No healthy node available in the cluster

Docker Engine Swarm
1

Hi,

I tried to create a simple Swarm cluster without using Docker Machine. So I launched docker daemon with:

docker daemon -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock

I got the Swarm token

docker run --rm swarm create
c9901d16a699397afa1f64122d9fba33

and let Docker engine join to itself.

docker run -d swarm join --addr=x.x.x.108:2375 token://c9901d16a699397afa1f64122d9fba33

Swarn list looks fine:

docker run --rm swarm list token://c9901d16a699397afa1f64122d9fba33
x.x.x.108:2375

Then I created manager on the same host and using 8888 port:

docker run -d -p 8888:2375 swarm manage token://c9901d16a699397afa1f64122d9fba33

I ran “hello world” container to test my Swarm cluster, but it showed an error:

docker -H tcp://0.0.0.0:8888 run hello-world
Error response from daemon: No healthy node available in the cluster

I used the docker info to check,and I got noting in my Swarm cluster:

docker -H tcp://172.20.2.108:8888 info
Containers: 0
Images: 0
Role: primary
Strategy: spread
Filters: health, port, dependency, affinity, constraint
Nodes: 0
CPUs: 0
Total Memory: 0 B
Name: 6300a3102044

I created a Swarm cluster with Docker Machine before, and it works fine.
Do I miss anything ?

3

Finally, I figure out what’s wrong in my lab.

Docker daemon should run with -H on each node of cluster.

docker daemon -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock

And make sure the firewall or iptables on each node allows 2375 port for Swarm manager accessing.

Then rebuild the Swarm cluster, it should be work now.

4

why we should have -H and use tcp 237?