic cards 18:19:21 and possibly regressions created, if you got new issues in hamburg :P 18:19:48 "As far as i read the communication, there are no inter-elements in voc2mix, so this is probably not applicable anymore." 18:19:58 so the fix on that bug issue seems to be "use voctomix 2" 18:20:22 eeek! 18:20:49 #info we've been have audio desync issues and it might be related to https://github.com/voc/voctomix/issues/58 18:21:00 I don't think we'll fix that today either :D 18:21:02 sounds like an exciting project for not-this-conference 18:21:15 it's good to have a suspect though 18:21:18 anything to add to this topic before we move on? 18:21:25 someone might want to experiment a bit during debcamp 18:21:28 (if there's time) 18:21:35 I'm happy to take a look 18:21:57 #topic DC19 - PC & server rental 18:22:23 so I haven't seen a quote for the PCs we need 18:22:26 but we have one for a server 18:22:28 phls: I agree with tumbleweed on the disks -- it's probably better to have 6x2TB rather than 1x10TB 18:22:51 some of the stuff might need to be translated a bit more though 18:22:53 the specs proposed are: 18:22:54 1 Servidor Dell Power Edge R620 com 2 Processadores Xeon Six Core 2.0GHz 18:22:55 what is a "Perc Array"? 18:22:56 32GB RAM 18:22:58 1 x Perc Array 18:23:01 6 x 2TB SATA Hot Plug (we are treying to change for 1 x 10TB) 18:23:03 6x2 sounds fine too me. 18:23:04 4 x 10/100/1000 18:23:07 perc is a dell hw raid 18:23:07 wouter: a raid controller 18:23:14 oh right 18:23:22 they're not bad 18:23:23 white label lsi crap 18:23:28 heh 18:23:32 lsi isn't too bad 18:23:39 configurable through linux, via non-free debs 18:23:42 you could run them in JBOD mode and then do ZFS on top 18:23:47 I think those specs are OK for our needs 18:23:52 or mdadm 18:23:55 wouter, actually lenharo is deal with pcs 18:23:58 or that 18:23:58 sorry I might be projecting my opinion of megacli :P 18:24:13 is this server for live streaming or for transcoding? 18:24:16 or for both? ;-) 18:24:18 I think this is for storage 18:24:21 and possibly transcoding? 18:24:28 for everything 18:24:35 I don't think we'll get another server onsite 18:24:36 storage and streaming backend (so downscaling, not the primary stream encode) 18:24:39 I'm very hestitent to have everything on one machine 18:24:43 yeah, me too 18:24:50 we had some bad experience with that during dc17 18:25:17 I'm very hestitent to have everything on one machine 18:25:19 I have the machine we bought onsite at dc18 as a router, than I can bring 18:25:34 however many machines we have someone is going to DoS the wrong one at the wrong time 18:25:36 ^^ didn't we say that a big enough machine running several VMs would be ok? 18:25:37 (assuming we don't need a router) 18:25:42 tumbleweed: brazil customs might make some trouble 18:25:50 RattusRattus: ish, but that's hardly a crazy big machine 18:25:57 olasd: I personally think we should have the live streams separate from everything else 18:26:18 * RattusRattus also preferat boxes for seporate tasks... 18:26:19 because you don't want to kill the live streams by accidentally doing something wrong 18:26:22 we don't need the live streams to be on site at all 18:26:25 if machine is under U$ 500, is ok bring it 18:26:44 there is that 18:27:07 olasd: not true - if people are watching talks from hack labs etc 18:27:21 so are we fine with the dc18 router PC + the rented server, ? 18:27:24 ^ depends on bw to / from sites 18:27:26 (unless of course you don't have the bandwidth to push 720p, but then...) 18:28:17 RattusRattus: the streaming backend still doesn't need to be on-site 18:28:41 olasd: it's probably a good idea, though 18:28:55 but yes, we do have some flexibility there 18:28:55 olasd: assuming there is enough bw that people watching ON site don't collapse the outgoing feed I agree 18:29:31 olasd: but nothing a bit of traffic shaping can't fix 18:29:45 which means we need a machine to shape traffic :P 18:29:48 what kind of bandwidth will there be? 18:30:17 tumbleweed: well the router PC *shoud* do that for us right:-) 18:30:24 should even! 18:30:31 also, isn't that something the network team can take care o? 18:30:34 depensd what we're using it for 18:30:36 *of 18:30:39 wouter: what network team? 18:30:48 ehh 18:30:51 at this point, we're the network team! 18:30:56 well, I was assuming there'd be one, but meh, okay 18:30:57 (as usual) 18:31:08 so from that I understand, dc18 router PC + the rented server doens't make us happy 18:31:19 at the very least we should ensure that outbound stream feeds and incomming acks for said feeds is reserved bw on the router 18:31:27 so we want another server machine 18:31:28 pollo: part of the issue here is that there are a lot of unknowns 18:31:40 pollo: and machines are like duct-tape, you use them to solve unexpected problems 18:31:44 yes, but if the local team needs to find something else, they need to know 18:31:47 i haven't contact with network team 18:32:24 phls: you looked for server rental. How expansive was that quote? 18:32:30 could you get 2 of them instead? 18:32:37 lenharo, ? 18:32:42 fwiw, there's no point having a server if we don't have PCs to put the capture cards in 18:32:59 I'm more worried about that than anything else 18:33:02 olasd: fair point 18:33:04 olasd: +1 18:33:11 I added a link to the detailed quote (which also has PCs) 18:33:12 we only need one server with loads of storage, really 18:33:29 everything else can just be (powerful) desktop machines 18:33:42 yep 18:33:44 so we really don't need two servers, but we need 5-10 desktops 18:33:55 (handwavy number there) 18:34:02 * olasd checks ivodd's link 18:34:05 problem is we don't have a source to borrow from 18:34:21 one for each room, one for the live streams, one for routing, plus a few spares 18:34:45 * RattusRattus did this work at FOSDEM sprint 18:35:02 what has changed that we are saking for somthing differant now 18:35:09 (at this late stage) 18:35:58 we can't find i7 to borrow 18:36:26 so the server rental is 3330 BRL, which is about 770 EUR 18:36:42 ok - but the SPEC hasn't changed... 18:36:43 (or 865 USD) 18:37:05 phls: would the company that is renting you the server also rent out desktops? 18:37:14 phls: alternatively, you can do the same trick that FOSDEM does: 18:37:15 wouter: that's for the server + 4 i7 desktop + 2 i3 desktop 18:37:17 phls: https://salsa.debian.org/debconf-team/public/data/dc19/uploads/4e702e4d90a0c7adda2a4f8f95023fa9/ICTL_281418_ATUALIZADA.pdf also has a quote for i7 18:37:18 they do, it's in the quote. 18:37:26 3330 BRL is to rent server R620 + 6 computers (4 I7 and 2 i3) 18:37:27 so it seems we do have PCs 18:37:33 ah, oops, missed that :) 18:37:35 the network cards seem to be 10/100 though? 18:37:41 I'm really dubious about the specs 18:38:00 these i3 are for hotel network, not for video team :-) 18:38:05 "i7 2 GHz" sounds really really really low end 18:38:08 can't we just borrow a few of the PCs standing around in "our" rooms anyways ... https://salsa.debian.org/debconf-team/public/share/debconf19/blob/master/photos/organization/utfpr/utfpr-lab-B108/utfpr-lab-B108-02.jpg ? 18:38:09 anyway, the FOSDEM trick is to not rent, but buy old Lenovo X series laptops on ebay 18:38:10 i have asked and they say that have 10/100/1000 network card. I have to check 18:38:11 so, R620 + 4x i7 18:38:17 and then after the conference, sell them at cost 18:38:36 wouter: if you can fit our capture cards in lenovo x series laptops, go right ahead 18:38:45 ah yes 18:38:48 * wouter slaps $SELF 18:38:52 sorry, didn't think of that bit 18:39:02 :) 18:39:21 is 4 i7s enough though? 18:39:36 it would leave us no spares or dev boxes 18:39:44 that's my point 18:39:51 2Ghz i7s sounds like a mobile chip 18:39:54 there's the two i3s, but I suspect those are for other purposes? 18:39:55 so, the question is: are those 6 desktops sufficient? 18:40:03 are all of them for the video team? 18:40:07 tumbleweed: 4, the i3s are not for us 18:40:22 OK, then I'd say we should pad that out with a few more 18:40:29 +1 18:40:30 wouter, i3 are for other purposes 18:40:34 we know we can alway use more 18:40:38 tumbleweed: 3x vocto, 1x gateway? 18:40:43 just i7 now are for video team 18:40:52 paddatrapper: that leaves 0 spare 18:40:59 paddatrapper: plus something for the live stream, and spares 18:41:15 tumbleweed: exactly,I was talking about the existig 4 18:41:16 phls: I'd add at least 2 more. i3/i7 - I don't really care too much but i7 doesn't hurt 18:41:18 existing too 18:41:55 ok, so to move forward: 18:41:55 the rent is 60 US$ for each i7, so adding a few shouldn't be an issue 18:41:56 #agree we wan't a least 2 more i7 (6 instead of 4) and want another server 18:42:02 tumbleweed, ok, we quoted 4x i7 + 1x server because pollo had asked this number 18:42:06 do we want another server ? 18:42:09 pollo: eh, why do we want another server? 18:42:12 I don't think so, no 18:42:23 really, that's what I had understoods, sorry 18:42:26 #undo 18:42:26 Removing item from minutes: 18:42:32 just the 2 extra i7s for my money 18:42:41 yeah, that's what I thought we were talking about 18:42:44 #agree we wan't a least 2 more i7 (6 instead of 4) (and more if possible) 18:42:51 is R620 newer than R710? :-) 18:43:04 yeah, I was wondering about that too, doesn't look like it to me 18:43:05 highvoltage: yes R620 is newer than R710 18:43:10 but Dell might be weird :) 18:43:10 can we move forward? we still have to talk about audio 18:43:14 sure 18:43:17 just one last thing 18:43:20 R620 is the 1U variant of R720, which is newer than R710 18:43:31 it'd be good to have the *full* specs of the machines ASAP 18:43:32 we have to check if these i7 has 2 slots PCI-E x1 slot? or 3? 18:43:49 phls: 2 slots 18:43:53 phls: we _need_ 2 but 3 is better 18:43:57 since we have the serial cards 18:44:09 ah, forgot about that 18:44:12 ok 18:44:42 also, is 28/7 ok as a return date for the server? 18:44:59 A few extra days is usually a good idea 18:45:02 that leaves very little time to make sure everything is done 18:45:09 sure, np 18:45:18 28 is sunday 18:45:27 could someone follow up on the changes to the quote in the next few dates? 18:45:32 s/dates/days 18:46:01 ah, is the server ok? 18:46:06 ivodd: can you do that? 18:46:10 if someone can go to the rental company with a debian live usbstick and get the output of dmidecode / lspci for the i7s it'd be great 18:46:13 just change to 1x 10TB? 18:46:16 pollo: no 18:46:23 phls: why do we want 1x10TB? 18:46:24 phls: 6 x 2TB is just fine 18:46:30 phls: no, we don't need a 10TB disk 18:46:31 I'd rather have the redundancy of 6x2 18:46:34 me too 18:46:36 way better than 1 x 10TB 18:46:37 ivodd: I will 18:46:51 paddatrapper: thanks! 18:47:00 #action paddatrapper to follow up with phls about the pc + server rental + full hw specs 18:47:03 the request was 1x 10TB. I am confirming if must be 1x 10Tb 18:47:22 no 18:47:25 but ok i got now 18:47:25 we need 10TB storage after RAID, but we want redundancy 18:47:26 then no, the current proposal for 6x2 is fine 18:47:29 :) 18:47:34 ok 18:47:35 I think we can move on now 18:47:38 let's 18:47:41 yeah 18:47:50 #topic DC19 - Audio rental 18:48:25 so, I got 2 quotes 18:48:41 one with all, and one missing equips 18:49:35 the beringer RX1202FX 18:49:38 as I explained to some of you before, its hard get rent quotes here. For a lot of reasons 18:49:38 the Behringer RX1202FX will not work for us 18:49:39 NOT SUITABLE 18:49:48 heh jinx 18:50:16 paddatrapper, why? just to know 18:50:23 We also need 2x Behinger DI100 per room (so 6 in total) as they are mono 18:50:23 not got sliders 18:50:34 yeah, we definitely need those :) 18:50:37 we need 3 mixers anyway 18:50:40 paddatrapper, the request was 3, one per room 18:50:46 pollo: 2 mixers at venue already 18:50:51 ah sorry 18:50:54 we need sliders and desk-top, not rack mount 18:51:09 phls: 3x stereo, or 6x mono 18:51:43 on your buy list (not that we want to buy) the Yamaha MG12XUK is also not a suitable mixer (same reason) 18:52:04 paddatrapper, about RX1202FX could yot explain me why is not ok (I have to explain to them) 18:52:30 phls: we need desk-top mixers with sliders, not rackmount with no sliders 18:52:31 phls: the RX1202FX does not have sliders to correct individual channel volumes 18:52:32 phls: we need sliders not knobs for each channel 18:53:11 RattusRattus: it looks like it has sliders 18:53:26 tumbleweed: no -- https://www.sweetwater.com/store/detail/RX1202FX--behringer-eurorack-pro-rx1202fx-rackmount-mixer-with-effects 18:53:29 (between the knobs, which is eww) 18:53:41 https://www.google.com/search?client=firefox-b-d&q=Yamaha+MG12XUK 18:53:56 also no submix/aux bus 18:53:59 wouter: I see sliders on this picture 18:54:08 oh yes, indeed 18:54:08 missing submix is a problem 18:54:08 ugly 18:54:13 rackmount is annoying 18:54:16 tumbleweed: you're right, missed them 18:55:01 anyway, phls: RattusRattus' link shows a mixing console that is similar to what we need 18:55:15 phls: we need submix (the ability to route some channels to a submix that is a separate output from the main out) 18:55:16 (doesn't have to be that one, but one that has the same functionality) 18:55:19 and the MG12XU? 18:55:27 wouter: NO that link was for the NOT suitable yamaha mixer (no sliders) 18:55:34 err the MG12XUK does not have sliders 18:55:37 oh? 18:55:56 ah yes, sorry, bad resolution here 18:56:05 It would be suitable for us but would really throw new volenteers 18:56:29 and Soundcraft signature12 ? 18:56:36 Soundcraft signature12 is great mixer 18:56:37 ah, eh, actually, the photos that I see don't have the XUK one, only the XU 18:56:40 which does have sliders 18:56:41 hence my confusion :) 18:56:42 The MG12XU has sliders 18:56:59 ok, lets split. First, the mixer 18:57:00 yup 18:57:19 the 2 mixers for rent are not good of us? 18:57:35 no 18:57:53 phls: the MG12XU is fine 18:58:00 the MG12XUK is not, but that's not in the quote 18:58:09