CentOS Reminds Everyone End-Of-Life Is Coming For CentOS Linux 7, CentOS Stream 8

[bash2bash]

No worries, I've already migrated all CentOS servers (dev/stage and production) to AlmaLinux.

Too bad redhat decided to kill CentOS... such a tragedy, but from its ashes we get a whole bunch of EL distros (Alma, Rocky, Oracle).

[You-]

I switched the centOS 7 machines to oracle linux 8.
Not sure at what point Oracle became the good guys, but yeah, that happened.

hint: they didn't.

They are just letting you rest in their jaws until the time comes to crunch.

[Chugworth]

I'm glad I never standardized my systems around CentOS. I've always preferred Debian over Red Hat. For years I used Debian, but within the past several years I gradually switched over to Ubuntu Server since that's the best way to get ZFS. These days my standard work server runs Ubuntu on two mirrored drives, and the additional drives are ZFS. The host OS does very little, and most of the work takes place in either LXC or KVM containers which all live on the ZFS drives. Backups are just a matter of taking a snapshot and using ZFS send/receive.

[mroche]

The point I was making was that for a resource constrained company of our size it's a massive undertaking that ultimately steals a lot of effort from revenue generation. Yes we have been working on this, fyi we are moving everything to debian,but it's no small feat.
Rocky is still unproven, RHEL licenses will liquidate us.

Switching from a RHEL derived platform to a Debian derived platform, in my eyes (former Red Hat solution architect), seems like a much larger undertaking than upgrading + sidegrading the original platform. Particularly if you do more than just "$packagemanager install <foo>" and have built your own deployment pipelines around RPM packages.

I totally understand the sentiment of the newer no-cost distributions being "unproven", but in that respect I would say Oracle Linux (I can't believe I just said that) would fit that bill. I would also recommend AlmaLinux, as it is sponsored and maintained by CloudLinux, and I believe used as the foundation for their CloudLinuxOS platform. CL isn't new to rebuilding, either.

Cheers,
Mike

[Veerappan]

Interesting timing. I spent part of this morning working on replacing our centos 7 vagrant images with rocky 9 as part of a planned move we've got coming in the next few months.

And then I see this article over lunch...

And yes, I know... Vagrant. We're working on it.

[kpedersen]

The point I was making was that for a resource constrained company of our size it's a massive undertaking that ultimately steals a lot of effort from revenue generation. Yes we have been working on this, fyi we are moving everything to debian,but it's no small feat.

If you don't actively benefit from RH support contract, a decent migration path could be to run Rocky, Alma, even old unsupported CentOS in a chroot ontop of Debian.
That way you can slowly phase out the legacy software requiring Red Hat as needed. Looking at the time span, it is fairly clear that your migration is planned to be very slow over a very long timeframe, which isn't a bad thing.

[WiR3D]

For the record upgrading every 3 years is about in line with our current schedules. And allows a nice long period of stability with enough time to test and make changes to support newer OS's and all the runtime versions and quirks they come with.

I will say I do appreciate the recommendations on AlmaLinux, thank you for that.

The big changes are things like orchestration, which needs to have a LOT of changes, and obviously build pipelines which are also tedious, but once it's done then it becomes business as usual.

That said the opportunity cost does hurt, especially in this climate (not that we use that as an excuse to not increase revenue), but you know we are all humans and want bonuses, and you only get that when the bank account looks really good.

[smotad]

ELevate yourselves

[mSparks]

I totally understand the sentiment of the newer no-cost distributions being "unproven", but in that respect I would say Oracle Linux (I can't believe I just said that) would fit that bill

Their servers have always been responsive for updates (perhaps more so than centos ever was tbh), their artwork is clear and unobtrusive. other than its just a reasonably well supported redhat clone, in fact they go one further, because they also have an "Unbreakable Enterprise Kernel" build.

I actually suspect OL may have played a role in IBM ending up canning centos, if behind the scenes they were losing to Oracle anyway.

[slalomsk8er]

For the record upgrading every 3 years is about in line with our current schedules. And allows a nice long period of stability with enough time to test and make changes to support newer OS's and all the runtime versions and quirks they come with.

I will say I do appreciate the recommendations on AlmaLinux, thank you for that.

The big changes are things like orchestration, which needs to have a LOT of changes, and obviously build pipelines which are also tedious, but once it's done then it becomes business as usual.

In my experience the Debian upgrades are way easier then ELevate from CentOS7 to AlmaLinux8 so I don't care if I have to do them 3 times as often and stay more "current" on Debian.

I switched to Ansible with lfops/debops for orchestration and this way we can abstract away the OS differences.