Restore Data - VMware ESX 2.5.x Server - Full System Restore

Topics | How To | Full System Restore | Related Topics

Overview

Perform a Full System Restore

Overview

You can use the Linux File System iDataAgent as one component for performing a full system restore of the VMware ESX Server 2.5. The full system restore procedure currently provided serves as only one example of how the restore can be completed (other methods may be preferable), and it includes two distinct restores, as follows:

• Base ESX (Linux) File System Restore
• Restore of the guest operating systems or virtual machines.

Perform a Full System Restore

Before You Begin

• Linux and ESX Server administrators should tailor the restore based upon the server configurations.
• This restore assumes a VMware installation with only a new single SCSI disk installed and that no other hardware changes have been made on the system. If other hardware changes are made post-backup and prior to the restore, the restore might not be successful.
• The following is an example of a map of the original VMware ESX Server disk layout. It is assumed that a new SCSI disk has been installed in the machine.
  
  [root vmhba1:0:0:6]# fdisk -l /dev/sda
  
  Disk /dev/sda: 255 heads, 63 sectors, 8908 cylinders
  Units = cylinders of 16065 * 512 bytes
  
  Device Mnt Start End Blocks Id System
  /dev/sda1 / 1 6 48163+ 83 Linux
  /dev/sda2 /boot 7 324 2554335 83 Linux
  /dev/sda3 swap 325 515 1534207+ 82 Linux swap
  /dev/sda4 516 8908 67416772+ 5 Extended
  /dev/sda5 vmcore 517 529 104422+ fc Unknown
  /dev/sda6 vmfs2 530 8908 67304286 fb Unknown
  [root vmhba1:0:0:6]#

  (/dev/sda5 is special partition type “fc” used by VMkernel for core dump. /dev/sda6 is special partition type “fb” used as vmfs2 file system.)
   

To perform a full system restore of the VMware ESX Server:

1. Boot from the ESX Server disk. Then enter the license information and accept the defaults until you come to the disk configuration screen. On the disk configuration screen, use “Custom”. Thereafter, use the Install GUI to do the following:
   • Install / in /dev/sda6 (temporary Linux install).
   • Install /boot in /dev/sda1 (default).
   • Install swap in /dev/sda3 (default).
   • Install vmcore in /dev/sda5 (default).
   • Also, skip /dev/sda2 (do not modify).

   Use the defaults to save some time later.

2. Reboot the system. At the lilo prompt, select linux-up. You now have a functional default ESX Server that will be used to restore the Linux portion of the original ESX Server.
3. Find and kill the klogd process. This is needed to allow /boot to be un-mounted. Then execute the following command to prepare the system to re-install the ESX Server from the backup.

   unmount /boot

   Create an ext3 filesystem on the original / partition (/dev/sda2). You will restore the backup to this partition. Set the label of the partition to /.

   mkfs.ext2 –j –L / /dev/sda2

   Then mount the new root (/) to a temporary mount point.

   mount /dev/sda2 /mnt

   Then create the new /boot directory. This filesystem will be overwritten (unconditionally) from the backup. Mount the existing disk to the new mount point.

   mkdir /mnt/boot
   mount /dev/sda1 /mnt/boot

   Thereafter, create a new /proc mount point. The system will not automatically create this directory.

   mkdir /mnt/proc

4. Reinstall the Linux File System iDataAgent from the installation disc. Use the default values. See Deployment - Unix File System iDataAgents.
5. In the CommCell Console, for the appropriate Linux File System iDataAgent default backup set, browse and select a full (default) restore from the existing backup in the "Latest Data" window. Also, de-select the /vmfs filesystem (this filesystem will be restored later). Then click Recover All Selected. From the Restore Options dialog box, select Unconditional Overwrite, and type /mnt in the Destination Folder field. Thereafter, click OK to start the restore.
6. Once the restore has completed, set the new root filesystem by completing a couple of tasks. First, execute the following command to set the root directory to be the restored root currently under /mnt.

   chroot /mnt

   Then execute the following command to update the boot loader with the new root (/) and /boot information.

   lilo

7. Shut down and reboot. Allow the system to come up normally.
8. Run the Virtual Machine Web interface and then check the VMware Web console. Any virtual machines that existed previously should be listed and shown as powered down in the console.
9. If the interface has a warning concerning the swap for the virtual machines, ignore this warning; the swap will be re-created automatically at the final reboot after all the virtual machines have been restored.
10. On the Options tab in the Virtual Machine Web interface, bring up the Storage Management screen. Then delete the #6 logical volume. This represented the original root (/) filesystem when the server was first installed. Once you delete this volume, create a new vmfs2 volume using the rest of the disk.
11. In the CommCell Console, browse and select only the /vmfs filesystem (since everything else has already been restored) in the "Latest Data" window. Then go to the Restore Options dialog box and select Unconditional Overwrite and Restore to Same Paths. Thereafter, click OK to start the restore.
12. Shut down and reboot the machine.

The system has now been restored and should behave as before the failure. The virtual swap will be re-created, and the virtual machines will start automatically if set to do so.

Back To Top