2022-12-01 10:53:42 <psykose> gonna upgrade deu7-dev1 and deu1-dev1 to 3.17 in a bit
2022-12-01 11:38:11 <psykose> deu7 seems to be fine
2022-12-01 11:38:21 <ikke> ok, dmvpn working?
2022-12-01 11:38:50 <ikke> A patch got removed that broke dmvpn, but ncopa put it back
2022-12-01 11:40:55 <psykose> not sure how i would check that :-)
2022-12-01 11:41:00 <ikke> ip route
2022-12-01 11:41:08 <ikke> do you see routes to the other sites
2022-12-01 11:41:18 <ikke> can you ping 172.16.6.3
2022-12-01 11:41:21 <psykose> ah
2022-12-01 11:42:09 <psykose> seems like i can ping it from deu1
2022-12-01 11:42:17 <psykose> and deu7
2022-12-01 11:42:20 <ikke> That it appears to be working
2022-12-01 11:42:30 <psykose> bunch of routes
2022-12-01 12:10:45 <psykose> algitbot: ping
2022-12-02 09:18:31 <clandmeter> ikke: which was the new dev box?
2022-12-02 09:18:34 <clandmeter> the ip
2022-12-02 09:29:18 <clandmeter> got it
2022-12-07 17:37:06 <ikke> 
ACTION needs to look at that

2022-12-08 20:08:33 <ikke> I guess OOM
2022-12-08 20:40:25 <psykose> the armhf is gone too
2022-12-08 20:41:00 <ikke> hmm, vm seems to be still running
2022-12-08 20:44:18 <ikke> Why do you think it's not running?
2022-12-08 20:49:04 <psykose> missing in builder list
2022-12-10 09:25:04 <ikke> hmm
2022-12-10 09:25:06 <ikke> I can still reach it
2022-12-10 09:43:42 <ikke> me whistles innocently
2022-12-10 09:43:54 <ikke> I had nothing to do with it at all
2022-12-10 12:43:25 <psykose> definitely not
2022-12-10 13:05:44 <ikke> hmm, that's for real
2022-12-10 19:26:02 <mps> psykose: !42206
2022-12-10 19:26:57 <mps> I know there is newer version but intentionally didn't upgrade it because kernel 'wants' older versions
2022-12-10 19:31:36 <psykose> no, there is a single argument name change and is fixed with a mere https://github.com/AsahiLinux/PKGBUILDs/commit/9e607035bab7fd5af25555602bce38e41e5f0c15
2022-12-10 19:31:41 <psykose> the actual versions don't matter
2022-12-10 19:32:45 <mps> don't look at asahi PKGBUILDs, they don't apply to alpine
2022-12-10 19:33:18 <mps> and these things at above url are outdated
2022-12-10 19:33:43 <psykose> they do, that is the only thing one needs to change to use bindgen 0.63 with the linux tree
2022-12-10 19:33:47 <psykose> and the asahi kernel is build with 0.63
2022-12-10 19:34:31 <mps> https://tpaste.us/Z4ld
2022-12-10 19:34:54 <mps> trust me, I've built it on alpine many times
2022-12-10 19:35:10 <psykose> trust you on what
2022-12-10 19:35:15 <psykose> all you have to do is add a sed -i into prepare
2022-12-10 19:35:21 <mps> but, please don't upgrade it
2022-12-10 19:36:12 <mps> I'm not in good health state to make long disscusions
2022-12-10 19:36:21 <psykose> there is nothing to discuss
2022-12-10 19:36:38 <mps> ok, please don't upgrade
2022-12-10 19:37:25 <mps> allow me to check it on next week, and I will inform you of result
2022-12-10 19:37:51 <mps> and yes, I would like if it could be upgraded
2022-12-10 19:38:17 <psykose> do whatever you prefer, i'm just saying it's 3 lines to fix and not broken :)
2022-12-10 19:38:19 <psykose> no rush
2022-12-10 19:38:37 <mps> I'm not against you and MR, just would like to be sure
2022-12-10 19:41:33 <mps> btw, what is strange I noticed that I built asahi kernel with gpu driver and forgot to use CLANG/LLVM and it works. I'm really amused 
2022-12-10 19:42:12 <psykose> hahw
2022-12-10 19:42:20 <psykose> was there a requirement for using clang in the kernel build for it?
2022-12-10 19:42:40 <mps> gpu author says yes
2022-12-10 19:43:09 <mps> gpu driver is in rust, first rust kernel driver afaik
2022-12-10 19:44:07 <mps> but working on it I see that we can build kernels for alpine with LLVM/CLANG without problem
2022-12-10 19:50:47 <psykose> i build mine with clang so yeah
2022-12-10 19:50:48 <psykose> works fine
2022-12-10 19:52:22 <mps> nice
2022-12-10 19:52:57 <mps> but I hope we will not get request to build two versions of same kernel for alpine :)
2022-12-10 19:54:18 <psykose> we did like a year ago, i closed the merge request
2022-12-10 20:03:44 <mps> yes, I remember that someone asked for using clang instead of gcc, and it was not from you
2022-12-10 20:04:18 <mps> ok, I checked rust-bindgen MR and approved it
2022-12-10 20:04:40 <mps> I think it is fine
2022-12-10 20:05:11 <mps> thank you for forcing me to look at it
2022-12-10 20:05:57 <mps> (but now I must take some good red wine and rest)
2022-12-10 20:07:57 <psykose> :)
2022-12-10 20:08:02 <psykose> did you rebuild with it and everything still works?
2022-12-10 20:08:08 <psykose> (just making sure)
2022-12-10 20:08:57 <panekj> that fix lol
2022-12-10 20:09:00 <psykose> i used it for rusticl and it seems to work there too, other unrelated issues aside
2022-12-10 20:09:35 <mps> no, I didn't, but looking at it I don't expect problems, and if they occurs we will fix it
2022-12-10 20:10:42 <mps> but now really have to go for wine, cheers ;)
2022-12-10 20:12:27 <psykose> enjoy :)
2022-12-10 20:19:43 <ikke> mps: I suppose there are no newer kernel versions for the starfive?
2022-12-10 20:23:31 <mps> ikke: there are but I'm busy with some other things, sorry
2022-12-10 20:23:34 <ikke> np
2022-12-10 20:23:37 <ikke> no priority
2022-12-10 20:24:32 <mps> because of hype around asahi and apple silicon I worked on it to make alpine ready for it
2022-12-10 20:25:27 <mps> and I finished most things, just didn't upgraded latest kernel
2022-12-10 20:26:44 <ikke> I just powered on my board again
2022-12-10 20:28:44 <mps> 6.1 kernel will be released in a day and I think Esmil will rebase on it, so maybe in 3-4 days we will have new version with some features added and some bugs fixed
2022-12-10 20:29:40 <mps> and my boars is also on the desk, but not powered for 2-3 weeks
2022-12-10 20:29:54 <mps> s/boars/board/
2022-12-10 20:29:54 <alpine-meetbot> mps meant to say: and my board is also on the desk, but not powered for 2-3 weeks
2022-12-11 18:55:52 <ikke> lesigh
2022-12-11 22:28:19 <ayakael> Build of dotnet7-build is stuck on `build-edge-s390x`. Feel free to kick it by killing the build. I'd appreciate access to the log of the failed build as to debug where it hung. :)
2022-12-12 05:24:54 <ikke> ayakael: https://tpaste.us/Mk8X 
2022-12-12 14:24:43 <ayakael> ikke: huh that's a new one, sourcelink has never hung during build. Noted, thanks!
2022-12-14 16:21:55 <kit_ty_kate> hi, i received an email about a package i maintain being out-of-date but it's not technically true. How are new releases detected? (in this case the release is still in a draft state on github so the notice is a bit early)
2022-12-14 16:23:08 <kit_ty_kate> btw I'm not sure if this channel is meant to be public but just in case it is: https://wiki.alpinelinux.org/wiki/Alpine_Linux:IRC does not list it
2022-12-14 16:23:54 <ptrc> they're usually detected via anitya (https://release-monitoring.org)
2022-12-14 16:24:13 <ikke> kit_ty_kate: yes, this is a public channel
2022-12-14 16:25:02 <ptrc> in case of opam, they just check the git tags
2022-12-14 16:25:08 <ptrc> ( https://release-monitoring.org/project/16645/ )
2022-12-14 16:25:43 <ptrc> and as 2.1.4 was tagged without any suffix like -beta/-alpha/-rc, it's treated as an actual release
2022-12-14 16:25:58 <psykose> the tag is also final and you can't retag it so that is actual 2.1.4
2022-12-14 16:26:13 <psykose> unless opam retags the same tag and changes the sha
2022-12-14 16:26:34 <psykose> ptrc: you should fix that you know
2022-12-14 16:26:43 <ptrc> fix what?
2022-12-14 16:26:48 <psykose> they're not filtered
2022-12-14 16:27:08 <ptrc> wdym? they are being filtered by the suffix on anitya
2022-12-14 16:27:17 <ptrc> ah, in aports-turbo?
2022-12-14 16:27:19 <psykose> ye
2022-12-14 16:27:22 <ptrc> sure
2022-12-14 16:27:59 <minimal> I guess anitya is the least-worst solution we have lol
2022-12-14 16:29:10 <kit_ty_kate> (the email doesn't really tell which IRC channel to go as well, i only guessed it)
2022-12-14 16:29:44 <minimal> there's a couple of package I maintain (from the same people) who have their CI stuff as part of their tarballs, so every-so-often there's a "release" made where no code has actually changed and my packages get tagged as out of date lol. They even managed to go through quite a few separate release numbers in a single day due to tinkering with their CKI stuff :-(
2022-12-14 16:30:14 <ikke> ugh, anoying
2022-12-14 16:30:39 <minimal> yeah I haven't (yet) brought the subject up with them
2022-12-14 16:31:11 <kit_ty_kate> yeah this is quite common in the ocaml world too. People will wait for it to be merged in opam-repository and will retag the releases if it doesn't pass CI and needs fixing
2022-12-14 16:31:54 <ikke> it is anoying
2022-12-14 16:32:09 <ikke> people who fetched the old tag already will not automatically receive the updated tag
2022-12-14 16:32:19 <ikke> so for them, if they checkout the tag they will get the old commit
2022-12-14 16:33:00 <kit_ty_kate> yeah that's why i'm surprised the github releases are not checked by anitya
2022-12-14 16:33:44 <ikke> this is plain git I'm talking about, and the result of retagging
2022-12-14 16:33:51 <ikke> https://gitirc.eu/git-tag.html#_on_re_tagging
2022-12-14 16:33:59 <psykose> it can check releases too
2022-12-14 16:35:11 <psykose> (so i made it do that)
2022-12-20 09:37:13 <clandmeter> ikke: looks like arm only has ipv6 addresses
2022-12-20 09:55:44 <ikke> Yes, telmich already mentioned that in the beginning. 
2022-12-20 09:55:53 <ikke> For me that's not an issue 
2022-12-20 09:56:29 <ikke> It should be able to connect to ipv4
2022-12-20 09:57:35 <clandmeter> he mentioned we could also get an ipv4 address before
2022-12-20 09:58:03 <clandmeter> ipv6 is kind of complicated when your provider does not support it
2022-12-20 10:20:44 <ikke> IPv4 address can
2022-12-20 10:20:46 <ikke>   point to an IPv6 address in the server and thus they are fully
2022-12-20 10:20:48 <ikke>   reachable from the IPv4 Internet - if needed
2022-12-20 10:20:59 <ikke> Emphasis on the last part 
2022-12-20 10:22:57 <clandmeter> i guess for now the easiest way is to route all ipv6 traffic over alpine openvpn connection
2022-12-20 10:23:22 <clandmeter> or better do this on wireguard
2022-12-20 10:23:39 <clandmeter> but our wg does not provide a ipv6 address
2022-12-20 10:23:52 <ikke> Need to set that up still
2022-12-20 10:24:30 <ikke> I've reserved a global but non-routable ipv6 prefix we can use for dmvpn / wg in netbox 
2022-12-20 10:26:11 <clandmeter> i guess i can route ipv6 traffic over openvpn right?
2022-12-20 10:55:06 <ikke> I haven't checked if the dmpvpn hubs have ipv6 enabled 
2022-12-20 11:17:23 <clandmeter> ikke: what i mean is, i dont have ipv6 but when using alpiine openvpn i am provided with a ipv6 address. I guess i could use that to route to the internet based on ipv6?
2022-12-20 11:19:02 <ikke> clandmeter: if it's a publically routable ipv6 address, that should work
2022-12-20 11:19:22 <clandmeter> yes, i think you set it up before :)
2022-12-20 11:19:32 <ikke> yes, but that was based on the equinix host
2022-12-20 11:20:13 <clandmeter> so you are saying the routhing should be broken now
2022-12-20 11:20:18 <clandmeter> routing*
2022-12-20 11:21:29 <ikke> clandmeter: yes
2022-12-20 11:22:35 <ikke> The prefix it anounces belongs to equinix, so that's not going to work
2022-12-20 21:35:09 <ikke> psykose: ptrc: I've deployed aports-turbo with the anitya change
2022-12-20 21:35:56 <ptrc> ikke: cool, thanks! i'll check today at 3 (CET) if any of the pre-releases get tagged
2022-12-20 22:03:42 <psykose> ikke: thanks :)
2022-12-21 00:45:56 <anonymous> hi
2022-12-21 00:47:53 <psykose> what's up
2022-12-21 00:48:41 <anonymous> f
2022-12-21 08:01:52 <psykose> ptrc: seems like it was fixed
2022-12-21 08:01:53 <psykose> :)
2022-12-21 08:02:06 <psykose> i checked a few and it didn't recommend the prerelease one over the other version
2022-12-21 08:02:14 <psykose> ikke: also riscv64 has been stuck for days
2022-12-21 11:08:49 <mps> interesting performance increase with MGLRU on Ampere Altra https://lore.kernel.org/linux-mm/20221220214923.1229538-1-yuzhao@google.com/T/#u
2022-12-21 11:09:13 <mps> is the our aarch64 builder this model
2022-12-21 11:09:20 <psykose> it is
2022-12-21 11:09:42 <mps> hah, so we have to upgrade it to linux-edge ;)
2022-12-21 11:15:15 <ikke> We can try it perhaps on the new HW that we are going to install
2022-12-21 11:17:34 <psykose> what's the new hw
2022-12-21 11:19:00 <ikke> Should be similar to what we alreayd haave
2022-12-21 11:19:02 <ikke> already have
2022-12-21 11:19:50 <psykose> ah
2022-12-22 12:18:53 <mps> do we have tool for checking pkg version in different alpine releases
2022-12-22 12:19:31 <ikke> aports-turbo :P
2022-12-22 12:20:30 <mps> apk search aports-turbo => nothing
2022-12-22 12:20:48 <ikke> It's not a command-line tool
2022-12-22 12:21:04 <mps> ahm, right
2022-12-22 12:21:25 <mps> I forgot to ask for 'cli tool'
2022-12-22 15:55:23 <clandmeter> arm1 is online :)
2022-12-22 16:02:51 <ikke> \o/
2022-12-22 16:31:54 <clandmeter> looks like arm2 is down
2022-12-22 16:32:01 <clandmeter> cant connect to it anymore
2022-12-22 16:32:04 <clandmeter> so i give up for now
2022-12-22 22:53:54 <clandmeter> arm2 is also online
2022-12-27 05:58:39 <ayakael> By curiosity, is there an APKBUILD for gitlab-ce implemented somewhere, or does our instance use docker composer or something along those lines?
2022-12-27 06:52:22 <ikke> ayakael: Yes, we build it in docker 
2022-12-27 06:56:44 <ikke> ayakael: https://gitlab.alpinelinux.org/alpine/infra/docker/gitlab
2022-12-27 07:58:44 <ikke> ayakael: it's not really feasible to package it. The gitlab front-end for example still officially only supports ruby 2.7, which we last shipped in Alpine 3.14
2022-12-27 08:00:51 <clandmeter> that, and its hell to manage ruby in apk
2022-12-27 08:01:47 <clandmeter> i think jirutka did manage to package it ones, not sure what the status is now. he had it in his personal repos.
2022-12-27 08:03:00 <ikke> https://github.com/jirutka/user-aports/tree/v3.16/user/gitlab-foss
2022-12-27 08:04:43 <ikke> still maintains it apparently
2022-12-27 08:10:35 <clandmeter> looks like it, think he did it for uni but what i read on twitter he was leaving.
2022-12-27 08:33:27 <ayakael> Many thanks for the details - does look like a pack of trouble to manage. I see why y'all took the docker route.
2022-12-30 11:38:27 <ikke> Sometimes I wish it was easier to find out where traffic is dropped in the linux kernel
2022-12-30 11:38:52 <ikke> Can ping one direction, but in the opposite direction, traffic does not arrive
2022-12-30 11:39:51 <ikke> I see the traffic arriving at the host, I see counters for the iptables rules that allow the traffic increase
2022-12-30 11:40:09 <ikke> but I don't see it arriving in a container
2022-12-30 11:40:26 <ikke> from the container, I can ping the other way around
2022-12-30 12:14:28 <mps> tcpdump -i any
2022-12-30 12:19:14 <mps> oh, run tcpdump in container to see if packets arrive
2022-12-30 12:19:51 <ikke> I did
2022-12-30 12:19:57 <ikke> On the host I see the packet
2022-12-30 12:20:01 <ikke> in the container I don't
2022-12-30 12:20:13 <ikke> Forwarding is enabled
2022-12-30 12:23:25 <mps> iptables filter maybe?
2022-12-30 12:23:54 <mps> or some routing rules
2022-12-30 12:26:19 <mps> ip route get oif $out_interface address
2022-12-30 12:29:59 <ikke> WHere do I run the latter?
2022-12-30 12:30:19 <mps> on the host
2022-12-30 12:30:53 <mps> you can also use 'from address' in combo
2022-12-30 12:31:00 <ikke> and $out_interface is the bridge of the container?
2022-12-30 12:32:11 <ikke> ip route get oif br-0da1bc49616a 172.16.242.2 from 172.16.252.5 -> RTNETLINK answers: Network unreachable
2022-12-30 12:32:11 <mps> yes
2022-12-30 12:32:30 <ikke> (without the from I get an entry
2022-12-30 12:33:36 <ikke> https://tpaste.us/BkrY 
2022-12-30 12:34:11 <ikke> Not sure why it's considered unreachable from that address
2022-12-30 12:34:33 <mps> my example: ip r g from 109.72.52.77 oif br0 10.5.1.5 => 10.5.1.5 from 109.72.52.77 dev br0 uid 0
2022-12-30 12:49:26 <ikke> Wondering why it would show as unreachable
2022-12-30 12:54:53 <ikke> Ok, I think I see what's going on
2022-12-30 12:54:55 <ikke> 172.16.224.0/27 via 172.16.1.1 dev gre1 proto zebra metric 20
2022-12-30 12:55:09 <ikke> It tries to route it via the vpn instead of locally
2022-12-30 12:55:45 <ikke> oh, wrong server, sorry
2022-12-30 12:56:09 <ikke> 172.16.224.0/27 dev br-0da1bc49616a proto kernel scope link src 172.16.224.1
2022-12-30 12:56:14 <ikke> that's the correct route
2022-12-30 19:02:04 <ikke> ah, I was missing a route in the container
2022-12-31 14:17:32 <ptrc> found another fun regression in gitlab: when browsing branches, the "merge request" button used to redirect to an already created MR; now it tries to open a new MR for the same branch
2022-12-31 14:24:02 <ikke> aports-qa-bot or Aports QA Bot as display name?
2022-12-31 14:24:06 <ikke> (Or something else)
2022-12-31 14:24:55 <psykose> probably latter even though i prefer the former
2022-12-31 14:25:32 <ikke> I think I prefer the former as well
2022-12-31 14:25:48 <psykose> nerd :p
2022-12-31 14:26:09 <ikke> Guilty as charge
2022-12-31 14:26:40 <ptrc> might be just me, but "Aports" feels weird to read
2022-12-31 14:28:43 <psykose> Ptrc
2022-12-31 14:30:11 <ptrc> 😠
2022-12-31 19:28:40 <ikke> Going to reboot gbr2-dev1
2022-12-31 21:52:32 <ikke> Got quite some things done today :)
2022-12-31 21:56:46 <psykose> :)
2022-12-31 21:56:47 <psykose> poggers