Failure and Recovery Scenarios

Describes several failure scenarios (using the Sample Backup and Recovery Plan setup), how recovery works, and implications for Sybase Unwired Platform operation.

Disk C has an unrecoverable error. The *.db files have been lost.

Recovery:

Install a replacement disk, and use standard file restore procedures to reinstall the SQL Anywhere software, and whatever else is needed. If the restore returns the *.db files, there is no harm, but do not rely on these files to be valid. Instead, copy the last backup version of the *.db files across from Computer B.
If *.log files are healthy and have all the latest transactions recorded, run:
```
dbeng11 <path>default.db -a <restored_path>default.log
```
This command ensures that the restored database applies transactions from the current log to the older *.db file. Otherwise, the database does not start.
Start the SQL Anywhere server, which detects that the *.db files are not up-to-date with the checkpoints in the *.log files on drives D: and E: (which are unaffected by the C: drive failure). The server automatically replays transactions recorded in the transaction log to bring the database back to the state of all committed transactions at the time of the C: drive failure.

Sybase Unwired Platform can then start up and run normally. Sybase Unwired Platform mobile device clients are not affected except by the inability to synchronize between the time of the failure, and the time at which the recovery process has completed.

Disk D: or E: failure. One of the *.log files has been lost.

Recovery: Install a replacement disk and restore from backups.

Once the disk has been restored, copy the *.log files from the drive that did not fail to the one that failed. Restart the failed drive.

Complete failure of Computer A, and disks lost.

Recovery: See Restoration of the Runtime Database for information. This should be a very infrequent event.

In this scenario, the database has lost all transactions since the previous backup to Computer B. Any Sybase Unwired Platform mobile device clients that synchronized between the time of the previous backup and the time of the failure cannot now sync. Clients must delete their remote device client database and start fresh. Any pending operations on these clients are lost. Clients that have not synchronized since the previous backup are unaffected.