Quartz.NET 3.x

These are the documents for the 3.x line, which is maintained. Quartz.NET 4.x is the current release; the migration guide says what changed and what to do about it.

Clustering

Clustering works only with the AdoJobStore (JobStoreTX or JobStoreCMT). It provides load balancing and job fail-over (when the JobDetail's "request recovery" flag is true).

To enable clustering:

  1. Set quartz.jobStore.clustered to "true".
  2. Use the same Quartz properties on every instance. The allowed differences are the thread pool size and quartz.scheduler.instanceId.
  3. Give each node a unique instanceId. AUTO does this without separate properties files.

See Clustering in the configuration reference for how load balancing and fail-over work.

Caution

Never cluster separate machines unless their clocks are synchronized by a time-sync service that runs very regularly (clocks within a second of each other). See https://www.nist.gov/pml/time-and-frequency-division/services/internet-time-service-its if you are unfamiliar with how to do this.

Caution

Never start (scheduler.Start()) a non-clustered instance against the same set of database tables that any other started (Start()ed) instance uses. You may get serious data corruption, and will see erratic behavior.

Caution

Make sure your nodes have enough CPU to complete jobs. A node at 100% CPU may be unable to update the job store, and other nodes can then consider its jobs lost and recover them by re-running.