Red Hat listed several Frequently Asked Questions around the issue, which, strangely, did not include the obvious "Why did you do this?"
They don't even mention jumping ship entirely. I wonder why that might be.
Red Hat has suggested that if customers are worried about the impending end of life for CentOS 7, they might wish to migrate to Red Hat Enterprise Linux (RHEL) via its Insights service. CentOS Linux 7 support ends on June 30, 2024, a decision Red Hat took in late 2020. Now the IBM offshoot is concerned about the fate of …
Following significant customer demand, "Why did you do this?" will be added but you will need to enter credit card details before you can see the answer after agreeing to IBM's terms and conditions.
The answer will be "you should have just paid for the RHEL licenses, it would have been cheaper..."
Centos 7 is the only OS that I can get a particular driver tarball for a tape drive to work. This means EOL or not, Centos 7 will be sticking around in the home lab; albeit in an off-network format.
Someone with the right knowledge probably could get it working on other host OS. It is such a fringe requirement that finding anyone with right hardware and interest in doing it is rather challenging.
Sensible me says I should probably go back to using hard drives plus burned discs for otherwise unrecoverable files.
Not sure if this will work but for all the stuff I had running on CentOS 7 I built a container with the same set of dependencies and that lets me run everything I built on CentOS 7 just fine on the new Rocky 8 hosts we moved to. RedHat lost all our respect with what they did and Rocky is working well for us. If a container doesn't work due to your driver needing a kernel stick CentOS 7 in a VM and work that way. Rocky 8 with KVM using the Cockpit Virtual Machines interface works really nicely too. Pass the hardware through to the VM, mount storage via NFS and you're off to the races.
"Centos 7 is the only OS that I can get a particular driver tarball for a tape drive to work."
Have you tried Slackware?
Yes, I am biased towards Slack, but in all reality I have never failed to get a known workable tarballed driver to work with it.
Give it a shot. At the very least you'll learn something. (So will I, and many other people, if you report back.)
wow, is that how much it costs? That's nuts. When I worked Corporate IT we ran all Microsoft Server and it was hard enough to get the CFO to approve the CAPEX for the hardware and server licence, even though we needed the extra server. There's no way on earth he would have approved GBP25 000 for a support contract, No. Way.
Red Hat is obviously out of touch with Medium Enterprise. We weren't massive but we were spread over 2 continents with 6 factories and supplying Military as well as Health Sector but no way we could afford that kind of money. Which is why CentOS was so popular.
Rocky and Alma will continue to service that market.
At MOST, the CFO might have agreed to spend GBP 1000 a year on support. No more than that.
I mean, when a staff member needed a new pc or laptop I had to do a CAPEX for each and every one that cost more than GBP 500. If you know the market, no pc worth buying costs less than GBP 500. Throw in a monitor or two and a dual-monitor arm and we're well over 500.
The decision taken in 2020 was to, effectively, end support for CentOS 8 (not 7) by migrating it to Streams, thereby making it upstream of RHEL instead of downstream.
Support for CentOS 7 was always going to end next year, at the same time as mainstream support for RHEL7 ceases.
A few hiccups along the way while I worked out a migration process on a spare server.
When we did it for real, the migration took just two hours. That all took place last August. We moved to Rocky 9.2 last week. Zero problems with the applications.
Proper Planning and Preparation Prevent Piss Poor Performance
One obvious reason for going for Alma,Rocky,Oracle Linux is that you have 1000s of cores/sockets in a HPC/cluster environment and outside large corporations they are often run on the smell of the proverbial oily rag.
Those deployments are typically pretty static - distro version (eg centos7) lasts the life of the hardware so major version upgrades are not normally contemplated which is a blessing even a rhel7->8 isn't without dramas (eg several sas hba drivers dropped.)
I am guessing many such facilities are planning hardware refreshes and OS selection for 2024 and many are considering alternatives ranging from an Debian/Ubuntu, OpenSuSE through to the Alma and Rocky offerrings (I don't think many would be considering enter Larry's version of Hell. :) I would also be keeping an eye on Springdale (Princeton.)
At the moment I would go for AL 8 or 9 and have both running quite nicely on various hardware (inluding 9.2 on RPi4) and VMs. Non Linux options like BSDs or Illumos unfortunately, aren't usually on the table although I would like to explore SmartOS as a basis of a HPC/cluster facility.
One thing I would look out for is RH/IBM pulling an Oracle/Java licensing stunt. Namely if your organisation has *any* RHEL licensed installations (developer or production) *and* have *any RHEL derivatives* running especially production workloads you will be liable for licencing fees on each. Audits to follow.
"So what are all those CentOS 7 users to do when end of life arrives?"
They are used to fending for themselves, so perhaps now would be a good time to switch to Slackware. Whatever they do, there will be a learning curve ... might as well switch to the oldest, and many would say the best distribution.
Spending two/three days over that period where other people shovel tonnes of food in their faces rsyncing ~5 TB of data from the in-house ML310 running CentOS 7 to the shed ML310 already running Rocky 8, then re-configuring UEFI/PXEBOOT to the second server, and re-installing the in-house server with Rocky 8. Prep-work is underway to check on config changes for Apache, Sendmail, Sendmail-milter, greylist-milter, spamassassin, spamass-milter, clamav-milter, procmail, Dovecot, chrony, tftpboot, NFS, OpenLDAP, DHCPd, Named... How many daemons is too many daemons? Asking for a friend...
Desktop and laptops are already on Rocky 8, although there are issues, such as the lack of GUI Bluetooth control, three faulty MATE packages in EPEL, and out-dated HP printer drivers - that I worked around by rebuilding the SRPM from Fedora. A Fedora LTS might be nice on the desktop/laptop systems as RHEL is clearly aimed at servers, although systemd, chronyd, and NetworkManager all appear to have been aimed at laptop users! Go figure!! I have looked into possibly using Ubuntu LTS on the desktop/laptops, but when you dig deep, there is always something missing that you want to use, but lack the software skillz to build yourself.
We've already started the migration to AlmaLinux at my place of work. Going to take a while as we've tens of thousands of hosts. Refactoring the automated builds for different host roles is at an advanced stage so expect this to pick up pace over the next several months.