无法访问 Docker Swarm 中服务的 HTTP 端点

Question

I have an akka clustered application deployed to 3 nodes via Docker swarm. One of the node exposes an HTTP endpoint. The problem now is that the HTTP endpoint cannot be accessed even though it looks as if the proper port mapping has been done.

Below is the output of running docker service ls

docker service ls
ID                  NAME                    MODE                REPLICAS            IMAGE                                          PORTS
prbk5icazzw8        test_seed      replicated          1/1                 registry.me:5000/seedroute:0.1.0-SNAPSHOT      *:1600->1600/tcp, *:8000->8000/tcp
ldudkew6ym6q        test_worker1   replicated          1/1                 registry.me:5000/workerroute:0.1.0-SNAPSHOT    *:1601->1601/tcp
e2oonp28rrs3        test_worker2   replicated          1/1                 192.168.0.10:5000/workerroute:0.1.0-SNAPSHOT   *:1602->1602/tcp
pe1xzog9qw4y        registry                replicated          1/1                 registry:2                                     *:5000->5000/tcp

But when I login into the server running the test_seed node and do curl localhost:8000/members It seems the connection just hangs and I do not get a response in the return:

curl -v localhost:8000/members
*   Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 8000 (#0)
> GET /members HTTP/1.1
> Host: localhost:8000
> User-Agent: curl/7.58.0
> Accept: */*
> 

using wget is the same, I get the following:

wget localhost:8000/members
--2019-11-30 22:34:18--  http://localhost:8000/members
Resolving localhost (localhost)... ::1, 127.0.0.1, 127.0.0.1
Connecting to localhost (localhost)|::1|:8000... connected.
HTTP request sent, awaiting response... 

Any idea what could be wrong? In summary, I have an HTTP server that is part of application deployed via Docker swarm, and I cannot access this HTTP server, even from the node it is running on.

Any suggestions on how to fix this would be appreciated!

Edit-1

Forcing the request to use 1Pv4 also did not help:

root@akka:~# curl -4 http://localhost:8000/members
curl: (7) Failed to connect to localhost port 8000: Connection refused

Answer 1

As you can see from the following line of your curl/wget output

Resolving localhost (localhost)... ::1, 127.0.0.1, 127.0.0.1

your system is first resolving localhost to ::1 , the IPv6 lookback, which is not supported by Docker at the moment.

Possible solutions are:

1. Use 127.0.0.1 : curl http://127.0.0.1:8000/members
2. Force IPv4 : curl -4 http://localhost:8000/members
3. Disable IPv6 : As root, open /etc/sysctl.conf and add the following:

    net.ipv6.conf.all.disable_ipv6 = 1 net.ipv6.conf.default.disable_ipv6 = 1 

   Then run sysctl -p to apply the new settings.

Answer 2

Bug's answer above helped, but curl was still hanging intermittently. The following option below was least flaky with swarm.

Creating an overlay network docker network create -d overlay my-net and in the docker-compose.yml file marking this network external as follows:

networks:
  my-net:
    external: true

Services including traefik which used this my-net network were now accessible via curl.