Synchronization of DBH, SESDCN and UTM restarts

To guarantee an error-free restart of a distributed database system in a UTM environment, all the systems involved are started with restart data (warm start).
A cold start of one of the systems after a system failure is not permitted and usually not possible. It can be triggered by deleting the backup files, for example.

The following table indicates what happens if one of the systems involved is cold started:

Start of	Coldstart	Warm start	Consequence
DBH	x		The UTM application is aborted.
SESDCN	x		The UTM application is aborted.
openUTM		x	The UTM application is aborted.
DBH		x	Restart information for openUTM is lost as a result of the cold start; data consistency is not guaranteed.
SESDCN		x	Restart information for openUTM is lost as a result of the cold start; data consistency is not guaranteed.
openUTM	x		Restart information for openUTM is lost as a result of the cold start; data consistency is not guaranteed.
DBH		x	Transactions in the state PTC are not known to the DBH and can be rolled back by means of administration statements. The session can be continued, but data consistency is not guaranteed.
SESDCN	x		Transactions in the state PTC are not known to the DBH and can be rolled back by means of administration statements. The session can be continued, but data consistency is not guaranteed.
DBH	x		The databases are not consistent and must be repaired.

Table 11: Consequences of a DBH, SESDCN or openUTM restart

Start sequence

The systems involved in a restart forward information on their current processing status to their partners. On the basis of this information the transactions affected by the system failure are rolled back or terminated. To guarantee complete consistency, you must start the systems in the following order:

1. DBHs

2. SESDCNs

3. UTM applications

4. other application programs

If you start a UTM application before the DBH or before SESDCN, it is aborted.