clickhouse-operator icon indicating copy to clipboard operation
clickhouse-operator copied to clipboard

Problem with changing default LB service to ClusterIP

Open warlock9600 opened this issue 4 years ago • 5 comments

After updating CHI to add a service template using the ClusterIP service instead of the default LB, the CHI is in the "InProgress" state

If I remove CHI manually the LB is not removed, and if I manually remove the LB and redeploy the manifest with the CHI descriptions everything is fine

Kubernetes Provider - GKE, master and working versions 1.18.15-gke.1500

warlock9600 avatar Mar 01 '21 09:03 warlock9600

could you share your CHI manifest before change a serviceTemplate and after? When chi in InProgress state you can run kubectl describe chi -n <chi_namespace> <chi_name>" and look to detailed description what exactly clickhouse-operator try to do with your chi`

Slach avatar Mar 01 '21 10:03 Slach

chi-examples.zip added my examples

When chi in InProgress state you can run kubectl describe chi -n <chi_namespace> <chi_name>" and look to detailed description what exactly clickhouse-operator try to do with your chi`

tried this, but too later after deploy, the "events" section was empty

warlock9600 avatar Mar 01 '21 11:03 warlock9600

both "before" and "after" manifests looks good

maybe you can share clickhouse-operator pod log? kubectl logs -n <clickhouse_operator_namespace> <clickhouse_operator_pod> --since=24h ?

which clickhouse-operator version did you use?

Slach avatar Mar 01 '21 11:03 Slach

maybe you can share clickhouse-operator pod log? kubectl logs -n <clickhouse_operator_namespace> <clickhouse_operator_pod> --since=24h ?

I will try to reproduce the problem again and collect more facts.

which clickhouse-operator version did you use?

we use clickhouse-operator 0.13.5

warlock9600 avatar Mar 01 '21 13:03 warlock9600

Hi @warlock9600 , please try with 0.14.0 version. It actively reconciles all k8s objects, that may fix your issue.

alex-zaitsev avatar Apr 27 '21 07:04 alex-zaitsev

Fully fixed in 0.23.0. There is a Kubernetes bug that caused LB type change would not work sometimes

alex-zaitsev avatar Feb 14 '24 17:02 alex-zaitsev