Hey friends! We're currently seeing an issue with local-path-provisioner where we are getting really long queues due to it attempting to delete volumes on nodes that no longer exist. This leads to spammy API requests to the control plane, and massive delays in provisioning pods. We have a kind-of spammy pattern where we are scaling up and down a ton of nodes. The node no longer exists b/c cluster-autoscaler scaled it down. Is there a way to tell it to just delete the pv and the pvc if the node is no longer up without exhausting its retry count? I have an error status in the 🧵 . Take note of

PodGC:  node no longer exists

near the bottom.

I’m not really sure what it’s supposed to do. local-path-provisioner is just sort of a dumb stub for when you don’t have real volumes. If you delete the node, it’s not like it has some way to bring that data back or reprovision it on another node.