Future of CentOS for scientific computing

I thought this was worth bringing to wider attention, for anyone who missed this announcement and is currently using CentOS.

Yesterday, CentOS made an announcement, “CentOS Project shifts focus to CentOS Stream”, which will have a significant effect for organisations which have already made commitments to deployment using CentOS 8. There is some additional context in this blog post.

You can find further discussion and information on HN, LWN and on the CentOS blog which provide a range of opinion and additional context surrounding the move.

To summarise: CentOS previously rebuilt RHEL without the branding, and trailed slightly behind RHEL but was otherwise identical. From 2021, it will lead RHEL and will be a rolling release used as a testbed for changes which are intended for RHEL. This is a completely different value proposition, and effectively cuts the support lifetime from 2029 to 2021 when considered for deployment for its original purpose. See the amended EOL dates here.

It may well be the case that a new distribution will be created, or an older one such as Scientific Linux restored, to fill the gap left by this change. In the interim, it’s not entirely clear to me what the best option is yet. For OME Files, we currently have continuous integration for both Ubuntu and CentOS Linux distributions, but it may make sense to drop CentOS if its future is limited, and add a replacement in the future.

Has anyone else had any thoughts about this?

CentOS 7 is still supported until mid 2024 so if you’ve not moved to 8 yet, you can afford to wait and see then prepare for a transition. An obvious alternative is to now pay for RHEL. An other alternative is Debian stable which is widely used in the same contexts as CentOS. This would have the advantage of being in the same distro family as Ubuntu.