Docker Community Forums

Share and learn in the Docker community.

Docker Swarm taking 20+ minutes to resolve service name

I have a set of services that I am deploying to Docker swarm on a single node via a docker-compose file. There is a particular service that the others cannot reach for 20+ minutes.

  • OS Windows Server 2016 Version 1607 (OS Build 14393.2828)
  • Docker version 18.09.3, build 142dfcedca
  • Windows containers

Reproduction steps

  1. 1 Init the swarm via docker swarm init
  2. 2 Deploy services via docker stack deploy -c docker-compose.yml stack1
  3. 3 Observe service status via docker service ls - service is running (Replicated 1/1)
  4. 4 Observe service is running via docker ps & docker logs <myServiceContainerId>
  5. 5 Observe the container on the overlay network via docker network inspect <networkName> under Containers
  6. 6 Choose another container and run docker exec -i <containerId> cmd
  7. 7 From within this container run ping tasks.myservice
  8. 8 Ping consistently fails until one of the following is done:
    1. 1 Observe myservice 's IP from the failed ping and run ping <myServiceIP> . This succeeds and I can then run ping tasks.myservice successfully going forward.
    1. 2 Wait 20+ minutes for the issue to resolve itself and re-run the ping tasks.myservice from within another container. It succeeds.

I also discovered that with the above setup ping tasks.stackname_myservice returns successfully. If I change the docker-compose to point to tasks.stackname_myservice instead of tasks.myservice I can ping task.myservice successfully but not tasks.stackname_myservice .

Is there something I am missing that is causing this odd behavior?

Example of the docker-compose:

services:
  myService:
      image: containerRegistry/image:tag

  otherService:
      image: containerRegistry/image:tag
      environment:
        - MyServiceAddress=tasks.myservice