Hello Experts (if you're out there)
I'm running Oracle 11gR2 on Rhel 6.2 with SFHA 6.0.x and have configured the SFDB and can take checkpoints and restore them. But this doens't really mean my database is up and running again. Once you restore from your checkpoint you get a nice little note that say you might want to use the control file located at /var/tmp/XXXXXXX/control01.ctl to recover your database.
And this is where it all goes very pear shaped. I don't think the documents tell us enough about what is really going on when you run the restore. Secondly, combined with a very limited understanding of Oracle recovery concepts its quite difficult to actually recover the database.
What is happening behind the curtains.
Anybody with any information please share it
From a SF point of view, an SFDB checkpoint is just a COW (copy-on-write) snapshot that is contained in the filesystem. From what I remember when I last used (over 5 years ago) there was an automated process where you could restore back to a checkpoint or you could just mount the checkpoint if you wanted to just check something or export a single table. As the snapshot is in the filesystem, then you have to start snapshot database on the same system as live and therefore a different SID is required, which if I remember correctly is created by the Veritas checkpoint commands.
I can perform the roll back (havent tried cloning or mounting). I specfically want to get this working smoothly (more from the Oracle point of view) so that I can demo it to customers.
Its such a cool feature for recoving large databases that would otherwise take hours to restore, but one has to know the tricks around the actual Oracle recovery once you've put the datafiles back as they were.
I'll keep on playing