Node discovery: unreachable bootnode in private distributed network

Question

I am trying to create a private distributed network using Docker containers on Virtual Machines running ubuntu.

When I run:

bootnode --genkey $KEY_FILE --writeaddress
bootnode --nodekey $KEY_FILE --addr :30301 --nat extip:$MY_IP

either on the docker container (exposing ports 8454, 30303, 30301:30301 and 30301:30301/udp) or outside of it, if I run telnet <VM-BOOTNODE-IP> 30301 I get telnet: Unable to connect to remote host: Connection refused as reponse. If I create the bootnode NOT on a docker container sudo netstat -tulpn | grep 30301 responds with

udp6       0      0 :::30301          :::*                      */bootnode   

However, when I create in the container mapping the ports sudo netstat -tulpn | grep 30301 responds with

tcp6       0      0 :::30301          :::*          LISTEN      */docker-proxy 
udp6       0      0 :::30301          :::*                      */docker-proxy

Either way I get the same "Unable to connect to remote host" error with telnet. I believe it is because the bootnode's address is unreachable that the other nodes cannot be discovered.

Just to be clear: I am using the same genesis file with a particular networkId and GEN_NONCE on all VMs. My nodes are using --nodiscover (discovery port = 0) and they are listening on 30303. Using verbose level 9 on the bootnode container shows no ping/pong as suggested here.

Answer 1

Well, apparently the reason why the connection is being refused is that "'Connection refused' is ok for the Go bootnode. [as i]t does not accept TCP connections", as suggested by fjl in the go-ethereum issue 380

Discovery uses UDP as protocol. If you want to test if it's working properly, use netcat -u -z -v <PUB-IP> 30301 instead.