I run following Setup in my Testlabor before migrating into production:
Dm-crypt/luks
Raid1 / btrfs
EFI
dracut / fedora Server
GRUB2
Systemd
I simulated a drive failure (the one without boot partition)
The boot process stops with:
Waiting for Job of disk by uuid - no limit
I suspect it’s waiting for the root fs to come up.
- I added the root flag “degraded” so the btrfs is mountable.
- Removed the second disk from the parameters “rd….”
Still boot stuck.
So I started an emergency shell to alter fstab, the root sun vol was mountable with the degraded option without any problems.
My question:
How can I manually stop dracut / init from trying to mount the second cryptdevice? I am fine going into the shell in the case of boot and rebuild the initramfs without it or just use another disk out of the server?
Also happily take another suggestion for a saner build.
My main goal is to ensure data integrity that’s why I run raid1 on btrfs for the root. All other Volumens have integrity check and regular backup.
Still to think about the best way to ensure I can still Boot when the other disks fails as afaik the boot or at least boot/efi won’t run in this config. The server only supports console via ipmi I cannot mount a rescue image directly so having plugged in a usb Stick which constantly needs update or PXE might be my best bet for now.
I read the redhat documentation man pages and did google research but I couldn’t find a suitable answer to this case.
Thanks in advance
