fixes loadbalancer port for healthcheck

[stefanhenseler]

It seems like the TLS Change a couple of commits ago broke the minio health check function and with that the status update of the Tenant CRD. This prevented the console from ever being deployed. I've changed the health function to use the LB ports depending on the TLS setting.

[harshavardhana]

Port 9000 should be still accessible since that is still valid. Port 443/80 are just service ports only.

[stefanhenseler]

Port 9000 should be still accessible since that is still valid. Port 443/80 are just service ports only.

That would work, but then we need to change the service name and use the headless one for this.

• minio ClusterIP 10.63.245.183 80/TCP 19s
• minio-hl ClusterIP None 9000/TCP 19s

[harshavardhana]

That would work, but then we need to change the service name and use the headless one for this.

• minio ClusterIP 10.63.245.183 80/TCP 19s
• minio-hl ClusterIP None 9000/TCP 19s

Oh I see the bug is that service DNS doesn't now route to 9000

[Alevsk]

Port 9000 should be still accessible since that is still valid. Port 443/80 are just service ports only.

That would work, but then we need to change the service name and use the headless one for this.

• minio ClusterIP 10.63.245.183 80/TCP 19s
• minio-hl ClusterIP None 9000/TCP 19s

using headless service for requests is not recommended because k8s will not load balance traffic
@harshavardhana port 9000 cannot be used anymore on the service since the only exposed ports are 443/80, targetPort its 9000

[dvaldivia]

I think we should only expose 9000 by default and let the users decide if they want a service with 80/443 since it would simplify operation, we could do this by either adding a service for internal k8s consumtion or making the service port configurable