Yesterday, our server came to a screeching hult, none of our users could access the DB, and if they could it was very slow.

I noticed in the Event Viewer on the server the following message:

----------------------------------
Event Type: Error
Event Source: DB2
Event Category: None
Event ID: 4
Date: 11/16/2004
Time: 11:44:32 AM
User: N/A
Computer: MAIN
Description:
An entry has been logged to the DB2 Diagnostic Log
for instance DB2.
Please refer to this instance's DB2 Diagnostic Log for additional information.

2004-11-16-11.44.32.750001 InstanceB2 Node:000
PID:980(db2syscs.exe) TID:2136 Appid:C0A8000C.FF0E.041116124751
access_plan_manager sqlra_cache_mem_please Probe:15 Database:ELDORADO

Package Cache overflow
--------------------------------------

Has anyone seen this error before? I cant seem to find why this happened. I was thinking that we might need to run a performance update on the tables??

Have you recently added new queries to the workload? It may be as simple as increasing the package cache size...

The package cache holds compiled sql statements so they don't have to be re-compiled. If you have a lot of SQL statements but not much space in the package cache, they will be "swapped" out and will need re-compilation.

When there is a package cache overflow, it steals from lock list and dbheap and the memory taken is not returned until reboot. This results in loss of concurrency and performance degradation.

Look up the package cache elements in the System monitor guide for more information on what happens and how to tune it.