#meeting:fedoraproject.org log

<@yselkowitz:fedora.im>

19:00:45

!startmeeting ELN SIG 29 May '25

<@meetbot:fedora.im>

19:00:47

Meeting started at 2025-05-29 19:00:45 UTC

<@meetbot:fedora.im>

19:00:47

The Meeting name is 'ELN SIG 29 May '25'

<@yselkowitz:fedora.im>

19:00:50

!meetingname eln

<@meetbot:fedora.im>

19:00:51

The Meeting Name is now eln

<@salimma:fedora.im>

19:01:04

!hi

<@zodbot:fedora.im>

19:01:05

Michel Lind (salimma) - he / him / his

<@tdawson:fedora.im>

19:01:22

!hi

<@zodbot:fedora.im>

19:01:23

Troy Dawson (tdawson)

<@gmoro:matrix.org>

19:01:37

!hi

<@zodbot:fedora.im>

19:01:39

Guilherme Moro (guilhermemoro)

<@sgallagh:fedora.im>

19:01:57

!hi

<@zodbot:fedora.im>

19:01:57

Stephen Gallagher (sgallagh) - he / him / his

<@yselkowitz:fedora.im>

19:03:33

!topic New business

<@yselkowitz:fedora.im>

19:04:01

!link https://github.com/fedora-eln/eln/issues/247

<@yselkowitz:fedora.im>

19:04:13

Troy Dawson ?

<@tdawson:fedora.im>

19:04:51

I'd say I'm about 2/3 of my way through all the logo's.

<@tdawson:fedora.im>

19:05:54

Some of them I'm very impressed with. But it's just a bit of a chore. There are so many different formats, shapes, and different ways to present the logo.

<@tdawson:fedora.im>

19:06:19

OK, I wouldn't call it a chore ... I really love doing it.

<@tdawson:fedora.im>

19:06:51

Anyway ... work is progressing, when I'm done, I'll put it up on github just like the backgrounds.

<@yselkowitz:fedora.im>

19:07:28

!link https://github.com/fedora-eln/eln/issues/249

<@nhanlon:beeper.com>

19:08:20

!hi

<@zodbot:fedora.im>

19:08:21

Neil Hanlon (neil) - he / him / his

<@yselkowitz:fedora.im>

19:08:22

a couple weeks back, the rc6 kernel and subsequent snapshots broke booting on media

<@salimma:fedora.im>

19:08:45

both Rawhide and ELN, or just ELN?

<@yselkowitz:fedora.im>

19:08:49

the bug has since been fixed

<@yselkowitz:fedora.im>

19:09:11

both, but the rawhide builds were caught by gating on the update

<@yselkowitz:fedora.im>

19:09:33

where the ELN builds made in parallel got through (since we don't gate) and then made compose tests fail

<@yselkowitz:fedora.im>

19:09:56

I am working with the kernel maintainer to switch kernel builds over to ELNBuildSync

<@salimma:fedora.im>

19:10:26

I wonder if we should just defer starting the build until the Rawhide build passed gating

<@salimma:fedora.im>

19:10:31

or will that slow things down too much

<@yselkowitz:fedora.im>

19:10:39

that's what using ELNBuildSync would do

<@salimma:fedora.im>

19:10:43

oh - is that what... yeah

<@salimma:fedora.im>

19:10:51

why is the kernel currently built in a special way then?

<@yselkowitz:fedora.im>

19:10:55

because the ELN build would not start until the rawhide build was stable

<@salimma:fedora.im>

19:11:20

right. but why is the kernel not using elnbuildsync already

<@yselkowitz:fedora.im>

19:11:25

1) secure boot

<@yselkowitz:fedora.im>

19:11:45

2) the maintainer wanted to keep a closer eye on the builds

<@yselkowitz:fedora.im>

19:11:55

but that dated back to the beginning of ELN

<@sgallagh:fedora.im>

19:12:47

Right, I think at this point the original reasons are no longer really valid

<@sgallagh:fedora.im>

19:12:58

What did Justin have to say when you spoke with him?

<@yselkowitz:fedora.im>

19:13:12

he agreed once the bug got fixed

<@yselkowitz:fedora.im>

19:13:17

which it since has

<@yselkowitz:fedora.im>

19:13:33

I just need to fix a bug in ELNBuildSync wrt reloading the config

<@sgallagh:fedora.im>

19:14:22

I was going to ask about that bug later

<@yselkowitz:fedora.im>

19:15:16

since we're on the topic already...

<@yselkowitz:fedora.im>

19:16:08

https://gitlab.com/redhat/centos-stream/ci-cd/distrosync/distrobuildsync/-/commit/6db3b915e06dd0bf85c20db3bf4184a93753d0da

<@yselkowitz:fedora.im>

19:16:52

When I updated EBS to F42 (from F37), it wouldn't work at first

<@salimma:fedora.im>

19:17:13

woof, from F37!

<@yselkowitz:fedora.im>

19:17:14

the reason being that /tmp was no longer writable as it was in previous builds

<@yselkowitz:fedora.im>

19:17:48

https://gitlab.com/redhat/centos-stream/ci-cd/distrosync/distrobuildsync/-/commit/2d97e3fcb0d54acb7e414303e0a764e03e19265d

<@sgallagh:fedora.im>

19:19:10

Hmm, we probably could have dropped htmlmin support entirely

<@yselkowitz:fedora.im>

19:19:10

anyway, afaics the bug was in gitlab-runner so I decided just to avoid it by using a different directory (/var/tmp) for writing

<@yselkowitz:fedora.im>

19:19:24

and that *seemed* to work fine, updates were flowing etc.

<@yselkowitz:fedora.im>

19:19:43

but the config update mechanism doesn't work now

<@sgallagh:fedora.im>

19:19:47

It's there to make the status.html endpoint a big faster to load, but it's not all that much of a gain

<@sgallagh:fedora.im>

19:20:06

It's there to make the status.html endpoint a bit faster to load, but it's not all that much of a gain

<@sgallagh:fedora.im>

19:21:10

Someone who knows web design a lot better than I do could almost certainly rework things to use a paged lookup of the status.json to speed it up far more efficiently without needing to minimize the JSON, but it's just not a real priority.

<@sgallagh:fedora.im>

19:21:23

Since very few people would ever look at the status.html anyhow

<@yselkowitz:fedora.im>

19:23:20

which only became apparent once I tried updating the config, which was last week iirc, and I was on PTO for part of this week, so haven't got back to that yet

<@yselkowitz:fedora.im>

19:25:02

Stephen Gallagher do you have any insight into the possibility that gitlab-runner would change the permissions of the working directory of an image it builds, and why that didn't happen with past builds?

<@sgallagh:fedora.im>

19:26:11

It shouldn't actually have the privilege to change the permissions.

<@sgallagh:fedora.im>

19:26:28

(I assume we're talking about the gitlab-runner operator in OpenShift?)

<@yselkowitz:fedora.im>

19:27:04

yes, some searches showed that it had happened before, and options were put in place to enable/disable that

<@yselkowitz:fedora.im>

19:27:30

but iiuc we switched from docker to podman runners sometime recently?

<@sgallagh:fedora.im>

19:28:37

yselkowitz: In OpenShift? I think they were always podman runners

<@sgallagh:fedora.im>

19:29:51

But if the problem is that /tmp isn't guaranteed to be writable, then we should probably just move the tempfile location to /var/tmp

<@yselkowitz:fedora.im>

19:30:16

hence https://gitlab.com/redhat/centos-stream/ci-cd/distrosync/distrobuildsync/-/commit/6db3b915e06dd0bf85c20db3bf4184a93753d0da

<@sgallagh:fedora.im>

19:30:19

https://gitlab.com/redhat/centos-stream/ci-cd/distrosync/distrobuildsync/-/blob/main/elnbuildsync/config.py#L322 needs to add the `dir=` kwarg

<@sgallagh:fedora.im>

19:31:37

`/var/tmp` is not mounted from the host, correct?

<@yselkowitz:fedora.im>

19:32:03

I don't think so

<@yselkowitz:fedora.im>

19:32:55

export TEMP=/var/tmp in run.sh?

<@sgallagh:fedora.im>

19:35:06

yselkowitz: Does Python honor that for `tempfiles`?

<@sgallagh:fedora.im>

19:35:16

yselkowitz: Does Python honor that for `tempfile`?

<@yselkowitz:fedora.im>

19:35:23

https://docs.python.org/3/library/tempfile.html#tempfile.mkstemp ?

<@yselkowitz:fedora.im>

19:35:42

> If dir is not None, the file will be created in that directory; otherwise, a default directory is used. The default directory is chosen from a platform-dependent list, but the user of the application can control the directory location by setting the TMPDIR, TEMP or TMP environment variables

<@sgallagh:fedora.im>

19:36:58

Ah, I was unaware of that. So yeah, I guess that would work.

<@yselkowitz:fedora.im>

19:37:14

ok I'll try that then

<@yselkowitz:fedora.im>

19:37:35

anything else on this?

<@yselkowitz:fedora.im>

19:38:15

!link https://github.com/fedora-eln/eln/issues/250

<@salimma:fedora.im>

19:38:43

ahh. interesting

<@yselkowitz:fedora.im>

19:39:06

since RHEL 10 was "reverted" to wget-1 after branching, now that wget1 is back in Fedora, I'm thinking ELN should probably switch back

<@salimma:fedora.im>

19:39:31

seems reasonable

<@yselkowitz:fedora.im>

19:40:13

!link https://github.com/fedora-eln/eln/issues/251

<@yselkowitz:fedora.im>

19:40:56

adamw was nice enough to not only let us know about this but also pinpoint the issue and file an upstream bug

<@yselkowitz:fedora.im>

19:41:40

does further point out the need for notifications for openqa failures though

<@salimma:fedora.im>

19:41:47

oh woops, ha, it was my PR to mutter that triggered this

<@salimma:fedora.im>

19:42:46

I'm glad we're flushign out issues caused by turning off X11 though. since EL10 is a bit weird

<@salimma:fedora.im>

19:42:54

that we build some packages with X11 support but they are not usable

<@yselkowitz:fedora.im>

19:43:13

agreed

<@salimma:fedora.im>

19:44:00

ELN, living in the future further ahead even of Rawhide

<@yselkowitz:fedora.im>

19:44:20

!link https://github.com/fedora-eln/eln/issues/252

<@yselkowitz:fedora.im>

19:44:49

notice was sent to devel@ that the python update is coming as soon as next week

<@yselkowitz:fedora.im>

19:45:02

that will be disruptive

<@yselkowitz:fedora.im>

19:45:30

between that and the F43 mass rebuild in July, I have the feeling it will be a long couple of months

<@gmoro:matrix.org>

19:46:27

I almost finished the buildsync work for bodhi at least

<@salimma:fedora.im>

19:46:29

hopefully easier than the tooling update in Q1

<@salimma:fedora.im>

19:46:37

after all the Python team has been filing bugs for months

<@gmoro:matrix.org>

19:46:49

But not sure if we can put that online on time tho

<@sgallagh:fedora.im>

19:47:07

gmoro: If you need a reviewer, feel free to ping me.

<@yselkowitz:fedora.im>

19:47:20

if we don't have it in time for the python mass rebuild, we'll just do what we did in January and bypass openqa

<@yselkowitz:fedora.im>

19:47:32

hopefully we can get this done for the f43 mass rebuild at least

<@gmoro:matrix.org>

19:47:37

Stephen Gallagher: will do

<@yselkowitz:fedora.im>

19:48:13

yes they have, but filing bugs isn't always enough unfortunately

<@yselkowitz:fedora.im>

19:48:35

but yes at least they've been doing their due diligence

<@salimma:fedora.im>

19:48:55

nods

<@salimma:fedora.im>

19:49:25

but at least us provenpackagers can fix bugs now without being accused of ignoring the packagers, since... the bugs have been open for a long time

<@yselkowitz:fedora.im>

19:49:39

was there a policy change?

<@yselkowitz:fedora.im>

19:50:01

oh you're just saying the bugs mean they already have been put on notice

<@salimma:fedora.im>

19:50:20

yeah

<@salimma:fedora.im>

19:50:35

no policy change, and in general I prefer bugging the maintainer first, but... if it's been months

<@yselkowitz:fedora.im>

19:50:42

anything else on this?

<@yselkowitz:fedora.im>

19:51:13

!topic Old business

<@yselkowitz:fedora.im>

19:51:13

!link https://github.com/fedora-eln/eln/issues/202

<@yselkowitz:fedora.im>

19:51:29

Michel Lind UTC-5: fsverity-utils (from the hyperscale Extras workload) is the last known dependent on openssl-devel-engine, but it is built for neither EPEL 10 nor in CBS. Is it still needed?

<@salimma:fedora.im>

19:52:29

ah right, we meant to check on this... let me report back after the meeting?

<@salimma:fedora.im>

19:52:42

I think tentatively not, but one sec

<@salimma:fedora.im>

19:52:51

I mean, don't wait, carry on first

<@salimma:fedora.im>

19:53:10

ah ok. yes, remove it please

<@salimma:fedora.im>

19:53:16

I just found the chat where we discussed this :P

<@yselkowitz:fedora.im>

19:53:29

could you file the CR PR?

<@salimma:fedora.im>

19:53:35

will do

<@yselkowitz:fedora.im>

19:53:40

thanks!

<@yselkowitz:fedora.im>

19:53:46

!topic Next meeting

<@yselkowitz:fedora.im>

19:53:57

the next two weeks are flock and devconf.cz iiuc

<@salimma:fedora.im>

19:54:11

fwiw Davide and I are not going so we'll be here

<@salimma:fedora.im>

19:54:20

we'll be at DevConf.cz in two weeks though

<@tdawson:fedora.im>

19:54:48

I won't be available for either of the next two Thursdays, so I won't be able to be here.

<@yselkowitz:fedora.im>

19:55:38

then on for next week and probably off for the following?

<@salimma:fedora.im>

19:55:58

that should work

<@salimma:fedora.im>

19:56:15

june 5th.. I'm flying out june 7th

<@yselkowitz:fedora.im>

19:56:47

!info next meeting is next Thursday 05 June at 15:00 EDT

<@yselkowitz:fedora.im>

19:56:52

!topic Open Floor

<@yselkowitz:fedora.im>

19:56:58

anything else for today?

<@tdawson:fedora.im>

19:59:00

Nothing from me.

<@yselkowitz:fedora.im>

19:59:35

ok then, seeing none, thank you all for joining the good discussions today, see you in channel or back here next week

<@yselkowitz:fedora.im>

19:59:38

!endmeeting