Datacom Server increases the CPU usage to the point where we must restart Datacom Server to prevent the system from slowing down.
How can we reduce this high consumption of CPU?
Component: Datacom Server
Shutdown Datacom Server with EOJ to generate statistics in the joblog.
Go to the DSV00136I-DB Thread Usage Statistics section and look for any performance related concern like:
DSV00136I-Total DB Requests delayed due to no thread available: 31958
The DB Thread Usage is a measure of the DBUSERS= allocation. For this scenario, this Datacom Server instance had DBUSERS=50. Over the life of job, almost 32K times when a DB thread was needed but there was not one available to handle a DB request. When there is no DB thread available, Datacom Server waits for a time period, then tries again. So, some of the CPU burn is likely in the looping to retry.
You have to increase he DB Thread Usage to DBUSERS=100 or 150. Then, monitor it to see if it needs to be increased further.
You can get a DB threads report while the Server region is active. The console command is:
/F CA11SRVW,DB_THREADS
You can also use the SVCOMPR utility with a similar command:
APPLID=id
PORT=1234
HOSTNAME=name of the host
STATUS
DB_THREADS
When the DBUSERS= option is set optimally, you'll see the following messages at the end of the DB_THREADS report:
DSV00136I-DB Threads xx to 100 are unused
DSV00136I-DB Requests delayed due to no thread available: 0
Monitor your Server address spaces and make appropriate adjustments as soon as possible.
Read the SVCOMPR Commands section in the Datacom Tools documentation.