Unable to connect to docker-manager after scale-set restart

This is similar to Unable to SSH to worker nodes from master but I didn’t see a fix that worked for me there.

Expected behavior

ssh -A -p 50000 docker@ should connect to my docker-manager node

Actual behavior

I get a connection refused error

Additional Information

This happens typically on my ‘developer’ swarm which I shut down each evening to save on cost. Each scale set is ‘deallocated’. Once restarted I am unable to connect to either manager or swarm nodes. Azure portal lists machines as being available and started etc. but SSH daemon appears not to be running on the machines.

Steps to reproduce the behavior

  1. Create a Resource group using the docker-azure CE edition template.
  2. Connect using SSH to manager node and confirm that connection works
  3. Navigate to docker-node scale set and select “deallocate”
  4. Navigate to docker-manager scale set and select “deallocate”
  5. Wait a few minutes
  6. Return to docker-manager scale set and select “start”
  7. Return to docker-node scale set and select “start”
  8. Attempt to once again connect using SSH to manager node. Connection is refused.

This looks to me like an issue with the either the “agent” container or possibly the docker daemon not starting up on a machine restart, or perhaps some missing metadata or something preventing these containers from starting. Any ideas on how to fix?


was it fixed? i am having the same issue

This appears the still be an issue. Any fix?

The issue is being tracked here: https://github.com/docker/for-azure/issues/48, but no there is no fix yet

1 Like