I am new to Quartz Scheduler Framework. We are using quartz 2.3.2 for scheduling jobs. I have around million scheduled jobs, which has different frequencies. But mostly I get 6000-7000 triggers every minute. These jobs are very light on computation side, they just publish message over rabbit. I am using postgres database to persist these jobs. I have a cluster of scheduler service instances to handle the scale.

I have all schedules as crontriggers with misfire policy as “MISFIRE_INSTRUCTION_IGNORE_MISFIRE_POLICY”.

Quartz configuration:

 spring:
  application:
    name: my-scheduler-service
  quartz:
    job-store-type: jdbc
    jdbc:
      initialize-schema: never # Let Flyway handle initialization
    properties.org.quartz:
      scheduler:
        batchTriggerAcquisitionMaxCount: 600
        instanceName: JobScheduler
        instanceId: AUTO
      threadPool:
        class: org.quartz.simpl.SimpleThreadPool
        threadCount: 600 # Todo: determine appropriate size here
        threadPriority: 5
      jobStore:
        maxMisfiresToHandleAtATime: 400
        acquireTriggersWithinLock: true
        isClustered: true
        driverDelegateClass: org.quartz.impl.jdbcjobstore.PostgreSQLDelegate
  resources:
    global:
      enabled: true
      url:
      hikari:
        poolName: GlobalDBPool
        minimumIdle: 20
        maximumPoolSize: 50

Problems

1. We have deployed the service on Kubernetes with 4 pods. Unfortunately, we are seeing huge delay in task executing. Resulting in huge number of misfires. We are not able to catchup to this misfires and live jobs are not getting triggered.
2. We also observed that quartz locking was taking time (70-74 sec’s). Mostly this is for “SELECT * FROM QRTZ_LOCKS WHERE SCHED_NAME = ? AND LOCK_NAME = $1 FOR UPDATE” query.

What are the things I can do to improve the quartz performance?