The end of Scientific Linux

Fermilab has maintained Scientific Linux, a derivative of Red Hat Enterprise Linux, for many years. That era is coming to an end, though: "Toward that end, we will deploy CentOS 8 in our scientific computing environments rather than develop Scientific Linux 8. We will collaborate with CERN and other labs to help make CentOS an even better platform for high-energy physics computing." Maintenance of the SL6 and SL7 distributions will continue as scheduled.

From:		Pat Riehecky <riehecky-AT-fnal.gov>
To:		"scientific-linux-announce-AT-listserv.fnal.gov" <scientific-linux-announce-AT-listserv.fnal.gov>
Subject:		Scientific Linux and EL8
Date:		Mon, 22 Apr 2019 08:37:42 -0500
Message-ID:		<ea737b05-e168-79ec-f75b-76b69ba95b87@fnal.gov>
Archive-link:		Article

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Scientific Linux is driven by Fermilab's scientific mission and focused
on the changing needs of experimental facilities.

Fermilab is looking ahead to DUNE[1] and other future international
collaborations. One part of this is unifying our computing platform with
collaborating labs and institutions.

Toward that end, we will deploy CentOS 8 in our scientific computing
environments rather than develop Scientific Linux 8. We will
collaborate with CERN and other labs to help make CentOS an even better
platform for high-energy physics computing.

Fermilab will continue to support Scientific Linux 6 and 7 through the
remainder of their respective lifecycles. Thank you to all who have
contributed to Scientific Linux and who continue to do so.

[1] For more information on DUNE please visit https://www.dunescience.org/

James Amundson
Head, Scientific Computing Division

Office of the CIO
Fermi National Accelerator Laboratory

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)

iEYEARECAAYFAly9vsQACgkQsLQYPxkqfX2RbACfZghqc7YKILcIwK9cGkq6mC+p
0N4AniBzteElb5p2jPuQQktEOgK2z4w+
=MQZ2
-----END PGP SIGNATURE-----



-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: GnuPG v1.4.5 (GNU/Linux)

mQGiBEpXadARBACHhOfMUuT/4iDvBRmm6bEsvnMN++L79aYhEUMNlrZ2TjKPjPvG
Y0vGk+I0JhUJWutkQRZVbqgVDsNjmnELnACK+xvdryvtxh50wCI9WUl7CT5EV7BS
/jD7JxTFbXyC/Xv0ixMB9vj6U9cySyE8PxONp0HzO6LTIr1OMPgDUsP4lwCgh8De
fmY8TN2m9a0huLdNrnmKw0cD/2bkt6rJAi3+BGHWNgQ9Nb/4wQff8BKGDtL/8acp
3yH91axuD2iYCKw0ZP5akBpRGv+4e30Plmbi1f5NaEDo9Ga1c4TDPopwgiYhrVLj
56efoTfP2AiZl3iBKFPI83/YOhrVZF8UiYoAoUnOFpOg8vmtCzgvYip5UZLTgbfJ
lcWvA/9vMb8By+1pHjW98d7GkzvZqzyMtWlbO7PXCn8P7bGQYjwvyTGiRNz3q22c
2Z29qQw4r1L1L1JGsUwuOMahkczWVdD4TRHc8mhVJEUEA6AkNAZc+Ymsfr/ip0kX
nSZLE3pYVifOhBRO8EbT0WhCMScmZNpwvZU//HKL/p+n3LArUrRZU2NpZW50aWZp
YyBMaW51eCAoUlBNIHNpZ25pbmcga2V5IGZvciBTY2llbnRpZmljIExpbnV4KSA8
c2NpZW50aWZpYy1saW51eC1kZXZlbEBmbmFsLmdvdj6IYAQTEQIAIAUCSldp0AIb
AwYLCQgHAwIEFQIIAwQWAgMBAh4BAheAAAoJELC0GD8ZKn19cXIAnA5R+EbaYr4/
IL6It/UxHXlBFIajAJ9bwmNDF14uvDnFigg1PLevLTBnTIhGBBARAgAGBQJKV6lf
AAoJENpq0AiC/ReyKLQAmwVC/Ii3sAKsptwZKHw/uk1kbupCAJ0eIzSaUo1hSa1V
fP7O/dqigu6JAbkCDQRKV2nZEAgAzAcaC7unRNdrIwAGGKqOIvI8WNwpftHY50Y5
zPSl7vtWVkp3N+2fynJR+tW4G/2xDChBbPzPz/TavRyBc21LKzAlym8qIGEE02cZ
U/YJAYnbAkNNiGMOAnAIjBw1KUcQamAxdk0glE7MP1JiXY1MO4tTW38UEcvQbSvg
Mh/eECqFOwiQXJmkPpZhPUwnwmZRCV4vlCZQM3CMExZ9pDV/V+kuhefw2WeheXyh
g4DC88gcrv2mO0I3sVmpxn3JLMayiMlQbOSYLQuNVKN/EFDwuAbS9Ane7vm6wF9X
NswMX0I/vO1IVvSN1fi5ZM71QzeYUGKBQv97kLO20hbRWZ1V+wADBggAys+jhlYH
mtFZQxV4an1ucqnVauKnstj0zF88Hiy7yivT3W5h3Zd067uOfcBQCJUlt7y8sYD2
q9htm5Rrxx+J29bl0zxwrEatnv0gLzprSa7Ei3wR6IrvBM3Ic0mGSzlsSxlzaFtt
Pwak5C47vX9+PwKEKXFdM1gVzHTuD6PXEYxA4YMlQGeGVA68FvTHxMHpf8POQWTV
QtjoI0flvFT7d4ozqUJdjJZxJDFQ7GO2YdIfF3sUdfn5kFxK0SUzqrmCYXeheniS
LKC4mpAR0PetWJ7r1gY5khHb2eHW1vdEBYUXlHjB+jLaOBns05MHMZYd4CHe8q/Q
gzMeVlh8YLSdZYhJBBgRAgAJBQJKV2nZAhsMAAoJELC0GD8ZKn19iU8AniUIFu32
VeRJ+VKL2vBQMVbFVZOMAJ434Bi99fN2CSh7T62oxtQvhw70fw==
=eL9H
-----END PGP PUBLIC KEY BLOCK-----

The end of Scientific Linux

Posted Apr 22, 2019 16:44 UTC (Mon) by rsidd (subscriber, #2582) [Link] (6 responses)

The vast majority of scientists never used it -- it was intended for one community only (high-energy physics). It should have had a more appropriate name.

The end of Scientific Linux

Posted Apr 22, 2019 17:24 UTC (Mon) by teknohog (guest, #70891) [Link] (4 responses)

I came across SL quite recently. I was looking for a live distro for a student workshop and I needed gnuplot, which was particularly hard to find in a live system. SL LiveDVD Extra was the only one that fit this bill.

Most scientists will probably need some specific software that doesn't come in any default install, but can be installed in a variety of distros (incidentally, I used to spend time with the HE physics community). The distro choice is much more crucial if you need something available in a LiveCD.

The end of Scientific Linux

Posted Apr 22, 2019 17:51 UTC (Mon) by dowdle (subscriber, #659) [Link]

Learn how to build your own live media (with your desired packge list) and it won't be a problem. It is actually fairly easy. Install livecd-tools, get a sample kickstart... and just augment the package list.

The end of Scientific Linux

Posted Apr 22, 2019 19:44 UTC (Mon) by markhahn (guest, #32393) [Link] (1 responses)

Were you fixated on your live distro being RO? Don't most people use live images from USB sticks (and can therefore trivially install new packages - and critically, update existing ones)?

Live USB is limited

Posted Apr 22, 2019 21:48 UTC (Mon) by jreiser (subscriber, #11027) [Link]

live images from USB sticks (and can therefore trivially install new packages - and critically, update existing ones)

The available space in the filesystem after booting a Live USB stick is limited: enough fora few handfuls of small or medium packages, but definitely less than one might think. I have run out a few times, such as "dnf update" of a Fedora Live USB two months later.

The end of Scientific Linux

Posted Apr 23, 2019 13:03 UTC (Tue) by nitroflow (guest, #131593) [Link]

The wonderful thing about linux is that you can install it on a USB stick and then customize it and update it.

The end of Scientific Linux

Posted Apr 22, 2019 18:38 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]

> It should have had a more appropriate name.

Kinda late to debate the naming when a distro is announcing EOL

The end of Scientific Linux

Posted Apr 22, 2019 17:08 UTC (Mon) by jccleaver (guest, #127418) [Link]

Sadly, the era of "white box" rebuilds of RHEL gaining any meaningful traction are probably behind us. Scientific Linux in earlier times had benefits, but by the late EL6 period there was very little it did that could not have been covered by "___ Extras" repositories on a binary-compatible RHEL rebuild (notably with a more finely-tuned kernel), and SL7 might as well have been a CentOS 7 Spin.

It's a shame for a number of reasons, including simply having heterogenous diversity within the EL ecosystem on the RedHat side. At this point, the people the most likely to want the stability of an EL environment to begin with have to have a *significant* justification to eschew the 100% binary-compatible guarantee goal of CentOS itself. After all, while rebuilds can often get unofficial support, once you're off the reservation you're sort of off the reservation. More than anything else, this is probably why a systemd-free EL version isn't easy to get off the ground.

Unless your name is Johnny Hughes, it doesn't seem likely than anyone's EL rebuild project will take off in the future given the resource re-investment required. I think the community will be less-strong as a result.

The end of Scientific Linux

Posted Apr 22, 2019 22:41 UTC (Mon) by mohg (guest, #114025) [Link] (1 responses)

I've used Scientific Linux for many years on a number of systems (to support a research group in a science field that wasn't high-energy physics). The Fermilab team have done a great job maintaining SL for as long as I've used it.

From memory, there was a time (several years ago now) when CentOS updates were pretty delayed, but I think SL has had a consistenly good track record in this regard. (Perhaps this is nostalgia.)

Given the current healthy state of CentOS (and of course the acquisition by Red Hat), and the seemingly ever-tightening funding situation for US national labs, this feels like an inevitable outcome. I don't know how much of their work week the people involved spent on SL, but I hope they will transition to other projects in the lab. It is great to hear that older versions will continued to be supported.

And no, I don't think there was anything special about the distribution from a "scientific" perspective. It was just a solid, professionally maintained distribution with its roots in the HEP community.

Thank you Scientific Linux!

The end of Scientific Linux

Posted Apr 23, 2019 1:25 UTC (Tue) by LightDot (guest, #73140) [Link]

From memory, there was a time (several years ago now) when CentOS updates were pretty delayed, but I think SL has had a consistenly good track record in this regard. (Perhaps this is nostalgia.)

I also think this is correct. SL has been solid for us, thanks to all that work or were working on it.

To reminiscent a bit, there was indeed a significant delay at the beginning of the CentOS 6 cycle in 2011.

The CentOS situation back then wasn't at all good. After the initial delay following the RHEL 6 release, CentOS 6 got released but than the updates for the OS were non-existent for some time and that was not the level of service CentOS users were used to.

If I recall correctly, no critical security issues were discovered while this was going on, except for a remotely exploitable Firefox bug. I'm sure not that many workstations were deployed and those who were could use a vanilla Firefox or rebuild this particular RedHat's SRPM to remedy the situation.

In any case, I remember Scientific Linux getting a big user boost at that time, although CentOS returned to its usual level of service later on.

By the time RHEL 7 and CentOS 7 were released, the situation has greatly changed and there were indeed questions if any other rebuilds besides CentOS make sense. I guess the end of Scientific Linux further answers this dilemma now that RHEL 8 is 'round the corner.

IMHO, this does make the community weaker in a sense, although on the other hand, the knowledge or the ability to make alternative RHEL rebuilds isn't going away. If CentOS falters or fails in some way, the community will respond.

The end of Scientific Linux

Posted Apr 23, 2019 1:20 UTC (Tue) by pabs (subscriber, #43278) [Link]

Glad to see a derivative distribution merging with its parent. That seems to validate the "work in the upstream distro" approach taken by the Debian Science team. OTOH there isn't any mention of Fedora/COPR in the announcement but I guess that is due to the length of support that Fedora releases get.

https://wiki.debian.org/Teams/DebianScience

The end of Scientific Linux

Posted Apr 23, 2019 15:38 UTC (Tue) by smoogen (subscriber, #97) [Link]

I have a sad on this.. I started working with Fermi Lab in 1998(?) on the first Fermi Lab Linux which later became Scientific Linux. The people who worked on the OS over the years did a lot of work in the open about problems and fixes they needed to make for their customers. You could get serious and spirited debates on everything from needs for particle physics to openssh working with various kerberos environments. Plus a lot of TeX..

The end of Scientific Linux

Posted Apr 23, 2019 21:22 UTC (Tue) by Kamilion (guest, #42576) [Link] (4 responses)

Welp, guess that closes the book on redhat for me.
No more community distros, and a focus on keeping expensive commercial software running?

After redhat ate Centos, the only RHEL-distro I could continue to recommend was Scientific.

Now there's nothing.

Go get debian buster. Forget about SRPMs, RPMs, DoesNotFinish, Yum, Extras repos, RPMForge, and all the rest of the redhat headaches.

Bye Redhat. You were nice twenty years ago; I still have fond memories of Redhat Linux 7.

... I wonder if I should hold a little funeral in my backyard with the old optical media?

The end of Scientific Linux

Posted Apr 24, 2019 1:40 UTC (Wed) by scientes (guest, #83068) [Link] (1 responses)

RedHat does an incredibly amount of development, that projects like Debian will never match, and they do it all under free licenses. I wish people would stop thinking that making money is a bad thing.

The end of Scientific Linux

Posted Apr 24, 2019 7:35 UTC (Wed) by sytoka (guest, #38525) [Link]

The objective of Debian is not to make software development but build a distro. The great job of Debian is to support multi-arch (many soft just don't compile by default outside of amd64 or arm64, thanks to Debian) and now Debian give more and more the key to build and verify your package (see Reproductible build project). Red-hat keep the key of their building infrastructure.

So Debian and Red-hat are just complementary...

The end of Scientific Linux

Posted Apr 24, 2019 10:29 UTC (Wed) by nilsmeyer (guest, #122604) [Link]

> ... I wonder if I should hold a little funeral in my backyard with the old optical media?

Since it's not bio-degradable, please don't ;)

The end of Scientific Linux

Posted Apr 26, 2019 19:16 UTC (Fri) by hkario (subscriber, #94864) [Link]

> After redhat ate Centos, the only RHEL-distro I could continue to recommend was Scientific.

except that CentOS is continuing as it was, with software being freely downloadable?

The end of Scientific Linux

Posted Apr 24, 2019 12:29 UTC (Wed) by walex (guest, #69836) [Link]

This is very sad, because an independent distro tied to a very long term project like the CERN Grid was very useful to have around for "continuity". But first CERN lost interest in SL, and decided to adopt CentOS, and then the end of SL was inevitable.

It is part of a general trend: universities and research organizations always cut first non-research engineering jobs, because they are "overheads", not "front line" and the result is inevitably a considerable shrinking of the ability to operate and maintain infrastructure, or later having to pay a fortune to vendors for remedial operation and maintenance.

But in the short term dumping/recycling a few (probably SL took 3-4 engineers, a drop in the ocean of HEP staff levels) staff jobs and letting a vendor fund the equivalent work seems to work.

At this point I guess that if we are to be limited to "in house" "not so independent" community distributions, OpenSUSE looks very attractive compared to CentOS, with much better support from their vendor, and much nicer update cycles, and SUSE has been growing robustly, so it is a viable long term sponsor.

The end of Scientific Linux

Posted Apr 25, 2019 0:04 UTC (Thu) by bwned (guest, #131622) [Link]

We have to discontinue SL because of DUNE. You know how DUNE gets. You can't reason with it.