scheduled-in-cluster

This is a scheduler implementation based on Spring scheduling which executes jobs only on one node in a cluster environment. Each node tries to register a scheduler; the first one wins, the rest ones are disabled. Once the node - which successfully registered the scheduler - is down, the other node replaces it.

Advantages:

• it uses only one table to synchronize nodes,
• works with a standard Spring @Scheduled annotation.

Disadvantages:

• it does not spread jobs execution among all nodes; all jobs are executed on one node.

Demo:

1. Start H2 server by running

java -Dspring.profiles.active=h2Server -jar scheduled-in-cluster-1.0-SNAPSHOT.jar

2. Start node 1 in a cluster by running

java -Dserver.port=8083 -jar scheduled-in-cluster-1.0-SNAPSHOT.jar

3. Start node 2 in the cluster by running

java -Dserver.port=8086 -Dscheduler.name=node2 -jar scheduled-in-cluster-1.0-SNAPSHOT.jar

Expected: Only node 1 executes jobs

4. Stop node 1

Expected: After ~90 seconds node 2 executes jobs