Enqueue and lock table issue analysis can be bit hard form time to time. They don’t regularly occur and when they do, they can have big system performance impact.
This blog will explain:
- How to detect enque issues?
- How to quickly analysze the enque issues?
Detecting enqueue issues?
Enque issues can be easily detected in SM50 and SM66 if workprocess get stuck long time with status ENQ.
First analysis on enqueue issues
The first analysis on enque issues can be done in tcode SM12. From the menu now select the options Extra / Diagnosis and Extra / Diagnosis in Update. This will run the diagnostics on the enqueue handling.
Result looks like:
To get statistics on the enque processing, on the same SM12 start screen select the menu Extra / Statistics.
Deeper analysis on enque issues
For deeper analysis on the lock issues, you might need to switch to the detailed error handling part of SM12. This is a hidden feature. To switch it on you must have the correct authorization (S_ENQUE with ALL in the activities). Switching can be done by keying in the word TEST in the GUI command line (where you key in the tcodes and the /n etc).
Now you will see an extra menu called Error Handling.
From this menu you can directly launch program RSMONENQ_PERF via the menu option Error handling/Diagnosis envorinment. This programs will check the performance of the enque handling:
The Error Handling menu will also give you option to trace the enque processing.
More backgrounds can be found in OSS note 2252679 – How to analyze an enqueue lock problem and OSS note 2126913 – ENQU: The enqueue log (specificly on the logging).
Lock table overflow
Lock table overflow can happen when more locks are set by programs then the available allocated memory for the locks. In a normal system this will hardly occur. But during a conversion that is operating on massive amount of data (sometimes even using parallel jobs) this lock table overflow can happen. If it happens this will effect ALL users. They will get lock table overflow error and cannot save their work. More then enough reason to have large conversion tested first on a test system with production like sizing and settings.
What can be done about lock table overflow?
Provided you have checked your system sizing, you can increase the lock table memory by increasing the parameter enque/table_size.