I have a CentOS virtual machine because I had to install it on a compute server in my office, and I keep it since it's such an interesting antithesis to the rolling release distribution I prefer for my daily computing environment. CentOS is by a large margin the most conservative of all Linux distributions, and it's sometime useful for me to have access to older software in its natural habitat. Just look at this table comparing the versions of some of the major packages on fully updated Arch and CentOS 7 installations:
Current Arch CentOS linux 5.1.15 5.1.15 3.10 libc 2.29 2.29 2.17 gcc 9.1.0 9.1.0 4.8.5 systemd 242 242 219 bash 5.0 5.0 4.2 openssh 8.0p1 8.0p1 7.4p1 python 3.7.3 3.7.3 2.7.5 perl 5.30.2 5.30.2 5.16.3 texlive 2019 2019 2012 vim 8.1 8.1 7.4 xorg 1.20.5 1.20.5 1.20.1 firefox 67.0.1 67.0.1 60.2.2 chromium 75.0.3770.100 75.0.3770.100 73.0.3683.86
You can easily see why I prefer Arch over CentOS as a desktop system.
But CentOS has it's merits, particularly for servers. There's no other distribution (except, of course, its commercial sibling RHEL) with a longer support span: CentOS 7 was released in July 2014 and is supported till end of June 2024. And that's not just a partial support as for the so-called LTS versions of Ubuntu.
Now, I've noticed that CentOS keeps old kernels after updates, befitting its highly conservative attitude. However, in view of the very limited hard disk space I typically give my virtual machines (8 GB), I got a bit nervous when I saw that kernels really seemed to pile up after a few updates. There were five of them! Turned out that's the default, giving the word “careful” an entirely new meaning.
But am I supposed to remove some of these kernels manually? No. I was glad to find that the RHEL developers had already recognized the need for a more robust solution:
And to make this limit permanent, I just had to edit /etc/yum.conf and set
Well thought out. 😉