#archlinux-ports | Logs for 2017-11-23

[00:00:19] -!- p71 has quit [Read error: Connection reset by peer]
[00:00:37] -!- yokel has joined #archlinux-ports
[00:06:27] -!- zaungast has quit [Quit: Too late now.]
[00:06:46] -!- p71 has joined #archlinux-ports
[00:16:05] -!- deep42thought has quit [Quit: Leaving.]
[00:44:02] -!- because has quit [Read error: Connection reset by peer]
[00:51:09] -!- because has joined #archlinux-ports
[01:44:17] -!- isacdaavid has quit [Ping timeout: 248 seconds]
[01:58:51] -!- isacdaavid has joined #archlinux-ports
[02:04:57] -!- guys has quit [Ping timeout: 260 seconds]
[02:25:54] -!- guys has joined #archlinux-ports
[02:48:09] -!- isacdaavid has quit [Quit: Leaving.]
[05:01:59] -!- hringriin_ has joined #archlinux-ports
[05:04:49] -!- hringriin has quit [Ping timeout: 248 seconds]
[05:04:50] hringriin_ is now known as hringriin
[05:35:32] -!- tyzoid has quit [Ping timeout: 260 seconds]
[05:37:30] -!- tyzoid has joined #archlinux-ports
[05:54:10] -!- bill-auger has quit [Ping timeout: 268 seconds]
[05:58:05] -!- tyzoid has quit [Read error: Connection reset by peer]
[06:00:16] -!- hringriin_ has joined #archlinux-ports
[06:02:57] -!- hringriin has quit [Ping timeout: 240 seconds]
[06:02:58] hringriin_ is now known as hringriin
[06:03:49] -!- tyzoid has joined #archlinux-ports
[06:05:58] -!- deep42thought has joined #archlinux-ports
[06:13:17] -!- tyzoid has quit [Ping timeout: 268 seconds]
[06:15:17] -!- tyzoid has joined #archlinux-ports
[06:59:20] -!- deep42thought has quit [Quit: Leaving.]
[07:14:17] -!- p71 has quit [Read error: Connection reset by peer]
[07:17:59] -!- p71 has joined #archlinux-ports
[08:11:32] -!- deep42thought has joined #archlinux-ports
[09:29:37] <mrkiko> good morning all! :)
[09:29:48] <deep42thought> good morning!
[09:30:47] <mrkiko> :D
[09:53:51] -!- abaumann has joined #archlinux-ports
[09:53:57] <deep42thought> Hi abaumann
[09:54:06] <abaumann> good morning :-)
[09:54:30] <abaumann> I'm currently working on fixes for qbittorrent and crystal..
[09:54:39] <deep42thought> skip crystal
[09:54:48] <abaumann> but it has 32-bit support.
[09:54:51] <deep42thought> yes
[09:54:58] <deep42thought> but I'm almost done with it
[09:55:01] <abaumann> it brings up an interesting issue around source_${arch}.
[09:55:08] <abaumann> ah. :-)
[09:55:08] <deep42thought> just didn't have the time to push the fix, yet
[09:56:18] <deep42thought> that's the issue if you work on three tasks in parallel ...
[09:56:55] <abaumann> crystal takes ages to build.. that's also a reason.. :-)
[09:57:17] <deep42thought> nah, I just wait, that it does not stumble over the initial error anymore :-)
[09:58:06] <abaumann> patching source_x86_64 by removing it with a sed in PKGBUILD doesn't seem to work, it still gets downloaded
[09:58:19] <abaumann> afterwards I had a source_i686=(), which run in the chroot correctly.
[10:00:05] <deep42thought> you need to 'unset source_x86_64'
[10:00:54] <abaumann> ah. obviously. :-)
[11:02:45] <abaumann> libreoffice-still: I see you put a 'make build-nocheck' into libreoffice-fresh.
[11:03:04] <abaumann> libreoffice-still fails in this unit test 'dbaccess_hsqldb_test' with a segfault.
[11:03:41] <abaumann> should we just do the same as in libreoffice-fresh, not run the test cases?
[11:03:53] <deep42thought> hmm, oh, yes, I did
[11:04:16] <deep42thought> yeah, just do it
[11:04:22] <abaumann> ok.
[12:55:46] <tyzoid> good morning
[12:58:41] <tyzoid> deep42thought: I'm wondering if we should make pool.mirror.archlinux32.org for all of the T1 mirrors.
[12:59:11] <tyzoid> motivating factor: https://bbs.archlinux32.org
[12:59:13] <phrik> Title: where can i get the rootfs of archlinux32 ? / Installation / Arch Linux 32 Forums (at bbs.archlinux32.org)
[12:59:21] <tyzoid> people who don't care too much about speed and just want uptime.
[13:19:08] <tyzoid> forum migrated
[13:19:11] <tyzoid> finally
[13:19:17] <tyzoid> now to do the news page
[13:23:52] <deep42thought> we had the discussion of providing dns aliases for mirrors, before, on the mailing list
[13:25:48] <tyzoid> I remember that. I was against it.
[13:26:11] <tyzoid> I was also against jpn.mirror.archlinux32.org
[13:26:27] <deep42thought> oh, I missed that
[13:26:33] <deep42thought> what alternative do you suggest?
[13:28:28] <tyzoid> to jpn.mirror.archlinux32.org?
[13:28:56] <deep42thought> to the naming scheme of these mirrors (in general)
[13:29:00] <tyzoid> pool.archlinux32.org I think is okay enough.
[13:29:12] <tyzoid> The main issue is jpn.mirror.archlinux32.org makes it seem like an official mirror
[13:29:28] <tyzoid> instead of a community T1 mirror.
[13:29:51] <abaumann> jpn rings a "Japanese mirror bell" for me :-)
[13:29:56] <deep42thought> right
[13:30:00] <tyzoid> I'd be fine with jpn.pool.archlinux32.org
[13:30:36] <tyzoid> since the 'pool' is imo a decent indication that it's provided for convenience.
[13:30:49] <deep42thought> Hmm, I don't know - to me "pool" vs "mirror" doesn't make a big difference in deciding wether it's official vs. community
[13:30:51] <tyzoid> much like pool.ntp.org
[13:31:39] <tyzoid> I'm only suggesting existing conventions used elsewhere. If we want, we could go with a different domain entirely
[13:31:52] <tyzoid> or something like jpn.community.mirror.archlinux32.org
[13:32:00] <tyzoid> but then it gets a bit long
[13:32:03] <deep42thought> "pool" sounds like "big container, you put a lot of the same things inside"
[13:32:10] <deep42thought> yeah, too long is bad, probably
[13:32:14] <tyzoid> pools can have one mirror :P
[13:32:23] <tyzoid> I'm thinking of it as a pool of all mirrors in that region
[13:32:34] <deep42thought> honestly, I don't like "pool.archlinux.org" - in the end they're mirrors, and that should probably be reflected in their domain
[13:32:42] <deep42thought> maybe pool.mirror.archlinux32.org
[13:33:03] <tyzoid> That'd be fine
[13:33:20] <tyzoid> I'm thinking of the pool having all T1/T2 servers (we've only got T1 now)
[13:33:32] <tyzoid> then <region>.pool.mirror.archlinux32.org can have the regional mirrors.
[13:33:35] <abaumann> ARM has: http://mirror.archlinuxarm.org for the pool one, http://br.mirror.archlinuxarm.org for specific ones.
[13:33:37] <phrik> Title: Index of / (at mirror.archlinuxarm.org)
[13:33:38] <tyzoid> Still a bit long, but I think it's nice.
[13:33:47] <tyzoid> interesting
[13:33:56] <tyzoid> we've set up our infra. to use mirror as the mastor
[13:33:57] <tyzoid> master*
[13:34:07] <deep42thought> and then we'd introduce stuff like 'de.pool.mirror.archlinux32.org'?
[13:34:20] <tyzoid> yea, or eu.pool.mirror.archlinux32.org
[13:34:29] <deep42thought> but alarm also has some degree of control over their mirrors
[13:34:29] <tyzoid> since I don't like putting the master in the pool
[13:34:33] <deep42thought> because they push
[13:34:43] <tyzoid> yeah. That also makes a difference
[13:35:12] <abaumann> The ARM way is more consistent, but it creates a management problem on the DNS level.
[13:35:13] <deep42thought> I'm currently considering 'pool.mirror.archlinux32.org' and '$location.pool.mirror.archlinux32.org'
[13:35:38] <tyzoid> That'd be my vote.
[13:35:40] <deep42thought> pool.mirror.archlinux32.org also creates a management problem on the DNS level
[13:36:04] <tyzoid> It does, but we can do that after we get a tracker, I think.
[13:36:59] <abaumann> tyzoid: how exactly is this pool-thingy working?
[13:37:00] <deep42thought> a mirror tracker?
[13:37:36] <deep42thought> abaumann: I guess, we just set up A/AAAA records for all the mirrors on the mirror list and the requesting client picks one?
[13:37:51] <abaumann> ah. ok.
[13:38:14] <tyzoid> Yeah, just round-robin DNS.
[13:38:31] <deep42thought> but on the other hand, we need to edit that list each time one of the mirror operators changes ips
[13:39:01] <tyzoid> Though if mirrors have different update frequencies, we'd need to make sure pacman can handle 404s by trying the next IP in the pool.
[13:39:25] <deep42thought> not good
[13:39:28] <tyzoid> since it'd be possible to get an updated DB in one request, and outdated files in others.
[13:39:42] <abaumann> How is mirror.archlinuxarm.org working?
[13:39:42] <deep42thought> I think, pacman considers one 'Server=' record to be one server
[13:40:04] <abaumann> It looks like a single-point-IP to me, no round robin.
[13:40:14] <deep42thought> they use geoip, afaik
[13:40:20] <tyzoid> I doubt they'd have anycast
[13:40:27] <tyzoid> My guess would be load balancer
[13:40:42] <abaumann> ah, ok. And a single point of failure, eventually.
[13:40:53] <deep42thought> WarheadsSE: ^
[13:40:59] <abaumann> yep. :-)
[13:41:00] <tyzoid> :)
[13:41:09] <deep42thought> https://archlinuxarm.org
[13:41:10] <phrik> Title: Mirrors | Arch Linux ARM (at archlinuxarm.org)
[13:41:39] <tyzoid> abaumann: Ah, so they have a custom DNS server that serves you a local IP.
[13:42:09] <tyzoid> aka none of the above :P
[13:42:31] <abaumann> Yeah. And it seems, they serve the same client IP the same mirror. This would solve the possible sync problems and getting outdated packages in the DNS round-robin case
[13:42:51] <tyzoid> I like that approach better.
[13:43:02] <tyzoid> Still requires mirror monitoring.
[13:43:12] <tyzoid> and more work to set up.
[13:43:19] <abaumann> ah. poking with curl, each session gets a different Location.
[13:43:47] <abaumann> I'm toggling between nl.mirror.archlinuxarm.org and it.mirror.archlinux.org :-)
[13:44:08] <deep42thought> sounds like it's not as stable as we'd need
[13:44:24] <abaumann> So they most likely have a small piece of software knowing about all mirrors which redirects based on GeoIP
[13:45:47] <abaumann> Upstream Archlinux seems to implement a similar scheme, just with 3 worldwide master mirrors.
[13:47:00] <abaumann> with with the difference that being a CNAME/A, mirrors.evowise.com a 5-peered redundant IP and mirror.rackspace.com a single IP... mmh.
[13:47:16] <abaumann> first mirror was archlinux.surlyjake.com
[13:49:08] <deep42thought> why don't we keep the mirror list's dns structure as is and add pool.mirror.archlinux32.org (plus $location variants) dns records for the purpose outside of pacman?
[13:49:12] <brtln> we don't implement anything
[13:49:49] <brtln> it's mirror providers who do, we don't do anything fancy except for having tiered mirrors
[13:52:02] <tyzoid> btw, news page is migrated over too
[13:52:21] <deep42thought> nice :-)
[13:55:46] <tyzoid> ha
[13:55:48] <tyzoid> ha
[13:55:50] <tyzoid> haha
[13:55:52] <tyzoid> hahahahahaha
[13:56:07] <tyzoid> I'm looking at giving you access to bbs/news again
[13:56:09] <tyzoid> but
[13:56:20] <tyzoid> I'm looking at the server I just migrated from
[13:56:30] <tyzoid> which didn't have access in the first place :/
[13:57:51] <abaumann> :-)
[14:21:54] <WarheadsSE> deep42thought: tyzoid technically our master _is_ in the pool. We just don't mention which ;) We then use PowerDNS + GeoIP + round-robin depending on the level you've requested via DNS
[14:22:57] <WarheadsSE> All tied into PlugBuild, so that it turns them off while we push to them, and failed sync's cause offline after the second failure.
[14:23:59] <WarheadsSE> The "downside" to that is SSL on the mirrors is neigh impossible.
[14:29:05] <tyzoid> WarheadsSE: If you're willing to comprimise and distribute ssl keys via scp/other secure tunneling, you can have ssh.
[14:29:09] <tyzoid> https*
[14:29:19] <tyzoid> WarheadsSE: Beats no ssl at all
[14:29:23] <tyzoid> imo
[14:29:48] -!- hringriin_ has joined #archlinux-ports
[14:34:56] <tyzoid> deep42thought: Do you have a minute to help me debug sshd?
[14:35:02] <deep42thought> yeah
[14:35:27] <tyzoid> I keep getting read: Connection reset by peer when I try to sshfs
[14:35:45] <tyzoid> deep42thought: It's set up with your keys too, bbs-archlinux32@bbs.archlinux32.org
[14:54:31] <deep42thought> I set up some dns for pool.mirror.archlinux32.org
[14:59:24] <deep42thought> this should get updated automatically from our mirrorlist on github
[14:59:36] <deep42thought> I skipped all https:// only mirrors
[15:00:04] <deep42thought> for this to work, the mirrors should additionally all have the same directory structure
[15:03:39] <tyzoid> one sec, I need to add that vhost
[15:05:51] <tyzoid> deep42thought: And it looks like you've got HSTS set on pool.mirror.archlinux32.org :/
[15:05:52] -!- sunshavi has quit [Remote host closed the connection]
[15:06:20] <deep42thought> hmm
[15:06:29] <tyzoid> Probably residual headers set from mirror.archlinux32.org
[15:06:52] <tyzoid> (Set the ip to my server in the hosts file, and got hsts. I don't send hsts for my mirror)
[15:07:14] <deep42thought> I don't for mine, too
[15:07:19] <tyzoid> hmm
[15:07:33] <tyzoid> anyway, my mirror now responds to pool.mirror.archlinux32.org and us.pool.mirror.archlinux32.org
[15:07:49] <deep42thought> Haven't set up locations, yet
[15:07:55] <deep42thought> I would do that later ...
[15:08:05] <tyzoid> yup. I figured I'd set up the vhost so it just works when you do :)
[15:08:35] <tyzoid> It's nice when things 'just work'
[15:08:41] <deep42thought> :-)
[15:08:47] <cypa[m]> firefox quantum constantly writes a few M/s to disk on youtube video
[15:08:54] <tyzoid> :/
[15:08:57] <deep42thought> sometimes, I want to implement something and then I realize, I already did
[15:09:13] <cypa[m]> while chromium do not
[15:09:21] <tyzoid> cypa[m]: Do you have enough ram to support the stream in-memory?
[15:09:36] <cypa[m]> 1Gb
[15:10:13] <cypa[m]> but chromium do not write that much to disk
[15:10:17] <cypa[m]> a few K/s
[15:10:43] <cypa[m]> and chromium uses 3fold less cpu
[15:11:15] <cypa[m]> some google trick with youtube
[15:11:21] <cypa[m]> &
[15:11:23] <cypa[m]> ?
[15:11:56] <cypa[m]> maybe hardware accel do not enabled in firefox?
[15:12:30] <tyzoid> cypa[m]: https://lifehacker.com
[15:15:58] <cypa[m]> OK
[15:21:09] <deep42thought> WarheadsSE: what's actually the issue with ssl? Doesn't let's encrypt like to have multiple certificates for the same domain or what?
[15:29:21] <WarheadsSE> deep42thought: yes, but if you don't control the mirrors, how do you securely place them?
[15:29:58] <deep42thought> well, It's the responsibility of the mirror operator to get ssl certificates, isn't it?
[15:29:59] <WarheadsSE> And if DNS reponds with the IP of br.mirror.archlinuxarm.org when queried for mirror.archlinuxarm.org, how do you handle that?
[15:30:08] <deep42thought> can't you just set up let's encrypt for that domain?
[15:30:13] <WarheadsSE> They can't get SSL certs for a domain they don't control
[15:30:50] <deep42thought> ah, I think, I got it
[15:50:49] <tyzoid> The way I see it, just get a letsencrypt cert for mirror.archlinuxarm.org on the master
[15:50:54] <tyzoid> then distribute that to the slave servers
[15:51:19] <deep42thought> actually, http is fine for mirrors
[15:51:22] <deep42thought> isn't it?
[15:51:31] <tyzoid> if they sign the packages, sure.
[15:51:35] <tyzoid> idk if they do
[15:51:47] <deep42thought> well, _we_ sign the packages
[15:51:51] <tyzoid> plus https is nice if you've got a corporate firewall, since it'll usually let those through.
[15:52:01] <deep42thought> yeah
[15:52:12] <tyzoid> case in point: I needed to set https://arlm.tyzoid.com instead of http://arlm.tyzoid.com to get through my corporate firewall
[15:52:14] <phrik> Title: Index of / (at arlm.tyzoid.com)
[15:52:18] <tyzoid> thanks, phrik
[15:52:30] <tyzoid> very helpful
[15:52:32] <tyzoid> lol
[16:14:11] -!- deep42thought has quit [Quit: Leaving.]
[16:19:42] -!- hringriin_ has quit [Quit: ^Cc12^CbSee ^Cb^Cc04^Cya^Cc!]
[16:32:04] -!- sunshavi has joined #archlinux-ports
[16:41:28] <tyzoid> Just posted a list of mirrors that need updating: https://bbs.archlinux32.org
[16:41:30] <phrik> Title: Mirror Pool / Mirrors / Arch Linux 32 Forums (at bbs.archlinux32.org)
[17:08:33] -!- because has quit [Ping timeout: 255 seconds]
[17:10:08] -!- because has joined #archlinux-ports
[17:39:01] -!- because has quit [Ping timeout: 268 seconds]
[17:39:22] -!- abaumann has quit [Quit: ChatZilla 0.9.92 [SeaMonkey 2.49.1/20171107234153]]
[17:39:58] -!- because has joined #archlinux-ports
[17:45:21] -!- because has quit [Ping timeout: 248 seconds]
[17:46:17] -!- because has joined #archlinux-ports
[17:56:27] -!- because has quit [Ping timeout: 240 seconds]
[18:36:47] <tyzoid> rewbycraft: Can you update your mirror not to redirect if it's accessed via pool.mirror.archlinux32.org?
[18:36:55] <tyzoid> https://bbs.archlinux32.org
[18:36:56] <phrik> Title: Mirror Pool / Mirrors / Arch Linux 32 Forums (at bbs.archlinux32.org)
[18:52:29] -!- bill-auger has joined #archlinux-ports
[19:10:23] -!- zaungast has joined #archlinux-ports
[20:09:01] -!- because has joined #archlinux-ports
[22:17:36] -!- deep42thought has joined #archlinux-ports
[22:22:06] -!- jonathon has joined #archlinux-ports
[22:23:58] <jonathon> hi! is anyone else getting "maximum file size reached" errors from the repos?
[22:24:11] <deep42thought> which mirror?
[22:24:21] <jonathon> archlinux32 repos, that is...
[22:24:22] <deep42thought> and which file?
[22:24:57] <jonathon> both archlinux32.vollzornbrot.de and mirror.archlinux32.org, a variety of files in each repo... let me paste and link...
[22:26:01] <jonathon> https://bpaste.net
[22:26:03] <phrik> Title: show at bpaste (at bpaste.net)
[22:26:13] <jonathon> it _could_ just be me
[22:27:38] <deep42thought> gawk-4.1.4-2-i686.pkg.tar.xz downloads just fine for me from mirror.archlinux32.org
[22:28:07] <deep42thought> same for vollzornbrot.de
[22:28:23] <jonathon> downloading directly or via pacman?
[22:28:38] <deep42thought> directly
[22:28:53] <deep42thought> let me boot my archlinux32 box
[22:28:56] <jonathon> wget works fine, if there's an issue it's with the package database itself
[22:29:41] <deep42thought> are these all [core] packages?
[22:29:59] <deep42thought> or are there packages from other repositories, which also fail?
[22:30:01] <jonathon> those in the paste are, those are just base-devel
[22:30:15] <jonathon> i've seen it for extra too, not cure about community
[22:30:17] <jonathon> sure*
[22:30:23] <deep42thought> hmm
[22:31:04] <deep42thought> looks good on my end (with pacman)
[22:31:20] <deep42thought> what pacman and icu version do you have?
[22:32:09] <jonathon> 5.0.2, 59.1
[22:32:35] <deep42thought> and behind the "-"?
[22:32:52] <jonathon> oops, sorry - 5.0.2-2, 59.1-2
[22:33:14] <deep42thought> ah
[22:33:41] <deep42thought> you most probably need to install icu-60.1-1
[22:33:54] <deep42thought> ah, wait
[22:34:15] <deep42thought> does 'pacman -Syu' also fail?
[22:35:02] <deep42thought> Might it be, that you did a partial upgrade - which is a bad idea in general?
[22:37:54] <jonathon> gah... quite possibly.. not sure how I've ended up with that...
[22:38:14] <jonathon> i can -Syu fine (which brought in icu=60.1), and then -S base-devel works fine
[22:38:25] <deep42thought> :-)
[22:39:18] <jonathon> thanks for your help :)
[22:39:31] <tyzoid> np.
[22:39:36] <deep42thought> np - I'm glad, it was not my fault (this time)
[22:39:40] <tyzoid> lol
[22:39:52] <tyzoid> deep42thought: mirror.archlinux32.org doesn't respond to pool.mirror
[22:40:37] <deep42thought> it's not in the dns either
[22:40:54] <deep42thought> intentionally
[22:41:05] <tyzoid> ah, nice.
[22:41:56] <tyzoid> makes sense
[22:42:04] <deep42thought> once we have some mirror statistics up, we can use that to seed the dns - currently it's seeded from the mirrorlist on github :-/
[22:42:44] <tyzoid> the mirrorlist on github has the master mirror, though
[22:42:52] <tyzoid> do you explicitly exclude that one?
[22:43:07] <deep42thought> yes, I do
[22:56:58] -!- jonathon has quit [Quit: Quit]
[22:57:21] -!- isacdaavid has joined #archlinux-ports
[23:11:22] -!- isacdaavid has quit [Ping timeout: 260 seconds]
[23:31:14] -!- zaungast has quit [Quit: Too late now.]
[23:34:01] -!- because has quit [Ping timeout: 240 seconds]
[23:34:51] -!- because has joined #archlinux-ports
[23:41:33] -!- deep42thought has quit [Quit: Leaving.]