Please enter search query.
Search <product_name> all support & community content...
Article: 100001005
Last Published: 2011-12-01
Ratings: 0 0
Product(s): NetBackup
Problem
How to backup local drives on a Microsoft Cluster Node
Solution
Introduction
This document discusses Best Practices Guidelines and Recommendations how to backup a cluster node in a Microsoft cluster.
Documentation
Documentation suggests that to backup a local node in a cluster, the ALL_LOCAL_DRIVES directive could be added to the Backup Selection list and then add the shared disks to the exclude list for each client. Documentation also states that when you add the shared disks to the Exclude List, it prevents the shared disks from being backed up with each client's local disks.
This is true, and even though the shared disks in this scenario will be excluded from backup, they will still be snapshotted when WOFB (Windows Open File Backup) is in use for the client. This is not desired when backing up a local node as this causes unnecessary overhead and uses more system resources than necessary to protect the local node.
The reason for these drives still being considered for snapshotting is that during backup using ALL_LOCAL_DRIVES the following will happen:
This document discusses Best Practices Guidelines and Recommendations how to backup a cluster node in a Microsoft cluster.
Documentation
Documentation suggests that to backup a local node in a cluster, the ALL_LOCAL_DRIVES directive could be added to the Backup Selection list and then add the shared disks to the exclude list for each client. Documentation also states that when you add the shared disks to the Exclude List, it prevents the shared disks from being backed up with each client's local disks.
This is true, and even though the shared disks in this scenario will be excluded from backup, they will still be snapshotted when WOFB (Windows Open File Backup) is in use for the client. This is not desired when backing up a local node as this causes unnecessary overhead and uses more system resources than necessary to protect the local node.
The reason for these drives still being considered for snapshotting is that during backup using ALL_LOCAL_DRIVES the following will happen:
- For an ALL_LOCAL_DRIVES backup, NetBackup needs to determine what mount points / drives are available on the server;
- Once NetBackup knows which drives are available and if WOFB (Windows Open File Backup) is in use, the bpfis process will make a call to Microsoft VSS (if VSS is configured as the snapshot provider) to snapshot the drives;
- Once the snapshot occurs, the actual backup will be performed and at this point the exclude list will be considered (which also adds unnecessary overhead to the backup)
Documentation also states that for Windows 2000 or later, always include System_State:\ or Shadow Copy Components:\ in the file list to back up the System State for each cluster node. This information could be misunderstood, because the directive ALL_LOCAL_DRIVES in itself will by default backup Shadow Copy Components.
Best practices
To backup / protect a local node in a cluster, the recommendation and best practice is to NOT use ALL_LOCAL_DRIVES for reasons mentioned above. Instead specify the actual local drive letters available to the node and to fully protect the node, also add the Shadow Copy Components :\ directive to the backup selection list.
Either use one policy per node or if adding two (or more) nodes in one policy, make sure to list all drive letters that are available.
For example:
Node 1 has one local drive which is C:\
Node 2 has three local drives which are C:\ D:\ and E:\
In this scenario, add to the Backup Selection list:
C:\
D:\
E:\
Shadow Copy Components:\
This document does not address backing up and protecting an entire virtual server but generally speaking, each virtual server should be added to its own policy and shared drives should be backed up using virtual servers as the client in the policy. So to backup a shared drive which is known as F:\ the virtual server associated with this drive letter should be the client configured for the backup.
Node 1 has one local drive which is C:\
Node 2 has three local drives which are C:\ D:\ and E:\
In this scenario, add to the Backup Selection list:
C:\
D:\
E:\
Shadow Copy Components:\
This document does not address backing up and protecting an entire virtual server but generally speaking, each virtual server should be added to its own policy and shared drives should be backed up using virtual servers as the client in the policy. So to backup a shared drive which is known as F:\ the virtual server associated with this drive letter should be the client configured for the backup.