Configuring jobrunner thread pools

In Black Duck, there are two job pools - one that runs scheduled jobs (called the periodic pool) and one that runs jobs that are initiated from some event, including API or user interactions (called the ondemand pool).

Each pool has two settings: max threads, and prefetch.

Max threads is the maximum number of jobs a jobrunner container can run at the same time. Adding together periodic and ondemand max threads should never be larger than 32 as most jobs use the database and there are at most 32 connections. It is very easy to saturate the jobrunner memory, so the default thread counts are set very low.

Prefetch is the number of jobs each jobrunner container with grab in each round trip to the database. Setting this higher is more efficient, but setting it lower will spread the load more evenly across multiple jobrunners (although even load is not a design goal of jobrunner in general).

In Kubernetes, you can override the thread counts settings using the following override file:

jobrunner:
     maxPeriodicThreads: 2
     maxPeriodicPrefetch: 1
     maxOndemandThreads: 4
     maxOndemandPrefetch: 2