Production migrated to CentOS 8 Stream
Daniel Nashed – 22 August 2021 20:23:09
As posted earlier I don't see any reason why I don't want to move to CentOS Stream.
It is the distribution becoming the next dot release of Redhat Enterprise Linux (RHEL).
I mostly see it positive! We are getting important updates earlier and the distribution will be as stable it used to be.
Redhat will have a bigger interest in CentOS than they had before if you read their FAQ carefully.
https://www.redhat.com/en/blog/faq-centos-stream-updates
So migrating to CentOS 8 Stream is the logical step for me. But of course I am also running a zoo of strange animals and well knowns. How can you really know other platforms without working with them..
Domino is supported in the same other distributions are supported!
If you read carefully HCL split the support statement into different sections.
A. RHEL & CentOS 7.4+ / RHEL 8.0+
B. SLES 12 SP2+ / SLES 15.0+
C. Everything else that meets the kernel, glibc and libstdc++ requirements listed in the technote receives basic platform support.
Reference: https://support.hcltechsw.com/csm?id=kb_article&sysparm_article=KB0086047#OS
It's not wise to install something really wild, because HCL support doesn't have those environments to test and to help you with OS specific administration questions.
But if problems occur, which are not specific to this distribution, this support statement helps you.
CentOS 8 Stream falls under category C.
And I don't see why this should not be supported in the same way a Ubuntu or Debian is supported -- which are more fare away from a comparable RHEL or SLES release.
So I took some external lab servers first before moving all production servers until all have been updated successfully today.
You basically only have to run those two commands
dnf swap centos-linux-repos centos-stream-repos
dnf distro-sync
The only cases where you should be a bit careful is if you have special kernel dependencies.
CentOS Stream might install a bit newer kernel versions and modules like ZFS and VeeamSnap shot driver in my case, have to checked for every kernel version.
Those servers had ZFS volumes and even the latest ZFS 2.0.x wasn't happy.
But you don't have to run the latest kernel in those cases!
The community is currently looking into how to better cover situations like this and I saw some discussions in the ZFS github project.
Specifying the kernel to load
You can figure our which kernel would be loaded on next start up.
And you can change the kernel to an older kernel with the same tool.
In my case the older kernel was still installed and I found it in the /boot directory.
Again this is a very special case. ZFS is an add-on that most customers and partners don't run.
If you do, I would like to hear from you! I am looking into performance testing ZFS for Domino and found some interesting and promising tweaks.
grubby --grub2 --default-title
grubby --default-kernel
grubby --set-default "/boot/vmlinuz-4.18.0-305.3.1.el8.x86_64"
Enterprise supported Linux is the only way to play completely safe
You can never be sure a next kernel release might cause an issue. But nobody forces you to use the latest kernel.
Of course you also have to look into other software on those machines that might have challenges and support statements.
And it is not always sure that a new enterprise distribution with support does not cause any conflict in future like they did in the past.
To be very clear and this is really something that causes some frustration on my side.
Basing your business on top of a free Linux OS without enterprise support contract has also risks!
So if you want to be on the safe fully supported side, you have to pay money and get a proper supported Linux OS from SUSE or Redhat!
You can't expect a completely free ride with Linux if you want enterprise grade support for your business!
Having said that CentOS Stream will be probably one of the distribution which will have a lot of community support once the dust cleared.
-- Daniel
- Comments [0]