You should have a well-defined backup plan that takes into account whether or not your data is replicated among a set of servers. A plan is the best insurance against significant downtime or data loss in the event of an unrecoverable database issue.

Note:

The server root directory should never be restored from a file system backup or snapshot. External backup methods, such as virtual machine (VM) snapshots, are not recommended, especially if data was corrupted during a transaction.

Keep in mind the following general points about database recovery:

Regular backup from local replicated PingDirectory server
Take a backup from a local replicated PingDirectory server and restore to the failed server. This is more recent than any other backup you have.
Restore the most recent backup
Restore the most recent backup from a local server. In some cases, this might be preferred over taking a new backup if that would adversely impact performance of the server being backed up although it takes longer for replication to play back changes.
Contact support
If all else fails, contact your authorized support provider and they can work with you, and possibly in cooperation with the Oracle Berkeley DB JE engineers, to try a low-level recovery, including tools that attempt to salvage whatever data they can obtain from the database.
Using VM snapshots
Although VM snapshots are not recommended for disaster recovery, remember the following in case you must use them:
  • Snapshotting the VM should never result in a corrupted PingDirectory server.
  • Shutting down the DB before snapshotting is only recommended for speeding up recovery time.