CSI takes so long to create volumeattachment

[m-niak]

TL;DR

When I Deploy a statefulset that includes multiple pvcs like postgresql , CSI creates volumes on the panel but
takes so long to attach it to the related worker because it does not create volumeattachment resource instantly.
kubectl get volumeattachment

k8s version 1.34.2

csiAttacher:
  name: registry.k8s.io/sig-storage/csi-attacher
  tag: v4.10.0 

---

csiNodeDriverRegistrar:
  name: registry.k8s.io/sig-storage/csi-node-driver-registrar
  tag: v2.15.0

Expected behavior

kubectl get pvc
shows all pvcs bound

Observed behavior

kubectl get pvc

pending

Minimal working example

No response

Log output

Additional information

No response

[lukasmetzner]

Hey @m-niak I need some more information to understand your problem, or what your goal is.

The VolumeAttachments are not created by the hcloud-csi-driver. They are created either by kubelet or the attach/dettach controller inside the control-plane. The hcloud-csi-driver controller Pod contains the external-attacher container, which is responsible for listening to VolumeAttachments and invoking the correct gRPC call on our side to attach or detach Hetzner Cloud Volumes. The external-attacher contains is developed at https://github.com/kubernetes-csi/external-attacher.

Either your issue relies on your cluster, or the Hetzner Cloud Volumes did not create or attach fast enough in the API. This is not our scope here, we are not responsible for the Hetzner Cloud Volumes product.

If your issue lies somewhere else, I would need log messages of the hcloud-csi-driver controller and node driver. Additionally, a kubectl describe on the csi-driver, the PersistentVolume, and the VolumeAttachment would be of interest. Please also provide the csi-driver version you are using.