Service placement on different node makes it unreachable

[dorinclisu]

I installed the basic traefik stack on 1 node and I was pleased with how it works.
Then I installed a custom application based on this config https://dockerswarm.rocks/thelounge/
All working well, dashboard available at https://traefik.mydomain.com, app available at https://app.mydomain.com

Then I added one more node to the swarm, and the next stack update placed the app container on the worker node. Traefik still runs on master node according to placement constraints, and the dashboard still runs fine and reports no errors.
However, trying to access the app doesn't work and the browser gets a Bad Gateway message (probably from traefik?)

The log line traefik produces for every request to access the app at https://app.mydomain.com

2021-01-25T18:16:07.965245081Z **client-ip** - - [25/Jan/2021:18:16:04 +0000] "GET / HTTP/2.0" 502 11 "-" "-" 419 "app-https@docker" "http://10.0.1.8:8000" 3057ms

Again, I didn't change anything except that I added a worker node and redeployed the stack.

[dorinclisu]

I think the problem was the firewall not allowing the docker mesh network to work properly between the nodes.
https://www.digitalocean.com/community/tutorials/how-to-configure-the-linux-firewall-for-docker-swarm-on-ubuntu-16-04

After opening ports 4789 and 7946 (2377 was already open, otherwise I couldn't join nodes to the swarm), the service works as expected regardless of placement.

[tiangolo]

Thanks for reporting back and closing the issue 👍

Sorry for the long delay! 🙈 I wanted to personally address each issue/PR and they piled up through time, but now I'm checking each one in order.