SL didn't have "support", but the mailing list provided excellent, real-world support. At least during the SL 3-5 timeframe, CentOS had nothing even close that I could find.
There's obvious value in the broader community involvement that comes through CentOS, and in providing a free alternative for those who don't need / can't afford RH licensing. Wiping out CentOS would hurt the ecosystem. That doesn't mean it can't happen, but
it seems unlikely.
One company I worked for never bought RHEL because it would have been too pricey under the circumstances. We found a cou0ple of bugs that got reported back upstream. Another company I worked for moved to RHEL from CentOS as soon as it could afford to, because
we needed the support. Both companies made the right decision for their situation, and both were good for RedHat, just in different ways.
RedHat has been fine with CentOS and SL. I see no reason for that to change. IBM is not micro-managing RedHat. Hopefully that won't change, either.
As we could not afford the license-for-fee model that RedHat started a number of years ago (prior to which, one could download and install production RedHat -- not the "Fedora" equivalent -- licensed for free but without RedHat
support -- but updates, etc., were available without fee), we too went with CentOS. Before RH, I used Debian, but there were issues of stability. RH was stable. The problem with CentOS was that it was more or less a volunteer deployment, and we did not
have the personnel to join the effort as our internal and external funding could not be used for that purpose. Once SL became a more-or-less "stock" version of RHEL, and given that SL had professional funded employed personnel (as required by HEP and funded
by the various governments that support Fermilab or CERN), this was the logical choice. SL came with no support, but as several of us (myself included) were at one epoch "kernels internals" persons, and were "systems persons", and not as "IT" but as scientists
and engineers, with the SL users list for "help", we had no significant issues -- see the recent exchange concerning a bug in EPEL that prevented an "easy" upgrade of the MATE desktop GUI environment.
However, RedHat is now owned by IBM, and CentOS is the RedHat "licensed for free" distro front end. The only reason IBM exists is not to support the goals of the Freesoftware Foundation (GPL), but to support profit -- it is a major for-profit (effectively,
trans-national) corporation. Thus, one cannot rely upon entities within such a corporation to do anything that will undermine or reduce the profits of the corporation (including the overall compensation package of the CEO and the like), except in those nation
states that have enforced regulations controlling the product deployments. The USA has very little compared to much of the EU. As Fermilab/CERN do not exist for the same purpose as IBM (individual scientists who may be the group leaders, etc., at such entities
notwithstanding), SL was a viable alternative. There is absolutely no reason to assume that IBM will be such an alternative unless one wants to pay. I am not going to argue with those who claim we are "freeloaders" despite paying the taxes that in part support
Fermilab and CERN, but not CentOS -- if we cannot pay, we should not use -- but the realities of much university-based academic research is that there is no money and we do what we can.
In the simplest terms. I trust IBM to maximize overall return-on-investment (e.g., profit), and a "free" CentOS that truly competes with licensed-for-fee products does not fit that for-profit model.
Yasha Karant
On 2/21/20 7:41 AM, Michel Jouvin wrote:
Hi,
I'm surprised by the so negative feeling against CentOS which is a great project too and has been working well since it was "acquired" by Red Hat. I see no official sign that it should change. Moving from SL to CentOS is straightforward, I don't think you
can speak about it as a migration as it is exactly the same product. And staying with CentOS will give you a chance to meet the DUNE people at some point and more generally the HEP community if you liked interacting with it!
Cheers,
Michel
Le 21/02/2020 à 16:32, Peter Willis a écrit :
Hello,
Thanks to everyone for clarifying the future status of SL.
I guess it’s time to start researching he docs for Ubuntu/Debian or something.
Looks like we need to revise our computing cluster plan.
The computer here is
pretty small with only two nodes and a controller totalling 112 CPUs.
We use it for numerical modelling of ocean and river currents and sediment transport (OpenMP/MPICH/FORTRAN).
The changeover will be
pretty small. We are still waiting for the OK for a new node or two.
The current nodes are ten years old. The update to a controller and SL7 was a
last ditch effort to join the two nodes and increase the scale of the models without costing too much more.
In other news, the link you shared has an article about ‘DUNE’ which seems like an interesting project.
I’d certainly frostbite a few toes to just stand around and watch that thing run experiments.
Thanks for the info,
Peter
>Hello Peter,
>
>> Is Scientific Linux still active?
>Scientific Linux 6 and 7 will be supported until they are EOL, but there will be no SL8.
>
>Here is the official announcement from last April: