Kacper Rzetelski
Kacper Rzetelski
What we've found so far (cc @zimnx): An error logged by the operator during scaledown: ``` E0420 15:59:23.721881 1 scyllacluster/status.go:112] only last member of each rack should be decommissioning, but...
`yq '.items.[] | select(.spec.claimRef | (.namespace == "e2e-test-scyllacluster-frb8h-c5jdt" )) | [.]' persistentvolumes.yaml` ``` - apiVersion: v1 kind: PersistentVolume metadata: annotations: hostPathProvisionerIdentity: 32a4dd97-e96a-4482-8dc2-8d12dbecde78 pv.kubernetes.io/provisioned-by: k8s.io/minikube-hostpath creationTimestamp: "2022-04-20T15:52:25Z" finalizers: - kubernetes.io/pv-protection name:...
Discussed offline, but just for the record: we are not gathering logs from previous iterations of containers running in non-e2e namespaces, so best I could find was container statuses from...
Reproduced in [Go #3252](https://github.com/scylladb/scylla-operator/runs/6208288355?check_suite_focus=true) [logs_6239.zip](https://github.com/scylladb/scylla-operator/suites/6284442666/logs?attempt=6) [e2e_artifacts.tar.lz4](https://github.com/scylladb/scylla-operator/suites/6284442666/artifacts/225589728) ``` I0428 09:35:16.761321 1 controller.go:1478] delete "pvc-2f1cc092-2e44-41be-a62f-60f0a0210548": volume deletion ignored: ignored because identity an notation on PV does not match ours I0428 09:35:24.860347 1...
@tnozicka I've changed the approach described in the proposal
@tnozicka I've addressed all your comments. I'm not entirely sure about how this enhancement could be affected by running the manual replacement. I don't see any potential issues/races atm but...
@tnozicka if you mean the PoC for fetching the manifests and extracting labels then this is the the code using [containers/image](https://github.com/containers/image) in a CLI tool mimicking skopeo for [`docker transport`](https://github.com/containers/image/blob/main/docs/containers-transports.5.md#dockerdocker-reference)...
@zimnx Should we close this issue? I believe it was fixed with #901 .
Should've been fixed by https://github.com/scylladb/scylla-operator/issues/971. Any other issues related to (minor) delays in gossip propagation should be covered by https://github.com/scylladb/scylla-operator/pull/1054. @tnozicka should we close it?
Service IP doesn't change in the meantime and cluster rolls out fine. Flaky behaviour might be an unhandled case in gocql lib. I'm currently trying to reproduce it with gocql...