#meeting-1:fedoraproject.org log

<@davdunc:fedora.im>

15:00:10

!startmeeting fedora_cloud_meeting

<@meetbot:fedora.im>

15:00:11

Meeting started at 2025-07-31 15:00:10 UTC

<@meetbot:fedora.im>

15:00:12

The Meeting name is 'fedora_cloud_meeting'

<@davdunc:fedora.im>

15:00:21

!topic roll call

<@davdunc:fedora.im>

15:01:58

!topic actions from last meeting

<@davdunc:fedora.im>

15:02:13

!topic action items from last meeting

<@conan_kudo:matrix.org>

15:03:30

!hi

<@zodbot:fedora.im>

15:03:33

Neal Gompa (ngompa) - he / him / his

<@davdunc:fedora.im>

15:04:29

Hey Neal!

<@davdunc:fedora.im>

15:05:01

So I am not going to fret over the action items because we have had a full two cycles.

<@jcline:fedora.im>

15:05:32

!hi

<@zodbot:fedora.im>

15:05:34

Jeremy Cline (jcline) - he / him / his

<@davdunc:fedora.im>

15:05:48

hey Jeremy Cline !

<@davdunc:fedora.im>

15:06:29

sorry I have been absent. we had a big reorg over the past couple of weeks and I lost some colleagues. It's been complicated to resorb that workload.

<@davdunc:fedora.im>

15:06:49

(before that I was in MX in the woods)

<@davdunc:fedora.im>

15:07:10

but I'm back now.

<@davdunc:fedora.im>

15:07:32

!topic Rawhide EC2 images failing to build because ec2-instance-connect was retired

<@davdunc:fedora.im>

15:07:45

#link https://pagure.io/cloud-sig/issue/443

<@davdunc:fedora.im>

15:07:59

!link https://pagure.io/cloud-sig/issue/443

<@davdunc:fedora.im>

15:08:10

:D old habits die hard.

<@davdunc:fedora.im>

15:08:22

I am planning to pick this package up.

<@conan_kudo:matrix.org>

15:08:32

you mean you already did

<@conan_kudo:matrix.org>

15:08:43

!link https://src.fedoraproject.org/rpms/ec2-instance-connect

<@davdunc:fedora.im>

15:08:53

haha. okay. thanks.

<@conan_kudo:matrix.org>

15:09:11

I revived it and transferred it to you

<@davdunc:fedora.im>

15:09:21

aha. thanks.

<@davdunc:fedora.im>

15:09:48

I did a business case forit internally to identify the impact and I think I have support for it.

<@conan_kudo:matrix.org>

15:10:02

excellent

<@davdunc:fedora.im>

15:10:59

Thank you Neal for your help and support.

<@adamwill:fedora.im>

15:12:00

i think it being unretired might mean it's fixed already, let me check

<@adamwill:fedora.im>

15:12:35

yeah, today's compose had the package in it and has EC2 images, so you can go ahead and close it

<@adamwill:fedora.im>

15:12:49

Or I will

<@davdunc:fedora.im>

15:12:52

@adamw yea. I think that neal did it. Thanks for reporting it.

<@davdunc:fedora.im>

15:13:01

I have it now.

<@davdunc:fedora.im>

15:14:20

!topic #442 Please push new Fedora 42 images to Hashicorp vagrant registry

<@davdunc:fedora.im>

15:14:42

I missed on this one as a result of the tumult. I'll get to this Friday.

<@davdunc:fedora.im>

15:15:47

!link https://pagure.io/cloud-sig/issue/442

<@davdunc:fedora.im>

15:16:09

!action davdunc to complete the vagrant image upload links by Friday Aug 1.

<@davdunc:fedora.im>

15:17:22

!topic #441 cloud images are missing `saved_entry` in grubenv

<@davdunc:fedora.im>

15:17:32

!link https://pagure.io/cloud-sig/issue/441

<@davdunc:fedora.im>

15:18:01

Neal Gompa (Fedora) do you know where we stand on this investigation?

<@davdunc:fedora.im>

15:18:50

I'll take that as a "busy right now"

<@nhanlon:beeper.com>

15:18:51

!hi

<@zodbot:fedora.im>

15:18:52

Neil Hanlon (neil) - he / him / his

<@davdunc:fedora.im>

15:19:00

hey Neil!

<@nhanlon:beeper.com>

15:19:06

heya Dave

<@nhanlon:beeper.com>

15:19:08

howzit

<@davdunc:fedora.im>

15:19:12

thanks for dropping in!

<@davdunc:fedora.im>

15:19:20

it's grubby right now! :)

<@nhanlon:beeper.com>

15:19:38

🙂 I've got my travel router plugged into my actual router providing internet right now

<@nhanlon:beeper.com>

15:19:50

because someone cut the fiber trunk thru town

<@davdunc:fedora.im>

15:20:09

whoa!

<@nhanlon:beeper.com>

15:20:11

who even needs path diversity, anyway

<@nhanlon:beeper.com>

15:20:14

who even needs path diversity, anyways

<@davdunc:fedora.im>

15:20:31

haha.

<@nhanlon:beeper.com>

15:20:35

i'm just struggling to understand how they cut a fiber bundle using a hydrovac, but.. idk. i'm not a real engineer 😛

<@nhanlon:beeper.com>

15:20:38

back to cloud things 😛

<@davdunc:fedora.im>

15:21:34

So Martin stated that there is some randomness to the kernel that boots first in the cloud images? Is that what I am to understand from his last statement?

<@davdunc:fedora.im>

15:22:19

Neal seemed to think that we were on the right path to fix the issue, but I don't see agreement from the reporter.

<@nhanlon:beeper.com>

15:22:59

/me looks

<@davdunc:fedora.im>

15:23:02

Jeremy Cline are you familiar with "the randomness"?

<@jcline:fedora.im>

15:24:31

No, I've not looked into this

<@conan_kudo:matrix.org>

15:24:42

I'm not sure what is still creating this problem

<@davdunc:fedora.im>

15:25:51

I can understand that.

<@davdunc:fedora.im>

15:27:17

I'll engage Martin in the ticket to see if he has an idea of what or why this is occurring and we have the reproducer here with his one command it looks like.

<@davdunc:fedora.im>

15:27:39

!action davdunc to update the ticket and followup with Martin on possible areas to explore.

<@conan_kudo:matrix.org>

15:28:18

oh now I understand what's happening

<@conan_kudo:matrix.org>

15:28:32

`grub2-editenv` is for some reason not in the cloud images

<@davdunc:fedora.im>

15:28:32

oh awesome.

<@davdunc:fedora.im>

15:28:58

hmm. okay. so we need to add it. it's helpful to have it.

<@conan_kudo:matrix.org>

15:28:58

[ DEBUG ]: 17:22:08 | /image/config.sh: line 46: [[Cloud-Base-AmazonEC2,CloudCore,BootCore: command not found

<@conan_kudo:matrix.org>

15:28:58

```

<@conan_kudo:matrix.org>

15:28:58

```

<@conan_kudo:matrix.org>

15:29:07

command not found on the line running that command

<@davdunc:fedora.im>

15:29:36

cool. That's easy enough to remedy.

<@jcline:fedora.im>

15:30:17

Hm, should that script be set to fail if a command fails?

<@conan_kudo:matrix.org>

15:31:03

yeah

<@conan_kudo:matrix.org>

15:31:10

but I thought I had already done that

<@conan_kudo:matrix.org>

15:32:02

`set -euxo pipefail` is supposed to make that fail

<@davdunc:fedora.im>

15:32:15

in a big way.

<@jcline:fedora.im>

15:32:21

Weird

<@jcline:fedora.im>

15:32:49

My bash skills are... not good, so I have no idea

<@davdunc:fedora.im>

15:34:26

yea. that's basically identifying that if any issues occur, get out.

<@davdunc:fedora.im>

15:36:18

okay. So we have a path to completion. That's awesome.

<@conan_kudo:matrix.org>

15:37:32

I'm confused though

<@conan_kudo:matrix.org>

15:37:42

it should be present since the grub efi packages require grub2-tools-minimal

<@davdunc:fedora.im>

15:39:15

yea. I'm seeing this a a glass half full moment Neal Gompa (Fedora) in that the issue is somewhat identified and we have a trajectory and a describable problem fix even if there is a new obstacle in the way of making that one work. :D

<@davdunc:fedora.im>

15:40:29

!topic QA changes in scope.

<@davdunc:fedora.im>

15:42:09

do we need to be working on building more of our own support? adamw is there something that we need to do for you as a SIG or as members of the QE

<@adamwill:fedora.im>

15:42:29

the main thing would be to implement automated real-cloud testing

<@adamwill:fedora.im>

15:42:35

because right now i always wind up doing that manually

<@adamwill:fedora.im>

15:43:10

if we had that i think all testing would be automated

<@jcline:fedora.im>

15:43:24

So I've talked to a couple folks internally, and we were already planning to run some set of LISA tests against the Azure images. Supposedly LISA also does AWS, but not GCP.

<@adamwill:fedora.im>

15:43:42

that's all i ever test on manually, so hey. but of course testing on more would be great.

<@adamwill:fedora.im>

15:44:05

it would also need to be hooked up to the wiki reportotron

<@adamwill:fedora.im>

15:44:11

i can help with that bit if you can get tests actually running

<@davdunc:fedora.im>

15:44:18

Jeremy Cline we can talk to @ajorgensen at GCP and get his assessment on what tools we need to use there or how they might be able to help us update the LISA tests.

<@jcline:fedora.im>

15:44:19

My initial plan is to add another consumer that either lives with the image uploader or in its own place (doesn't really matter for this discussion) that runs the test suite on every new image and publishes the results via AMQP

<@davdunc:fedora.im>

15:44:36

I personally like the public test suite from Azure.

<@jcline:fedora.im>

15:44:53

We could also tag the images that pass all the tests and promote them automatically

<@davdunc:fedora.im>

15:45:13

I like that.

<@davdunc:fedora.im>

15:46:30

There are a lot of validations that only need one pass on one provider.

<@jcline:fedora.im>

15:46:54

One of my teammates has already written a little fedora-messaging consumer that runs the LISA tests on the Azure images, we just need to get that reviewed and running in Fedora infra and then we can build it out to cover AWS and GCP

<@davdunc:fedora.im>

15:47:40

I am happy to help with that!

<@davdunc:fedora.im>

15:47:51

(for as much help as I can be)

<@davdunc:fedora.im>

15:48:21

I don't want to slow you down too much, but I also don't want you to be the SPOF.

<@davdunc:fedora.im>

15:49:05

okay. I think we have a path forward here.

<@davdunc:fedora.im>

15:49:42

I just don't want to drop in quality beause Adam has enough responsiblity.

<@jcline:fedora.im>

15:49:44

My plan is mostly to help my teammate handle a lot of it, and mostly handle the deployment and maintenance

<@davdunc:fedora.im>

15:49:56

okay. awesome.

<@jcline:fedora.im>

15:50:39

We'd probably run the tier 1 test suite for each image (https://mslisa.readthedocs.io/en/main/run_test/microsoft_tests.html#test-tier) and then maybe higher tier test suites less frequently

<@jcline:fedora.im>

15:51:50

And if there's anything that's not covered by the tier 1 tests that are in the manual tests we've been running we can add that. My understanding is the LISA folks are happy to take PRs.

<@davdunc:fedora.im>

15:52:04

yea -- just for beta and rc or where there is a specific test related to a feature change or addition.

<@davdunc:fedora.im>

15:53:46

okay.

<@davdunc:fedora.im>

15:54:16

So what I am seeing here is that we think that LISA is the right tool to adopt for testing. adamw any objections?

<@davdunc:fedora.im>

15:54:33

(sorry to keep pinging you.)

<@davdunc:fedora.im>

15:54:57

he can raise them any time, I suppose.

<@davdunc:fedora.im>

15:55:17

!link https://mslisa.readthedocs.io/en/main/quick_start.html

<@davdunc:fedora.im>

15:55:52

!info Cloud SIG is officially adopting the LISA testing tools for platform validations.

<@jcline:fedora.im>

15:56:00

My plan for the higher tiers was maybe once a month or something, depending on how much it ends of costing. Some of the tests are for VMs with several hundred cores or 2TB of RAM, for example.

<@davdunc:fedora.im>

15:56:14

ah gotcha.

<@davdunc:fedora.im>

15:56:48

yea. I just ran into an issue (not with Fedora) on an extra large instance type with a gazillion vcpus.

<@davdunc:fedora.im>

15:57:38

but they are really immediate and not typically an issue unless there is a major change in a subsystem for the kernel or deprecation in the glibc.

<@davdunc:fedora.im>

15:58:06

!topic open floor

<@davdunc:fedora.im>

15:58:28

Anyone have something that they want to cover in the next "one minute". :D

<@davdunc:fedora.im>

15:58:44

once . . .

<@davdunc:fedora.im>

15:58:49

twice . . .

<@davdunc:fedora.im>

15:58:56

!endmeeting