17:04:22 <h01ger> #startmeeting snapshot.debian.org march 2024
17:04:22 <MeetBot> Meeting started Mon Mar  4 17:04:22 2024 UTC.  The chair is h01ger. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:04:22 <MeetBot> Useful Commands: #action #agreed #help #info #idea #link #topic.
17:04:41 <h01ger> do we have any agenda?
17:04:52 <weasel> rollcall?
17:04:55 * weasel present
17:04:58 <ln5> not really, at least i didn't prepare an agenda
17:05:02 * ln5 present
17:06:00 * h01ger is here too
17:06:05 <weasel> so, who's driving this?  ln5?  h01ger?
17:06:15 * h01ger is re-reading https://lists.debian.org/debian-snapshot/2024/02/msg00003.html
17:06:35 <weasel> right, let's go over the work items from that list?
17:06:40 <ln5> one question i'd like to try to figure out is what would be useful to do *without* having any new shiny hardware to run stuff on
17:06:40 <weasel> re 1) set up dev/staging/test environment
17:06:50 <weasel> any progress on that?
17:06:56 <h01ger> weasel: nope
17:07:01 <h01ger> (afaik)
17:07:02 <ln5> haven't seen any for 1
17:07:10 <weasel> any blockers or just lack of tuits?
17:07:29 <h01ger> ln5: can you queue your question for later, so we can go through the work items from last meetings
17:07:43 <ln5> weasel: i kinda connected it with access to new system
17:07:43 <h01ger> lack of tuits and probably lack of hw too
17:07:49 <ln5> h01ger: yup
17:08:06 <weasel> ok, fair enough.  so it's behind new hosting/infra.
17:08:29 <weasel> re 2) find new main storage
17:08:31 <ln5> but instead of waiting for new stuff, dev env could be put together anywhere ofc
17:08:48 <h01ger> nothing happened on 2) neither TTBOMK
17:08:50 <weasel> jcristau indicated before meetbot arrived that he didn't get around to asking hetzner or cloud team
17:09:04 <weasel> so unless anyone else can report anything, that's that about 2)
17:09:12 <h01ger> wait
17:09:17 <weasel> go ahead
17:09:28 <h01ger> except: jcristau: do you intend to do that still?
17:09:35 <ln5> i don't have enything for 2)
17:09:41 <h01ger> (talking to hetzner & cloud team that is)
17:10:19 <h01ger> lets move on, jcristau can answer anytime..
17:10:19 <jcristau> yeah.  i'll try to draft something after this.
17:10:31 <weasel> great, thanks
17:10:41 <h01ger> awesome, thank you
17:10:48 <h01ger> 3) S3 for serving web content
17:10:56 <weasel> I assume 3) (s3 support) and 4 (indexing) and 5 (sha256) are blocking on 1 (dev environment)?
17:11:07 <h01ger> yeah
17:11:19 <h01ger> which brings us back to the question ln5 brought up
17:11:23 <h01ger> [18:09] < ln5> one question i'd like to try to figure out is what would be useful to do *without* having any new shiny hardware to run stuff on
17:11:46 <weasel> right, so in summary: 3, 4, 5 block on 1,  1 might (or might not) block on 2.
17:12:04 <weasel> we can probably clarify the might or might not :)
17:12:11 <jcristau> i don't think 1 should block on 2
17:12:15 <weasel> I agree
17:12:19 <h01ger> yeah
17:12:26 <ln5> if we separate dev, test and staging we can be more precise
17:12:45 <weasel> the question is, what kind of resources do we need for a dev system?  does it have to be full-scale or would something a lot smaller also be good for a starting point?
17:12:47 <ln5> depending on what one means with it ofc
17:13:47 <h01ger> weasel: depends on the dev to be done :/
17:13:55 <jcristau> weasel: i was hoping it could be smaller, at least initially, e.g. just mirror (and maybe just mirror for a subset of archives), and expiring data after set time
17:14:47 <jcristau> to make it easier to get hosting secured and something going
17:15:30 <weasel> what are the active repositores still in snapshot?  debian proper, and security.  and we want ports back?
17:15:38 <weasel> (and -debug)
17:15:39 <h01ger> we want ports back
17:16:01 <h01ger> just spoke with someone today who wants to do riscv64 ci tests and needs ports snapshot..
17:16:01 <weasel> so if the dev system can be useful with just one of these, we could just do -security to start with.
17:16:07 * h01ger nods
17:16:08 <jcristau> we want ports back and we want to add security-debug
17:16:22 * h01ger wonders whether we should keep this meeting very short & have another one next month, when we hopefully got a positive reply from hetzner or cloud team?
17:16:28 <weasel> would that be useful?
17:16:29 <kpcyrd> if s3 is something debian can get for this amount of data, one could try to build a structure that allows the kind of lookups #debian-reproducible needs. you wouldn't necessary need significant data or hardware to develop with.
17:16:38 <kpcyrd> (sorry late to the meeting and slow typing)
17:16:41 <weasel> hey kpcyrd
17:16:53 <kpcyrd> hello! :)
17:16:54 <h01ger> hi kpcyrd
17:17:08 <weasel> a box with a security mirror and some disk to have it in a farm and a PG might be a good starting point
17:17:14 <jcristau> weasel: agree
17:17:19 <h01ger> yup
17:17:36 <ln5> how much of vcpu/ram/storage would that be?
17:17:43 <weasel> if we (<DSA hat>) set that up, who wants to bring snapshot-dev online there?
17:18:13 <ln5> i wouldn't mind doing that
17:18:26 <kpcyrd> if I recall correctly #debian-reproducible needs a lookup for release-datetime -> `Release` file and `<filename>.deb` -> <file content> (sha256 instead of filename might work too)
17:18:33 <weasel> debian-security is currently ~200gb
17:18:37 <ln5> might need some help, but happy to try
17:18:50 <weasel> so we'd need at least 250g for the mirror, 300g for the farm, some 50g for the DB and misc OS stuff.
17:19:08 <weasel> not crazy
17:19:25 <jcristau> rounding up to 1t :)
17:19:26 <h01ger> ln5: i can probably help with advise or maybe testing, but i definitly cannot drive this. etoobusy
17:19:35 <h01ger> kpcyrd: lets leave such details for later..
17:19:46 <kpcyrd> ok :)
17:20:01 <weasel> jcristau: conova-node0[34]?
17:20:09 <ln5> h01ger: sounds good
17:20:19 <h01ger> :)
17:20:23 <jcristau> weasel: looks plausible
17:21:03 <weasel> if ln5 and h01ger would find that useful, I might even try setting that up if that works for you jcristau
17:21:09 <jcristau> wfm
17:21:17 <weasel> great.
17:21:22 <ln5> weasel: useful indeed
17:21:37 <weasel> ln5, h01ger: we can coordinate oob once I have something started.
17:21:50 <ln5> weasel: splendid!
17:22:16 <h01ger> awesome
17:22:17 <weasel> I'm not good at keeping track of things today.  any questions that popped up that are still open?
17:22:53 <h01ger> i think we are good for today, except maybe finding another meeting date? maybe just monday in 4 weeks, same time?
17:23:35 <weasel> 4 weeks from now is easter monday
17:23:45 <ln5> the first of april? very easter, not great here
17:24:03 <weasel> tuesday, or just the week after?
17:24:15 <weasel> (or the week prior?)
17:24:32 <ln5> week prior or weak after would be prefered for me
17:25:07 <weasel> the week prior is dirauth sync
17:25:11 <h01ger> april 8th then?
17:25:19 <weasel> wfm
17:25:36 <ln5> looks good
17:25:47 <weasel> 1700 UTC or 1600UTC?
17:25:57 <h01ger> hehe. 16 UTC
17:26:03 <ln5> i like 1600
17:26:09 <weasel> jcristau?
17:26:12 <weasel> kpcyrd?
17:26:24 <weasel> going once
17:26:39 <weasel> going twice
17:26:40 <jcristau> i don't know yet, but no objection
17:26:51 <weasel> sold
17:26:57 <ln5> i like this meeting
17:26:59 <h01ger> \o/
17:27:20 <h01ger> #agreed next meeting monday, april 8th, 16 UTC
17:27:36 <h01ger> and obviously we can use this channel until then too :)
17:27:52 <weasel> https://v.noreply.org/2024-03-04-hO4NhaR8Kx4/8003a68e27acbc837284642e7ae00792fe2ddb47.ics
17:28:08 <h01ger> endmeeting? :) (happy to discuss more, but also happy to end this now..)
17:28:20 <weasel> I think we're done for now.
17:28:25 <ln5> yup agreed
17:28:31 <weasel> my action item is set up a vm and ping you both
17:28:49 <ln5> nice, thank you
17:29:01 <h01ger> #info weasel will setup a vm and ping ln5 and h01ger
17:29:07 <weasel> don't have a timeline yet.  please ping if I disappear again
17:29:13 <ln5> ack
17:29:13 <h01ger> #info jcristau will ask hetzner & cloud team about hosting
17:29:19 <h01ger> great great
17:29:30 <h01ger> thank you all!
17:29:34 <weasel> cheers
17:30:09 <h01ger> #endmeeting
17:30:24 <h01ger> why no links with logs, meetbot?
17:30:40 <weasel> we trusted you, meetbot
17:30:56 <h01ger> there are no logs on http://meetbot.debian.net/ neither
17:30:59 <ln5> they were behaving weirdly in another channel recently, and refused to end a meeting
17:31:27 <h01ger> http://meetbot.debian.net/debconf-hamburg/2024/ still worked nicely two weeks ago
17:32:04 <ln5> tnx h01ger
17:34:13 <ln5> lmk if i should mimic the software and summarise the meeting in an email to list, or if you'd rather try to feed it the log by hand
17:34:40 <h01ger> i'd appreciate a summary to the list
17:34:49 <ln5> ack
17:35:08 * h01ger is juggling a bit too many things currently
17:38:21 <weasel> cheers
10:48:20 <h01ger> jcristau: thanks for the mail!
12:53:02 <jcristau> h01ger: np, thanks for the prod
18:38:59 <jcristau> fwiw, new development from this weekend is that sallinen is out of space for the mirrors
18:39:22 <h01ger> sallinen is doing what re: snapshot again?
18:39:54 <jcristau> it mirrors all the archives and imports them into snapshot
18:40:02 <jcristau> so now that doesn't work
18:40:10 <h01ger> ah. thanks.
18:40:32 <jcristau> (now trying to find a workaround)
18:40:38 <h01ger> :)
18:40:56 <h01ger> & obviously thanks for that too! :)
19:14:09 <jcristau> the temporary workaround is to move the debian-security import from sallinen back to sibelius, which is a bit backwards, since sibelius is old and we want to get rid of it, but sibelius is where there's space right now...
19:14:42 <jcristau> that'll hopefully buy us a few days until the debian and debian-debug archive grow too big again
19:18:32 * h01ger nods
19:35:41 <jcristau> I, [2024-03-11 19:04:16#27545]  INFO -- : New mirrorrun 86408 for debian-security
19:35:44 <jcristau> I, [2024-03-11 19:31:36#27545]  INFO -- : [run #86408] Mirrorrun 86408 for debian-security completed.
19:35:47 <jcristau> I, [2024-03-11 19:31:36#4808]  INFO -- : Indexing mirrorrun #86408 of debian-security from 2024-03-11 19:04:16
19:35:50 <jcristau> I, [2024-03-11 19:31:37#4808]  INFO -- : Indexing done
19:35:52 <jcristau> not very fast...
19:35:55 <jcristau> but at least that seems to work
21:49:35 <ln5> import of debian-security now runs on sibelius? the tired host with less RAM but the satabeast connected?
09:01:18 <jcristau> ln5: yes
18:44:51 * h01ger cheers another mail by jcristau
06:33:17 <jcristau> h01ger: thanks for the positive reinforcement :)
08:55:24 <h01ger> :)
16:52:19 <weasel> autojoin fixed, I hope
15:00:01 <jcristau> bah.. we're back to ENOSPC on sallinen
16:36:25 <jcristau> adsb: pondering moving the debian mirror to sibelius.  i think /srv/mirrors/debian-archive on sibelius is dead and could be removed to make space?
17:00:16 <adsb> jcristau: yeah, the live copy is /srv/archive-mirror, which is what's then mounted on sallinen
17:00:50 <adsb> jcristau: if you care, although it's not in the archive.d.o DNS rotation, sibelius still has /srv/mirrors/debian-archive exported via apache and rsync atm
17:03:15 <adsb> (on the archive-master.d.o name, as that's still pointing at sibelius)
17:32:30 <jcristau> hmm right
17:32:44 <jcristau> not sure i'll do anything about it today tbh