[Veritas-bu] disaster recovery techniques
2004-06-28 20:31:51
At 10:05 AM +1000 6/24/04, Nick Lefebvre wrote:
>The client?s root file system was intact but in a moment of panic
>the engineer tried to roll back to the previous night?s full backup
>and restore the entire root file system on the client.
>
>The restore was initiated from our master server and directed to the
>original location on the Solaris client, that being everything under
>/ was to be restored to its original question.
>The restore failed when Netbackup tried to restore a library file
>libc_psr.so.1, and from then on Netbackup kept repeating messages
>failed to restore fie.
>
>My feeling is that Netbackup tried to overwrite a system file that
>was being used by the operating system and put the system into a
>spin.
What Bare Metal uses, and you can too, is to boot off the network (or
a CD) and then restore to the disk.
Basically you have to set up a situation where you can restore root
and /usr without restoring on top of the current running OS.
I have jumpstart configured on a multihomed machine. I have all our
servers configured to be able to boot off it. I haven't used it for
netbackup restores (I'd have to install the client on the image but
that should be doable) but I have used it to break out of major
problems with booting (eg losing the root password, bad rcS.d file,
etc).
You could also configure a CD with jumpstart + a flash archive of the
system, or a basic system.
The possibilities are endless.
You could install a stripped down OS on the swap partition and then
restore the other partitions.
|
|
|