Text Log: https://meetbot.fedoraproject.org/meeting_matrix_fedoraproject-org/2024-07-24/virtualization-sig.2024-07-24-16.00.log.txt

Meeting started at 2024-07-24 17:00:05 UTC

The Meeting name is 'fedora-server'

Let's wait a moment. I'll post the agenda in 2-3 minutes with a new business to deal with in addition to the agenda as sent earlier today.

!hello

Emmanuel Seyman (eseyman) - he / him / his

Hi Emmanuel!

Hi, Peter

!hi

John Himpel (jwhimpel)

Hi John !

I read your request too late, bat early enough to add it to the agande

Thx.

I think, we start now. Hopefully some of us join later.

Anything to add?

Nope

OK Let's start

Nothing new here today. So I will not repeat the items of the last session.

Anyhing to announce here?

OK, let's move on.

John, the floor is open. Probably we need a short time for reading?

I received an response in our pagure issue ticket from Release Engineering. They collectively agree that documentation of the current build process is in terrible shape.

They are raising the importance of improving the build documentation.

It was suggested that we wait until the documentation improvement is complete before we attempt to rename our distribution media.

They also admit that there is no one person (or possibly even a group of persons) that understands the entire build process.

It was strongly suggested in another response from Release Engineering that we migrate away from lorax (I think) to KIWI for building VM and container images.

KIWI does not currently build iso images to be used as input to anaconda. However, several people though SUSE has an extension to KIWI that could do that. Further research and packaging would be required to make that available in Fedora.

There is to be a KIWI presentation at FLOCK. However it seems to be targeted towards VM and container images.

It was strongly suggested we continue the conversation in the Pagure issue.

For VM images we currently use ImageFactory to create the image itself. The further processing is the current standard build process. Unlike Workstation, our image process is working without any problems so far.

What is RelEng recommending for all the many images we currently build with ImageFactory?

I don't know if there's a tool to locate all of the places in the WIKI that would need updating. Release Engineering seemed to think changes would be required in many places on the WIKI and docs if we rename our distribution media.

Peter Boy: I did not ask about any ImageFactory replacement. The entire Fedora Project seems to be migrating to KIWI. But that's only my impression from following mailing lists.

The big advantage of Image Factory is that it uses the same key files as the creation process for iso. So we only have to maintain it once.

If we migrate to Kiwi, it would be nice to create the iso images with kiwi as well. So we can use the same config files.

Peter Boy: +1

And I don't like to rush and would like to follow the lead of RelEng.

I don't want to become an expert in image creation, but simply use proven tools (= recommended and used by Releng)

Peter Boy: Agreed. I will ask in the ticket about what are the future plans for ImageFactory. I will also ask if they are interested in adding the SUSE extension to KIWI to the Fedora Package Repository.

OK. So we will defer until RelEng has decided about the future Fedora tool set?

Let's talk about it at Flock

nirik: Are you online and can comment on this?

I would add that we defer until RelEng has decided about the future Fedora tool set AND has updated the documentation.

++1

proposal !agreed we defer the renaming until RelEng has decided about the future Fedora tool set AND has updated the documentation.

Patrik Polakovic, Steve Milner and Neal Gompa are the folks responding in the ticket. Patrik is the resource assigned to the ticket from RelEng.

For anyone who is interested, the pagure issue# is 12098

OK. any objection to the proposal? We will follow the process in the releng ticket.

no objection here, therefore

- we want to synchronize our manual installation guide, the Ansible doc/guide, and the Ansible process results (i.e. Ansible should create the same as the manual doc describes).

Our goal as discussed last meeting:

Current status so far:

FTR, I've packaged the robertdebock roles in an rpm : https://eseyman.fedorapeople.org/ansible-robertdebock-roles/

sorry was in the middle of things. Happy to discuss at flock/other... but yeah, getting tooling in one/less places would be nice before renaming things.

I have the four ansible roles (nfs_client, nfs_client_mount, nfs_service and nfs_service_mount) running successfully on my home network. However, when I try to use the client from sisyphos.resdigita.eu, the client fails to connect to the server on my local network. I suspect a firewall issue or a missing package on sisyphos.

nirik: Thanks, let's do it at Flock or later

I tried to upload the ansible roles to pagure.io/fedora-server, but can't seem to get to push the changes. My git knowledge and pagure knowledge is quite limited. I'm sure it's something simple, but I don't know what.

John Himpel: any error message? We won't get far without one

When I try git push, I get the authentication questions. I input my userid and FAS password, I get rejected for bad credentials.

Let me have a look at pagure

you should add a ssh key and push via ssh (easiest), or if you must use https, you have to use a pagure api token... let me find the doc

Earlier, I did a fork of fedora-server. But my local .git points to the original, not my fork. Is that currently a problem or will it become a problem in the future?

https://docs.pagure.org/pagure/usage/http_push.html?highlight=token

I just see, we don't have the WG members as committers in our own repro. What a sh....

oh yeah, will need to make sure you have perms for the repo

Current write enabled are sgallagh, coppedfri, pboy and salimma.

copperi and salimma don't contribute for years now. We should remove them.

I'm not "fluent" in admin pagure. But I can figufre it out later and add all our core members.

Peter Boy: Thanks! I was about to pull out what little remaining hair is left on my head. :-)

yes, please :-)

And I will add eseyman as a system user to sisyphos, wo we have a test bed.

a test bad we all can work and test on.

Well, so I think we should continue next week, when John uploaded the Ansible code and we all can test on sisyphos.

So we change to the next topic?

- We want to split the current NFS draft in 2 parts: Installation and (Basic) Configuration.

Current status so far:

-- Do we want to have a separate Administration guide / page?

What would the Admin guide/page contain?

Not sure at the moment, maybe troubleshooting, issues with clients, ???

Well, the configuration part would we sort of a admin guide, I think.

I would leave it off for now. If we later perceive a need, we can always add it later.

John Himpel: agreed!

What would be the next step and who will working on it?

I can begin adding bullet points to the docs AFTER flock.

Ok. bullet points are a good start and would provide the ideas.

Would it help if I use my docs working routine to split the current guide in 2 parts, installation and configuraition?

I would support that idea. If you could add empty docs for installation & configuration to the repo, then I can use markdown to add section/topic headers.

OK, I'll add empty docs we can fill it and migrate the current docs into the new structure.

I know we decided to document the process using ansible and by using cli. I don't recall what we decided about documenting usage via cockpit.

We didn't yet. The current Cockpit module has some issues. I think, we want to decide when we have the other parts done.

Sorry to have used most of the hour. Emmanuel, do you have anything you wanted to cover today?

I'm exchanging with Robert de Bock over the rpm of his roles. I'ld like him to update his collection

Once that's done, I'll submit it for review to be included in Fedora

and, in the meantime, I'm swapping my homemade stuff for his roles

that's pretty much it

Does anyone have an objection to me upgrading sisyphos from F39->F40 in a few hours?

No. Go for it

No, do it!

The last backup ot the VM ist from yesterday. Just in case :--)

OK, I think, we are done for today!

Anything to add here?

(besides out time is up)

I'm good

OK, let's close for now