Forum Discussion

RCBrown_DKH's avatar
13 years ago

7.1 Cluster Configuration

Could someone point me in the direction of a "How-To" configure a cluster backup in Netbackup 7.1?

  • Depends what you are backing up.

    If the clustered system has a shared disk (that mounts to the active node), then you use the cluster name as the client name. That way it will be backed up when either nodename is active.

    For any other filesystems that stay on the node (local), you should use the nodename as the client name.

    There is nothing more complicated to it than that.

9 Replies

  • How to configure a backup for a clustered system or how to configure NetBackup in a clustered system?

  • Depends what you are backing up.

    If the clustered system has a shared disk (that mounts to the active node), then you use the cluster name as the client name. That way it will be backed up when either nodename is active.

    For any other filesystems that stay on the node (local), you should use the nodename as the client name.

    There is nothing more complicated to it than that.

  • Just to clarify my previous post

    -> If the clustered system has a shared disk (that mounts to the active node), then you use the cluster name as the client name. That way it will be backed up when either nodename is active

    It will backup the shared disk when either node is active and "presented" the shared disk.

     

  • If you use the actual name for both nodes of the cluster and all local drives, you will backup anything see on that node at the time of the backup.  The file systems on the active node will only be seen on one or the other, you should not need to use the cluster name unless you are running a special backup (SQL or Oracle for example) that only needs to run on the active node.  If you use all three names in the backup policy, you will backup the active node twice for OS type backups.

  • One policy called CLSTR_system

    physical server names

    backup

    C:\

    system state and or shadow copy components

    One Policy called CLSTRV_<virtual server name>

    backup the drives that are shared resources  like the E:\ or F:\  - this way no matter which physical server E or F is currently assigned to it will find it, and as you are not trying to backup E or F from the sever it is NOT on you will not get failures because it cannot find them.

    If the server has a database you will have to make your normal plans for that.

     

  • If you use all three names in the backup policy, you will backup the active node twice for OS type backups.

    Only if using ALL_LOCAL_DRIVES as the file selection directive.

    If you have a shared drive and use the nodename, then you have to know which node was active when you did the backup and want to restore. Not a major issue, you just select the node on restore and if the restore selection is not found, you select the other node.

     

  • There are several ways to skin this cat.  I'm a fan of using the all_local_drives directive, so that is why I do it the way I do.  I've had too many instances of admins adding new drives or mount points that end up not getting backed up since they never told me about them.  I like that NBU will go figure out what is on the server and back it up.  J.Hinch's method is perfectly acceptable as well.  The size of your environment and the amount of changes you have to deal with may dictate how YOU want to do this.