Meeting started at 2024-07-17 17:00:21 UTC

The Meeting name is 'fedora-server'

!hello

Emmanuel Seyman (eseyman) - he / him / his

Well, let's wait a moment for everybody to shop up.

Hello, Peter Boy

Welcome Emmanuel!

!hi

John Himpel (jwhimpel)

Hello, John

Good afternoon here. Heat finally subsided for a while.

Yeah, it's a nice evening here, too.

I think, we should start.

Hoping for some more perticipants comming a bit late.

Anything else?

Let's start

Nothing here

Lets' go

We have basically the same state as last week.

Still open because of an ongoing discussion about improving Server first boot process

All these will be longer running items, I suppose.

For details see:

Anything to add here?

OK, let's move on.

I was stalled for several days dealing with the new format of /etc/exports.d/* files. Jinga templates are not my favorite things to wrestle with. Finally figured out the last issue about an hour ago. Hopefully will upload to repo in the next couple of days.

FTR, I've decided to package the robertdebock roles in rpm format (including the nfsserver one)

Emmanuel Seyman: very much appriciated!

They'll probably end up in EPEL as well so that I can use them at $WORK as well

EPEL support it crucial, I think.

it will provide more feedback, at the very least

In the long term we should watch ELN modifications. But Stephen Gallagher will notice us in case of important items.

I get more feedback over my Perl modules from EPEL than from Fedora

Not surprisingly, Perl is for long-term and stable projects.

nowadays

agreed

Well, I think we'll discuss specific code next week, then.

But we still have other outstanding issues:

- What exactly are we aiming at?

- What is the end product (a playbook, per service?)

- what about manual documentation and synchronization?

to name a few.

the end product is the ability to write a playbook that will configure one or more services

Sorry a little late to the party

this playbook will contain or link to your inventory (the list of your machines) and variables (your "personal" data)

hello

mowest: Doesn't matter, welcome!

hello, mowest

Hi jednorozec!

Emmanuel Seyman: just the ability, or will we distribute ready to use, or better ready to configure, playbooks for every service?

I would prefer thje latter

We can ship sample playbooks, I guess

I don't think we can cover all possible usecases

The readme.md file should show example usage along with instructions for setting up inventory files and playbook variables

I'm not familiar with Ansible, but my son has written I think Ansible playbooks to bring up each of the services on our homelab server with podman. I'm in favor of having (whatever the correct Ansible term is) in Server documentation or easily installable so that people can be up and running services on a Fedora server in as little time as possible.

I think the playbooks should restrict their scope to "small office/home office" scenarios. Anything else becomes quite complicated and interdependent in a hurry.

We definitely can't cover all usecases. But we should provide a ready to use (and configure, e.g. the inventory, or a list of dirs to export) default use case or common use case.

that sounds fair

And each playbook could contain a "hook" where local extensions can be connected

I agree that we can't cover every instance, but 3-6 good examples that demonstrate different concepts like on bare metal, in a podman container, etc. would be good.

mowest +1

so I think we should have the common use case like: podman/virthos connected to NFS shares for app data

Ideally I just setup my inventory and keys and run the playbook to get a libvirthost with all the belles and whistles connected storage and so on...

I think, on the server side we should provide a list of directories to export, for whatever usage.

same for podman, also some simple web proxy

jednorozec: agreed!

with those 3 scenarios I think we will have a great entry point for people interested in running fedora server

My work was geared towards four different ansible roles: 1) NFS server install/configure 2) NFS client install/configure 3) Maintain list of directories to export from server 4) Maintain list of directories to import from server

John Himpel: Yeah, but you probably mix server and client playbooks?

For server wi would provide 1) and 3) ?

of perobvably 4 to import from another server

For server run 1 (probably only once) then 3 for each time you change list of directories served. Same concept for 2 & 4

that is a great start!

I can help with libvirt host/podman playbook,

hm my network is glitchy :/

jednorozec: Let me finish cleaning stuff up and uploading to repos. Hope to finish that this week. Then we can extend/modify them for libvirt host (if necessary) and podman (if necessary).

Just a proposal: let's concentrate here on NFS. And open a new topic for libvirt next. For libvirt we have a working manual documentation. I think we should try to get the same via Ansible.

sorry for the de-rails

Spontaneity is fruitful!

You just have to systematize it in a follow-up :-)

OK, I think

we agree about the goal for NFS. I will try to put it together in the ticket. So we can improve it.

Should we forward to the next topic?

I see no objections, therefore

As we discussed earlier, as a firsrt step we wanted to run a survey

to collect actual usages and probably demands for specific improvements or extensions.

Do we want to stick with it?

I think so

OK, I think, we agree on this.

So, in a next step we would have to collect questions. I think, we should use hackmd.io?

I can offer to make a first draft, if nobody else steps up.

But can someone else do a research how we can conduct such a survey in fedora?

And about the best way to di it?

mowest, are you still typing?

discourse has a survey function, I believe

I would imagine that the best way to get responses from the community would be to publish the survey in discussion.fedora

As Emmanuel Seyman pointed out

there was a 'what can we use AI for?' survey recently

Yes, i remember the discussion on mailing list.

@mowest do you think, you could take over the lead, how to orginize and to do it, when we have collected our questions?

I will look into creating surveys on https://discussion.fedoraproject.org

Yes, thanks!

So we have for now:

It looks like I will need to use another tool "Limequery" and link to that, which is setup on Fedora Infra apparently by the website address. I will see what I come up with.

mowest: Thanks.

So we can proceed with the next topic?

No objections here

yes, let's go

This is a longstanding plan, together with the next. Any idea how to bring this forward?

OK, I think we're out of steam now, let's switch to open floor for a moment.

I ahve one here, Peter Boy can you please close https://pagure.io/fedora-server/issue/114

Anythink to comment, to ask, to contribute here?

and also maybe grant me ticket permissions on the fedora-server repo?

Still looking forward to meeting you all at Flock...

Does anyone know if there will be KIWI presentations at FLOCK? That will probably determine if I attend or not.

I wrote comment there and also in the BZ and created change draft for the change...

jednorozec: Sorry, I read your mail, but couldn't act because of a bike accident.

I will elevate your permissions. And comment on this

not sure about that but the guys who did the move of KDE and containers will be there

John Himpel: This is one presentation that is planned Fedora Cloud Edition Image Building Workshop - Using Kiwi Definitions

oh that will be the same one as on devconf, really good entry point into kiwi!

This is the whole schedule = https://cfp.fedoraproject.org/flock-2024/schedule/

Would the KDE move create a media to be consumed by anaconda for a new install? Inquiring minds want to know :-)

I'm going to be there with my son.

not sure about that but both Neal and David will be there so we can put together hallway session about kiwi and server

OK, we are out of time now and should close. Any objections!

3

Sounds good

nope

2

1