My short story of dealing with failed grub2 install which rendered my server unbootable.
It was all good till recent update of grub2. Apparently it no longer fits into MBR and I missed related warnings. Or may be I had grub-legacy. I do not know honestly, but in result I had un-bootable remote server.
Using rescue mode (which is like Live-CD, but instead of GUI gives you SSH access so you could restore your server) I was able to get into server and try to update grub. Here is what I got for grub-install /dev/sda:
# grub-install /dev/sda /usr/sbin/grub-setup: warn: This msdos-style partition label has no post-MBR gap; embedding won't be possible!. /usr/sbin/grub-setup: warn: Embedding is not possible. GRUB can only be installed in this setup by using blocklists. However, blocklists are UNRELIABLE and their use is discouraged.. /usr/sbin/grub-setup: error: will not proceed with blocklists.
Basically, it is saying that it needs to put its installer somewhere and can’t find place for it. For grub2 there is relatively small part of code which get executed after code in MBR finish. This code run different modules to detect file systems, software raids, etc. It is small, few kilobytes. There are two places where it can be placed on MBR partitioned disk – either right before first partition (called embedding) or placing all code in the regular files system and store specific physical position of these files on the disk in the MBR (called blocklists.) Blocklists approach is not reliable due to the fact that files might move and if physical location in MBR will not update – grub code will never get a chance to start.
After I figured out all of the above (which took a while – I have problem like this for the first time, so if I explained something wrong – please let me know,) I decided to find a place before first partition to use embedding option. I’ve found there is about 700Kb of free space between 1st and 2nd partition:
Number Start End Size Type File system Flags
0.00B 511B 512B Free Space
1 512B 512MB 512MB primary ext2 raid
512MB 513MB 753kB Free Space
2 513MB 2560MB 2047MB primary linux-swap(v1)
3 2560MB 954GB 951GB primary ext4 raid
954GB 1000GB 46.3GB Free SpaceSomehow my server was setup with first partition starting with sector 1 while usually first partition starts with sector 63 specifically for purposes to give enough space for all kind of pre-boot software. So in theory I can move 1st partition closer to 2nd and get these necessary few kilobytes to put grub2 code. Easiest for me was to mount /boot partition (that’s what first partition used for currently,) copy all files in a safe location, delete this partition and recreate it on a new place.
To do precise positioning on the disk, I switched to “sector” units in “parted”:
# parted
(parted) unit s
(parted) print free
Model: ATA WDC WD10EARS-003 (scsi)
Disk /dev/sda: 1953525168s
Sector size (logical/physical): 512B/512B
Partition Table: msdos
Number Start End Size Type File system Flags
0s 0s 1s Free Space
1 1s 1000000s 1000000s primary ext2 raid
1000001s 1001471s 1471s Free Space
2 1001472s 4999167s 3997696s primary linux-swap(v1)
3 4999168s 1863024639s 1858025472s primary ext4 raid
1863024640s 1953525167s 90500528s Free SpaceSo now I remove old partition, create new one, format it with same file system, mount and copy all files over:
(parted) mkpart primary 1471s 1001471s Warning: The resulting partition is not properly aligned for best performance. Ignore/Cancel? Ignore (parted) set 1 raid on (parted) quit # mke2fs /dev/sda1
At this point I have empty ext2 partition at /dev/sda1 which starts at sector 1471. Now copy all files back over to new boot partition and run grub-install to finally get no errors!
# mount /dev/md1 /mnt # mount /dev/md0 /mnt/boot # mount --bind /proc /mnt/proc # mount --bind /dev /mnt/dev # mount --bind /sys /mnt/sys # grub-install --boot-directory=/mnt/boot /dev/sda
Reboot and after that it likely will boot :)