sig-storage-lib-external-provisioner
sig-storage-lib-external-provisioner copied to clipboard
kubernetes-sigs
On migrating from v6 to v8, my provisioner crashes with the following error: ``` F0720 16:59:52.441682 5606 controller.go:878] Error creating lock: endpoints lock is removed, migrate to endpointsleases ``` I...
I see https://github.com/kubernetes-sigs/sig-storage-lib-external-provisioner/blob/a35bec6eb688ef7a0e31b340ed77bdb84010ace7/controller/controller.go#L1452 set `annDynamicallyProvisioned` from StorageClass.Provisioner, and test it using `knownProvisioner` with comments https://github.com/kubernetes-sigs/sig-storage-lib-external-provisioner/blob/a35bec6eb688ef7a0e31b340ed77bdb84010ace7/controller/controller.go#L1372-L1378 But in `updatePersistentVolume` and `shouldDelete` just use plain `ann != ctrl.provisionerName` to make comparison https://github.com/kubernetes-sigs/sig-storage-lib-external-provisioner/blob/a35bec6eb688ef7a0e31b340ed77bdb84010ace7/controller/controller.go#L1233...
The prow jobs don't check that examples/hostpath-provisioner is in a healthy state. Either "make test" or "make build" should cover that. I'm leaning to "make test", but that's debatable.
Invoking "make" multiple times in examples/hostpath-provisioner does not rebuild the binary even if the source code has changed.
Copy of https://github.com/kubernetes-incubator/external-storage/issues/912 https://github.com/kubernetes-incubator/external-storage/pull/902 introduces topology plumbing through the library, which includes a Node get call for every volume provision operation. Since CSI depends on this library and all volumes...
If a node is deleted while a pod is scheduled on a node (but before a claim is provisioned), a pod can become indefinitely stuck in a Pending state. Typically...
Follow up to https://github.com/kubernetes-csi/external-provisioner/pull/792 This PR adds a label to the existing metrics to determine the data source of the PVC being monitored **Testing** 1. Deploy host path driver with...
fix pvInformer cannot trigger delete event, The reason is that after the list watch to the delete event, the flector will judge whether the object exists in the cache store...
**What** The MR sets a log level for `k8log.Info` calls in the module. I used a few sources to choose what level to set: - https://github.com/kubernetes-csi/external-attacher - https://github.com/kubernetes/community/blob/master/contributors/devel/sig-instrumentation/logging.md#what-method-to-use It resulted...
Unit test `TestTopologyParams/provision_with_selected_node,_but_node_does_not_exist` fails frequently with log: ``` W1221 13:41:19.810378 1289631 shared_informer.go:544] resyncPeriod 100ms is too small. Changing it to the minimum allowed value of 1s W1221 13:41:19.810457 1289631 shared_informer.go:544]...
