### BEGIN /etc/grub.d/00_tuned ###

set tuned_initrd=""

set tuned_params="skew_tick=1"

### END /etc/grub.d/00_tuned ###

However, every time I update the F27 kernel, it mangles the params line, to something like:

set tuned_params="skew_tick=1"=1"=1"=1"=1"

The =1" can be repeated many times. (I saw it go up to something like a gigabyte once).
I do not know if this is related to the "broken pipe" error I see during every kernel update, but it does seem to be a problem with tuned and grub configuration updates during postinstall for kernel-core.

Every time this happens, I have to edit /etc/grub2.cfg (which points to '../boot/grub2/grub.cfg') to remove the extra =1" characters before I reboot. If I forget, the configuration file can't be read properly, and grub stops at the. grub> prompt (this is a big pain on AWS EC2, because I have to detach the drives and mount them elsewhere to fix, and then reattach).

I have not seen the same problem on F27 where I run grub2-efi... I've only seen this in AWS EC2, which has grub2 installed in the MBR.

I've tried to track this down to the post-install scripts, and grubby, and such, but I'm not really sure how all those tie together, and there's a lot of pieces. I did try to run shellcheck:

[root@localhost ~]# shellcheck /usr/lib/kernel/install.d/* /etc/kernel/postinst.d/* /etc/grub.d/* /etc/tuned/bootcmdline

It found a lot of things, but nothing stood out to me as obviously broken.

I'm not sure whether to file a bug against tuned, the kernel, or grub2 (for failure to tolerate the config error), or if I should do more troubleshooting first. If more troubleshooting, I'm not sure where to start.

Has anybody else seen this or know what I should try next? It does not appear to be a problem with F26. It's annoying to manually edit the grub configuration file every time I update my kernel... but currently, it's my only workaround.

Thanks,

Christopher