I/o timeout trying to join swarm


I’m using docker-machine (OpenStack) to provision a swarm-master (using a token).
I’m then doing the same to provision a swarm node, but I can see in the log that it does an i/o timeout on:
level=error msg="Post https://discovery-stage.hub.docker.com/v1/clusters/my-tocken-goes-here?ttl=60: dial tcp i/o timeout

Because I’m behind a corporate firewall, I use --engine-env to pass http_proxy and https_proxy during the docker-machine create. I can validate this on both master and node1, by issuing a docker run hello-world.

More surprising, I can do a wget on https://discovery-stage.hub.docker.com/v1/clusters/my-tocken-goes-here so it gets there.

So I’m confused, of what the i/o timeout could be.

Any idea on how to troubleshoot this?




This is not a solution for your issue but if you are behind corporate firewall why cant to you other discovery services that are available with docker like etcd, zookeeper or consul.

This will help you to achieve the same. https://docs.docker.com/swarm/discovery/


Hello Ranjan, yes I guess it could be a last option.

But the problem is that I’m using docker-machine to build the master and the nodes, and I don’t know how to do alternate discovery in that context. And I find the token quite nice, just need to understand why the port 443 call fails.