Open marcopaggioro opened 3 months ago
The certificate has a label hcloud-ccm/service-uid
that must match the kubernetes service uid. You have to update this manually right now.
$ service_uid=$(kubectl get service -n traefik traefik -o=go-template --template='{{ .metadata.uid }}')
$ echo $service_uid
$ hcloud certificate add-label --overwrite https-certificate hcloud-ccm/service-uid=$service_uid
It sounds like a good workaround but, in fact, is a workaround. No? It sounds no-sense that I have to "patch" the label of the exist certificate every time I re create the Service
Yea, we need a better way to associate the cert with the service and potentially clean it up when necessary.
Suggestions by @micheljung in #596:
This issue has been marked as stale because it has not had recent activity. The bot will close the issue if no further action occurs.
problem still exists
TL;DR
It seems that HCCM is not able to see that a certificate already exists (already created by HCCM from a re-created Service or from another Service). If it already exists then it fails and the Services in the load balancer are not produced
Expected behavior
I expect HCCM not to fail if the certificate already exists and was created by itself.
Observed behavior
When I create the Service (annotations below) for the first time I can see the new certificate in the Hetzner Certificate section
Due to that I can see that HCCM populates even the Services section of my Hetzner Load Balancer. Everything works fine
If i destroy and recrete my service, than HCCM reports these errors
Seems like it can't detect that the certificate already exists and it fails.
If I delete the certificate nothing more happens but then If I delete and re-create the Service so HCCM "wake up" and recreated the certificate correctly (with Services in LB).
Minimal working example
Services with these annotations
Log output