I’ve got Nexus 3.x latest running on the default port and working via an http url to the server name.
We’ve set up an F5 load-balancer config to accept/force https via the url someservice.mydomain.com
it talks over http port 80 to the server’s Apache
Apache passes the request to 8081
ProxyPass / http://localhost:8081/
The problem is the web UI shows mixed content errors on .js and most images.
I tried setting Admin > Capabilities > Base Url to the https url (to no effect).
I think we have the header: proxy_set_header X-Forwarded-Proto=http (not https)
Any suggestions how I can fix this?
Yup confirmed that’s the problem.
I still have this issue behind F5 proxy after following all instructions I am aware of… this is a kubernetes cluster stood up by RKE (Rancher). We have two F5 proxies, one appliance and one VM, broken on both for each permutation of the test cases described next.
I have two versions of k8s and it is broken mixed content in both of them (v 1.14 and v 1.15). The ingress is whatever nginx out of the box, two CNIs, weave and canal, broken in both.
The odd thing is, we have an old cluster 1.14, and it is working just fine behind the proxy. I must have missed a step. I don’t know why this one old cluster has the wisdom to send https links (that old cluster also works for a brand new nexus deployment with nothing in addition to serving the production repo).
Since it does not affect the docker pulls and pushes, only the UI, I have gone ahead with migrating to one of the broken clusters, leaving the working one behind.
The UI for now can function with http. Not ideal but we are internal network and I’ve already spent a day on this… will revisit if there are troubleshooting tips here. We don’t have support so that’s on us.
Thank you community.