BWA Best Practices, How to, Troubleshooting Notes
As the best practice for BWA it is always important to be at the most recent revision of BWA. SAP notes below clearly mention that:
Unlike for ABAP systems, the implementation of local fixes with small “Advanced Correction” is not possible for BWA.
Instead, all changes get delivered in new Revisions. Normally a new Revision gets released when there is at least one major fix that should be made available.
The coding of BWA gets constantly improved:
- Software errors get fixed as soon as they become known.
- Improvements (Performance, Usability) get developed were this seems
- The latest version of the complete code gets constantly checked by many
automatic and manual tests.
Therefore the SAP Support Strategy strongly recommends to keep a BWA installation on a as new as possible Revision. If a software error gets reported to SAP, the members of SAP Support may in some cases ask to first upgrade the BWA installation before a detailed investigation can get done. This is necessary to avoid that an already solved error gets investigated again and again. The SAP Support team is not staffed for this.
Afterwards, if issues persists please check in detail the information as per notes shown below as they are critical for troubleshooting and analyzing general BWA issues. All of them are a combination of very useful notes with detailed screenshots.
– Troubleshooting: BWA How to verify if your BWA is overloaded?
1. Verify how much memory you have available on the BWA
2. Find out how much index data you already created on the BWA
3. Check if you have any backup blades
4. Get the memory of the backup blade
5. How to calculate if your system is overloaded or not
– Troubleshooting: BWA Wrong data returned by a query
1. Verify if the problem is still reproducible and find out which BWA indexes are involved
2. Note down the cubes that were from the BWA-index
3. Find out how to solve the problem
1. Simplify the query-result
2. Perform the keyfigure-definition & provide the TREX-SQL to SAP-support
3. Create the listcube-report
– Troubleshooting: BWA Rollup/Change Run Issues
1. Parallel issues
2. Performance issues
3. Lock issues
– Troubleshooting: BWA Query Performance
1. Check whether the query uses BWA or not
2. Check where most of the runtime is spent
3. Check for any errors using the checkBIA.py python script
4. Check if a reorg is proposed by BWA
5. Check for partitioning of a huge F-Index
6. Review BWA Statistics index
7. Review the Alert server for any errors/warnings
8. Execute a Performance trace
9. Execute Python Trace
– Troubleshooting: BWA Hardware or Network Issues
1. Check the hardware availability.
2. Validate the all services are up and running.
3. Analyze the relevant operating system logs.
4. Check the write performance to the disk.
5. Check the network performance.
6. Check that network connection between the BWA blades and the BW application servers
7. Use the traceroute command to check the network route.
– Troubleshooting: BWA High CPU Usage
1. Verify which process utilizes most CPU.
2. Check the CPU usage from TREXAdmin Tool -> Services -> Load.
3. Get a list of running threads
4. If CPU is exhausted during indexing jobs.
5. If CPU usage is high when queries are executed
– Troubleshooting: BWA Indexing Performance
1. Check the BWA Alert Monitor.
2. Check the BW logs -> SM37 job log, SM21 system log, and the SLG1 application log
3. Check the Indexing settings in the BWA Monitor using RSDDBIAMON2
4. Run a BWA Check (System Check) from within RSDDBIAMON2
5. Maintain indexing statistics data in table RSDDSTATTREX and RSDDSTATTREXSERV
6. Ensure DB statistics for underlying tables and indexes are up to date
7. Check the BWA TREXAdmin tool for the load on BWA during indexing
8. Check the BWA TREXAdmin tool IndexServer trace file for any relevant error information about the process
9. Check the BWA TREXAdmin tool TREXIndexserver.ini file parameter settings
10. Extra information
– Troubleshooting: BWA index server crashes:
1. Verify whether the TREXIndexServer process has crashed and was restarted.
a. Check the BWA alerts in transaction RSDDBIAMON2
b. Check runtime of BWA processes
c. Check the trace files
2. Perform Root cause analysis of the crash.
a. Analyze the BWA system load
b. Analyze existing core files.
3. Crash Analysis using Python trace.
– Troubleshooting: BWA Indexing Errors
1. Check the hardware configuration, including the I/O for each blade and the network settings between BW and BWA
2. Investigate the job Logs
3. Check the application log – transaction RSDDB / alternatively SLG1
4. Check if a service was recently started in Services -> Services
5. Check if there is a high load and if this is balanced on all blades or it only happens on a specific blade
6. Trace files Analysis
7. Check the core files
8. Check the Python Trace
9. Check the var/log/ messages
– Troubleshooting: BWA High Memory issues:
1. Check which service causes the huge memory consumption.
2. Check for long running threads.
3. Check the ratio of disk usage to main memory.
4. Verify the landscape reorganization algorithm.
5. Check for alerts in RSDDBIAMON2/TREXAdmin tool.
6. Check operations in BWA
7. Check the trace file.
8. Check memory relevant parameters of TREXIndexServer.ini
9. Search for known issues for your BWA revision
– Troubleshooting RFC Connection to TREX/BWA Engine:
1. Troubleshooting TREX Engine.
2. Troubleshooting Server Programs.
3. Troubleshooting Gateway.
4. Confirm notes and settings are implemented.
– Best Practices When to rebuild BWA indexes after an upgrade?
1. BW systems upgrade.
2. Database upgrade.
3. BWA revision upgrade.
IMPORTANT: Please note, in order to view the contents of the guides and SAP Notes, you will need to be logged into Service Marketplace.