Quartz scheduler maintenance and performance overheads

Question

We are currently evaluating quartz-scheduler to use in our project. For our use case, we need only one time trigger to be fired at some point in future, it need not to be a repeatable or cron trigger.

So in my POC, I'm creating a new simple one time trigger when business event occurs. I can see in clustered environment (using JDBC store of quartz), triggers are being balanced/distributed among multiple nodes.

Desired behaviour is observed from POC, but, from performance standpoint, how expensive will it be if we create a new one time trigger each time when we run at scale. From my understanding, one bottleneck would be bloating of database with triggers, possible solution for database cleanup is to add a background task to cleanup old triggers.

I am interested in hearing about experiences and pain points on maintaining scheduler with our design and any inputs for improving design.

Answer 1

You can safely use one-time triggers and they will be automatically removed by Quartz after they have fired. What happens is that Quartz checks all triggers and determines if these triggers are going to fire at some point in the future. If they do not, Quartz simply removes them from the store because it makes no sense to keep them.

A somewhat similar principle applies to jobs. If a job has no associated triggers, Quartz automatically removes it from the store unless the job has the durability flag set to true.

So in your case, you will probably want to register a bunch of durable jobs and then your app will create one-time triggers for these jobs on as needed basis. The jobs will remain in the store and the triggers will be automatically cleaned up when they are done.