How to re-generate rescue kernel image for current kernel?

leniviy · 05-20-2016, 06:32 AM

Hi.

Initially after installing Centos 7 1511 I had 2 entries in grub:

Code:

CentOS Linux (3.10.0-327.el7.x86_64) 7 (Core)
CentOS Linux (0-rescue-0175d9b02fad463da9f6044b7ae92091) 7 (Core)

After updating to kernel 3.10.0-327.18.2 additional hostonly entry appeared, but rescue image still boots the previous kernel version 3.10.0-327. Even after uninstalling kernel-3.10.0-327.el7.x86_64 the old rescue image is still there.

Who creates the rescue image? A kernel post-install script or anaconda?

How to fix this? Should I manually run dracut with "--no-hostonly -a 'rescue'" arguments?

Who should create /boot/vmlinuz-0-rescue* ?

syg00 · 05-20-2016, 11:57 PM

A rudimentary search sent me to look at /usr/lib/dracut/dracut.conf.d/02-rescue.conf (on Fedora in my case).

Code:

dracut_rescue_image="yes"

Looks like it should do the job - but doesn't. My rescue entry is from December 2013 - might have been broken for a while.

Edit: I might add, in all the years I've used Fedora (and occasionally RHEL/CentOS) I've never even thought of using this kernel.

leniviy · 05-21-2016, 01:52 AM

I think I found it:

Code:

rm -f /boot/vmlinuz-0-rescue-* /boot/initramfs-0-rescue-*.img
/etc/kernel/postinst.d/51-dracut-rescue-postinst.sh $(uname -r) /boot/vmlinuz-$(uname -r)

This script does not overwrite existing rescue image, this is why it's never updated.

syg00 · 05-21-2016, 02:56 AM

So you only get one per install - maybe ok for Servers, but as in my case, version updates leave it horribly out of date.
I guess they don't want to risk replacing it each kernel update in case you need to use the immediate prior one for a recovery. Like I said, not much chance of that IMHO.

syg00 · 06-18-2016, 03:39 AM

And just to clean this up, yes the next kernel update did re-generate the rescue one it had been deleted.

Milan Kerslager · 05-03-2020, 05:27 AM

Quote:

Originally Posted by syg00

And just to clean this up, yes the next kernel update did re-generate the rescue one it had been deleted.

To regenerate rescue kernel/initramfs immediately, reinstall kernel in CentOS 8 by:

Code:

dnf reinstall kernel-core

In CentOS 7 by:

Code:

yum reinstall kernel

2ndAmendment · 08-28-2020, 12:48 PM

If you use the post install dracut script, you may get that command new-kernel-pkg is not found. It is a deprecated command. The rescue image files will be created, but the loader/entries/*-0-rescue.conf file will not be updated.

Quote:

Originally Posted by leniviy

I think I found it:

Code:

rm -f /boot/vmlinuz-0-rescue-* /boot/initramfs-0-rescue-*.img
/etc/kernel/postinst.d/51-dracut-rescue-postinst.sh $(uname -r) /boot/vmlinuz-$(uname -r)

This script does not overwrite existing rescue image, this is why it's never updated.

If you use:

Code:

rm -f /boot/vmlinuz-0-rescue-* /boot/initramfs-0-rescue-*.img
/usr/lib/kernel/install.d/51-dracut-rescue.install add $(uname -r) "" /lib/modules/$(uname -r)/vmlinuz

you will not get the error and the rescue config file will be properly updated.