-
Story
-
Resolution: Unresolved
-
Undefined
-
None
-
None
-
None
-
RHOAI, Training
-
False
-
False
-
None
User Story:
As a PSAP engineer, I want to run the scale test with multiple MAX_CONCURRENT_RECONCILES values in the DSPO environment to find out the most efficient setting.
For the time being, this requires stopping the RHOAI operator, otherwise the env value will be automatically reverted.
The first tests with 100 users, with the default value showed that the DSPO is under-utilizing the CPU.
the dspo also has an env var called MAX_CONCURRENT_RECONCILES if you're willing to scale down the ODH/RHOAI operator and update this env var in the DSPO deployment to more workers, we can see if this has impact on performance (assuming the pod isn't resource starved)