Hi All,
I have a Veritas Cluster consisting of two Sun E450s that use a NetApp Filer as
the central repository of binaries, data, and databases. The system is
configured properly, with only one E450 acting as the NBU Master (with a link
to netapp:/bin/openv/netbackup...) and the other E450 acting as a client using
a local install of NetBackup (/usr/openv/netbackup...). The other E450 runs
Oracle 8i, again with the binaries and database on the NetApp.
Everything has been working great, except that we are not shutting down Oracle,
before backing up the database. I am aware of the Oracle extension for NBU, but
we decided it would be less expensive and simpler to just shut down oracle
before backing it up. Since Oracle is part of VCS, all I need to do is offline
the Oracle group, in the bpstart_notify.oracledb script. The script would also
check to ensure it is offline, then let the backup run. The
bpend_notify.oracledb script would online the Oracle group, and make sure it is
back up.
THE QUESTION
------------
So, has anyone implemented a script that offlines a VCS resource, and checks to
ensure it is offline? Possibly erroring the script out if Oracle does not go
completely offline within five minutes?
Detail Summary:
---------------
'Sone' - Sun E450 running NBU 3.4.1
'Stwo' - Sun E450 running Oracle 8.1.7.x
NetApp - Shared Network Storage for both PR2 and PR3
Everything is configured to work with Veritas Cluster Server 1.4.1
Rough Procedure to Implement:
bpstart_notify.oracledb
hagrp -offline oraclegrp -sys pr3
(* Some way of doing error checking to ensure Oracle is down *)
bpend_notify.oracledb
hagrp -online oraclegrp -sys pr3
(* Some way of doing error checking to ensure Oracle is back up *)
Thanks in advance,
- Miles Weissman
Prepress Systems Administrator
USA TODAY
mweissman locatedAT usatoday.com
|