On Thu, Jan 25, 2024, 08:15 Jeroen Baten via lpi-examdev <
[email protected]> wrote:

> My 2 cents
>
> On 25-01-2024 14:35, Fabian Thorns via lpi-examdev wrote:
> > * Shall we stick with Postfix, switch to exim or reduce the mail topic
> > to no longer including configuring an MTA?
> I would stick with Postfix.
>

RE (my argument for Exim as a #2 consideration):
https://list.lpi.org/mailman/private/lpi-examdev/2023-October/004326.html

I often play 'devil's advocate' ... i.e., I only made an argument for Exim
being a consideration (in addition), because half (or more) of the Internet
relays are using it, not because I use it (although I have).  I'm a
[former] red Fedora** so I'm clearly in the 'I use Postfix' camp,
especially being that I made (and still make) most of my money on RHEL**,
and have always loved Stream (even before it was public).  But ...

The reason why Exim is so popular is because a lot of hosting providers run
Debian, not just CentOS/Stream.  I'm also a huge Debian fanboi, especially
when I worked as a maintainer (of no note) building DPKG alongside RPM, and
even I love Exim out-of-the-box.  And that's likely only going to
increase as Red Hat (*cough*IBM*cough*) keeps making overnight decisions
without any lead time or community consideration.**

>
> > * Shall we reduce Wireguard to awareness level and re-focus on OpenVPN?
>
> Why? What is the reasoning behind this question? I really like to know
> before I send an opinion.
>

RE (my argument against Wireguard):
https://list.lpi.org/mailman/private/lpi-examdev/2023-October/004304.html

Wireguard's crypto sucks ... badly.  I won't allow it, period, in any
enterprise I touch, even putting US NIST FIPS 140-2 (or 140-3) aside --
although in that latter case,CMMC 2.0 is starting to affect US
companies, not just those around defense and critical infrastructure (but
totally them too).  Wireguard designed for speed at the expense of
security, to a horrendous level.

I'm willing to listen to anyone who differs, if they have an alternative,
crypto view.  But I don't know anyone who takes a trip to those realities
who defends it for any enterprise or even SOHO use these days.  If you are
one, I would love to discuss further, in case I'm ignorant or oblivious of
something.

- bjs

**FURTHER DISCLAIMER (way off-topic ... and long):   I see Debian as the
future for a lot of the community.  I don't think OpenELA is the future,
not at all (more at the end).

I don't bite my tongue on this any more, and will argue with any red
Fedora, former or current, especially anyone saying it's not IBM.  It's IBM
darn it!  I mean, I think we all have a history of not always agreeing with
Red Hat decisions.  My first go back to the '90s, but the paywall that
happened in the '00s, and other things, are largely because of Oracle.
Back then, Red Hat usually wouldn't make major moves, and only minor ones,
to thwart direct copyright/IP theft from Oracle (I mean, they've removed my
personal copyright before, let alone others I've seen), because Oracle was
much bigger, and has lots of lawyers, even when they're the ones that
shoudld be sued.  That changed once IBM purchased Red Hat in late 2018, and
the community continues to suffer.  ERGO and ...

MOST NOTABLY:  Even when Red Hat decisions were unpopular, Red Hat was
always good about *'lead time' *... until 2020, and definitely *not *in
2023+!

In fact, even though the move to make Stream public alongside
*'rebuild' *(traditional)
CentOS, which I strongly feel is far better, more secure, and with less
bugs, especially for hosting providers and Internet servers (as both
Facebook and Twitter can long attest, as they were running it before it was
public), was not only a decision made *before *the IBM purchase, and CentOS
and CentOS Stream co-existed for awhile, but many of us argued for (my
background is FSI / Wall Street, where we had been running it since the
late '00s -- yes, it's that old of a workflow).  BUT ...

It was not only the decision 2020 to nuke CentOS (non-Stream)
next-to-overnight after CentOS 8 had already been out well over a year
(either nuke it before release, or support it to term -- not nuke it after
release!), with only short grace period (EoL by 2022), but it was followed
by the 2023 decision in both Red Hat layoffs (which Red Hat growth was
still outstanding) and total end of Source RPM releases -- the latter of
which really got to most of us.  That was obvious when Red Hat
(*cough*IBM*cough*) decided to kill (i.e., moderate) internal Memo-list,
which literally flies in the face of both Whitehurst's *'An Open Culture'* and
one of the last things I contributed to before I left as a FTE (I would be
contracted afterwards, as I did not want to be an employee in Services any
more, only Engineering) ... The *'Open Source Decision Framework*' which
was purposely designed to stop such non-sense from happening, as it had
increasingly as Red Hat grew from sub-1,000 employees at my start, to well
over 10,000.  Most were acquisitions with employees not accustomed to open
source values.

Understand that US-based Red Hat was not first with *'Enterprise Linux,' *but
German-based SuSE AG.  It was quite the value, even though SLES was 2-6x
the cost of RHL EE at the time.  SuSE grabbed 5x the paid share, but once
Red Hat refocused with RHAS (lka RHEL), because SuSE did not release SRPMs
(and wouldn't until after the Novell purchase)-- and there wasn't something
like CentOS, Red Hat flipped to 6x the paid share over SuSE and the rest is
history.  That's why I don't see a net positive for Red Hat.

And so IBM is going to cross Oracle as it sees fit, and the community will
suffer.  Although I don't think OpenELA is going to work out either.  Both
Oracle and CIQ have poor open source histories, and SuSE attempted to sell
Red Hat RPMs and SLAs, with Microsoft backing, a decade ago, and it was a
colossal failure -- all 3 have either open source and/or sustainment
issues.  HENCE ...

Debian is likely the future of the community, even putting Canonical
Advantage and SLAs aside (but that will factor in as well).  I love Stream,
and would love to see Stream (first 6+ years) + SRPM rebuilds after (late
4+ years), but that's now dead. IBM killed it, and really put everyone in a
pickle.  And I will argue with the top levels of Red Hat over how they
killed themselves.  It will take awhile, maybe not until the 2030s, but a
shift is coming.  Again, I don't think the OpenELA is it.  But as always,
we'll see.


--
Sent from my phone, apologies for any brevity as well as autocorrect
Bryan J Smith - http://linkedin.com/in/bjsmith
_______________________________________________
lpi-examdev mailing list
[email protected]
https://list.lpi.org/mailman/listinfo/lpi-examdev

Reply via email to