Backup / restore of VMware VADP backup sets

DS-Client can use CBT (Changed Block Tracking) for VMware VADP backup sets. If enabled, DS-Client attempts to reconfigure the virtual machine to allow tracking of the blocks of data that have changed. This option can improve the speed of incremental backup activities, but can also slow down the virtual machine since it consumes additional resources.

Subsequent backup activities will track the changes for the virtual machine files and will only retrieve the blocks of data that have changes. These blocks will be processed for master/delta generations and sent and saved to DS-System and/or Local Storage.

If CBT fails to be enabled (even if configured for use in the backup set), DS-Client will continue its incremental backup activities by reading the entire data of the virtual machine disks, processing the data for master/delta generations and sending the backup data to DS-System.

Backup of virtual machines on clustered hosts (VMware cluster) should be done through the VMware vCenter Server. Make sure to enter the IP address of the vCenter Server when creating the backup set. Then you can browse and select the specific virtual machine cluster to backup.

If you back up an individual virtual machine in a VMware cluster by specifying the host IP address when creating the backup set, subsequent backups can fail, depending on the configuration of the cluster.

Normally, a VMware DRS (Distributed Resource Scheduler) cluster’s automation level is configured to be Fully automated for High Availability. This means the individual virtual machines that make up the cluster will be dynamically assigned to the best suited host at startup. This means the virtual machine might be automatically moved to another host when the VMware cluster (or virtual machine) is restarted.

Since you can back up individual virtual machines in a VMware cluster (by specifying the host IP address when creating the backup set), if the virtual machine moves to another host, it will not be found by the backup set and errors will occur. However, if the virtual machine is selected through the vCenter Server, the location of the virtual machine is always known (even if it moves to another host).

If you have to back up a specific virtual machine in a VMware cluster (using a dedicated backup set to a specific host machine’s IP address), you must configure the cluster to not move that virtual machine to different hosts.

• In the Virtual Center, edit the VMware DRS cluster settings: “Enable individual virtual machine automation levels”.

• For the specific virtual machine that you want to back up, change its Automation Level setting to Disabled. This will prevent the vCenter Server from migrating the virtual machine.

NOTE: It is easier to create the backup set by specifying the vCenter Server’s IP address as the source in the New Backup Set Wizard. This allows you to browse and select the specific virtual machine from the cluster and you will not need to change the automation settings for the cluster.

DS-Client automatically locks the target virtual machine during VMware VADP backups and restores. This will prevent any attempts to delete or migrate the target until the backup or restore process has completed.

DS-Client cannot back up virtual machines that have fault tolerance (FT) enabled since it depends on making a snapshot of the virtual machine to back it up and VMware does not support snapshot creation on FT virtual machines.

VMware has documented the full procedure for backup / restore of FT virtual machines using templates in their knowledge base article # 1016619.

• You must select the Attempt incremental restore check box in the Restore Now Wizard when you start a restore session for a VMware VADP backup set. This configures the backup set for Incremental virtual machine restore. Otherwise, Linux DS-Client will perform Full virtual machine restore by default.

In the first restore session after a VMware VADP backup set has been configured for Incremental virtual machine restore, Linux DS-Client will perform a Full virtual machine restore, and backup set information will be saved with the restored virtual machine. Then, when you configure a subsequent restore session for the backup set, the Attempt incremental restore check box is selected in the Restore Now Wizard, and Linux DS-Client will attempt Incremental virtual machine restore for the backup set in that restore session and onward until you clear the check box.

• The virtual machine must already exist in the target restore destination as a result of a previous virtual machine restore performed by DS-Client.

When a virtual machine restored by DS-Client is powered on, VM configuration parameters will likely change. As a result, DS-Client will disallow an Incremental virtual machine restore and force a Full virtual machine restore in the next restore session. For more information, see “VM configuration changes and incremental virtual machine restore:”.

To aid a successful Incremental virtual machine restore in the next session, practice the following recommendations:

• To test a restored virtual machine, clone the virtual machine using VMware tools, and use the cloned virtual machine for testing. Do not test the restored virtual machine directly. This recommendation applies whether the most recent restore of a virtual machine was Incremental or Full.

• Do not power on a restored virtual machine manually or directly in the vCenter Server unless you have decided to use the restored virtual machine as a production machine.

Before initiating a restore for a VMware VADP backup set configured for Incremental virtual machine restore, DS-Client compares the backup generation to be restored and the virtual machine at the destination location. If DS-Client detects any differences in VM configuration parameters in a restore session, DS-Client will not perform an Incremental virtual machine restore but force a Full virtual machine restore instead.

NOTE: In the backup session immediately after the FLR feature is enabled, a full backup of the virtual machines is performed regardless of whether the CBT option is enabled for the backup set. All subsequent backups will be incremental.

The FLR option must be enabled in the New Backup Set Wizard or on the Option tab of the Backup Set Properties dialog box before performing the backup.

NOTE: To use FLR to restore from local storage when using a Linux DS-Client, you must enable both the FLR and CBT options when creating the backup set.

• When using a Linux DS-Client, FLR can only restore to the local machine on which DS-Client is running. FLR cannot restore to a remote location through NAS, UNIX-SSH, or NFS.

• When using a Linux DS-Client, drive letter information will be missing in DS-User. Only Windows DS-Client with NTFS virtual machines will display drive letter information.

• FLR is supported when creating Online backup sets with the Save on Local Storage or Save in instant recovery vault option. However, when performing a File Level Restore (FLR) of an instant recovery backup set, the files are always restored from the DS-System online storage.

FLR is supported on Windows NTFS and Linux Ext4 file systems installed on one of the following Partition IDs for the target virtual machine:

Table 1 Supported partition types
Partition ID	Description
00h	Empty partition.
05h	Extended partition.
07h	Installable file system.
83h	Native Linux file system.
8Eh	Single disk Linux LVM

NOTE: You can check the type of the LVM volume by using the “lvdisplay -m” command in a Linux environment. Also note that FLR does not support any type of mixed mapping (for example, if you have a logical volume using both Linear and Stripe mapping).

• FLR can fail on corrupted partitions (for example, due to manual changes to a partition type name). To view the configuration of disk partitions of the target backup virtual machine, use any application that provides disk configuration information (like the Disk Management tool in Windows or the Disk Utility tool in Red Hat environments) and view the following:

• In the case of an unsupported partition type, FLR shows an empty partition in the DS-User with the name starting with “Unsupported ... partition ...”.

A known issue for VMware VADP backup sets using the CBT feature occurs in a very specific scenario where an incorrectly configured virtual machine is backed up. The following steps describe the situation when this error occurs:

Error opening file disk2000.vmdk for reading QueryChangedDiskAreas: SOAP 1.1 fault: "":ServerFaultCode [no subcode] "Error caused by file /vmfs/volumes/4ac65eb0-8b20da10-c550-0030483062fe/rhel61thin-test/rhel61thin-test.vmdk" Detail: <FileFaultFault xmlns="urn:vim25" xsi:type="FileFault"><file>/vmfs/volumes/4ac65eb0-8b20da10-c550-0030483062fe/rhel61thin-test/rhel61thin-test.vmdk</file></FileFaultFault> (rhel61thin-test) (VMware VADP\10.20.30.111\\DataCenter\Test\rhel61thin-test\disk2000.vmdk).

This is because after migration, even though the virtual machine has a property setting of CTKenabled=true, it has not been properly reconfigured by the vCenter Server. If this error occurs, it means the virtual machine has a configuration problem and probably should not be used.

One possible fix is to make a modification that triggers the vCenter Server to run a re-configuration cycle on the virtual machine. This can be done by performing the following steps:

2. Using the vCenter Server, select the virtual machine and select Edit Settings. The Virtual Machine Properties dialog box appears.

4. Select the virtual machine again, and then click Edit Settings. The Virtual Machine Properties dialog box appears.

The vCenter Server should perform a re-configuration of this virtual machine (this can take a few minutes). After re-configuration is finished, power on the virtual machine. You should now be able to back up the virtual machine with the Use CBT feature.

Whenever you migrate a virtual machine using the vCenter server, always perform the steps in the Fix (above) that will trigger the vCenter server to run a re-configuration cycle on the virtual machine.