On the whole CentOS thing

On the whole CentOS thing

Earlier this year, I faced the problem of a machine that was growing older and was approaching the end-of-life for the OS that was installed on it. My old server for this stuff, magpie.rachelbythebay.com, was running RHEL 6. It was 6 because at the time (2015), I was working at a place that had *ahem* hundreds of thousands of RHEL 6 machines, and it was familiar.

It didn't hurt that my old machine (oasis) was running RHEL 5, and the one before that (beach) was RHEL 3. I picked /that/ up back in the days of doing web hosting support work, in which most of our world was RHEL boxes, and the occasional OG pre-RHEL Red Hat 7.x boxes which were still hanging around.

As a result, this spring when I faced the EOL for RHEL 6 on magpie, it wasn't much of a stretch to go to another RHEL type system. I wanted the most amount of life from it, and RHEL 7 was set to retire at the end of June 2024. That pushed me onto version 8, which had a final drop-dead date of May 2029, but it'd be prudent to migrate before that.

When it was time to provision the new machine, I found out that SoftL^H^H^H^H^HIBM didn't do RHEL 8 installs through their kickstart system. This was certainly unexpected, given that they *bought* Red Hat before that point, and RHEL 8 had been out for something like nine months by then (February 2020). They also didn't offer CentOS 8 for some reason.

Well, I guess we know what's up with that now. The latest news this week is that CentOS 8 is having the rug pulled out from under it. Instead of having a final EOL of 2029, it now hits the wall in *six months* and will be completely dead in twelve. By the end of 2021, it will be all over.

What's coming is something using the same name to push a different kind of product - one I'm not looking to run. I want boring. I *like* boring. I want a machine that just sits there and occasionally drops in a new patch and never picks up any new features. This is a good thing to me.

Why does this matter? Ah, yes, when faced with the lack of a kickstart for either RHEL 8 or CentOS 8, I specified "no OS at all" and received a machine with blank disks. I did all of the rest of it by hand over the remote access stuff, which is surprisingly capable these days (it only took 20 years, but I digress).

Given that I had to do everything manually up to this point, I figured I wasn't going to pay someone for the privilege of doing it myself. That's why I picked CentOS instead of figuring out how to do RHEL as an individual apart from the usual hosting provider RHN/satellite/whatever stuff they generally have.

Of course, here it is ten months later in the worst year ever, and I get to figure this out *again*. My options seem to be:

* Stay put, do nothing, and I guess I wind up on this "stream" business, and who knows what kind of cruft will come down the pipe and hose things. Yeah, not a fan.

* Attempt to do the "migrate to RHEL from CentOS" thing on a running machine, which will probably break something, and almost certainly will try to demand a completely unnecessary reboot or two. Then I also get to figure out the whole "RHEL as an individual" thing, and deal with even more IBM/RH corporate stupidity... while rewarding them for breaking their promises. Not so much.

* Do something really awful and migrate the existing CentOS *live, production server* to something else. I mean, I've done such stupidity with Raspberry Pis just to prove a point, but a machine that's 1500 miles away and is actually running real stuff? No thanks. Just because I can doesn't mean I should.

* Pop *yet another machine* somewhere, and load some other stable OS on it (Debian?), and deal with the fallout of that, and then migrate to it. I guess I'd need another bird name? IBM is also not going to sell this particular hardware for much longer, at which point the prices go up another $50-60/month for some stupid reason. Not looking forward to that.

* Pop another CentOS machine somewhere, migrate everything to it temporarily, then take old snowgoose here and reinstall it as something else, and migrate everything back and cancel the temporary machine. This keeps me on the same box at the same price point, but sheesh, work.

* Wait a little bit, and see what happens with Rocky Linux. This has promise. It's the least amount of goofiness. Also, my general bad luck in terms of tripping over every broken thing can feed back in terms of bug reports and hopefully higher quality for everyone else.

All I can say to the people who decided this is: pissing off a bunch of sysadmins right before the holidays? Not a good choice. I expect a lot of people to get little RH/IBM voodoo dolls in their stockings this year.