SCIENTIFIC-LINUX-USERS Archives

February 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:
"~Stack~" <[log in to unmask]>
Reply To:
~Stack~
Date:
Mon, 24 Feb 2020 17:41:32 -0600
Content-Type:
multipart/signed
Parts/Attachments:
text/plain (4064 bytes) , signature.asc (847 bytes)
On 2/24/20 8:09 AM, Peter Willis wrote:
> Hello,

Greetings!

> The variation in uses of t Scientific Linux is quite interesting.
> 
> As mentioned before, we are using it for fluid dynamics modelling and
> oceanography, in the context of parallel computing with OpenMP and MPICH.
> 
> I am curious to see what everyone else have been using it for.
> 
> Perhaps, if it’s not too much trouble, people on the list might give a
> short blurb about how they use it and why.

I had been using CentOS 5.x (and 4.X before that) for the base image in
a large High Performance Cluster (many hundreds of nodes) with Red Hat
as the infrastructure and important nodes but couldn't afford to pay for
that many nodes, especially since they were so minimalist. It was a
government institution that did everything from weather forecasts to
military parts modeling.

Red Hat released RHEL6 in November of 2010. I had a /really/ strong need
to move packages to EL6 for a new workload but again couldn't afford the
price tag (I can't remember how many nodes we had at that time...I think
over 300...we grew even more over the years). SL6 released 4-ish months
later in March 2011. I had been pestered to move to it but I wanted to
stick with CentOS because I knew it and was a little bit involved in the
community at the time. But there were a lot of issues and they were
re-tooling their build scripts. Finally in May, I gave in and made the
switch with the plan to revert back to CentOS6 when it released.

During the installations, I realized I goofed and my scripts that
figured out if it was on a RH host or a CentOS host were "failing" the
check and defaulting to RH...except they weren't failing...I thought it
was a fluke or something else was really broken so I dug into it. That's
when I realized that SL was actually closer to RH then CentOS was!

A lot of my "proofs" for this claim require quite a bit of setup or
configuration, but the easiest one that anyone can test is simply "yum
update --security". SL publishes a security channel, CentOS doesn't!
Which something even that simple means less work as I no longer needed
to scrape and parse out a massive list of CVE's to determine which
packages I needed to install (at the time I had to apply security
patches daily but I didn't like patching/updating packages just because
it was an update...if something broke I wanted as few things to check
for as I could. I'm at a different job now and still have the same
restriction though). Soon I ditched a TON of custom scripts for CentOS
because it all just worked great on SL the same as it did the RH hosts!
Bonus, RH6/SL6 was a lot more stable for us and let us do a few things
even better so I expanded the cluster a few hundred more nodes by the
end of 2011.

By the time CentOS 6 released in July 2011, I had zero desire to go back.

Today, I'm still the admin of big High Performance Clusters for a well
known economic modeling and research institution. Things that management
really cares about that they want to be able to pickup the phone and
yell at someone or get warm fuzzys about support contracts (eg: Ceph),
those are still RH. All the servers I care about being close to RH but
can't justify in the budget for (aka management won't pay for) RH are SL
6/7. At home, I run SL7 for all my servers (Lubuntu for my
desktop/laptops because I really like LXQT). I've even done the CERN
charity donations before where I send thank you notes to the SL devs in
the notes fields (no idea if they got them or not). :-D

We are just now exploring RH8/CentOS8. I've got a single RH8 VM I'm
doing testing in and I'm building a CentOS 8 later this week. There's
little reason for us to move to 8 at this moment...the bigger push is
that we still have a MASSIVE system (~100 nodes and quite important)
that is SL6 based and we need to get off of it by end of summer (both
hardware support ending and EL6 being EOL in November). So I'm trying to
figure out if I am going to take the easy path to SL7 that I know I can
do or if I jump it to 8... *shrug*

So that's more than just a short blurb...guess I will shut up now. :-D

~Stack~



ATOM RSS1 RSS2