Docker-swarm >> Cannot connect to the docker engin

问题:

docker version 1.9.1
swarm version 1.0.1

why on connecting 3 VMs (bridged net) to swarm. "docker info" shows me all nodes

Status pending.

1 of 3 hosts is manager all output is from this host. I don't know where to look for.

On running swarm --debug manage token://XXXXX

output >>

*INFO[0000] Listening for HTTP addr=127.0.0.1:2375 proto=tcp

DEBU[0000] Failed to validate pending node: Cannot connect to the docker engine endpoint  Addr=10.32.1.38:2375

DEBU[0000] Failed to validate pending node: Cannot connect to the docker engine endpoint  Addr=10.32.1.4:2375

DEBU[0000] Failed to validate pending node: Cannot connect to the docker engine endpoint  Addr=10.32.1.33:2375

Then

root@ubuntu:~# ps -ef | grep swarm

root 2158  1391  0 12:28 pts/2 00:00:00 swarm join token://xxxxxxx --addr 10.32.1.4:2375

root 2407  1213  0 13:57 pts/1 00:00:00 swarm manage token://xxxxxxx -H 0.0.0.0:4243

root 2413  1391  0 13:57 pts/2    00:00:00 grep --color=auto swarm

Then

root@ubuntu:~# swarm list token://xxxxxxxxxxx

10.32.1.4:2375
10.32.1.33:2375
10.32.1.38:2375

Then

root@ubuntu:~# ps -ef | grep docker

root      2330     1  0 12:52 ?        00:00:00 /usr/bin/docker daemon

root      2421  1391  0 14:10 pts/2    00:00:00 grep --color=auto docker

heartbeat sorted - runs in background, checked ports, name resolution, pingable from manager.

回答1:

The docker daemon can listen on three different types of Socket: unix, tcp and fd.

By default, docker daemon just listen on unix socket.

If you need to access the Docker daemon remotely, you need to enable the tcp socket.

When creating docker swarm cluster, the swarm manager need to access the docker daemon of swarm agent nodes remotely.

Therefore, you need to re-configure the docker daemon

vim /etc/default/docker

Add following line:

DOCKER_OPTS="-H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock"

Restart docker daemon

sudo restart docker

By doing this, the docker daemon can be accessed remotely.

References:

1. Docker document: docker daemon
2. Docker document: create a swarm for development

回答2:

I have added DOCKER_OPTS values in /etc/default/docker

DOCKER_OPTS="-H <>host IP<>:2375 -H unix:///var/run/docker.sock"

to be more precise << Host IP >> is same host IP you editing your /etc/default/docker file.

Maybe it will help someone.

回答3:

Configuration methods differ depending on the host OS on which you're running Docker.

See Docker Daemon configuration instructions (scroll down to find your host OS):
Docker Configuration Documentation

I'm running Red Hat Enterprise Linux 7.2 : therefore I followed config instructions for my OS form the link above and did the following:

1. mkdir /etc/systemd/system/docker.service.d
2. Create a docker.conf file
   vi /etc/systemd/system/docker.service.d/docker.conf

3. Add the following to the docker.conf file then save.

   [Service]
   ExecStart=
   ExecStart=/usr/bin/docker daemon -D -H unix:///var/run/docker.sock -H tcp://0.0.0.0:2375
   

4. Flush Changes

   systemctl daemon-reload
   

5. Restart Docker

   systemctl restart docker
   

The key line of configuration (and where I deviated slightly from the documentation) is:

ExecStart=/usr/bin/docker daemon -D -H unix:///var/run/docker.sock -H tcp://0.0.0.0:2375

Let's go through this piece by piece:

• /usr/bin/docker start from the docker binary
• daemon start the Docker Daemon
• -D start in debug mode (not required)
• -H unix:///var/run/docker.sock creates the default Docker Socket
• -H tcp://0.0.0.0:2375 creates a tcp Docker Socket listening on port 2375 on all network interfaces

After making these changes, I restarted my docker swarm containers (and in my case my Consul containers as well) and ran docker -H tcp://<IP_OF_SWARM_MASTER>:<PORT_YOU_TOLD_SWARM_MASTER_TO_LISTEN_ON> info to see if I was no longer getting the Cannot connect to the docker engine endpoint error.

回答4:

Worked for me by adding in swarm workers service config /usr/lib/systemd/system/docker.service where ExecStart is located this line:

ExecStart=/usr/bin/dockerd -H unix:///var/run/docker.sock -H tcp://0.0.0.0:2375

Flush changes and restart Docker:

sudo systemctl daemon-reload
sudo systemctl restart docker.service

But this approach is unprotected, need additional security configs.

回答5:

I tired the solutions mentioned in this thread. Editing /etc/default/docker did not work for me.

Finally what was the following solution.

Edit /lib/systemd/system/docker.service file Find the line which starts with ExecStart and add -H=tcp://0.0.0.0:2375 to make it look like

ExecStart=/usr/bin/docker daemon -H=fd:// -H=tcp://0.0.0.0:2375

Save the Modified File Reload the docker daemon

systemctl daemon-reload

Restart the container

sudo service docker restart

Test if it is working by using following command

curl http://localhost:2375/images/json

If everything is fine below command should return a JSON.

To test remotely, use the PC name or IP address of Docker Host.