Hacking Grub To Control Number Of Retained Kernels.

Home » CentOS » Hacking Grub To Control Number Of Retained Kernels.
CentOS 4 Comments

I’ve recently had this problem on two C7 systems, wherein when doing “yum update”, I get a warning about /boot being low on space.

both systems were installed using the partition size recommended by Anaconda, right now “df -h” shows /boot as 494M, with 79M free.

I don’t store unrelated crap on /boot, I assume that yum and/or grub will manage it for me. So, why, after over a year, is it running low on space on two different systems?

Is there some location in /boot where junk piles up, but shouldn’t, that I have to know about so I can clean it out?

I see EIGHT initramfs files in /boot, two per kernel, same name but one has a kdump just before the .img suffix. do I need those for old kernels that I may or may not ever boot? (they’re 30 to 50 MB each).

For the moment I’ve edited /etc/grub.conf and changed installonly_limit from 4 to 3. (related question: do I need to manually remove the oldest kernel, having done this, or will yum/grub clean it up the next time there’s a kernel to install?)

thanks!

Fred

4 thoughts on - Hacking Grub To Control Number Of Retained Kernels.

  • I may be off-base here, but isn’t that more a yum configuation issue?
    What is the installonly_limit in /etc/yum.conf?

    Jon

  • Oh, senior moment… that actually is the file I changed. :(
    (getting old ain’t that much fun sometimes…)

  • I recommend the package-cleanup command from the yum-utils package. See the ‘–oldkernels’ option in particular; it will remove obsolete kernels, i.e., those which are no longer installable from your yum repositories.

  • I think jump using /boot is a bad idea. I wonder if that’s really necessary? Anyway, long term solution from the anaconda list is increasing
    /boot size to 1GiB.

    Chris Murphy