Hangs in a data sharing environment

In a data sharing environment, there can be two types of hangs: single system hangs and group hangs (multi-system hangs).

A single system hang means that the agent that is hanging is contending for resources that are held by other agents on the same Db2 member. A single-system hang can be diagnosed with a console dump of only the one Db2, and its associated IRLM.

A group hang means that the agent or agents that are hanging are contending for resources that are held by other agents on other members. Diagnosing a group hang requires console dumps from multiple Db2 systems, and their associated IRLMs.

If a hang is in a sysplex and there is any doubt of the extent of the hang, take dumps of all instances of Db2 and their IRLMs.

If applications hang in the data sharing environment, there are several places where the problem can occur. The problem location might be:

• The local Db2/IRLM
• A peer member Db2/IRLM
• In the z/OS® XES component within the sysplex

The probable cause for a hang in the data sharing environment is a problem with one of the following functions:

• P-lock negotiation
• Global locking
• IRLM Notify message sending
• Making the transition from group buffer pool simplex mode to duplex mode, or from duplex mode to simplex mode

Regardless of the type of hang or the symptoms of the hang, you can diagnose it by using the procedures in WAIT/LOOP keywords.