2026-03-01 18:11:33 <achill> on it
2026-03-01 18:15:43 <ikke> algitbot: ok, thanks
2026-03-01 18:15:46 <ikke> achill: *
2026-03-01 18:15:54 <achill> yup
2026-03-01 18:17:10 <PureTryOut> Strange, it succeeded on CI 😅
2026-03-01 18:17:22 <PureTryOut> Ah you literally just merged the glslang update, ok
2026-03-01 18:20:05 <achill> yeah :))
2026-03-01 18:20:20 <achill> and my MR rebuild doesn't get any runners :((
2026-03-01 18:20:36 <achill> all already busy
2026-03-01 18:22:39 <ikke> Hmm
2026-03-01 18:28:27 <ikke> There should be plenty of space for the smaller jobs, but they are not picked up
2026-03-01 18:44:04 <achill> amend-commit and force push worked /shrug
2026-03-02 11:12:15 <achill> algitbot: retry master
2026-03-02 12:47:14 <omni> algitbot: retry master
2026-03-03 08:17:24 <Ariadne> algitbot: retry master
2026-03-03 08:17:28 <Ariadne> algitbot: retry master
2026-03-03 08:17:38 <Ariadne> hmm, s390x is out to lunch?
2026-03-04 16:10:46 <achill> ikke: ^
2026-03-04 16:16:33 <ikke> achill: tailscale?
2026-03-04 16:16:37 <achill> yea
2026-03-04 16:16:40 <ikke> listen unix /home/buildozer/aports/community/tailscale/src/TestDoesNotOverwriteIrregularFiles465813604/001/TestDoesNotOverwriteIrregularFiles: bind: invalid argument
2026-03-04 16:28:08 <ikke> achill: the path is sus
2026-03-04 16:28:23 <ikke> tailscale/src/TestDoesNotOverwriteIrregularFiles3807131128
2026-03-04 16:28:55 <ikke> I guess it uses GOTMPDIR?
2026-03-04 16:42:44 <ikke> yes, it has to do with us overriding GOTMPDIR
2026-03-04 16:53:14 <ikke> achill: ah, apparently the path of unix sockets can be max 108 characters, and the tmpdir location makes it 130
2026-03-04 16:53:20 <ikke> https://stackoverflow.com/questions/79646557/is-there-any-restrictions-on-unix-socket-path-length
2026-03-04 16:54:35 <ikke> achill: I suppose we could use the same logic as the test does now for macos
2026-03-04 17:34:53 <ikke> lesigh
2026-03-04 19:10:37 <ikke> achill: The current test failure is due to go1.26
2026-03-04 19:11:05 <ikke> On my local build container, I still had 1.25.* and the test passed
2026-03-04 19:11:21 <ikke> There is some difference in httptest or something like that that makes it fail
2026-03-04 19:34:45 <ikke> d'oh, go 1.26.0 explicitly sets a DialContext now, which tailscale does not expect and panics
2026-03-04 19:58:41 <ikke> achill: so tailscale was never rebuilt against go 1.26..
2026-03-04 21:07:48 <ikke> algitbot: retry master
2026-03-04 21:11:46 <ikke> algitbot: retry master
2026-03-04 22:33:33 <andypost[m]> algitbot: retry master
2026-03-05 08:36:54 <omni> algitbot: retry master
2026-03-05 09:33:13 <omni> algitbot: retry master
2026-03-05 09:36:11 <omni> =(
2026-03-05 09:36:34 <omni> oh, OOM
2026-03-05 09:36:46 <omni> algitbot: retry master
2026-03-05 10:04:29 <omni> algitbot: retry master
2026-03-05 10:06:10 <omni> ikke: the aarch64 rust build got butchered by the OOM murderer again, is there anything that can be done to avoid that? like in CI
2026-03-05 10:13:12 <ikke> omni: https://gitlab.alpinelinux.org/alpine/aports/-/blob/master/.gitlab/job-config.yaml
2026-03-05 10:15:41 <omni> ikke: does that also affect the package builders?
2026-03-05 10:15:52 <ikke> omni: no
2026-03-05 10:16:04 <ikke> If it fails on the builders, then it's actually OOM
2026-03-05 10:16:11 <ikke> (or address space)
2026-03-05 10:16:30 <omni> right
2026-03-05 10:17:48 <ikke> According to our monitoring, the amount of memory available on the aarch64 host has not been < 200GB
2026-03-05 10:19:21 <ikke> So I wonder why it says "out of memory"
2026-03-05 11:08:21 <omni> \o/
2026-03-05 11:08:41 <ikke> omni: still, would be good to find out why it even happens
2026-03-05 11:09:07 <ikke> On a server with 256GB of memory
2026-03-05 11:53:39 <omni> ikke: I was thinking too many concurrent jobs, and perhaps limiting that would help, but if you're saying it always had a lot of free memory idk...
2026-03-05 11:53:59 <omni> and it couldn't have peaked in memory usage between samples, right?
2026-03-05 11:54:48 <ikke> For a moment, the server went from ~64GB memory used to ~96GB. Not enough to end up in OOM land
2026-03-06 01:19:35 <huajingyun> algitbot: retry master
2026-03-06 17:41:37 <ikke> achill: bestline is in testing.l.
2026-03-06 17:41:39 <ikke> testing..
2026-03-06 17:41:42 <ikke> I was just working on a fix
2026-03-06 17:44:06 <achill> ikke: oh oops
2026-03-06 17:50:30 <ikke> Anoying tests: "expected false to be true"
2026-03-06 17:50:34 <ikke> Absolutely clear
2026-03-06 17:50:46 <ikke> How can you ever expect false to be true :P
2026-03-06 18:02:23 <ikke> ginkgo: https://github.com/onsi/ginkgo/issues/1640
2026-03-06 18:48:51 <andypost[m]> algitbot: retry master
2026-03-06 19:32:52 <ikke> sq: fails due to go vet complaining about %w not being supported by a function that adds support itself
2026-03-06 19:35:02 <ikke> grafana writes a response body in a test with a response code that does not allow one (1.26 added a check to disallow that)
2026-03-06 19:55:31 <ikke> lol: "crypto/tls bug: where's my session ticket?"
2026-03-07 10:10:26 <omni> wait, is aws-lc still failing only on arm?
2026-03-07 10:11:28 <omni> or.. it moved on on x86 at least..?
2026-03-07 10:12:10 <f_> seems like it https://build.alpinelinux.org/buildlogs/build-edge-x86/community/aws-lc/aws-lc-1.68.0-r1.log
2026-03-07 10:15:08 <omni> ok, perhaps there's some architecture specific reason for it not to Werror on x86..
2026-03-08 20:01:57 <omni> ikke: do you know what that "invalid cross-device link" for pulumi is about?
2026-03-08 20:02:08 <omni> perhaps not what it is failing on
2026-03-08 20:02:20 <omni> or due to
2026-03-08 20:03:11 <ikke> omni: My guess would be they try to create hardlinks across filesystems?
2026-03-08 20:03:29 <ikke> https://unix.stackexchange.com/questions/79132/invalid-cross-device-link-while-hardlinking-in-the-same-file-system
2026-03-08 20:05:07 <ikke> omni: or I guess in this case a rename (instead of move?)
2026-03-08 20:44:20 <omni> ikke: sure, but it must have worked at some point? this is supposed to just be a rebuild with newer go
2026-03-08 20:44:43 <omni> or is it fetching something that is not the same last time?
2026-03-08 20:45:08 <omni> last rebuild was a month a go
2026-03-08 20:51:47 <omni> how about we unbump it and wait for it to be sorted in !96438 ?
2026-03-08 21:05:11 <omni> !98728
2026-03-08 22:20:14 <omni> fossdd: I'd like to merge !98735 unless you have any objections
2026-03-08 22:20:37 <omni> it would be nice to let aarch64 and x86_64 upload to community/
2026-03-08 23:17:43 <omni> fossdd: sorry to not give you more time, I thought you were around since you had just gotten other things in
2026-03-08 23:28:14 <omni> !98737
2026-03-08 23:51:47 <omni> !98738
2026-03-09 00:29:15 <omni> !98741 !98742
2026-03-09 06:48:03 <ikke> omni: mio I don't have more time, but tofutf seems to be missing a trailing / in the URL, causing a 307
2026-03-09 06:48:26 <ikke> request: Path: "/api/v3/repos/acme/terraform/tarball",
2026-03-09 06:49:49 <ikke> resp header: Location: "/api/v3/repos/acme/terraform/tarball/",
2026-03-09 10:08:31 <omni> ikke: thanks! I don't have time either right now
2026-03-09 14:52:32 <mio> ikke: thanks for the hint, !98758
2026-03-09 21:27:03 <ikke> andypost[m]: thanks!
2026-03-09 21:27:27 <andypost[m]> home it faster then clean-up builders
2026-03-09 21:27:33 <andypost[m]> s/home/hope