Meeting Minutes 2025

AGREED: We switch to winter time in sync with US, again. And we modify the UTC time to keep local time unchanged.

AGREED: Server documentation will be modified according to proposal https://hackmd.io/@pboy/BJnkUjA3el

AGREED: Each WG member selects an open issue on Forge and either makes a suggestion on what to do or deals with the issue themselves until next week meeting.

We have a provisional repo structure according to our previous discussions and ticket 173.

AGREED: We start with a repo structure as proposed in ticket 173

AGREED: We keep the naming as is today, for now.

AGREED: For the time being we use one issue list for everything. We will examine whether it proves effective.

AGREED: We keep the tickets repo as is

AGREED: We adjust the docs repos to the Antorra required structure.

AGREED: We restructure the Ansible repo later

AGREED: We switch as soon as we can switch the docs pages to forge and freeze the docs pages on pagure in the meantime

ACTION: pboy will restucture the Docs repos as aggreed

ACTION: jwhimpel will make a proposal for the Ansible repo structure.

Since today we have a new forge:

https://forge.fedoraproject.org/server

AGREED: If the current installation problems are not fixed by the release date, no SBC version of the Server should be released at all.

AGREED: WG proposes the Server SBC installation medium as release blocking. Details to be coordinated with other editions.

Pboy made contact with Ryan Lerch about new forgejo Fedora forge. We’ll get a stg account fist to be able to explore our options.

Paul repoerted, the remote interactive installation via ks file and OEMDRV USB drive is back. So again, you don’t need to edit the kernel command line (clumpsy with a non_US keyboard layout) nor any termnal connected.

A severe issue is the SBC. It still doesn’t work at all.

Eseyman got a reply from the Tomcat group regarding the SELinux issue with ajp proxy. Blocking localhost access is intentional. The policy is no default network access. And localhost is seen as network access. We will pick up this later.

We want to clearly distinct between the Server Edition core use case and special uses such as ARM SBCs. In particular, a Raspberry Pi 4, the current focus in Fedora, is unsuitable for Server. We will reorganize the documentation to clearly separate the SBC articles from the main topics.

AGREED: we create a PR for httpd to include the config files as of stg article 2025-09-03

The change regarding bios boot systems as discussed at last meeting is now active.

Proposal: Limit release-blocking status of BIOS systems to just certain scenarios

There is a new issue with the ajp proxy protocol. The default configuratin is blocked by SELinux.

We want to put forward our Webservice configuration. pboy will update the guide in stg to the latest development.

Folllowing a discussion about the status and differentiation of editions

AGREED: We agree to keep server as a package based system for the foreseeable future and don’t intend to start an alternative planning any time soon.

AGREED: We publish the extended SW Raid doku now as is (adding F42 in meta data). Further refinement probably after F43 release.

As discussed with QA, this provides them with a concrete test path and decision criterion for the SW RAID item in the biosboot release blocker criterion.

The main installation media seem to be OK so far. With SBCs an issue with arm-image-installer on LVM are back again.

After a longer discussion:

After an intensive discussion:

According to our voting we have 2 new members:

After a longer discussion about various ways to support NFS file sharing:

AGREED: In about 14 days, Emmanuel and John will come up with a plan about implementation details.

ACTION: Jocelyn will write a summery of the discussion with the status as of next Wednesday, so that we have something to base our decision on.

Eseyman completed an RPM of most interesting Ansible roles, specifically the NFS role (currently on COPR). Now we have to test the RPM and combine with jwhimpels work.

AGREED: we will use the next two weeks to test the available tools and decide how we can best achieve our goal of supporting services. And we will consider how we can optimize the distribution.

We will make our FAS Server Group (server-wg) official, containing the authoritative list of WG members. The current WIKI list will then be removed..

AGREED: We postpone the NFS Ansible project until eseyman has a robertdebock rpm ready and eseyman and jwhimpel have a proposal how to (re)use it for our purposes.

Eseyman reported, he was pleasantly surprised. Lots of people seem satisfied with what we ship. The main call for improvement is documentation, use cases, best pratices, …​

Eseyman and pboy will prepare a finer analyses of the data.

AGREED: We try to create a Server Edition test day (or better test week)

ACTION: jwhimpel will create a Beta RPM that will establish the "base" filesystem and related documentation for our ansible setup (on admin’s workstation)

The final release actually no longer contains the possibility of a remote interactive installation initialized without terminal by kickstart file. This is the result of a system-wide switch from VNC to RDP initiated by the Workstation WG. The discontinuation is a side effect of the changes in Anaconda and was unannounced. We did not expect this omission and unfortunately discovered it too late.

AGREED: We will create a list of minimal manual tests and try to establish a test week for F43.

A particularly intensive checking require all changes that are in any way related to the Workstation WG.

There is a pending PR.

ACTION: Paul Maconi (Aggraxis) will take care or the PR regarding the size change for aarch64 architecture.

Testing so far revealed some minor issues. Really bad, the remote interactive installation doesn’t work anymore with the kickstart file.

The reasons are obviously specific to the characteristics of aarch64 architecture. The x86_&4 version is not affected.

AGREED: If the ongoing attempts to reduce the size are not successful, Server WG is in favor of an increase to 1.2 GB as an exception for now.

Pboy will perform the hardware relataed testing, eseyman takes care of SBCs and jwhimpel of VM installation. We will track the results in an issue.

During US daylight saving time, we adjust the beginning of our meeting to 17:00 UTC, so that local daily routines remain unchanged.

The podcast scheduled for yesterday was delayed, due to staff shortage. Mowest will try to join the new meeting together with pboy.

AGREED: Server WG emphasizes that we need access without official certificates for private networks

Eseyman will track this.

A review of the current state of the 'countme' statistics is to become a regularly recurring agenda item, approximately monthly or quarterly.

AGREED: We will let the survey closed as decided before.

AGREED: Server WG will limit NFS support to NFS 4

AGREED: We create a supported web service following the current description of a basic setup in the server docs. This is also the basis for developing an Ansible playbook.

We have a "brainstorming collection document" at https://hackmd.io/@pboy/SJzz7VcFJl. We agreed that everybody will enter their idea about data analysis and questions we try to answer by data analysis.

Once the survey is complete, pboy will export the data in a secure form and make it available to interested WG members.

AGREED Server WG decides to start work on a honelab / homeserver spin-off.

AGREED Eseyman, Mowest and Pboy will finalize suitable contributions to Flock 2025 on the basis of the current collection of ideas and submit the application(s).

We currently have 4077 participants, of whom 548 have fully answered the survey. The others have left out several questions. We will continue to promote the poll. The currently scheduled end date is February 18.

ACTION Emmanuel Seyman will post on redit forum, Fedora general user on Discussion, Fedora Devel on Discussion and the corresponding mailing lists about the next 2 weeks to participate

AGREED mowest takes over the conduct and organization of the data analysis. Support from Gwmngilfen, Emmanuel, pboy and anyone who is interested

At Fosdem there was a discussion with abbra about freeIPA and localKDC (local Key Distribution Center) that is planned for F43. It may help us to resolve the UID issue with NFS.

ACTION pboy will ping Abbra and the freeIPA / localKDC people to advice and help us to integrate the right thing into Fedora Server.

Regarding Flock we discussed several ideas for contribution. We will continue the discussion on the mailing list.

Nothing new so far. But we have a new issue to deal with, managing firewall portmapper port (111).

Nothing new so far.

Just for information see Current Actions summary below.

Fosdem runs from February 1-2. Fedora will be represented by a booth, eseyman and pboy will be there.

The poll is open and running. We don’t have any data on an interim result yet.

Because of the current issues with NFS which may take some tie to get resolved we open another path with web services.

We postpone this until we have evidence of our server poll.