The schedule table of our SM9.21 implementation gets a lot of records under the linker class for setting the interaction status to Open-Linked.
Because of this high load, the linker process starts to lag out and we get a backlog of a few hours worth of unprocessed schedule records.
Our is a horizontally scaled environment and we have a load balancer and two other nodes running. The linker is currently only running on the LB.
Will it help if we start more linker processes - one each on the other two nodes? Or will it cause more trouble?
Is there any other way to speed up the rate at which linker is crunching these records?
Is there any other way for us to speed up the linker processing?
In Horizontal based environment, I hope all SM background process would be running on load balancer port, not on other ports.
Anyhow you can add up new linker process and check whether that would speed up the process.
First of all, I would suggest you to find why linker process is taking lot of time to process records.
Good starting point could be the tracing the linker process.
The following command is for the linker background process:
sm -bg scheduler linker 60 1 -RTM:3 -debugdbquery:999 -log:linker.log
Before applying, please have a look at: