SCIENTIFIC-LINUX-USERS Archives

December 2020

SCIENTIFIC-LINUX-USERS@LISTSERV.FNAL.GOV

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Yasha Karant <[log in to unmask]>
Reply To:
Yasha Karant <[log in to unmask]>
Date:
Tue, 8 Dec 2020 16:23:11 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (88 lines)
For those who want to be nauseated, here is the essential quote of the post:

The future of the CentOS Project is CentOS Stream, and over the next 
year we’ll be shifting focus from CentOS Linux, the rebuild of Red Hat 
Enterprise Linux (RHEL), to CentOS Stream, which tracks just ahead of a 
current RHEL release. CentOS Linux 8, as a rebuild of RHEL 8, will end 
at the end of 2021. CentOS Stream continues after that date, serving as 
the upstream (development) branch of Red Hat Enterprise Linux.

Translation -- as a for-profit vendor, IBM does not want to subsidize a 
competitor to RHEL that is without fee.  Building RHEL from the source, 
that IBM RH is required to distribute under the terms of the license 
from which the source is obtained, is resource prohibitive.  I do not 
know the fate of the next Princeton clone of RHEL. What will the various 
HEP collaborations do?  Will Fermilab/CERN provide internal professional 
person power (not just grad students and postdocs for whom such support 
is only as much as their research supervisor requires) to maintain an 
internal RHEL 8 clone from RHEL source?  Given that the public 
pronouncements were to use CentOS 8 as the RHEL 8 clone in the HEP 
production environments, and that this is now not "long term" possible 
(CentOS stream is beta at best -- more or less a Fedora like unsupported 
cycle), one may be curious as to the future of HEP.  It is possible that 
IBM (that has branches almost everywhere in the nations from which HEP 
collaborators are housed) will decide for a publicity-gain and 
tax-write-off to partner with Fermilab/CERN and license RHEL 8 (and 9 
and ... ) at either a very reduced fee or for "free".  But what about 
those of us who are not in such a HEP collaboration?

I too have heard some nasty comments about Oracle EL 8 in terms of 
Oracle really using it more or less as a lure with the eventual goal of 
fund extraction from those who attempt to use the executable distro 
licensed for free.  Also, what about the various professional add-on 
distros, such as EPEL or ElRepo?

I suspect that I made the "correct" planning decision to switch to 
Ubuntu LTS (until such time as Canonical follows the RH IBM path ...). 
For those contemplating such a move, the changes are not that drastic, 
particularly if one "debugs" on a single sample of each class of machine 
(workstation, server, etc.).  I am willing to provide my notes (howtos) 
that I have garnered for Ubuntu LTS (my machine currently is 20.04.1 
LTS, and there is a 18.04 LTS machine that shortly will upgrade-in-place 
to 20.04 LTS -- both are laptop workstations, not "enthusiast home use" 
machines, one Dell, one HP).  For applications that are standardized for 
EL (we had these on a high performance compute server with a particular 
Infiniband implementation, but that machine largely is now obsolete), I 
am not certain what would be involved in porting -- if the libraries 
(typically, .so) are available in LTS, this should be not too difficult 
-- particularly on a set of replicated installs.

An additional large question for the community is the future of the 
X86-64/Nvidia GPU architecture.  The latest Fujitsu HPC is ARM based, as 
are the latest Mac OS machines.  Is ARM coming of real use beyond "smart 
phones" and the like, but as "real computers"?

Take care.  Stay safe.

On 12/8/20 3:38 PM, ~Stack~ wrote:
> Anyone else on the verge of tears after reading today's CentOS blog post?
> https://urldefense.proofpoint.com/v2/url?u=https-3A__blog.centos.org_2020_12_future-2Dis-2Dcentos-2Dstream_&d=DwICaQ&c=gRgGjJ3BkIsb5y6s49QqsA&r=gd8BzeSQcySVxr0gDWSEbN-P-pgDXkdyCtaMqdCgPPdW1cyL5RIpaIYrCn8C5x2A&m=t2J9jUFVgun90FIMquH4QRfvlPyoP8v5iYSZEcA87_g&s=-5u1jeYbTrmg0sZScxVN-0qJ1ifC2BEGlmW4_B70SYw&e= 
> 
> If you don't know CentOS Stream, it's "upstream RHEL". No, not Fedora. 
> Yes, that too is "upstream RHEL". CentOS Stream a rolling release (so 
> good luck getting long term steady kernels/packages) that is trying to 
> be Arch like but with RHEL flavor. It sits in between RHEL and Fedora. 
> It isn't and won't track steady releases like RHEL. It will have things 
> before RHEL, except for security patches which will still come in 
> whenever someone gets around to it. And, no, they still won't tag their 
> security patches as such because they expect you to apply patches (and 
> potentially reboot) at their whim.
> 
> For those of us in the scientific community who have packages from 
> vendors that standardize on RHEL dot releases, I'm not sure what we're 
> going to do. We have RHEL licensing on the important infrastructure 
> nodes but the hundreds of compute nodes, VM's, dev systems, and misc? 
> Going all RHEL would kill our budget. And I don't care if Oracle Linux 
> is free or how good of a clone it is, you only get burned by Oracle once 
> (and you are usually to broke to be burned a second time).
> 
> I suppose we can shift nearly all of our infrastructure to Ubuntu LTS 
> but there's a lot still left that I'm not sure we can move to CentOS 
> Stream nor can we afford to go to RHEL. Guess we are freezing our 
> conversations about moving away from SL7 and have year to figure it out 
> then make it happen...
> 
> *sigh*
> 
> ~Stack~

ATOM RSS1 RSS2