Docker must be configured not to use the HTTP proxy for all managers when connecting via internal IP addresses. To fix this you must add the advertise addresses of all Swarm managers to each manager’s NO_PROXY variable.
The /swarm-init and /swarm-join scripts scripts can be run with the
additional-no-proxy flag set to a comma-delimited list of all manager IP addresses in the Swarm.
curl -sSL https://get.replicated.com/swarm-init | sudo bash -s \
curl -sSL https://get.replicated.com/swarm-join | sudo bash -s \
Additional managers can be added later by manually editing the Docker NO_PROXY environment variable and restarting the Docker service.
You can change this by editing the file
/etc/systemd/system/docker.service.d/http-proxy.conf. You will then need to reload the systemd service configuration (
systemctl daemon-reload) and restart the Docker service (systemctl restart docker). More information can be found here on configuring the Docker service.