Information Gathering for Support
10. Information Gathering for Support
You can use the Must Gather Tool in the IBM® TRIRIGA® Administrator Console. For more information, see Gathering troubleshooting data for IBM Support.
Introduction
- A. As part of your overall performance assessment, review each tier of the stack as a potential source of the performance bottleneck: (1) TRIRIGA, (2) Software services, (3) Operating systems, (4) Hardware servers, and (5) Network architecture, from section 1.8 Infrastructure Pyramid in the 1 Introduction best practices.
- B. If possible, identify the segments of the pyramid where you isolated the performance issue.
System Architecture
- A. Do you have a Basic, Typical, or Advanced system configuration? Review the sections 3.5 Basic System Configuration, 3.6 Typical System Configuration, and 3.7 Advanced System Configuration in the 3 System Architecture and Hardware best practices.
- B. Attach a diagram of the system architecture with the specifications, application server types, and versions for each component, which includes the web server, load balancer, application server, process server, database, and so on.
Application Server
- A. Specify all application server and process server JVM heap sizes, initial and max.
- B. Specify any additional JVM arguments added.
- C. Specify the JDBC connection pool values, min and max.
- D. If you are using IBM
DB2, confirm that the
REOPT(ONCE)
bind option and isolation level is set at the application server level from section 5.3.1 IBM DB2 Database Server Tuning in the 5.3 IBM DB2 Database best practices, and section 6.3.3 IBM DB2 JDBC Data Sources in the 6 Application Server Tuning best practices.
Web Server & Load Balancing
- A. If you are using IBM HTTP Server (IHS), confirm that you followed the tuning recommendations from Web Server Tuning.
- B. Confirm if HTTP compression is being used.
- C. If you are using a load balancer, confirm that you reviewed the recommendations from section 6.7 Load Balancing in the 6 Application Server Tuning best practices.
Database Server
- A. If you are using Oracle, confirm the values for
CURSOR_SHARING
,NLS_LENGTH_SEMANTICS
, PGA, and any other settings from section 5.4.1 Oracle Database Server Tuning in the 5.4 Oracle Database best practices. - B. If you are using IBM DB2, confirm the values of all registry, database manager, and configuration settings that were recommended from section 5.3.1 IBM DB2 Database Server Tuning in the 5.3 IBM DB2 Database best practices
- C. Cite what database monitoring was done by using the standard database tools, and what database indexes were added or customized based on the recommendations.
- D. Specify the tablespaces that are used, and whether you added a third tablespace for binary large objects (BLOBs).
- E. Are statistics being updated regularly?
- F. Was any reorganization of tables or indexes performed as recommended from section 5.2.2 Reorganizing Tables and Indexes in the 5 Database Server Tuning and Maintenance best practices?
- G. Is the database using multibyte character sets (MBCS)?
- H. Is there an archiving strategy in place and how frequently is it performed?
- I. If you are using IBM DB2, confirm the use of automatic buffer pool size and auto extends as recommended from section 5.3.1.a DB2 Automatic Buffer Pool Size and Auto Extends in the 5.3 IBM DB2 Database best practices.
Network
- A. Confirm if hardware compression is being used.
- B. List the results of a network Speed Throughput Test on a sample client machine that exhibits slow performance. See the section 2 Network Considerations best practices.
IBM TRIRIGA
- A. Provide a list of values for every property (such as
WF_INSTANCE_SAVE=ERRORS_ONLY
) from section 7.1 System Properties in the 7 IBM TRIRIGA Tuning best practices. - B. Provide a list of every running agent on each TRIRIGA server that is indicated in the system architecture provided.
- C. Confirm the thread count values of all the multi-threaded agents from section 7.1.2 Agents in the 7 IBM TRIRIGA Tuning best practices.
- D. Confirm all Workflow Agent settings and properties (such as
WFAgentMaxThreads
andWF_AGENT_MAX_ACTIVE_PER_USER
) from section 7.1.2 Agents in the 7 IBM TRIRIGA Tuning best practices. - E. Confirm that the Platform Maintenance Scheduler (Cleanup Agent) is running,
note the activities it runs, and note the value of the
CLEAN_HOUR
property. - F. If the performance issue is determined to be at the TRIRIGA tier, performance logging must be run and analyzed. To determine the primary TRIRIGA components that are running slowly and why, see the Performance Analyzer or the presentation on Performance Log Analysis. If the slow-running components are customized in any way (reports, workflows, formulas), these customizations should be identified in the logs and optimization should be evaluated based on the following items.
- G. Confirm that any slow-running reports identified in the performance logging were optimized (in areas such as reverse associations, filters, and nested queries) from section 7.3 Query and Report Performance in the 7 IBM TRIRIGA Tuning best practices. If the report is still slow, include information from the database about the query execution plan for any SQL running during the report execution (for example, an AWR report or execution plan XML file).
- H. Confirm that fields and extended formulas identified in the performance logging were optimized (or are truly necessary for the business process) from section 7.3.3 Fields and Extended Formulas in the 7 IBM TRIRIGA Tuning best practices.
- I. If you are running advanced reporting (BIRT), confirm that BIRT reports are being offloaded to a separate BIRT process server. Confirm that custom BIRT reports use the appropriate filtering to limit the result set.
- J. Confirm that workflows identified in the performance logging were optimized (in areas such as Query Task vs. Retrieve Records Task, trigger an extended formula only when needed) from section 7.4 Workflow Performance in the 7 IBM TRIRIGA Tuning best practices. Also, was the Workflow Analysis Utility run? What were the results?
- K. Detail any integrations that run during slow performance periods. Identify what APIs are used, what they do, and when they run. Confirm that you followed the tuning recommendations from section 7.5 Integration Framework in the 7 IBM TRIRIGA Tuning best practices.