How to set hazelcast tcp strategy? - hazelcast

Question

We are running services on a cluster as docker containers.
Since we cannot use eureka or multicast, we are trying to use hazelcast TCP discovery.Currently, configuration is like this (example):
cluster:
enabled: true
hazelcast:
useSiteLocalInterfaces: true
discovery:
tcp:
enabled: true
members:
- 10.10.10.1
- 10.10.10.2
- 10.10.10.3
azure:
enabled: false
multicast:
enabled: false
kubernetesDns:
enabled: false
During service start, we get the following log message:
Members configured for TCP Hazelcast Discovery after removing local addresses: [10.10.10.1, 10.10.10.2, 10.10.10.3]
That means, the service didn't discover its local ip right.
Later in the log, the following message appears: [LOCAL] [hazelcast-test-service:hz-profile] [3.12.2] Picked [172.10.0.1]:5701, using socket ServerSocket[addr=/0.0.0.0,localport=5701], bind any local is true
Obviously, the services determines its local ip to be 172.10.0.1. We have no idea, where this ip comes from. It doesn't exist on the cluster.
Is there a way to give hazelcast a hint how to discover its local ip?

Answer 1

The address 172.10.0.1 must be one of the network interfaces inside your container. You can ssh into your Docker container and check the network interfaces (e.g. with ifconfig).
If you want to use another network interface, you can configure it with the environment variable HZ_NETWORK_PUBLICADDRESS. For example, in your case, one of the members can be started with the following docker command.
docker run -e HZ_NETWORK_PUBLICADDRESS=10.10.10.1:5701 -p 5701:5701 hazelcast/hazelcast
Please read more at Hazelcast Docker Image: Hazelcast Hello World