10:11:53	 From Alex Iribarren : Please raise your hands, I'll ask you to unmute
10:19:48	 From Nils Hoimyr : This just strenghtens the case for specific security updates to CC7 if need be during that window.
10:21:33	 From Andreas Pfeiffer : Wouldn’t puppet completely abstract the question of package manager ??
10:22:09	 From Jan Iven : (similar to Diana, for IT-ST: if we have an upgrade-in-place between major versions, we can more easily update even during RUN3)
10:22:14	 From Marc Dobson : I do not think so as everyone has to move their software to the new package manager
10:23:01	 From Thomas Oulevey : And we have Ansible ;-)
10:23:13	 From Andreas Pfeiffer : Indeed ! :D
10:27:45	 From Thomas Oulevey : why do we change the naming of repositories from the upstream "8-stream" to "s8" ? 
10:27:58	 From Thomas Oulevey : and then cs8 in the image name ?
10:28:44	 From Thomas Oulevey : (not important for discussion just a note)
10:28:50	 From Alex Iribarren : to try to maintain consistency with what we already had for c8
10:30:03	 From Piotr Golonka : For WinCC OA we are in close contact with the vendor to find out what is possible. Currently their recommendation for the former CetnOS customers is to go with Oracle Linux... And wtill they keep RHEL
10:31:07	 From Marc Dobson : Thanks Pitor
10:31:12	 From Marc Dobson : Piotr
10:34:56	 From Piotr Golonka : FYI, possibility of running RHEL for dedicated online systems (WinCC OA) with extended support was asked about by experiments controls coordinators yesterday (JCOP Coordination Board).
10:35:08	 From Thomas Oulevey : I can't raise my hand, give me the floor when you can
10:36:09	 From Roberto Divia : Same in ALICE.
10:36:45	 From Roberto Divia : Snapshots at well-specified moments and with well defined sets. No rolling updates.
10:39:03	 From Andreas Pfeiffer : So you keep every one of the ~1500+ daily snapshots over the next five years ??
10:39:04	 From Jan Iven : If you snapshot already, I don't see much difference with current model (minor releases+updates also being snapshotted) and new (ongoing updates being snapshotted)? the installtime-issue is valid but keeping old images on install server around would work?
10:40:09	 From Alex Iribarren : @Andreas, we keep the last 397 days
10:40:14	 From Jan Iven : (having one install image per month (3month? for each new kernel?) would not eat up much space..)
10:40:46	 From Andreas Pfeiffer : OK, so you can’t install a machine with a snapshot from 2yr ago.
10:42:37	 From Andreas Pfeiffer : How is the selection of which one to keep done ?  So that the experiments know which one is the “long term” one …
10:42:39	 From Alex Iribarren : not with our snapshots, but you're free to mirror the ones you need for longer-term storage
10:43:54	 From Andreas Pfeiffer : OK, that would work, thanks. :)
10:44:09	 From Jan Iven : RedHat negotiations  - are these "HEP-wide" or just for CERN?
10:44:23	 From Ben Morrice : hep wide
10:45:01	 From Jan Iven : (thanks - somebody from HEP software foundation involved as well?)
10:47:15	 From Ben Morrice : Jan: https://github.com/HSF/Linux4Science/discussions/1#discussioncomment-309928
10:48:40	 From Nils Hoimyr : If we decide today that CC7 will be used for Run3, then there is time to take into account a variety of requirments and hence consider other Linux distros
10:50:15	 From Marc Dobson : Not that much time, Phase-2 hardware is being produced now and an OS for it would be needed fairly quickly to see where we are going.
10:51:47	 From Piotr Golonka : Is RPM or deb actually that much relevant? Podman/docker? Flatpak? Fedora SilverBlue?
10:52:21	 From Federico Vaga : it shouldn't
10:52:28	 From Piotr Golonka : ... I mean: think what Stream9 is going to be forked-off
10:52:30	 From Roberto Divia : For ALICE: yes, it does. We are RPM-based.
10:52:55	 From Marc Dobson : There is work to move from RPM to something else
10:53:13	 From Piotr Golonka : Absolutely! I would hated to do this.
10:53:24	 From Marc Dobson : The extent of this is often not seen as once the experiment world, it involves many small groups.
10:53:34	 From Marc Dobson : Once in the...
10:55:06	 From Andreas Pfeiffer : SuSE has a (slightly) longer support history :)
10:55:52	 From Andreas Pfeiffer : Then the next question is: What is the profit (in terms of cost) to use a commercial distro ?
10:56:30	 From Jan Iven : I think we have two issues: a short-term "what to do for RUN3", and long-term "where should we be in 50..10years"?
10:56:42	 From Jan Iven : "5..10". sorry.
10:56:54	 From Roberto Divia : Indeed.
10:57:28	 From Piotr Golonka : ... meaning the very next one, for the LS3. We need to plan for it now already
10:57:44	 From Nils Hoimyr : The trends is towards more diverse hardware (ARM, GPUs.... ) so several Linux distros may be unavoidable in the medium to long term
10:57:52	 From Andreas Pfeiffer : Yep. Though they are not completely independent, in the sense that the “short term solution” may/can/will be affecting the “longer term” one
10:59:02	 From Andreas Pfeiffer : For long-term I think Debian (stable) is unbeatable, though it’s not necessarily the one with the most recent versions (at least not a while into it)
10:59:17	 From Jan Iven : are experiments really tied to a distro? they ship their own compilers, interpreters, libraries.. can we decouple the "kernel = hardware" from experiment-level app layer?
11:01:21	 From Zach Marshall : apologies for the naive question, but is someone present representing the accelerator itself? The experiments seem well represented, but of course we need the LHC team itself on board as well.
11:02:11	 From Marc Dobson : Jan, yes we are. Only some parts of the systems use their own compilers etc… Most in the online use the default one.
11:02:44	 From Marc Dobson : And also the kernel and kernel modules which they build for the HW that is connected to it.
11:03:43	 From Jan Iven : (kernel: on systems that have no untrusted users, and have been certified for the actual hardware, would you ever need to upgrade?)
11:04:21	 From Thomas Oulevey : Thanks !
11:04:36	 From Andreas Pfeiffer : Depends, I guess, some security updates may come for that one too.
11:04:42	 From Marc Dobson : Yes because users do log into those. And as you know one can never trust users. They sometime do stupid things or bring in things from unknown sources.
11:04:47	 From Andreas Pfeiffer : thanks