Performance: Deadlocks (DB2 UDB for UNIX and Windows)

The following section is only valid for RFC-monitored systems.

Deadlocks are situations where two or more applications lock each other, each waiting for the other(s) to release the lock.

Deadlock situations are detected automatically by DB2 and resolved after a specified time. This time period is determined by database configuration parameter DLCHKTIME.

The event monitor records the occurrence of deadlocks. To make sure that no data is lost, two event monitors, DB6_EDLMON1 and DB6_EDLMON2, are used. The system switches them on and off alternately, so that one event monitor is always active. The current status of both event monitors is displayed in the Deadlocks group box.

You can access the Deadlocks screen by calling the DBA Cockpit and choosing Performance ®Deadlocks in the navigation frame of the DBA Cockpit. The system displays information from the event monitor that was active before you called this function.

If there was a deadlock, the system provides, for example, the following information:

·        Number of applications involved

·        Deadlock detection time

·        Application ID

·        Table and tablespaces of lock waited on

·        Schema of lock waited for

·        Mode and type of lock

·        Mode application requested on lock

If there was no new deadlock found, the system displays a message No Deadlock found and lists the:

·        Name of the event monitor used

·        Database name

·        Event monitor start time

To display the latest information on deadlocks, choose  Refresh. During the refresh process the status of both event monitors is switched.

To display a list of previous deadlocks, choose  Display History. The system displays a list of all deadlocks detected within the last 180 days starting with the oldest. Information on deadlocks is deleted after that period of time, but twenty entries are always kept in the system no matter how old the entries are.