<@pboy:fedora.im>
17:00:16
!startmeeting fedora-server
<@meetbot:fedora.im>
17:00:18
Meeting started at 2025-04-16 17:00:16 UTC
<@meetbot:fedora.im>
17:00:18
The Meeting name is 'fedora-server '
<@pboy:fedora.im>
17:00:25
!topic Welcome / roll call
<@pboy:fedora.im>
17:00:35
I'll post the agenda in 2-3 minutes.
<@pboy:fedora.im>
17:00:35
As usual, let's wait a moment for everybody to show up.
<@jwhimpel:fedora.im>
17:00:51
!hi
<@zodbot:fedora.im>
17:00:52
John Himpel (jwhimpel)
<@pboy:fedora.im>
17:01:37
Hi John!
<@jflory7:fedora.im>
17:02:19
!hi
<@jwhimpel:fedora.im>
17:02:20
Peter Boy: Good afternoon here. Sunny but chilly. Normal "spring" weather for us.
<@zodbot:fedora.im>
17:02:20
Justin Wheeler (jflory7) - he / him / his
<@jflory7:fedora.im>
17:02:28
Mostly here to lurk 😉
<@eseyman:fedora.im>
17:03:06
!hello
<@zodbot:fedora.im>
17:03:07
Emmanuel Seyman (eseyman) - he / him / his
<@eseyman:fedora.im>
17:03:17
hi, folks !
<@pboy:fedora.im>
17:03:44
@Justin: You are welcome! Later today I have a urgent question. Would be nice if you could watch your DM
<@pboy:fedora.im>
17:03:58
Hi Emmanuel!
<@jflory7:fedora.im>
17:04:21
I will 👍🏻
<@pboy:fedora.im>
17:04:35
Justin Wheeler: Thanks.
<@pboy:fedora.im>
17:05:17
I would say, we start now! We have our quorum and Nirik is lurking, too, as usual (I think)
<@pboy:fedora.im>
17:05:25
!topic Agenda
<@pboy:fedora.im>
17:05:32
!info Follow-up actions & announcements
<@pboy:fedora.im>
17:05:41
!info Server user poll
<@pboy:fedora.im>
17:05:53
!info Follow-up of the F42 server development and test cycle.
<@pboy:fedora.im>
17:06:02
!info Ansible assisted installation and configuration of NFS service
<@pboy:fedora.im>
17:06:09
!info Open Floor
<@pboy:fedora.im>
17:06:27
Anything to add / to delete / to modify?
<@pboy:fedora.im>
17:06:51
OK Let's start
<@pboy:fedora.im>
17:07:01
!topic 1. Follow-up actions & announcements
<@pboy:fedora.im>
17:07:16
Nothing new here from me (as usual :-) )
<@pboy:fedora.im>
17:07:25
Anything to announce here?
<@pboy:fedora.im>
17:07:54
OK. let's proceed.
<@pboy:fedora.im>
17:08:01
!topic 2. Server user poll
<@pboy:fedora.im>
17:08:17
no tracking issue yet
<@pboy:fedora.im>
17:08:30
Maybe some (longer) input from me.
<@pboy:fedora.im>
17:08:43
<@pboy:fedora.im>
17:08:43
3.
<@pboy:fedora.im>
17:08:43
<@pboy:fedora.im>
17:08:43
2.
<@pboy:fedora.im>
17:08:43
Further options?
<@pboy:fedora.im>
17:08:43
<@pboy:fedora.im>
17:08:43
We need to decide how to proceed. Possible options are:
<@pboy:fedora.im>
17:08:43
As a first step we should put together the substantive questions that we would have liked to answer with the data analysis. They may be as trivial as the question of how many home server users we have. It may be possible to spice it up by considering the consequences that the result could have for our plans.
<@pboy:fedora.im>
17:08:43
1.
<@pboy:fedora.im>
17:08:43
a) Everyone works on the data set individually and extracts the desired results.
<@pboy:fedora.im>
17:08:43
Gwmngilfen is ready to help us with data analysis and date presentation. But that requires that we tell him what questions we have. Therefore, we should prepare our evaluation goals. I have prepared a brainstorming document on hackmd.io.
<@pboy:fedora.im>
17:08:43
<@pboy:fedora.im>
17:08:43
b) We create a shared workspace. This would allow us to install R (a statistical program) on our test server (sisyphos). We could then work together to specify and optimize the corresponding analysis steps.
<@pboy:fedora.im>
17:08:43
For write permission you have to login to hackmd.
<@pboy:fedora.im>
17:08:55
These are my thoughts so far. Any more ideas?
<@pboy:fedora.im>
17:08:55
<@jwhimpel:fedora.im>
17:10:54
I need to study the raw data for a while before I could begin formulating a meaningful set of questions.
<@pboy:fedora.im>
17:11:56
In scientific theory, it should be the other way around. 😃
<@pboy:fedora.im>
17:12:41
I think, from our discussions during the preparation, we had some ideas and some questions which made us make the pull questions as they are.
<@pboy:fedora.im>
17:13:01
Unfortunatley, we missed o write them down.
<@eseyman:fedora.im>
17:14:14
There's a 4-day weekend soon in France. I plan to use it to formulate questions, study the data and see where that takes me
<@eseyman:fedora.im>
17:14:33
but I don't have the brain cycles to spare until then
<@pboy:fedora.im>
17:16:14
OK, I think it is a good idea, to collect a first set of substantive questions.
<@pboy:fedora.im>
17:16:32
Can we do somethink to inspire ideas?
<@eseyman:fedora.im>
17:17:11
I doubt we're the first group to have done this. There's got to be precedent
<@pboy:fedora.im>
17:18:55
Emmanuel Seyman: indeed. But I know of any experiences from other groups. So, that would / could one part of out talk in Prague: Our experiences and what can others maybe learn from it. :-)
<@pboy:fedora.im>
17:19:12
Emmanuel Seyman: indeed. But I don't know of any experiences from other groups. So, that would / could one part of out talk in Prague: Our experiences and what can others maybe learn from it. :-)
<@eseyman:fedora.im>
17:20:02
I suspect we're the first in Fedora to have done this but other groups from other distributions must have...
<@mowest:fedora.im>
17:20:13
!hello
<@zodbot:fedora.im>
17:20:14
Steve Daley (mowest)
<@pboy:fedora.im>
17:20:26
Hi Steve !
<@pboy:fedora.im>
17:21:03
mowest: Any comment on our current discussion?
<@mowest:fedora.im>
17:23:07
A few basic questions would be: 1. A pie chart perhaps of the make up of our users (percentage that home lab with Server, SMB with Server, and Enterprise with Server). 2. 10 most common use cases for Fedora Server.
<@mowest:fedora.im>
17:23:38
Top 3 most requested improvements to Fedora Server or to our Documentation?
<@pboy:fedora.im>
17:24:03
Yeah, I see you have ideas !!
<@mowest:fedora.im>
17:24:06
Top 5-10 apps that are run on top of Fedora Server
<@pboy:fedora.im>
17:24:11
mowest: ++1
<@zodbot:fedora.im>
17:24:12
pboy gave a cookie to mowest. They now have 9 cookies, 1 of which were obtained in the Fedora 42 release cycle
<@mowest:fedora.im>
17:24:47
Answers to those few questions would help us make decisions of where to invest our time.
<@mowest:fedora.im>
17:25:27
Probably should have a Pie Chart of hardware installs, vs. VM's, vs. VM's on Cloud Plateforms...
<@pboy:fedora.im>
17:26:55
mowest: Agreed to all your items. Please, could you start to enter these into the hackmd.io page?
<@mowest:fedora.im>
17:27:12
Sure, I just saw that document today.
<@pboy:fedora.im>
17:27:26
Thanks.
<@pboy:fedora.im>
17:27:50
So I think we agree to start now a concentration on possible substantive questions.
<@eseyman:fedora.im>
17:27:51
Will do
<@pboy:fedora.im>
17:28:44
OK, any first comment on the 2. item, how to work? Each for themself or try so set up a workspace?
<@pboy:fedora.im>
17:29:00
OK, any first comment on the 2. item, how to work? Anyone for themself or try so set up a workspace?
<@pboy:fedora.im>
17:29:57
OK, obviously none so far.
<@pboy:fedora.im>
17:30:15
Any additional comments? ideas?
<@pboy:fedora.im>
17:31:19
OK, obviously none at all at the moment. Let's proceed.
<@pboy:fedora.im>
17:31:30
!topic 3. Follow-up of the F42 server development and test cycle.
<@pboy:fedora.im>
17:31:38
no tracking issue yet
<@pboy:fedora.im>
17:31:46
Some input from me (again :-) )
<@pboy:fedora.im>
17:32:01
The F42 cycle was not particularly brilliant for us in my opinion.
<@pboy:fedora.im>
17:32:01
<@pboy:fedora.im>
17:32:01
For example, we discovered far too late that the Anaconda crew had also removed the Kickstart capability for interactive installation, when they converted to RDP. Server installation now requires indispensable a terminal, except for a fully automatic, non-interactive installation – at least I hope so. I didn't test it. That's really really bad! In some server racks, this is quite a hassle. While plugging in USB sticks is a breeze anywhere.
<@pboy:fedora.im>
17:32:01
<@pboy:fedora.im>
17:32:01
What are the causes of this? On the one hand, we started testing this feature too late. On the other hand, we didn't check the change proposals thoroughly enough. I certainly couldn't have imagined that such a thing was even remotely possible. But we should have realized that backward compatibility and uninterrupted, continual operation do not play a major role for at least many Fedora developers – as can be clearly seen from the discussion about Gnome, Wayland and X11.
<@pboy:fedora.im>
17:32:01
<@pboy:fedora.im>
17:32:01
One of our main tasks in the coming months must be to get this function back into Anaconda. Given the resource constraints, this could prove difficult.
<@pboy:fedora.im>
17:32:01
And we need to examine the change proposals for negative side effects much more thoroughly.
<@pboy:fedora.im>
17:32:01
<@pboy:fedora.im>
17:32:01
And we need to start some smoke tests with Rawhide right now. F42 regularly announced test releases on the devel list. This has obviously stopped with F42. At least I didn't notice any mail. We need to get a replacement for this.
<@pboy:fedora.im>
17:32:01
<@pboy:fedora.im>
17:32:01
This is just one example. We should discuss this in more detail.
<@pboy:fedora.im>
17:32:14
So much from me for now.
<@pboy:fedora.im>
17:34:14
The F42 cycle was not particularly brilliant for us in my opinion.
<@pboy:fedora.im>
17:34:14
<@pboy:fedora.im>
17:34:14
For example, we discovered far too late that the Anaconda crew had also removed the Kickstart capability for interactive installation, when they converted to RDP. Server installation now requires indispensable a terminal, except for a fully automatic, non-interactive installation – at least I hope so. I didn't test it. That's really really bad! In some server racks, this is quite a hassle. While plugging in USB sticks is a breeze anywhere.
<@pboy:fedora.im>
17:34:14
<@pboy:fedora.im>
17:34:14
What are the causes of this? On the one hand, we started testing this feature too late. On the other hand, we didn't check the change proposals thoroughly enough. I certainly couldn't have imagined that such a thing was even remotely possible. But we should have realized that backward compatibility and uninterrupted, continual operation do not play a major role for at least many Fedora developers – as can be clearly seen from the discussion about Gnome, Wayland and X11.
<@pboy:fedora.im>
17:34:14
<@pboy:fedora.im>
17:34:14
One of our main tasks in the coming months must be to get this function back into Anaconda. Given the resource constraints, this could prove difficult.
<@pboy:fedora.im>
17:34:14
And we need to examine the change proposals for negative side effects much more thoroughly.
<@pboy:fedora.im>
17:34:14
<@pboy:fedora.im>
17:34:14
And we need to start some smoke tests with Rawhide right now. Before F42, there was regularly announced test releases on the devel list. This has obviously stopped with F42. At least I didn't notice any mail. We need to get a replacement for this.
<@pboy:fedora.im>
17:34:14
<@pboy:fedora.im>
17:34:14
This is just one example. We should discuss this in more detail.
<@mowest:fedora.im>
17:35:02
I still find it confusing what to test manually, and what gets tested automatically. If automatic, we don't want to reinvent the wheel, but if manual testing is needed, then we would need to get a list of those tests and see if we have hardware available to do the testing in our group.
<@mowest:fedora.im>
17:35:55
What you discovered Peter Boy I don't know if I would even run into it with my current use cases. But I see why it would be a serious issue in a datacenter.
<@eseyman:fedora.im>
17:36:21
I suspect there are not enough of us to cover all use cases
<@eseyman:fedora.im>
17:36:53
So we either set expectations lower or try to recruit more people
<@mowest:fedora.im>
17:37:30
I'm wondering how other "spins" or "editions" get their testing done, and if they have methods that work that we can adopt.
<@pboy:fedora.im>
17:37:47
Or both :-)
<@pboy:fedora.im>
17:37:47
<@pboy:fedora.im>
17:37:47
; So we either set expectations lower or try to recruit more people
<@pboy:fedora.im>
17:39:12
Yes, we have a resource problem. We need to set priorities. For example, testing our five installation guides first and early.
<@jwhimpel:fedora.im>
17:39:34
Several of the other groups are having the same issues with testing. Desktop and FreeIPA seem to be exceptions. The "virt-*" pages are tested extensively by their developers during development, but not as much during Beta, testdays, etc.
<@pboy:fedora.im>
17:40:43
Yeah, with the "virt-*" thingies we didn't experiencnd problems so far.
<@jwhimpel:fedora.im>
17:41:36
It is my personal opinion that the Workstation (Gnome) folks are making decisions for the distribution as a while. They seem to not consider Server, etc in their decision making.
<@pboy:fedora.im>
17:41:37
And maybe we should establish a test day or a test week to get more people involved.
<@pboy:fedora.im>
17:41:56
l
<@pboy:fedora.im>
17:42:09
John Himpel: Agreed very much !!!
<@mowest:fedora.im>
17:42:21
I like the idea of setting a test week from Thursday to Tuesday, reports of experiences at our next Wednesday meeting.
<@pboy:fedora.im>
17:42:25
That's really a problem.
<@pboy:fedora.im>
17:43:15
That's really a problem. (I meand Johns statement)
<@eseyman:fedora.im>
17:44:25
+1 for a test week (announced well in advance so we can plan for it)
<@mowest:fedora.im>
17:45:06
Yes, John Himpel , makes a good point. I don't think this is as big of an issue in Fedora as it is for the Ubuntu Spins and what the Flagship decides, but this is worth bringing up to the Coordinating Council. How can Workstation Decisions be weighed a little more against the other additions and goals of the spins in Fedora.
<@mowest:fedora.im>
17:46:19
Test week could be followed and added to by the wider community through a thread in "discussion.fedoraproject.org"
<@pboy:fedora.im>
17:47:24
One problem here is that Server is (still) heavily integrated into the general Fedora structure. CoeeOS, for example, is much more independent. They have a completely separate system for generating distributables.
<@pboy:fedora.im>
17:48:11
We share far too many elements with Desktop that we don't even use and that only cause us problems.
<@jwhimpel:fedora.im>
17:50:04
We could fix some of this by migrating from the current build tools to the new build tools. From the feedback I've seen, making changes to Server edition would become much easier. Generating test images much earlier in the development cycle would become possible.
<@pboy:fedora.im>
17:50:34
John Himpel: That sounds great!
<@mowest:fedora.im>
17:51:26
I like the idea of sharing a base with Desktop. If we are most heavily used in the home lab space, I think this makes sense and perhaps encouragement for us to adopt some of the Desktop defaults a little quicker too, even though I know this is not ideal for Enterprise use. If we had to have our own infra separate from Desktop, that sounds like more than our little group could hande.
<@mowest:fedora.im>
17:51:33
I like the idea of sharing a base with Desktop. If we are most heavily used in the home lab space, I think this makes sense and perhaps encouragement for us to adopt some of the Desktop defaults a little quicker too, even though I know this is not ideal for Enterprise use. If we had to have our own infra separate from Desktop, that sounds like more than our little group could handle.
<@jwhimpel:fedora.im>
17:52:25
Unfortunately, not all of our artifacts are supported by RH kiwi. But it seems to me (and I'm not an expert) SUSE is able to generate all of the artifacts we currently generate. I'm not good enough of a Python geek to know what it would take to evaluate the functionality and/or do the porting to Fedora.
<@mowest:fedora.im>
17:52:34
John Himpel: That sounds interesting. I'm not familiar at all with the current or new build tools, but that sounds promising.
<@eseyman:fedora.im>
17:52:53
I agree that having both the server and client come from a common base is a good thing for us
<@pboy:fedora.im>
17:53:58
mowest: We don't need our own infrastructure. For starters, it would be enough to have our own installation groups and not share them with Desktop. So no more group base, but base-desktop and base-server. etc.
<@pboy:fedora.im>
17:54:39
Emmanuel Seyman: Exactly that cause currently several issues.!
<@jwhimpel:fedora.im>
17:55:03
There have been some presentation on kiwi at recent Flock gatherings. Kiwi would generate our image from a common set of rpm(s) as used by Workstation. But it would allow us to add (and keep separate) our modifications to the Workstation settings.
<@pboy:fedora.im>
17:55:16
Specifically, that change the base installation group without any coordination with us, at least as far as I norice.
<@eseyman:fedora.im>
17:55:47
Separating both would solve some issues and cause others, imho. It is not obvious that it would be a huge win for us
<@pboy:fedora.im>
17:57:01
Emmanuel Seyman: Yeah, we would have to check in detail, indeed. But doing nothing about it doesn't help either.
<@eseyman:fedora.im>
17:57:09
True
<@pboy:fedora.im>
17:58:28
Proposed !agreed: We will create a list of minimal manual tests and try to establish a test week for F43.
<@pboy:fedora.im>
17:58:28
Out time is going up. I suppse we could agree:
<@jwhimpel:fedora.im>
17:58:33
The builds currently are separate now. As I understand it, kiwi makes it possible for each edition/spin to apply their own changes to the base. Currently almost noone knows how to apply changes for editions/spins using the old tools.
<@pboy:fedora.im>
17:58:35
Proposed !agreed: We will create a list of minimal manual tests and try to establish a test week for F43.
<@pboy:fedora.im>
17:58:35
Our time is going up. I suppse we could agree:
<@pboy:fedora.im>
17:59:29
John Himpel: thanks for the info. I didn't know that details so far.
<@pboy:fedora.im>
18:00:19
Any objections regarding my proposal?
<@pboy:fedora.im>
18:00:28
Or additions / modifications
<@jflory7:fedora.im>
18:00:56
Seems reasonable to me.
<@pboy:fedora.im>
18:01:05
Yeah, Ok
<@jflory7:fedora.im>
18:01:13
Coming up with some test cases and working with folks like Sumantro Mukherjee to get them scheduled is a good approach
<@pboy:fedora.im>
18:01:16
!agreed: We will create a list of minimal manual tests and try to establish a test week for F43.
<@jflory7:fedora.im>
18:01:28
He might also know others who could help from Fedora QA
<@pboy:fedora.im>
18:01:51
!Agreed: We will create a list of minimal manual tests and try to establish a test week for F43.
<@pboy:fedora.im>
18:02:15
Zodbot ??
<@pboy:fedora.im>
18:03:00
Justin Wheeler: Yeah, thanks for the hints. I think, we'll try to get it running.
<@pboy:fedora.im>
18:03:50
OK, time is up and Zodbot seems to sleep :-)
<@pboy:fedora.im>
18:04:32
I'll close and let us continue on mailing list and hackmd as diskussed.
<@pboy:fedora.im>
18:04:44
By Bye!
<@mowest:fedora.im>
18:04:58
Bye
<@pboy:fedora.im>
18:05:21
And we will start next week with Ansible / NFS to make up for missing it this meeting (again) I propoe
<@pboy:fedora.im>
18:05:23
se
<@eseyman:fedora.im>
18:05:25
bye all
<@pboy:fedora.im>
18:05:52
!endmeeting