Hello,
I’m trying to use OpenFlow in my lab in docker, i did modify the docker-compose.yml file in order to reflect my local DNS as per the documentation and change the port to access the application on port 8080.
Once the change done, i started the application and tried to access it, but the following error encore about the websocket
Websocket Connection to ‘ws://myhost.local’ failed:
I can’t access the application to the login page, do you have any idea ?
The documentation does not include anything about changing port.
Changing port is not supported. you can use either port 80 for http or 443 for https
hey Allan,
Thanks for your reply, is saw in the documentation that changing port is not supported but what i did is not an ultimate change, i did only from 8080 to 80 … like 8080:80.
Technically the backend port is not changed, does it still matter ? meaning could it be an issue with the websocket error ?
If i have multiple application that listen on 80 or 443, then i need to route to the destination port by leveraging a customer frontend port.
It is going to give you a lot of problems, everything is build around using 443 with fallback to 80.
You can override the uri, the websites uses by setting api_ws_url to use a custom port, but as i just said, it’s not supported, and you will most likely run in to other issues.
Thanks for the reply Allan, so if the api_ws_url will give me problems with custom port and port direct change is not supported… what would be your recommendation or guidance? Ngix or Apache reverse proxy in front ? On the same server I have Minikube + Docker, and Ngix is already deployed.
Thank you
That is what traefik does. That is the reverse proxy that distributes traffic between the pods, based of hostname
i feel i sound arrogant or not helpful, that is not the intent.
I’m happy to help help with almost any setup. But on the free forum I will only support the well tested and well documented setups on github/doc. For special requirements I do offer consulting work on a per/ hourly basis.
So, if you want to test the platform there are multiple variation of docker-compose files that show how you can easily get it up and running. They do however all require i relative clean install, and is not intended to co-exist with other services, If you need to use your docker for other things or have other applications using the ports, you can always spin up an ubuntu in most cloud providers for almost no money and run it there. I have multiple youtube videos showing that (like this one ). That also have the added benefit of making it easy to to add certificates using letsencrypt HTTP01 challenges.
Hey Allan,
No worry, i know you try to help here so don’t worry 
Since i did not want to take any chance by getting conflict port or misconfiguration, i did setup a new vm with Docke and docker compose installed on. I went through the install process as per the link for docker
I did update DNS entriesto reflect my local DNS in the config … and guess what 
i have the error message:
Bad Gateway
While accessing the the OpenFlow url … it is very frustrating since it’s a very minor setup.
All my docker service are up and not error in the log … i’m a bit confused.
there must be an error somewhere if you get bad gateway.
but if you are using the same dns entries, my first suspicion would be toward TTL in the dns, so dns is still pointing to the old ip.
but then i re-read what you wrote. Is this due to english not being your first language or did your realy mean what you wrote “update DNS entriesto reflect my local DNS” are you trying to run with some kind of split dns or updating etc/hosts ?
Hey Allan,
No worry for your English not being your primary, same for me 
My point about my DNS, is that i have update the entries in the docker-compose.yml file in order to resolve my internal AD DNS. I have replace all “localhost.openiap.io” entries with my server fqdn “myserver.local.lab”, it’s an internal purpose usage.
Sounds good 
Can you validdate dns is working.
From both the server and your localmachine run this to validate both primary and the * / wildcard domain names are working and pointing to the ip of the server.
ping myserver.local.lab
ping test.myserver.local.lab
Can you restart the server and the check the logs of treafik and api
docker compose logs traefik
docker compose logs api
Hey Allan,
I finally managed to get it working, nothing was in the log but something was acting up with my AD caching wrong host information … i was keeping giving to me the correct ip and while calling from the application some i was hit a different ip than expected.
Anyhow, your help was really appreciated.
Thank you very much.
I am glad you found the solution.
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.