Richard C. Dempsey wrote:
> We have a RAID cluster that is shared between two HP servers.
> At any given moment in time, one of the servers owns the cluster
> and has write access; the other has readonly access. Who has
> what access changes occasionally. Each server has some disks
> that are its exclusive property.
>
> Currently, we back up on both servers on the writable mount
> point. Thus, every night, the filespace backup fails on one
> of the servers. How do I get the scheduler not to try unless
> the node has control of the RAID?
The only mechanism I can think of offhand is to write a script that obtains a
list of file systems, checks the status of each file system, and issues 'dsmc
incremental' commands for the ones under its host's control.
|