Installation on nginx ingress leads to infinite redirect on dashboard

[tarioch]

Describe the bug

After installing on my cluster, the ingress that was generated for the dashboard looked like this

spec:
  rules:
  - host: che-eclipse-che.che.mydomain.com
    http:
      paths:
      - backend:
          service:
            name: che-dashboard
            port:
              number: 8080
        path: /dashboard/*
        pathType: ImplementationSpecific
  tls:
  - hosts:
    - che-eclipse-che.che.mydomain.com
    secretName: che-tls

When trying to access the dashboard, there was an infinite redirect to the dashboard.

After dropping the * at the end of /dashboard/, everything works fine.

Che version

7.32.0-SNAPSHOT
installed with chectl 7.31.0

Runtime

• k3s Server Version: version.Info{Major:"1", Minor:"21", GitVersion:"v1.21.1+k3s1", GitCommit:"75dba57f9b1de3ec0403b148c52c348e1dee2a5e", GitTreeState:"clean", BuildDate:"2021-05-21T16:12:29Z", GoVersion:"go1.16.4", Compiler:"gc", Platform:"linux/amd64"}

Installation method

• chectl

chectl server:deploy --installer=helm --platform=k8s --domain=che.mydomain.com --telemetry=off --multiuser
chectl/7.31.0 linux-x64 node-v12.22.1

Environment

• my computer
  • Linux

[l0rd]

cc @tolusha @sleshchenko

[sleshchenko]

Thanks, @tolusha and @skabashnyuk for investigating the problem. Investigation result:
It stops working after #19542 (7.31.0) but actually these changes are not cause of failure, while the original failure cause is #18720.

Which basically mean - path are ingress controller implementation specific, for nginx it's / or /.* (if we enable regexp with the corresponding annotation). For GCP ingress controller it's /*.
The solution here could be:
In addition to the ability to configure ingress annotations, we also should ask users to configure value for * for a specific ingress controller. The default value should be nginx specific one.