Comment 2 for bug 362001

Revision history for this message
Jon Tai (jontai) wrote :

So as it turns out, you can restore an instance's backup to another instance, i.e., you can load a backup of production into the training instance, but journal files appear to have the path of the database file embedded inside them, so you can't apply production's journals into the training instance. (At least, not easily...) This means we can get rid of a lot of fancy logic trying to detect and handle this case. And -j doesn't need an argument anymore.

Also, since old journal files are timestamped, the default behavior does not need to be "remove existing journals". We just need to shut off journaling and move the current journal file out of the way.

Here's the final usage, as of rev 81, when I proposed merging this into mainline:

Usage: ovrestore [-y] [-q] [-j] [-f] INSTANCE_NAME BACKUP_FILE
   or ovrestore -h

  -y do not ask for confirmation
  -q quiet; suppress status messages
  -j after restoring the backup, apply journal entries
  -f force; kill any processes accessing the database before restore
  -h display this help text