I've created 3 different configurations to cover the different possible environments

1. Local (disable multicast, enable tcp-ip and overwrite interfaces to use local 127.0.0.1)
2. Development (disable multicast, enable tcp-ip and overwrite interfaces with the nodes)
3. Test (enable multicast, overwrite interfaces to fix problems when multiple network interfaces on the server)

1. Local: mvn clean install -Plocal

2. Dev: mvn clean install -Pdev

3. Test: mvn clean install -Ptest

1. Copy distribution/pong-1.0-SNAPSHOT-fat.jar in one node
2. Copy distribution/ping-1.0-SNAPSHOT-fat.jar in other node
3. Copy distribution/bare-instance-1.0-SNAPSHOT-fat.jar in a different node

• Pong:

java -jar pong-1.0-SNAPSHOT-fat.jar -ha -cluster-host XXX.XXX.XXX.XX

for example

java -jar pong-1.0-SNAPSHOT-fat.jar -ha -cluster-host 127.0.0.1

or

java -jar pong-1.0-SNAPSHOT-fat.jar -ha -cluster-host 192.168.112.10

• Ping:

java -jar ping-1.0-SNAPSHOT-fat.jar -ha -cluster-host XXX.XXX.XXX.XX

for example

java -jar ping-1.0-SNAPSHOT-fat.jar -ha -cluster-host 127.0.0.1

or

java -jar ping-1.0-SNAPSHOT-fat.jar -ha -cluster-host 192.168.112.9

• Bare instance

java -cp 'distribution/*' com.jspcore.BareInstance bare

1. Scenario 1.

• Do a kill -9 of some process ping or pong (kill -9 PID), then the bare instance deploy the verticle killed

2. Scenario 2.

• Do a kill -9 of both process (kill -9 PID1 PID2), then the bare instance deploy sometimes one of them or none.

3. Scenario 3 (Deploying ping and pong in same machine and bare instance in a different machine)

• Do a kill -9 of some process ping or pong (kill -9 PID), then the bare instance deploy the verticle killed in the machine where is running

4. Scenario 4 (Deploying ping and pong in same machine and bare instance in a different machine)

• Shutdown the machine where ping/pong is running to simulate a crash, then the bare instance after 5 minutes deploy sometimes one of them or none.