Troubleshooting
Problem
You are experiencing a performance problem in WebSphere MQ (WMQ) for z/OS. You would like information about common problems and ways to find the root cause of the problem. The IBM Software Support Handbook states that analyzing performance is one of the activities that often require some form of Advance Support Offering. If analysis reveals a suspected defect in the product, refer to the "Collect troubleshooting data..." link below.
Symptom
High CPU
Slow throughput or response
Hangs
Environment
This information applies to IBM MQ V8.0 and WebSphere MQ V7.1, V7.0.1, V7.0 and V6.0 on z/OS. Refer to the IBM MQ Read First page for instructions on other operating systems:
Resolving The Problem
Troubleshooting hints and tips
- Review Webcast replay: WebSphere MQ for z/OS - Performance Topics
- See developerWorks Blog: MQ Key Performance indicators
- Check for CSQP020E messages logged in the MQ MSTR joblog. This would indicate a problem with the buffer pool.
- Set the INDXTYPE for queues if appropriate. See Performance issues getting messages from a queue with a large CURDEPTH.
PI16027 for MQ 7.1.0 and PI19991 for MQ 8.0.0 issue a warning message CSQI004I once per checkpoint when a local queue is accessed by message id or correlation id without an index configured and a large number of messages are skipped. - Avoid large units of work with many uncommitted gets or puts. See Why am I getting high CPU in MQ module CSQLLPLM?
- Check your z/OS workload management importance and velocity goals
- Search the WebSphere MQ support site for known problems using symptoms such as an error or informational message number or error codes or using the term "performance".
- Review RMF reports to look for CPU, I/O, or storage constraints.
- Review SMF data to look for problem areas or narrow down the scope of the problem. PM61284 for V7.0.1. and V7.1.0 provides enhanced log manager statistics.
See Monitoring performance and resource usage. Use a performance monitor or SupportPac MP1B to interpret accounting (SMF 116) and statistics (SMF 115) data. Documentation provided with the SupportPac provides assistance with interpretation.
- If there are delays on the z/OS queue manager, use the REXX utility to do compact LISTCAT for WebSphere MQ datasets and report the size and number of DASD stripes. This information will be useful if I/O to active logs is a factor in the problem.
Learn more about this component
- SupportPac MP16: WebSphere MQ for z/OS - Capacity planning & tuning
- SupportPac MP1B: WebSphere MQ - Interpreting accounting and statistics data
- WebSphere MQ Family - Performance Reports
- Library and Product Documentation
Additional Information
Related Information
Product Synonym
IBMMQ MQ WMQ
Was this topic helpful?
Document Information
Modified date:
15 June 2018
UID
swg21683568