Hi all,
I'd be grateful for some help getting my server booting again. It's
running debian jessie, with software RAID and LVM, running xen (or
was).
Yesterday one of the disks failed, so I ordered another, and, when it
arrived today, removed the dead one from the array, put the new disk
in, and.....landed at the grub rescue prompt.
So I booted up into a rescue disk, assembled the RAID, mounted the
root on /mnt, and did a chroot.
Running install-grub on the disk (/dev/sdb) gave me this error:
warning: your core.img is unusually large. It won't fit in the embedding area.
error: embedding is not possible, but this is required for RAID and LVM install.
So I looked at the partition table. I have two partitions, /dev/sdb1
and /dev/sdb2. The latter is nearly 1TB, and is the root. Both are
of type 'linux raid autodetect'. /dev/sdb1 started at 63, so I
deleted this partition, created a new one as type linux raid
autodetect, and it was automatically assigned the start of 2048.
Then I ran grub install again, calling the raid and ext2 and lvm
modules, but now I keep getting the error
grub-install: error: unknown filesystem
I'm not entirely sure what I'm actually doing here....been searching
for help for hours. I found one page that suggested grub2 gets
confused by a single disk in an array.
Here's the current partition table:
Device Boot Start End Sectors Size Id Type
/dev/sdb1 2048 996029 993982 485.4M fd Linux raid autodetect
/dev/sdb2 * 996030 1915060454 1914064425 912.7G fd Linux raid autodetect
/boot is part of the root on /dev/sdb2 rather than a separate lvm one.
I'm not entirely sure how it even managed to boot before, but I only
had the /dev/sda2 and /dev/sdb2 in a raid array, so there must have
been a subtle difference in /dev/sda that had allowed grub to install
to it. Maybe grub was installed before I created the RAID array - I
originally only had one disk in there?
Anyway, any suggestions very welcome. The trouble with a xen server
being that all my eggs are in one proverbial basket.
Jenny