Timeouts and deadlocks in a data sharing environment

A timeout or deadlock in the data sharing environment is similar to one in the non-data sharing environment, except that the problem might be caused by a lock that is held by a peer group member rather than a local or remote user on the same subsystem.

When a deadlock or timeout occurs, you receive one or more of the messages DSNT318I, DSNT375I, DSNT376I, or DSNT377I. Those messages display the holder of the lock and its Db2 member name and provide detailed diagnostic guidance.

Use the command DISPLAY DATABASE to gather information about the lock status of all members of the data sharing group. Specify the LOCKS keyword for information about transaction locks for table spaces, tables, index spaces, and partitions. Specify the CLAIMERS keyword for information about claims on table spaces, tables, index spaces, and partitions.

Issue the IRLM MODIFY irlmproc,DIAG,HANG command to collect SVC dumps for all the IRLM instances in the sysplex. This command should be used only under the direction of your IBM® service representative. This command prevents the TIMEOUT/LOCK SRB from running until z/OS® dump services reschedules it. After this command runs, you must ensure that IRLM returns to normal processing.

You can activate statistics trace class 3 or performance trace class 6 to get information about timeouts and deadlocks. IFCID 0196 gives timeout statistics, and IFCID 0172 gives deadlock statistics. Start the trace on the member where the application with timeout or deadlock problems runs. When a timeout or deadlock occurs, Db2 gathers information from all other members of the data sharing group.