Difference between revisions of "Grub Boot Hole Crash"
Jump to navigation
Jump to search
Michael.mast (talk | contribs) |
Michael.mast (talk | contribs) |
||
| Line 2: | Line 2: | ||
<br> | <br> | ||
<br> | <br> | ||
| − | Before I knew what was going on I got frustrated and deleted the contents of my /boot, figuring I needed to learn how to recover this kind of failure. Then I went through the process. | + | Before I knew what was going on I got frustrated and deleted the contents of my /boot, figuring I needed to learn how to recover this kind of failure. Then I went through the process.<ref>https://www.centlinux.com/2019/12/how-to-recover-grub-2-bootloader-centos-8.html</ref> |
*First I booted off install media and entered into recovery.<ref>https://www.linuxsysadmins.com/grub-rescue-in-centos-and-rhel-7/</ref><ref>https://access.redhat.com/solutions/1521</ref> | *First I booted off install media and entered into recovery.<ref>https://www.linuxsysadmins.com/grub-rescue-in-centos-and-rhel-7/</ref><ref>https://access.redhat.com/solutions/1521</ref> | ||
*Then I bound the install media to sysimage | *Then I bound the install media to sysimage | ||
Latest revision as of 08:53, 3 August 2020
I was hit with the BootHole patch crash[1] on my personal server. Before I really knew what to do about it I had destroyed my boot partition trying to recover. The official steps from Red Hat[2] is to downgrade grub, shim, and moktuil, then exclude them from updates. Though they quickly released a fixed shim file[3] which was really the only problem package. In the following I downgraded the others as well.
Before I knew what was going on I got frustrated and deleted the contents of my /boot, figuring I needed to learn how to recover this kind of failure. Then I went through the process.[4]
- First I booted off install media and entered into recovery.[5][6]
- Then I bound the install media to sysimage
mkdir /mnt/sysimage/install mount --bind /mnt/install/repo/BaseOS /mnt/sysimage/install chroot /mnt/sysimage/install
- I then removed all the old kernels, leaving just the latest one installed (4.18.0-193.14.2.el8_2). This was to simplify the process as I didn't need the old kernels anyway.
- After creating a test.repo config file pointing to the /install directory, I then downgraded the packages and re-installed the bootloader
dnf --repo install downgrade grub2\* shim\* mokutil grub2-install /dev/sda
- Though I was going to rebuild initramfs and vmlinuz, I realized the scripts for doing this run when installing a kernel. Though I tried to reinstall the kernel nothing would happen. Eventually I realized there is a difference between kernel and kernel-core
dnf reinstall kernel-core -y
- Great, now I have the basic folder structure and initial boot files. However, if I try to reboot I get the grub command line. At which point I have to manually boot.
set root=(hd0,gpt2) linuxefi /vmlinuz-4.18.0-193.14.2.el8_2.x86_64 root=/dev/mapper/OS-root initrdefi /initramfs-4.18.0-193.14.2.el8_2.x86_64.img boot
- Silly me, I didn't run the grub2-mkconfig to output to the correct directory on boot. After adding the following I was able to boot again.
grub2-mkconfig -o /boot/grub2/grub.cfg grub2-mkconfig -o /boot/efi/EFI/centos/grub.cgf
- ↑ https://arstechnica.com/gadgets/2020/07/red-hat-and-centos-systems-arent-booting-due-to-boothole-patches/
- ↑ https://access.redhat.com/solutions/5272311
- ↑ https://access.redhat.com/errata/RHBA-2020:3262?sc_cid=701600000006NHXAA2
- ↑ https://www.centlinux.com/2019/12/how-to-recover-grub-2-bootloader-centos-8.html
- ↑ https://www.linuxsysadmins.com/grub-rescue-in-centos-and-rhel-7/
- ↑ https://access.redhat.com/solutions/1521
