i am testing ignite with a node (node1) on local PC and a node (node2) on a vserver "on the internet" i tried this with ignite 2.16 and ignite 3-beta1. both nodes show only 1 server in the cluster.
the configured ports are accessible. (tested using telnet)
since i am interested in ignite 3 i tried to dig deeper in to the issue. the issue seems to be in the ip addresses.
there is no way to specify the host (netty bootstrap bind is invoked only with port) the bind returns some internal ip which is not accessible from the network. node1 connects to node2 and seems to publish this internal ip. it seems that node2 is trying to connect to this internal ip and fails:
logging on node2: Timeout getting GetMetadataResp from 192.168.99.1:3344
BTW: i also noticed that both nodes are logging: sendWithRetry resp=ActionResponseImpl [result=false] from=192.168.99.1:3344 to=192.168.99.1:3344 err=null
at a high rate (approx 20/sec)
-- Ron
To answer the question directly, there are two possibilities. First, you can specify the localAddress (in your Discovery and Communication SPIs). This can be useful when you have multiple network cards.
Or you can provide a mapping between internal and external addresses using an AddressResolver.
But the real answer is that it's generally not a good idea to spread a cluster across multiple networks.