Data nodes on windows, Master node in Rancher (docker/k8)

(Dan Petter Jacobsen) #1

Hi

We are setting up a cluster where two data nodes are in windows.
3 master nodes are inside rancher.

Data (two nodes):
discovery.zen.ping.unicast.hosts: ip.address.to.master, ip.to.other.data

Master (3 pods in on workload called master)
discovery.zen.ping.unicast.hosts: master, ip.to.data1, ip.to.data2

Master nodes can contact data nodes but it send internal ip address and port.

Master is available in network on 10.5.1.0:45621 (random port generated by rancher for testing, port 45621 is a proxy for port 9300 on internal rancher ip:port), and internal rancher ip is 10.50.5.57:9300
The master will then publish 10.50.5.57:9300 to data node and the result is a connection exception when master is trying to answer on that address.

is there a solution to this problem? or do i need to have the entire cluster on the same network?

Regards
Dan-Petter Jacobsen

(David Turner) #2

You only need to put the master-eligible nodes in the discovery.zen.ping.unicast.hosts setting on each node. There's no need to mention the data nodes there.

Yes. At least, you need each node to be able to access each other node in the cluster using its transport.publish_host address.

(system) closed #3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.