I’m trying ot upgrade from Mender 2.3.0 to 2.4.0. All works well, except that the deployments service stays unreachable. It doesn’t have an ip on the internal network, and the API gateway cannot reach it.
Is this a known thing? What did I goof up?
I already found out it has to do with my trying to put the storage stuff from 9000 onto a different domain and port 443.
Is that something that’s just impossible to do?
I forgot to mention: I have mender running on a ubuntu server, with nginx in front to ensure proper certificates.
When I tell storage-proxy to use port 443, ./run up -d shows that 443 is already in use (which makes sense, as that’s where nginx is running), putting it in another port, means the deployments service cannot access it (if i tell deployments to just access https://my-domain/)
My goal is to have client devices only access stuff on port 443 on my server.
I’m using mender. and mender-storage..
I adjusted the DEPLOYMENTS_AWS_URI to ready https://mender-storage.:443
I’m using nginx as a reverse proxy. When I set the storage-proxy to use “443:443” as ports, that conflicts with nginx running on port 443. Leaving storage-proxy at 9000:9000 and DEPLOYMENTS_AWS_URI at 443, means the deployments container (for example) will try to access the storage-proxy on port 443, using the internal (docker) IP.
Somehow I feel I need to change the port the storage-proxy listens to in the docker container.
sorry for late come back to this.
have you solved the issue?
if you run mender and storage proxy on the same machine they have to use different ports or different IPs, there is no other way.
I’m pointing the DEPLOYMENTS_AWS_URI to the external domain name. Then, that server (which also runs mender) uses nginx to pass different domains to different containers.