2021-02-01 18:12:36 <clandmeter> ping ikke 
2021-02-01 18:12:43 <ikke> pong
2021-02-01 18:12:49 <ikke> was just reading the e-mail
2021-02-01 18:12:53 <clandmeter> see mail?
2021-02-01 18:12:56 <ikke> :P
2021-02-01 18:13:33 <ikke> arm32 "will require additional modification / configuration"
2021-02-01 18:14:12 <clandmeter> yes not sure what that means exactly
2021-02-01 18:14:20 <clandmeter> but i think we only get 1 machine in return
2021-02-01 18:14:27 <ikke> hmm
2021-02-01 18:15:13 <ikke> so it means we will have to do armhf, armv7 and aarch64 on a single machine?
2021-02-01 18:15:27 <clandmeter> yup
2021-02-01 18:15:56 <ikke> will be fun
2021-02-01 18:20:23 <clandmeter> ok i emailed them back
2021-02-01 18:20:27 <clandmeter> would be nice to keep 2
2021-02-01 18:20:35 <clandmeter> even if its less powerfull 
2021-02-01 18:37:40 <ikke> ok, so yes, 1 machine
2021-02-02 12:02:16 <ncopa> I merged https://gitlab.alpinelinux.org/alpine/infra/docker/secdb/-/merge_requests/3 do I need something more for 3.13 to show up on https://secdb.alpinelinux.org/ ?
2021-02-02 12:03:25 <ikke> I think we need to pull that image and restart the container
2021-02-02 12:05:57 <ikke> yes, pulled the latest version of the repo
2021-02-02 12:06:00 <ikke> now it shows up
2021-02-03 08:47:25 <Cogitri> ikke: Could you move all `icons-*-aarch64-*.tar.gz` to `icons-*-*.tar.gz` (as in remove the `-aarch64` from the file name) and restart appstream-generator after pulling the new image?
2021-02-03 08:48:08 <Cogitri> Turns out one of my downstream patches wasn't quite up to speed so icon archives weren't generated for all arches (now it only generates one icon archive for all arches and names it `icons-community-128x128.tar.gz`)
2021-02-03 08:48:12 <ikke> Cogitri: can you prepare a `rename` command for that?
2021-02-03 08:56:02 <Cogitri> ikke: Sure thing: https://paste.gnome.org/prkkjqj0e
2021-02-03 08:56:34 <Cogitri> Err https://paste.gnome.org/pcjnch4go, whoops :D
2021-02-03 08:56:56 <ikke> Free dry-run option :P
2021-02-03 08:57:06 <Cogitri> Yup ^^
2021-02-03 08:57:28 <Cogitri> Oh, seems like the new image isn't uploaded yet
2021-02-03 08:57:42 <Cogitri> https://gitlab.alpinelinux.org/alpine/infra/docker/appstream-generator/-/jobs/311065
2021-02-03 08:58:05 <Cogitri> Also, please pull the compose so https://gitlab.alpinelinux.org/alpine/infra/compose/appstream-generator/-/commit/735b2d583062a0c8d0e5e396da3d10b0ebdcb2b7 is included :)
2021-02-03 20:40:37 <ikke> Cogitri: for aarch64, there are only 64x64 icons
2021-02-03 22:28:23 <Cogitri> Hm?
2021-02-04 09:33:36 <ikke> Cogitri: the script you provided failed 
2021-02-04 09:50:03 <Cogitri> rip
2021-02-04 09:50:10 <Cogitri> ikke: What doesn't work?
2021-02-04 10:04:19 <ikke> It cannot find any of the source files 
2021-02-04 10:06:05 <Cogitri> You're in the export dir of asgen?
2021-02-04 10:10:37 <ikke> Yes 
2021-02-04 10:10:54 <ikke> I can check again later 
2021-02-04 11:17:53 <ikke> Cogitri: https://tpaste.us/x1Rw 
2021-02-04 11:20:52 <Cogitri> Oh yes, should've been 64x64 and 128x128 in the loop, not 128x128 and 256x256, my bad
2021-02-04 11:21:00 <Cogitri> Could you run the loop again with just 64x64? Should work then
2021-02-04 11:21:17 <Cogitri> (Might still throw errors for 3.12 testing since I didn't bother adding checks if the file actually exists :)
2021-02-04 11:22:26 <ikke> ok, done
2021-02-04 11:22:57 <ikke> pulled the latest images, and recreated the containers
2021-02-04 11:23:10 <Cogitri> Alright, thanks :) 
2021-02-04 11:23:49 <Cogitri> Let's hope it runs through, should make support for GNOME Software and KDE Discover in Alpine (and pmOS) pretty neat 
2021-02-04 11:24:45 <ikke> btw, you could provide these kind of migration scripts through docker-compose as well
2021-02-04 11:25:08 <ikke> docker-compose run --rm migrate-icons
2021-02-04 11:25:40 <ikke> (For this it's probably overkill)
2021-02-04 11:26:15 <Cogitri> Oh, fancy, I'll have to take a look at that
2021-02-04 11:27:24 <ikke> It creates a new container for the command being executed with the given volumes
2021-02-04 11:57:43 <ikke> Cogitri: testing building gitlab on gitlab :P https://gitlab.alpinelinux.org/kdaudt/gitlab/-/jobs/312250
2021-02-04 12:51:38 <Cogitri> Oh, we build Gitlab ourselves :o
2021-02-04 13:15:13 <ikke> Yes, slightly less "everything in one container" and based on Alpine 
2021-02-04 16:49:39 <ikke> clandmeter: ping
2021-02-04 17:49:30 <clandmeter> 
ACTION hides

2021-02-04 17:51:11 <ikke> I'm working on the gitlab image for 13.6
2021-02-04 18:04:33 <clandmeter> Ok
2021-02-04 18:05:03 <clandmeter> Anything you need my help with?
2021-02-04 18:08:03 <ikke> We don't ship the grpc version specified in the gemfile
2021-02-04 18:08:12 <ikke> 3.12 is too old, 3.13 is newer
2021-02-04 18:08:59 <ikke> We need to switch already to ruby2.7
2021-02-04 18:09:25 <ikke> clandmeter: anything on the gitlab test server you need to keep, otherwise I'll redeploy it
2021-02-04 18:25:03 <ikke> oh, it seems like it just uses the gem
2021-02-04 18:25:18 <ikke> for grpc
2021-02-05 11:12:03 <Cogitri> ikke: Are you sure you also pulled the docker-compose of asgen? Seems like v3.13 isn't generated
2021-02-05 11:14:54 <ikke> Cogitri: very sure
2021-02-05 11:17:24 <ikke> https://gitlab.alpinelinux.org/alpine/infra/compose/appstream-generator/-/commit/735b2d583062a0c8d0e5e396da3d10b0ebdcb2b7
2021-02-05 11:18:11 <Cogitri> huh, wtf
2021-02-05 11:18:24 <Cogitri> Seems like it did regenerate the data here: https://appstream.alpinelinux.org/data/ but 3.13 is nowhere to be found
2021-02-05 11:19:43 <clandmeter> ikke: yes you need to set the version of grpc to build it manually
2021-02-05 11:20:53 <ikke> clandmeter: it pulled the 1.30.2 gem in, as specified in the Gemfile
2021-02-05 11:21:08 <ikke> Installing grpc 1.30.2 with native extensions
2021-02-05 11:21:20 <clandmeter> ok, its not ok?
2021-02-05 11:24:07 <ikke> clandmeter: I think it is
2021-02-05 11:24:12 <ikke> Just wanted to verify
2021-02-05 11:24:32 <ikke> The readme said to verify the version, but I see that later the code to build grpc manually had been removed
2021-02-05 11:24:39 <clandmeter> i think the troubles previously are solved
2021-02-05 11:25:41 <ikke> right
2021-02-05 11:25:54 <ikke> so my plan was to try ruby:2.7-alpine3.13
2021-02-05 12:03:30 <ikke> clandmeter: recreating the test instance
2021-02-05 12:38:53 <clandmeter> Didn't i try it already? Maybe I'm my dreams
2021-02-05 12:58:13 <ikke> clandmeter: I no you tested it, but it's not available afaics
2021-02-05 12:58:27 <ikke> I think*
2021-02-05 17:21:44 <ikke> clandmeter: https://gitlab.alpinelinux.org/alpine
2021-02-05 17:21:54 <ikke> https://gitlab-test.alpinelinux.org/alpine
2021-02-06 14:45:26 <mps> machine 172.16.4.44 (x86_64 developers macchine) can't be accessed over vpn? or it something problem on my side
2021-02-06 14:46:55 <mps> vpn.alpinelinux.org answers pings
2021-02-06 14:51:36 <ikke> hmm, no, indeed
2021-02-06 14:52:38 <mps> who is vpn admin, kunkku?
2021-02-06 14:52:51 <ikke> yes
2021-02-06 14:53:07 <ikke> I cannot reach anything on the dmvpn
2021-02-06 14:53:23 <mps> also I can't ;)
2021-02-06 14:54:44 <mps> we have two 'one bus factor'
2021-02-06 14:55:13 <ikke> I do have access to the dmvpn hosts
2021-02-06 14:55:18 <ikke> checking now
2021-02-06 14:55:43 <mps> ah, that is good
2021-02-06 14:58:35 <ikke> fixed
2021-02-06 14:59:30 <mps> yes, thanks
2021-02-06 15:07:54 <ikke> I see my trigger for services on Zabbix is not working
2021-02-06 15:08:53 <ikke> it should've caught this
2021-02-06 15:15:10 <ikke> service crashed about 1h ago
2021-02-06 15:16:09 <mps> maybe kernel should be upgraded on x86 devs lxc
2021-02-06 15:16:20 <mps> it is still 4.19
2021-02-06 15:16:22 <ikke> yeah
2021-02-06 15:16:29 <ikke> I want to upgrade all our infra to Alpine 3.13
2021-02-06 15:17:00 <mps> for testing start with devs boxes :)
2021-02-06 15:17:02 <ikke> nld5 (x86_64) dev is a good candidate to start with
2021-02-06 15:17:05 <ikke> yes
2021-02-06 15:17:32 <ikke> But I first want to upgrade gitlab
2021-02-06 15:17:58 <mps> ok, I don't force anything
2021-02-06 15:18:30 <mps> just sometimes have issues when testing/building some low level pkgs
2021-02-06 15:19:25 <ikke> "auth.info sshd[27509]: Invalid user mps" :P
2021-02-06 15:20:17 <ikke> /usr/sbin/nhrpd, pid 27157, terminated by signal 11
2021-02-06 15:20:49 <mps> hehe
2021-02-06 15:20:54 <ikke> nhrpd[27038]: segfault at 556143014ef0 ip 0000556143014ef0 sp 00007fff65a297f8 error 15
2021-02-06 15:21:38 <ikke> kunkku: fyi ^
2021-02-06 15:21:39 <mps> so i discovered bug in nhrpd ;)
2021-02-06 15:21:57 <ikke> I think it happened before
2021-02-06 15:22:16 <ikke> https://gitlab.alpinelinux.org/alpine/aports/-/issues/11755
2021-02-06 15:22:20 <mps> aha, then confirmed it
2021-02-06 15:23:26 <ikke> old version is still running
2021-02-06 15:24:08 <ikke> the fix was not bacported
2021-02-06 15:33:22 <ikke> backporting the fix
2021-02-06 15:36:59 <ikke> aaaand it faild again
2021-02-06 15:37:08 <ikke> let's see if the trigger works now
2021-02-06 15:42:07 <ikke> Yaay
2021-02-06 20:06:41 <kunkku> ikke: was that crash on dmvpn1?
2021-02-06 20:07:54 <ikke> yes
2021-02-06 20:08:22 <ikke> kunkku: I've backported a fix
2021-02-06 20:08:30 <ikke> that fabled did on edge
2021-02-06 20:09:02 <kunkku> the old core dump was there, so we didn't get a new one
2021-02-06 20:09:38 <kunkku> renaming it
2021-02-06 20:10:38 <ikke> I've also upgraded sudo for the CVE
2021-02-06 20:11:01 <kunkku> but did it crash again after backport?
2021-02-06 20:11:14 <ikke> no, not yet
2021-02-06 20:11:28 <ikke> I've upgraded it after the last crash that you see mentioned in this channel
2021-02-06 20:11:37 <ikke> before, it crashed twice again in short succession
2021-02-06 20:17:59 <kunkku> ok
2021-02-06 20:19:46 <kunkku> seems BRA1 and GBR1 are running old dmvpn-tools versions
2021-02-06 20:22:11 <kunkku> it's the combination of buggy nhrpd and dmvpn-tools that cause these crashes
2021-02-06 20:25:41 <kunkku> probably dmvpn2 should also get the backport
2021-02-06 20:30:15 <kunkku> which AL versions are running on bra1 and gbr1?
2021-02-06 21:32:36 <ikke> kunkku: gbr1 is 3.10
2021-02-06 21:33:13 <ikke> bra1 is even 3.9
2021-02-06 22:16:16 <ikke> clandmeter: ugh, there is an anoying issue in gitlab 13.5/13.6 (error bar on merge requests). This is fixed in 13.7, but not backported to 13.5/6
2021-02-06 22:16:53 <ikke> https://gitlab.com/gitlab-org/gitlab/-/issues/272983
2021-02-06 22:25:57 <ikke> So I guess we'll go straight ahead to 13.7?
2021-02-06 22:53:36 <clandmeter> ikke: hi
2021-02-06 22:53:51 <clandmeter> 13.7 sounds ok?
2021-02-07 07:17:19 <ikke> clandmeter: will do that
2021-02-07 10:49:49 <ikke> clandmeter: 13.7 is running now on gitlab-test.alpinelinux.org
2021-02-07 10:55:43 <mps> \o/
2021-02-07 10:56:16 <mps> yet heavier JS is on horizon, I guess :)
2021-02-07 10:56:44 <ikke> If I must believe the release notes, each release promises to be more performant :P
2021-02-07 10:57:24 <mps> lies, damned lies and marketing :)
2021-02-07 10:57:41 <ikke> hehe
2021-02-07 11:42:11 <clandmeter> Nice
2021-02-07 11:42:29 <ikke> I think I'll upgrade tonight
2021-02-07 11:42:29 <clandmeter> I guess without issues
2021-02-07 11:42:34 <ikke> yes
2021-02-07 11:43:02 <ikke> at least, not that particular one
2021-02-07 11:43:05 <ikke> api is working
2021-02-07 19:44:37 <ikke> clandmeter: gitlab has been upgraded to 13.7 :-()
2021-02-07 19:45:31 <clandmeter> nice
2021-02-07 19:45:41 <clandmeter> lets hope without any.bugs
2021-02-08 07:01:12 <ncopa> good morning! thanks for a good job with the gitlab upgrade
2021-02-08 09:06:29 <maxice8> please kill x86_64 builer 
2021-02-08 09:06:33 <maxice8> builder*
2021-02-08 09:06:38 <maxice8> stuck on mopidy
2021-02-08 10:05:56 <ncopa> ok
2021-02-08 11:24:09 <ikke> ncopa: thanks, still get nervous a bit each time I do it :)
2021-02-08 11:49:20 <clandmeter> yes, we get depended a lot on gitlab.
2021-02-10 09:22:54 <Cogitri> BTW, maybe don't upgrade Gitlab to 13.8.x, it's kind of buggy for GNOME
2021-02-10 09:23:21 <Cogitri> When merging MRs it just says there has been a merge conflict on merge and doesn't close the MR when it actually did apply the changes to master after all
2021-02-10 10:57:18 <clandmeter> Cogitri: we always stay 1 version behind
2021-02-10 10:57:25 <clandmeter> just to overcome such issues
2021-02-10 10:59:33 <Cogitri> Yup, thanks for being careful with Gitlab :)
2021-02-10 10:59:55 <ikke> and we can test it in advance
2021-02-10 10:59:59 <ikke> that's why we skipped 13.6
2021-02-10 11:00:19 <ikke> because of an innocent but anoying error message that appears on merge requests
2021-02-10 11:01:39 <Cogitri> Yup, it's really annoying when Gitlab does weird stuff, that bug on GNOME's gitlab makes me manually merge instead of merging via MRs
2021-02-10 16:42:46 <ikke> clandmeter: some new activity on our gitlab issue regarding developers being able to rebase internal projects
2021-02-11 08:05:19 <clandmeter> ikke: yup ive seen nit
2021-02-11 10:58:19 <ikke> Cogitri: seems that aports-qa-bot seems to fail enabling the collaboration checkbox on some MRs
2021-02-11 10:58:36 <Cogitri> Huh, weird
2021-02-11 10:58:52 <ikke> it's running, and I see that it received the webhook for that MR
2021-02-11 10:58:57 <ikke> and it did add labels
2021-02-11 10:59:26 <ikke> It happened to be the bitwarden_rs MR
2021-02-11 11:00:11 <Cogitri> Does it log `Enabling 'allow_collaboration' and 'remove_source_branch' for MR $SOME_NUM`?
2021-02-11 11:00:45 <Cogitri> It only enables collaboration if the MR status is open and it was just opened, but I don't see how a MR wouldn't do through that state at least once
2021-02-11 11:01:25 <ikke> https://tpaste.us/g6vy 
2021-02-11 11:02:08 <Cogitri> Hm, seems like it didn't go into the `if m.Inner.State == Opened && m.Inner.Action == Open` conditional then, how curious
2021-02-11 11:03:16 <ikke> Maybe I have to look further back
2021-02-11 11:03:23 <ikke> yeah
2021-02-11 11:05:47 <ikke> Cogitri: 2021/01/25 14:24:53 Enabling 'allow_collaboration' and 'remove_source_branch' for MR 17446 of project 1
2021-02-11 11:07:24 <Cogitri> Ah yes, the MR hook is triggered for every MR event, but only when opening the thing allow_collaboration is set
2021-02-11 11:07:40 <Cogitri> Did this happen for more MRs than bitwarden_rs? Maybe that one network request just went to /dev/null or something
2021-02-11 11:08:08 <ikke> maxice8 pinged me twice about this
2021-02-11 11:08:11 <ikke> for 2 MRs
2021-02-11 11:16:47 <Cogitri> Hm, weird. Seems like aports-qa-bot did send the API request for this and it didn't send an error back (otherwise `Failed to enable collaboration and removal of source branch for MR %d of project` would be in the logs)
2021-02-11 11:16:55 <Cogitri> It being gitlab here
2021-02-11 11:20:14 <maxice8> I guess we need aports-qa-bot to be more specific in its failures
2021-02-11 11:20:17 <maxice8> print the `err`
2021-02-11 11:21:47 <Cogitri> Hm sure, but I don't think it printed an error at all?
2021-02-12 17:57:24 <mps> looks like it is not bad idea to have '@' badge prefix for nick for some channels :)
2021-02-12 18:01:41 <ikke> freenode advise is to not have ops all the time
2021-02-12 19:43:50 <mps> now I think I didn't wrote this properly, I meant to say 'to have badge' but not to be ops, just psychological deterrent for some
2021-02-12 19:44:29 <mps> if I'm ops I doubt i would ever kick anyone
2021-02-15 08:38:46 <clandmeter> ikke: ping
2021-02-15 08:59:39 <clandmeter> ikke: have you seen instance_review
2021-02-15 08:59:44 <clandmeter> gitlab thjingy
2021-02-15 09:02:28 <ikke> No?
2021-02-15 09:05:31 <clandmeter> if you click your avatar menu on right top
2021-02-15 09:05:41 <clandmeter> maybe its only for me
2021-02-15 09:05:49 <clandmeter> get a free gitlab review
2021-02-15 09:06:34 <clandmeter> https://docs.gitlab.com/ee/administration/instance_review.html
2021-02-15 09:08:45 <ikke> I see it, just didnt notice it
2021-02-15 09:09:23 <clandmeter> there is also a security update
2021-02-15 09:09:32 <clandmeter> are you subscribed to sec emails?
2021-02-15 09:11:16 <ikke> apparently not
2021-02-15 09:14:36 <clandmeter> maybe a good thing to do :)
2021-02-15 09:16:14 <ikke> yes
2021-02-15 09:16:48 <clandmeter> https://about.gitlab.com/company/contact/
2021-02-15 09:21:24 <ikke> Do we want to request an instance review?
2021-02-15 10:06:34 <clandmeter> i dont know
2021-02-15 10:07:09 <clandmeter> i dont see major issues in doing so
2021-02-15 11:42:20 <clandmeter> ikke: ok?
2021-02-15 11:42:35 <ikke> clandmeter: both armv7 and aarch64 ci hosts run aarch64 atm
2021-02-15 11:42:46 <ikke> (vms)
2021-02-15 11:42:59 <clandmeter> right
2021-02-15 11:43:30 <ikke> I have no idea if it would make any difference, but it's a bit more eating our dogfood
2021-02-15 11:49:23 <clandmeter> ikke: the only thing you want to change is boot the armv7 kernel on qemu-system-aarch64?
2021-02-15 11:49:50 <ikke> there is no qemu-system-armv7?
2021-02-15 11:50:54 <ikke> qemy-system-arm
2021-02-15 11:51:09 <clandmeter> there is
2021-02-15 11:51:18 <clandmeter> but, from my experiance its limited to 4cpus
2021-02-15 11:51:23 <ikke> aha ok
2021-02-15 11:51:26 <clandmeter> i could not find any reference to why
2021-02-15 11:51:37 <clandmeter> I have been looking into it yesterday
2021-02-15 11:51:50 <clandmeter> mps: do you know what im talking about?
2021-02-15 11:52:57 <clandmeter> ikke: i know that setting one switch on aarch64 will make more cores available
2021-02-15 11:53:15 <clandmeter> qemu has a gaziilion options 
2021-02-15 11:58:20 <Hello71> probably -machine something
2021-02-15 11:58:31 <Hello71> assuming that -smp didn't help
2021-02-15 12:04:10 <clandmeter> machine -virt -cpu max
2021-02-15 12:04:33 <clandmeter> for aarch64 i set Sébastien
2021-02-15 12:04:36 <clandmeter> lol
2021-02-15 12:04:38 <clandmeter> wrong copy
2021-02-15 12:04:43 <clandmeter> gic-version
2021-02-15 12:04:51 <clandmeter> gic-version=max
2021-02-15 12:41:23 <mps> to boot qemu arm32 (armhf/armv7) with virtio-net-pci option is '-machine virt,highmem=off', without highmem it doesn't recognize net driver
2021-02-15 14:49:34 <clandmeter> mps, how did you know about highmem issue?
2021-02-15 14:53:38 <Hello72> this sounds suspiciously like a kernel config issue
2021-02-15 14:55:11 <mps> clandmeter: I tested it, without highmem=off virtio-net-pci doesn't work, with it works. it is related to PCI address space
2021-02-15 14:55:23 <clandmeter> ikke: regarding your question about arm vs aarch64
2021-02-15 14:55:33 <clandmeter> QEMU can emulate both 32-bit and 64-bit Arm CPUs. Use the qemu-system-aarch64 executable to simulate a 64-bit Arm machine. You can use either qemu-system-arm or qemu-system-aarch64 to simulate a 32-bit Arm machine: in general, command lines that work for qemu-system-arm will behave the same when used with qemu-system-aarch64
2021-02-15 14:55:40 <mps> and maybe building kernel in LPAE mode could fix this
2021-02-15 14:56:07 <clandmeter> mps i dont know what LPAE is
2021-02-15 14:56:30 <clandmeter> and qemu docs are not so easy to consume
2021-02-15 14:56:48 <mps> something about address space translation, i.e. use bigger address space
2021-02-15 14:57:13 <mps> address space extension?
2021-02-15 14:57:42 <mps> I had a hope that LPAE is not needed nowadays
2021-02-15 14:58:12 <mps> yes, qemu docs are black/white magic
2021-02-15 14:59:59 <mps> probably we can run armv7 with qemu-system-aarch64 but we then 'lost' some features which is nearer to real hardware
2021-02-15 15:00:41 <mps> but running it in qemu-system-aarch64 makes sense for our use cases in CI and maybe even builders
2021-02-15 15:03:35 <clandmeter> hah
2021-02-15 15:03:47 <clandmeter> its not qemu that is limiting the cpu count
2021-02-15 15:03:53 <clandmeter> its our armv7 lts kernel
2021-02-15 15:03:57 <clandmeter> which is set to 4
2021-02-15 15:05:38 <ikke> Oh, lol
2021-02-15 15:06:02 <clandmeter> https://git.alpinelinux.org/aports/tree/main/linux-lts/config-lts.armv7#n553
2021-02-15 15:06:04 <clandmeter> :)
2021-02-15 15:06:19 <mps> huh, my link is unstable
2021-02-15 15:06:42 <mps> I think I will soon be cut out of internet
2021-02-15 15:07:24 <clandmeter> elon will resque you
2021-02-15 15:08:16 <mps> https://en.wikipedia.org/wiki/ARM_architecture#LPAE
2021-02-15 15:08:16 <alpine-meetbot> [WIKIPEDIA] ARM architecture#LPAE | "ARM (stylised in lowercase as arm, previously an acronym for Advanced RISC Machines and originally Acorn RISC Machine) is a family of reduced instruction set computing (RISC) architectures for computer processors, configured for various environments. Arm Ltd. develops the architecture and licenses it..."
2021-02-15 15:08:43 <mps> yes, our kernel
2021-02-15 15:09:21 <mps> because that I wrote maybe to build kernel with LPAE
2021-02-15 15:10:18 <mps> and, nr of cpus
2021-02-15 15:10:41 <mps> for my armv7 exynos chromebook I build with 8 cpus
2021-02-15 15:12:56 <clandmeter> i think the starlink package is cheaper then my new fiber connection 
2021-02-15 15:13:01 <mps> https://git.alpinelinux.org/aports/tree/testing/linux-edge/config-edge.armv7#n542
2021-02-15 15:13:13 <mps> 8 in linux-edge
2021-02-15 15:13:39 <clandmeter> should probably up to 128 or so
2021-02-15 15:15:40 <Hello72> i wonder if the config help is even still accurate
2021-02-15 15:16:09 <mps> 128? :D
2021-02-15 15:16:38 <clandmeter> mps: yes
2021-02-15 15:16:46 <clandmeter> our new box will have more i think
2021-02-15 15:17:19 <mps> https://git.alpinelinux.org/aports/tree/testing/linux-edge/config-edge-virt.armv7#n434
2021-02-15 15:17:38 <mps> 32 in linux-edge virt armv7
2021-02-15 15:17:40 <clandmeter> Some 32-bit guests on the "virt" machine may need "-machine highmem=off" if they are unable to support LPAE and are confused by the PCI ECAM now being above the 4GB mark in the address space
2021-02-15 15:18:56 <mps> will test later in evening what will happen when starting it with 32 cpu
2021-02-15 15:19:10 <clandmeter> i would opt for at least 128 for virt if we want to run that in qemu on our new box
2021-02-15 15:19:30 <clandmeter> i think it will have 160 cores
2021-02-15 15:19:56 <clandmeter> https://www.anandtech.com/show/16315/the-ampere-altra-review
2021-02-15 15:20:09 <mps> ok, but I think BDFL will be frowned if we run linux-edge in our infra for anything
2021-02-15 15:20:33 <clandmeter> i talking about linux-virt
2021-02-15 15:20:56 <mps> aha, ok then
2021-02-15 15:21:12 <clandmeter> but you are free to up the count on edge too
2021-02-15 15:21:25 <clandmeter> i think it will only increase the kernel size a bit
2021-02-15 15:21:41 <mps> true
2021-02-15 15:22:08 <mps> though I forgot how much for each cpu, but not much
2021-02-15 15:22:13 <clandmeter> im having prolbems booting ppc64 in qemu
2021-02-15 15:22:20 <Hello72> i think it claims 4k per cpu or something
2021-02-15 15:23:04 <clandmeter> and it has 2 ways of booting
2021-02-15 15:23:54 <Hello72> 8k. also the max for arm is 32
2021-02-15 15:24:43 <clandmeter> yup 32
2021-02-15 15:24:48 <mps> hah, trying to set 64 on armv7 in kernel config says ' You have made an invalid entry.'
2021-02-15 15:25:02 <clandmeter> prompt: Maximum number of CPUs (2-32)
2021-02-15 15:25:03 <clandmeter> :)
2021-02-15 15:25:08 <mps> max is 32
2021-02-15 15:25:14 <mps> yes
2021-02-15 15:27:20 <mps> also, CONFIG_HIGHMEM is not set
2021-02-15 15:28:11 <clandmeter> is that needed when you only use 2Gb?
2021-02-15 15:28:16 <clandmeter> like in my test?
2021-02-15 15:33:09 <mps> I'm not sure would it help
2021-02-15 15:37:26 <mps> but there are many interdependent options in kernel and hard to comprehend fully for someone who is not kernel hacker
2021-02-15 15:38:04 <clandmeter> we can always peek at other dists what they do :)
2021-02-15 15:40:43 <mps> yes, I can install debian and forget alpine :D
2021-02-15 15:41:34 <mps> (and, tbh I think alpine will finish as worse debian 'blend')
2021-02-15 20:47:15 <ikke> clandmeter: seems like no stuck gitaly-ssh processes anymore
2021-02-15 20:47:24 <clandmeter> nice
2021-02-15 20:47:46 <clandmeter> did we get something interesting in return?
2021-02-15 20:48:58 <ikke> like?
2021-02-15 20:49:37 <clandmeter> dunno, some random issue
2021-02-15 20:49:52 <ikke> haven't noticed anything yet
2021-02-16 09:17:35 <clandmeter> ikke: what was dns server for alpin.pw?
2021-02-16 09:19:54 <clandmeter> ah i think ive found it in my ssh config
2021-02-16 09:30:14 <ikke> 8.3?
2021-02-16 09:39:56 <clandmeter> yes 
2021-02-16 09:39:58 <clandmeter> thx
2021-02-18 08:33:14 <ikke> up here
2021-02-19 08:32:36 <clandmeter> Hi chenguoqi 
2021-02-19 08:34:37 <clandmeter> chenguoqi: what is the formfactor of your hardware?
2021-02-19 08:42:53 <chenguoqi> clandmeter:   loongson 3A4000
2021-02-19 08:43:42 <clandmeter> chenguoqi: do you have a link to the actual hardware, the board?
2021-02-19 08:44:37 <chenguoqi> Yes
2021-02-19 08:46:54 <chenguoqi> I am a developer of Loongson company. Our company especially hopes Alpine Linux can run on Loongson CPU      
2021-02-19 08:48:02 <chenguoqi> Our company is particularly willing to donate Loongson CPU machines to alpine Linux community      
2021-02-19 08:48:21 <clandmeter> chenguoqi: that is very nice to hear
2021-02-19 08:48:50 <chenguoqi> We have ported Alpine Linux 3.11 to Loongson 3a4000 (mips64el), and we are in the process of  porting the edge version. We very much want to share our results with the community.
2021-02-19 08:49:18 <chenguoqi> If you want to experience the Alpine Linux 3.11 for mips64el, you can visit   alpine.loongnix.org  Or you can use docker pull loongnix/alpine:3.11
2021-02-19 08:49:29 <ikke> chenguoqi: in what kind of case does this board come?
2021-02-19 08:49:44 <clandmeter> chenguoqi: if it can boot linux with alpine edge userland, im sure we can find a place to host it.
2021-02-19 08:50:12 <clandmeter> chenguoqi: so its nice to know detailed specifications of the board, and also the case (formfactor).
2021-02-19 08:51:22 <clandmeter> to intoduce it into our infrastructure it need to "try" to keep up with our current buidlers, and we also need hardware to use for our CI.
2021-02-19 08:52:18 <clandmeter> do you provide something that can be rack mounted?
2021-02-19 08:56:02 <chenguoqi> ikke:  Our company is very friendly to the community
2021-02-19 09:00:47 <chenguoqi> clandmeter:  Now I will consult our colleagues about the hardware specifications, and then reply to you      
2021-02-19 09:01:29 <chenguoqi> clandmeter: :)
2021-02-19 09:01:42 <clandmeter> chenguoqi: ok, you can also email me at: clandmeter@alpinelinux.org
2021-02-19 09:01:55 <chenguoqi> clandmeter: OK
2021-02-19 09:21:02 <clandmeter> chenguoqi: if we would introduce mips64el to alpine, will your team support this port?
2021-02-19 09:37:00 <chenguoqi> clandmeter:   Our team is particularly willing to support and maintain Alpine Linux running on the mips64el platform.      
2021-02-19 09:38:06 <clandmeter> chenguoqi: thats great!
2021-02-19 09:39:35 <chenguoqi> clandmeter: Our company also hopes that more OS can run perfectly on our CPU machine !!.
2021-02-19 09:43:12 <ncopa> the loongson 3A4000 seems to be quadcore, 8GB ram?
2021-02-19 09:47:12 <chenguoqi>  ncopa:  Yes,  loongson  3a4000 is a  quad core, RAM is 8GB  or 16GB
2021-02-19 09:49:22 <ikke> chenguoqi: does it run a standard linux kernel?
2021-02-19 09:51:42 <ncopa> the slowest/smallest alpine builder is mips64 (big endian) 16G ram, 16 cores Cavium Octeon III V0.3  FPU V0.0
2021-02-19 09:52:22 <ncopa> our (current) arm builder has 64 cores, 128G ram
2021-02-19 09:52:30 <ncopa> htat is one of our arm builders
2021-02-19 09:53:14 <ncopa> this longsoon machine may become the bottle neck
2021-02-19 09:54:33 <clandmeter> i guess other distributions use cross?
2021-02-19 09:54:43 <ncopa> probably
2021-02-19 09:55:03 <ncopa> the downside with crosscompiling is that you cannot run the unit tests
2021-02-19 09:55:15 <ikke> and emulating?
2021-02-19 09:55:19 <ncopa> so we dont hit the bugs til until its out in production
2021-02-19 09:55:44 <ncopa> i dont know how good the mips64le emutaion is, but for arm it was not good enough
2021-02-19 09:55:45 <ikke> I think we will have the same issue with riscv
2021-02-19 09:56:25 <clandmeter> what about using icecream?
2021-02-19 09:56:50 <ncopa> but i think the most important bit is help with supporting and maintaining the platform
2021-02-19 09:57:35 <clandmeter> or distcc
2021-02-19 09:58:12 <ncopa> chenguoqi: do you know what the status is for rust on mips64el?
2021-02-19 09:58:21 <ncopa> distcc is cumbersome
2021-02-19 09:58:46 <clandmeter>  Icecream takes compile jobs from a build and distributes it among remote machines allowing a parallel build. But unlike distcc, Icecream uses a central server that dynamically schedules the compile jobs to the fastest free server. 
2021-02-19 10:02:26 <chenguoqi> ikke :    It can run standard Linux kernel  (Versions after 5.10)
2021-02-19 10:03:00 <ikke> chenguoqi: ok, good
2021-02-19 10:19:58 <ncopa> awesome
2021-02-19 10:20:28 <ncopa> havent heard about icecream before. sounds pretty cool
2021-02-19 10:24:28 <ncopa> chenguoqi: we are looking for a place where you can send the machine. would be good to know the form factor of the machine. can it be mounted in a rack?
2021-02-19 10:33:23 <clandmeter> ncopa: yes icecream sounds a bit like what we were talking about before. but i dont know the details.
2021-02-19 10:48:25 <Ariadne> rust for mips64el is broken
2021-02-19 10:48:29 <Ariadne> well
2021-02-19 10:48:32 <Ariadne> for us anyway
2021-02-19 10:48:36 <Ariadne> because cross is broken
2021-02-19 10:49:07 <Ariadne> i can help with mips in general, i think that mips64el should be hardfloat though
2021-02-19 10:49:19 <Ariadne> loongson CPUs are known to always have FPU
2021-02-19 10:49:50 <Ariadne> verses Octeon which has either no FPU (octeon1), a broken FPU (octeon2) or a working FPU (octeon3)
2021-02-19 10:51:09 <Ariadne> fwiw it should be possible to simply run sh bootstrap.sh mips64el and get a mips64el distro
2021-02-19 10:55:00 <Ariadne> but i think it should be switched to being hardfloat because a lot of the performance issues on mips64 is the softfloat
2021-02-19 10:55:26 <Ariadne> also, technically, cavium octeon3 can be booted as little endian
2021-02-19 10:55:44 <Ariadne> though i think the drivers are completely broken for that case
2021-02-19 10:57:55 <Ariadne> chenguoqi: would it be possible to get a loongson 3c5000 machine instead?  i think loongson has a SKU which is a 1U server for that
2021-02-19 10:58:00 <Ariadne> which is 16 cores
2021-02-19 11:02:35 <Ariadne> anyway i am kinda disappointed with our octeon3 board
2021-02-19 11:02:37 <Ariadne> hmm
2021-02-19 11:02:46 <Ariadne> chenguoqi: can loongson run in big-endian mode?
2021-02-19 11:04:43 <Ariadne> ikke: the different with riscv is that the qemu emulation for riscv is maintained by riscv.org
2021-02-19 11:04:49 <ikke> aha, ok
2021-02-19 11:04:52 <Ariadne> so i would consider it reliable
2021-02-19 11:05:19 <Ariadne> would prefer real silicon though, of course
2021-02-19 11:06:08 <ikke> yes, but ddevault mentioned that the available hardware is too slow
2021-02-19 11:08:07 <Ariadne> yes
2021-02-19 11:08:50 <Ariadne> but qemu-user emulation should be good enough i think
2021-02-19 11:09:02 <Ariadne> so we can just throw a threadripper at it and go to town :)
2021-02-19 11:09:02 <ikke> ok
2021-02-19 11:09:07 <ikke> haha
2021-02-19 11:09:19 <Ariadne> qemu-user emulation was mostly passable when i built the mips64 port btw
2021-02-19 11:09:24 <Ariadne> but there are some problems
2021-02-19 11:09:27 <Ariadne> random hangs
2021-02-19 11:09:51 <Ariadne> i think that was ultimately found to be a busybox bug though
2021-02-19 11:10:17 <ikke> they tend to happen, apparently :)
2021-02-19 11:10:31 <Ariadne> my plan for the m68k port is to do qemu-user
2021-02-19 11:10:36 <Ariadne> its the only way the port is viable
2021-02-19 11:11:05 <Ariadne> also qemu-user is getting better all the time
2021-02-19 11:11:44 <clandmeter> qemu-user is nice
2021-02-19 11:15:59 <Ariadne> anyway, mips64 keeps up with the other builders well enough
2021-02-19 11:16:05 <Ariadne> quad core is scary tho :)
2021-02-19 11:18:32 <ncopa> yeah
2021-02-19 11:19:11 <ncopa> would be nice with 2 machines actually. one dedicated for official builds, and one for developer use
2021-02-19 11:19:22 <Ariadne> we technically have a second octeon3 board
2021-02-19 11:19:36 <Ariadne> its just frustrating because the octeon3 NIC driver was never upstreamed
2021-02-19 11:19:40 <clandmeter> we would need 2 anyways if we want to use CI
2021-02-19 11:19:44 <Ariadne> and marvell does not care :D
2021-02-19 11:19:58 <ncopa> clandmeter: good point
2021-02-19 11:20:04 <Ariadne> clandmeter: well go might still be broken on mips64[el]
2021-02-19 11:20:10 <ncopa> i assume that go works on mips64el?
2021-02-19 11:20:18 <Ariadne> not sure
2021-02-19 11:20:26 <Ariadne> that's another problem with qemu-user
2021-02-19 11:20:36 <Ariadne> qemu-user + go doesn't mix
2021-02-19 11:20:43 <Ariadne> or at least didn't 2 years ago
2021-02-19 11:20:52 <ncopa> sounds like chenguoqi's team is motivated enough to get go working if it isnt
2021-02-19 11:21:18 <clandmeter> if they can boostrap edge, why not.
2021-02-19 11:21:31 <clandmeter> including go and related ofc
2021-02-19 11:22:14 <Ariadne> ncopa: yeah, loongson is basically driving all mips development at this point
2021-02-19 11:22:24 <Ariadne> since wave is uhh... well that's complicated topic :
2021-02-19 11:22:25 <Ariadne> )
2021-02-19 11:22:36 <clandmeter> its hard to find info about loongsons hardware
2021-02-19 11:22:52 <clandmeter> i dont see much ref to the 16 core one
2021-02-19 11:23:07 <Ariadne> yeah i don't know if that SKU is actually shipping yet
2021-02-19 11:23:27 <Ariadne> i just know 3c5000 is 4-16 core parts
2021-02-19 11:25:34 <Ariadne> what would be cool is if loongson 3 is bi-endian
2021-02-19 11:25:47 <Ariadne> then we could just ditch the octeon3 crap and do mips64[el] on 4 machines
2021-02-19 11:26:09 <ncopa> that would be great
2021-02-19 11:26:12 <Ariadne> but traditionally loongson is little-endian only
2021-02-19 11:26:15 <Ariadne> so not holding my breath :)
2021-02-22 16:37:53 <ikke> clandmeter: https://tpaste.us/Yo1Q 
2021-02-22 17:15:25 <clandmeter> Network down?
2021-02-22 17:16:48 <ikke> Don't know, but but that didn't seem to be the issue I was investigating
2021-02-22 17:17:02 <ikke> lately algitbot stops responding in #alpine-devel
2021-02-22 21:01:40 <ikke> gitlab 13.9 released
2021-02-23 15:39:58 <clandmeter> ikke: did you fix the bot?
2021-02-23 15:40:21 <ikke> I had restarted it, but it might happen again
2021-02-24 20:21:06 <clandmeter> ikke: what happend which caused the disks to go full?
2021-02-24 20:21:16 <clandmeter> i also see traffic alerts
2021-02-24 20:22:53 <ikke> New releases 
2021-02-24 20:23:28 <ikke> nld lv can be expanded, didn't check nld5 yet
2021-02-24 20:23:40 <ikke> nld3*
2021-02-25 10:13:41 <Cogitri> ikke: re https://gitlab.alpinelinux.org/alpine/aports/-/issues/12447#note_144359, I thought that if I ping folks they'll automatically get access to that specific issue
2021-02-25 10:15:52 <ncopa> I'd like to move ancient.alpinelinux.org to new host. Do we have somewhere we can move it?
2021-02-25 10:16:03 <ncopa> its the archive for older releases
2021-02-25 10:17:14 <ikke> Cogitri: nope 
2021-02-25 10:17:58 <ikke> Check the notification on top of the comment box 
2021-02-25 10:18:11 <ncopa> i dont see any problem with giving maintainers access to the confidential issues
2021-02-25 10:18:50 <ikke> ncopa: I'll add them as reporters then. We cannot give them access per ticket 
2021-02-25 10:19:14 <Cogitri> Hm, that'd give them access to all issues though
2021-02-25 10:19:29 <Cogitri> IMHO if we can't do it per issue it's better to not add them as reporters
2021-02-25 10:19:44 <Cogitri> SInce that'd mean having a huge amount of folks with permissions to open/close issues etc.
2021-02-25 11:26:38 <ikke> Cogitri: not every maintainer
2021-02-25 14:01:13 <Hello71> bugzilla allows adding users per ticket :/
2021-02-25 14:02:16 <Hello71> <bikeshed> honestly idk why people hate bugzilla. it's so much better than phab and it's at least faster than gitlab
2021-02-26 05:49:42 <ikke> clandmeter: should I just add 50G to the mirror volumes on nld5/nld3?
2021-02-26 05:52:13 <clandmeter> Ok
2021-02-26 05:56:05 <ikke> /sbin/fsadm: execvp failed: No such file or directory
2021-02-26 05:56:26 <ikke> need lvm2-extra
2021-02-26 05:57:37 <clandmeter> Are you resizing in single shot?
2021-02-26 05:58:04 <ikke> yeah, i did add -r
2021-02-26 05:58:24 <clandmeter> I never do that
2021-02-26 05:58:27 <ikke> now did it in 2 steps
2021-02-26 05:59:15 <ikke> done
2021-02-26 15:56:27 <ikke> Hmff
2021-02-27 20:42:05 <mps> x86_64 builder is in 'pulling git' state, whatever it means
2021-02-27 21:01:09 <ikke> That we need to clean our stuff up
2021-02-27 21:29:28 <ikke> I cleaned up logs yesterday, but apparently that was not enough
2021-02-27 21:29:59 <ikke> I wonder where all that space went
2021-02-28 10:06:32 <mps> usa4 disk is full
2021-02-28 13:13:17 <ikke> And nld9 (x86_64 builder) as well
2021-02-28 15:00:51 <ikke> grumble
2021-02-28 15:03:48 <ikke> Something just ate 15G in one hour
2021-02-28 15:08:52 <ikke> ceph build apparently
2021-02-28 15:09:03 <ikke> 17G     src
2021-02-28 15:11:10 <ikke> sigh
2021-02-28 15:13:10 <ikke> why..
2021-02-28 15:15:02 <mps> why? modern programming in interest of big corporations