Troubleshooting DR-Linux

From SEPsesam
Redirect page
Jump to: navigation, search

Troubleshooting during disaster recovery - Linux

The recovered System does not boot

The System refused to boot because /root/dev/console cannot be found

Certain distributions rely on the existence of the directory /dev/ while booting, or certain static devices which have to exist BEFORE the udev Daemon creates them.

In this case

* Include the /dev/ file system in your backup

If the restore cannot restore /dev/

* Boot from the SEP Sesam LIVE CD
* Mount the ROOT Partition of the restored system
* Manually create the /dev/ directory
* Manually create the /dev/console entry with:
mknod /path/to/target/mount//dev/console c 0 0

No bootable Operating system is found

If the system is not able to find a bootable OS instance after the restore, there may have been problems during the installation of the GRUB boot loader.

The restore protocol includes a statement whether or not the installation of the boot loader was successful:

2009-12-14 14:48:27: sbc-3500: Info:     Reinstall boot manager
[/sesam/bin/sesam//sbc_grub_auto /mnt/disk/ AUTO]

It is, of course, always possible to boot the system again from the Live-CD, mount the target partitions and use grub-install in order to install the boot loader correctly.

Device does not have any corresponding BIOS drive.

During the restore the following error occurs:

/dev/sda1 does not have any corresponding BIOS drive.

Please check the file /boot/grub/ on the target system. If there are entries, referring to the disk through /dev/by-disk/..., as shown in this example:

hd(0) /dev/disk/by-id/ata-SAMSUNG_SP2504C_S09QJ1GLA14263-part1

This entry is most likely the reference to the hard disk partition of the broken system. GRUB will not find the proper device. The problem can be solved by:

  • Booting again from the LIVE-CD
  • Mounting root and boot partitions to /mnt/disk (and /mnt/disk/boot, if necessary)
  • Starting grub-install again with the following options:
grub-install --root-directory=/mnt/disk --recheck hd0


grub-probe: error: Cannot open `/boot/grub/'
/usr/sbin/grub-install: line 374: [: =: unary operator expected
Installation finished. No error reported.
This is the contents of the device map /mnt/disk/boot/grub/
Check if this is correct or not. If any of the lines is incorrect,
fix it and re-run the script `grub-install'.

(hd0)   /dev/hda
(hd1)   /dev/hdb

You can ignore the error line 374: [: =: unary operator expected. More important is the result Installation finished. No error reported.

Case /dev/cciss/c0d0p2 does not have any corresponding BIOS drive

In case you see the message /dev/cciss/c0d0p2 does not have any corresponding BIOS drive in restore log.


fsck.ext3: Filesystem has unsupported feature(s)

During a restore of a system with kernel version 2.4 the system may not boot because the Live-CD creates a file system with features which are not supported by kernel 2.4.

Most likely the file system options resize_inode,dir_index,large_file,ext_attr are causing the problems and thus make the system unbootable.

This can be solved by booting again from the Live-CD image, which includes the tool debugfs.

Show the file system features with debugfs:

root@recover#: debugfs -w /dev/sda2
debugfs 1.41.1 (01-Sep-2008)
debugfs:  features
Filesystem features: has_journal ext_attr resize_inode dir_index filetype needs_recovery sparse_super large_file

/dev/sda2 has to be replaced with the right partition names on your system!

Removing file system features:

root@recover#: debugfs -w /dev/sda2
debugfs:  features -resize_inode -ext_attr -dir_index -large_file -needs_recovery -sparse_super
Filesystem features: has_journal filetype

After removing the options the system should boot correctly.

incorrect inode size (256)

After a successful restore the boot process stops with incorrect inode size (256)

Older kernel versions (2.4) may use a different inode size than the one the file system's created through the Live-CD (which includes kernel 2.6). For example, this happens during the restore of SLES8 based systems which use an inode size of 128k.

This can only be solved by formatting the devices manually from the Live-CD, using the proper mkfs options:

mkfs.ext3 -I 128 /dev/sda1

After this step, mount the partition again to /mnt/disk and repeat the restore operations.

Changing the inode size is only possible by reformatting the devices.

The restored system can't find a root filesystem and fails during resume

If the /etc/fstab file was configured with the root file system as UUID. You will have to specify the root file system device name in conventional device names if you are using a different physical disk. After booting, use YAST to reconfigure your boot loader or edit your /boot/grub/menu.lst manually.


Network problems

The restored system does not find any network-cards

If the restore was done to dissimilar hardware, SLES based distributions may not configure the network devices correctly. Unfortunately SLES based systems save their network configuration by using the systems MAC address.

Most likely the system will not use eht0 as device name, but eth1, as it has another MAC address.

In order to solve this problem, please use Yast and reconfigure your network interfaces.