Can some one direct me to the performance tuning tips/guides for the UCM. We have an instance of IBR and UCM running on the same linux box, the Java heap size for UCM is set to 1GB per Oracle's recommendation.
The release of the documents after checkin is taking quite a bit of time. Some times the documents are getting stuck in the queue and I had to restart the instance to get things going. I want to trace the flow of the document after I checkin and check which step is taking longer time, but have no clue on how to do that. How do I profile my UCM instance and identity the bottle necks?
Any help or pointers are appreciated.
Performance tuning can be done on different aspects :
1. WCC memory assigned :
This cannot be provided as a certain value until full performance analysis is done . 1GB would work for some instances and not for other and it depends on load , usage etc .
2. Indexer :
What search mechanism have you enabled ? Depending on the kind of indexer mechanism set relevant tables and indexes would need to be tuned and optimization done to get best performance .
3. DB Health :
This is critical to be monitored / fine-tuned at regular intervals to make sure that WCC db operations are not creating any performance bottlenecks .
I would suggest to turn on the following trace sections :
requestaudit , systemdatabase*, indexer* along with Full verbose tracing and then seeing which all particular cases / actions are having performance issues .
Based on that each of them could be tackled .
Hope this helps .
What Linux (RedHat or OEL) what kernel version?
We had an issue with performance on checkin and IBR etc that was directly related to a kernel version mismatch.
I will check my notes for the affected version and the resolution version
you may also want to chck your IBR instnace thinkgs like queue size, scheduling employed by IBR can impact the time taken to release the item.
also the UCM Configuration page you would see the queries that are taking longer than a minute to execute.
If you can optimize those query you would see a performance gain.
This is the current configuration on my server. Maximum jobs to Queue -- 1000* No.of seconds between successive transfer attempts -- 10* Compression threshold -- 1024 MB*
Not sure if this is optimum since it depends on the volume of the content items getting checked into the server. Can you clarify on the second config option above? Does that mean if 2 content items are checked in, the second one will be sent to IBR after 10 seconds the first is sent? or is it more like if the transfer of the first file failed ... re-attempt after 10 secs ? If it is the second case, is there another config option that says how many times CS should try to transfer the file?
I did go through the Admin guide for IBR ... but I find all Oracle documents to be very vague.
Thanks so much for your help.
Here are a few articles about performance tuning: