What do you think of the recent Red Hat announcement about CentOS Linux/Stream?

[bluegroper]

Thanks a lot scottro for your time and detailed answer.

+1

I've recently "upgraded" a soon-to-be-deployed server from CentOS 8 to CentOS 7, since 7 is perfectly suitable and now has a longer shelf-life.
Its a strategy designed to buy plenty of time, while we wait to see how these changes all play out, and what suitable alternatives emerge.
We're especially interested to see whether Linux Cloud (aka Lenix) and/or Rocky develop into viable alternatives.
But there's no rush. Hoping that patience will reward those who are prepared to wait.

[aks]

As a first lesson: Redhat can't do any kind of community engagement - even if it hit them in the face (laughable since that's what they (Redhat) proport to be waiting to do. The way this was announced is the evidence. It's obvious from what was announced and the way it was announced, that the reaction they would get, is the reaction they got.

If RH wants to accelerate the distribution, then okay. One of my main gripes about the distribution is that is extremely slow to get the new features. While this is great for stability, it sucks when you look at the feature set other distributions offer (even AWS Linux!)

I've been a Fedora user for a long time and have first hand experience of new software breaking what I already have, which is why I don't run Fedora on any production system (although to be fair, it has got a lot better than it was some years back). Linux's kernel interfaces have been quite stable, userland ofcourse is a different story. The really nice thing in Fedora is that I can upgrade across point releases and whole versions, can't do that with RH or CentOS (well you can, but that's a different story) - maybe that's what should be tackled?

When I do yum update -y, will it f**k up what I've already got? That's the real question. My (somewhat limited) understanding of stream is that it'll be software that's been QA'd and the community used to test, before the paying customers. That's okay, I'm prepared to put it on a staging or "experimental" (easy to rollback/throw away) kind of environment, but really, how am I supposed to feedback (previous attempts have all failed - seemingly I'm not as big as facebook).

My view of Redhat is that it's the 10,000 pound gorilla sitting in the corner, somewhat antiquated, but exerts (possibly) undue power and influence when it wants it's own way.

I think I'll try stream for a while and see how it goes. If I get big, obvious (well at least to me) problems I'll just move to another distribution. It's not 2003 anymore.

[optikab]

This is nothing short of diabolical.

If RHEL thinks this is a clever move to onboard new customers from Centos they are sorely mistaken.

Computer engineers are not regular consumers we don't get bullied into upstreaming because we can't work out an alternative, we are problem solvers.

Our company has just spent the last 6 months upgrading to and acquiring new Centos 8 servers, a costly and time consuming process. Giving less than a year to about-turn this work is a disgraceful and shameful act in behalf of RHEL. If RHEL wanted to encourage new customers then they should have the decency to support Centos 8 for at least 2 years. Given there are plenty of viable alternatives (paid and free) why on earth would we pay money to a company that has just completely shafted us?

Centos has been a mainstay of the the internet for a very long time, if RHEL want to kill it that is fine, but do so gracefully, intimidation does not go unnoticed and is not forgotten. What else do they have their sleeve for existing customers I wonder. If they have any sense they will be joining the exodus.

[svennd]

Well, for whatever we use Centos, it will remain a great (/best?) option, but dropping support from Centos 8 seems like a bad move; It certainly is a showstopper for me to become customer with RH/IBM. But what is most annoying is that the community might be split up even further; Some people are looking to Debian variants, some to Suse variants, but nobody is now going to want to jump on this sinking ship ...

[nmusatti]

I haven't answered the above poll because I would give several answers at once, both from a personal standpoint and from a professional one. While Red Hat are entitled to make such a choice, they are going about it in the worst possible way and the official posts aren't really redressing the situation.

There are several problems: first and foremost, one year is way to short a time for any company with more than a handful of CentOS servers. It means that they must start looking for an alternative now and cannot afford to wait to see if CentOS Stream is stable enough for their needs. These are likely to move away from CentOS, and RHEL as well. This is the position the company I work for is in. We are looking at Ubuntu right now. The fact that it is way cheeper than Red Hat is an added bonus.

Then there's the million dollar question: will CentOS Stream ever be stable enough to trust it with important loads? I doubt it. From how it was described CentOS Stream is likely to get the bugs before RHEL and the fixes after. It may be acceptable for people developing for RHEL, to be ready in advance for new minor (and major!) releases, but I don't think I would trust it for serious use. I don't feel like adopting it either for the kind of in-house servers that are running CentOS now or for my home PC, which is running CentOS 7 right now. I might move to CentOS 8 just to gain a year before making a strategic decision, given CentOS 7 is becoming a bit too old.

Another important question is whether Red Hat can be trusted to have CentOS 8 Stream ready in time. I have to say that the delay with which they made OKD 4 available, way after OpenShift 4 was released, makes me wary about this.

Is there anything Red Hat can do to improve the situation? I believe there is. First, they should extend CentOS 8 support to the end of the year in which RHEL 9 is released. That would give them time to complete the transition to the new CentOS Stream and give current CentOS users time to evaluate if CentOS Stream fits their needs. Then they have to guarantee that each CentOS Stream X release will be supported for as long as the corresponding RHEL X release. If it is really meant to be upstream, how can they do it differently anyway? In this way users will have the choice between staying with the older but stabler version or move on to the bleeding edge.

Would this be enough to make me stay with CentOS/RHEL? Frankly, right now, I don't know, but it will certainly give me time to find out.

[penguinpages]

Will this forum.. which I have been a member of for some time pick up a dedicated section for "streams"?

I realize it is more of a fluid platform... which makes helping other even more nessisary... but... also means harder load for those who are regular posters. I went to the CentOS streams site to find out if their was a "Streams Forum" and I did not see one .. just still the base "CentOS" one. Such as this group.

[DrDave]

Very unhappy. Will probably switch to Debian. One of the nice things is that CentOS is that it is stable, but that is going to go out the window.

I'm getting a trial, (and are likely to buy a copy), of a program called HFSS. That is pretty fussy about what it supports, but CentOS 7.4 to 7.7 are supported.

https://www.ansys.com/-/media/ansys/cor ... 020-r2.pdf

I know those are old versions, and I think Ansys could do better, given this is a costly (staring at £21,000 (GBP) / year) product of theirs. But I can't see them wanting to support CentOS, which is effectively going to be an alpha release of Redhat. That said, none of the Linux distributions they support apart from CentOS are free.

Anyway, I will move from CentOS and avoid Redhat after this. If I buy HFSS, I will probably have to buy SUSE Linux, but I sure wont buy Redhat. For other applications, like a web server, I will use Debian.

Dave

[penguinpages]

I made a post earlier (viewtopic.php?f=54&t=76600&start=90#p322560)with some "guesses" as to what could be going on in the mind of the RH support organization.

That a move to streams is a CICD shift for OS, not some kind of sinister push to get more licenses. I say that after 20+ years in IT and seeing the writing on the wall of modernization of applications going down this path. Like VMWare did with Virtualization of x86 back in 2000 ... Infrastructure as code and the entire DevOps methodologies have already begun to work though the datacenters.

What I was tossing out as a "guess" ... maybe has a bit more legs to it:

https://blog.centos.org/2020/12/centos- ... -delivery/

I don't know who all is posting this.. But I would offer it as food for thought. What OS will be left that does not implement CICD and is still profitable and viable in today's technology.. I use to hold out that things like storage, network, and OS would be foundations not touched by such methodologies.. but no more. Infrastructure as code is taking over datacenters... and the OS has to map to this.

Thoughts? Maybe someone within the know can weight in on this.

[ChristianVirtual]

Very unhappy with the kill-switch on CentOS as stable cornerstone of my few server.
Lucky most of them are on 7 and only one is on 8. The one on 8 can be replaced rather easy and will likely get Debian or Ubuntu as next OS.

Hopefully RockyOS get enough tracking as I don’t have a good feeling moving to Oracle. They are in my eyes the next risky candidate to do a similar move.

Wish IBM/RH would change course and leave CentOS as stable release; and going with Stream if they want. Fine for me. But yeah; seems not to be the case so back to Debian/Ubuntu. Or maybe a BSD variant for some of the non-graphical server stuff. Not yet sure.

[Kingston]

Cloud Linux is planning a version to be RH compatible in the first quarter of 2021, and one of the original CentOS developers is going to start creating Rocky Linux. However, neither of those two have anything available yet.

Yet another rebuild is Springdale (formerly PUIAS). This project actually started before CentOS.

I'm a work from home desktop user with multiple systems and thoroughly recommend Springdale partly because there is still a KDE4 option via the V7 netinstall. Plasma 5 might be an 'upgrade' in some respects but I miss decent weather apps for one example and nothing beats YAWP. I also like the flexibility to mix theme components so my desktop clock can be a different choice to the main theme. I currently have Sprindale running on everything including and ancient Samsung RV415 where it's the fastest performer. My main machines are Dell and LG and I've never had a single issue with Springdale on any of them. You won't find all the extras you might want via a simple 'yum install' but these and any missing dependencies can mostly be downloaded as Centos7 packages and installed that way. I keep mine in a folder and it has 17 items so no big deal to run through these for new installations.