Setting persistentVolumeReclaimPolicy to DELETE

[atrzcinski]

Hi,

I'm wondering what effect setting persistentVolumeReclaimPolicy of pv from RETAIN to DELETE will be?
Will the data be deleted or not and why?

[andyzhangx]

the mounted folder would be deleted if set as DELETE

[andyzhangx]

then you should have the /DeleteVolume logs in driver controller:
https://github.com/kubernetes-csi/csi-driver-smb/blob/master/docs/csi-debug.md#case1-volume-createdelete-failed

[atrzcinski]

there are only startup logs, no new logs provided after mount / unmount

`> cat csi-smb-controller.log
W0112 12:06:24.625992 1 main.go:62] nodeid is empty
I0112 12:06:24.631401 1 main.go:92] set up prometheus server on 0.0.0.0:29644
I0112 12:06:24.631704 1 smb.go:91]
DRIVER INFORMATION:

Build Date: "2022-08-20T07:34:36Z"
Compiler: gc
Driver Name: smb.csi.k8s.io
Driver Version: v1.9.0
Git Commit: ""
Go Version: go1.17.3
Platform: linux/amd64

Streaming logs below:
I0112 12:06:24.631842 1 mount_linux.go:208] Detected OS without systemd
I0112 12:06:24.631860 1 driver.go:93] Enabling controller service capability: CREATE_DELETE_VOLUME
I0112 12:06:24.631867 1 driver.go:93] Enabling controller service capability: SINGLE_NODE_MULTI_WRITER
I0112 12:06:24.631874 1 driver.go:112] Enabling volume access mode: SINGLE_NODE_WRITER
I0112 12:06:24.631883 1 driver.go:112] Enabling volume access mode: SINGLE_NODE_READER_ONLY
I0112 12:06:24.631887 1 driver.go:112] Enabling volume access mode: SINGLE_NODE_SINGLE_WRITER
I0112 12:06:24.631892 1 driver.go:112] Enabling volume access mode: SINGLE_NODE_MULTI_WRITER
I0112 12:06:24.631896 1 driver.go:112] Enabling volume access mode: MULTI_NODE_READER_ONLY
I0112 12:06:24.631910 1 driver.go:112] Enabling volume access mode: MULTI_NODE_SINGLE_WRITER
I0112 12:06:24.631928 1 driver.go:112] Enabling volume access mode: MULTI_NODE_MULTI_WRITER
I0112 12:06:24.631933 1 driver.go:103] Enabling node service capability: STAGE_UNSTAGE_VOLUME
I0112 12:06:24.631939 1 driver.go:103] Enabling node service capability: SINGLE_NODE_MULTI_WRITER
I0112 12:06:24.631944 1 driver.go:103] Enabling node service capability: VOLUME_MOUNT_GROUP
I0112 12:06:24.631949 1 driver.go:103] Enabling node service capability: GET_VOLUME_STATS
I0112 12:06:24.632154 1 server.go:118] Listening for connections on address: &net.UnixAddr{Name:"//csi/csi.sock", Net:"unix"}
I0112 12:06:24.699739 1 utils.go:76] GRPC call: /csi.v1.Identity/GetPluginInfo
I0112 12:06:24.699768 1 utils.go:77] GRPC request: {}
I0112 12:06:24.703908 1 utils.go:83] GRPC response: {"name":"smb.csi.k8s.io","vendor_version":"v1.9.0"}
I0112 12:06:25.386019 1 utils.go:76] GRPC call: /csi.v1.Identity/GetPluginInfo
I0112 12:06:25.386055 1 utils.go:77] GRPC request: {}
I0112 12:06:25.386147 1 utils.go:83] GRPC response: {"name":"smb.csi.k8s.io","vendor_version":"v1.9.0"}
I0112 12:06:25.387656 1 utils.go:76] GRPC call: /csi.v1.Identity/GetPluginCapabilities
I0112 12:06:25.387675 1 utils.go:77] GRPC request: {}
I0112 12:06:25.387837 1 utils.go:83] GRPC response: {"capabilities":[{"Type":{"Service":{"type":1}}}]}
I0112 12:06:25.388829 1 utils.go:76] GRPC call: /csi.v1.Controller/ControllerGetCapabilities
I0112 12:06:25.388846 1 utils.go:77] GRPC request: {}
I0112 12:06:25.388883 1 utils.go:83] GRPC response: {"capabilities":[{"Type":{"Rpc":{"type":1}}},{"Type":{"Rpc":{"type":13}}}]}

`

[andyzhangx]

then what's the pv status? kubectl describe pv pv-name

[atrzcinski]

after install (mount) the pv is in Bound status, after uninstall:

Error from server (NotFound): persistentvolumes "my-pv" not found

[andyzhangx]

try following example directly:

kubectl apply -f https://raw.githubusercontent.com/kubernetes-csi/csi-driver-smb/master/deploy/example/pvc-smb.yaml
kubectl apply -f https://github.com/kubernetes-csi/csi-driver-smb/blob/master/deploy/example/pvc-smb-static.yaml
# wait about 10s to make sure pvc is in Bound state
kubectl get pvc pvc-smb

# delete pvc and then pv will also be deleted
kubectl delete -f https://github.com/kubernetes-csi/csi-driver-smb/blob/master/deploy/example/pvc-smb-static.yaml

• in csi driver controller logs, it would how have following logs which means DeleteVolume happens, and since subDir is not in volumeID, it's skippped

I0113 09:42:04.021816       1 utils.go:76] GRPC call: /csi.v1.Controller/DeleteVolume
I0113 09:42:04.021851       1 utils.go:77] GRPC request: {"volume_id":"smb-server.default.svc.cluster.local/share##"}
I0113 09:42:04.021997       1 controllerserver.go:174] DeleteVolume(smb-server.default.svc.cluster.local/share##) does not delete subdirectory

[andyzhangx]

this doc PR would fix the issue: #572