Volumes stuck in Released state about k3s

[brandond]

yes, that sounds like a separate issue. I don't see any difference in the volume permissions or contexts between the two versions though? It is expected that the numeric portion at the end will differ.

[VestigeJ]

##Environment Details
Attempted to reproduce but didn't hit it this time using VERSION=v1.29.5+k3s1 nor VERSION=v1.30.1+k3s1
Validated using COMMIT=cff6f7aa1d7987a658b030b2bc69df7c25f515c8

Infrastructure

• Cloud

Node(s) CPU architecture, OS, and version:

Linux 5.14.21-150500.53-default x86_64 GNU/Linux
PRETTY_NAME="SUSE Linux Enterprise Server 15 SP5"

Cluster Configuration:

NAME               STATUS   ROLES                       AGE   VERSION
ip-3-3-8-8         Ready    control-plane,etcd,master   23m   v1.30.1+k3s-cff6f7aa

Config.yaml:

node-external-ip: 3.3.8.8
token: YOUR_TOKEN_HERE
write-kubeconfig-mode: 644
debug: true
cluster-init: true
embedded-registry: true

Reproduction

$ curl https://get.k3s.io --output install-"k3s".sh
$ sudo chmod +x install-"k3s".sh
$ sudo groupadd --system etcd && sudo useradd -s /sbin/nologin --system -g etcd etcd
$ sudo modprobe ip_vs_rr
$ sudo modprobe ip_vs_wrr
$ sudo modprobe ip_vs_sh
$ sudo printf "on_oovm.panic_on_oom=0 \nvm.overcommit_memory=1 \nkernel.panic=10 \nkernel.panic_ps=1 \nkernel.panic_on_oops=1 \n" > ~/90-kubelet.conf
$ sudo cp 90-kubelet.conf /etc/sysctl.d/
$ sudo systemctl restart systemd-sysctl
$ COMMIT=cff6f7aa1d7987a658b030b2bc69df7c25f515c8
$ sudo INSTALL_K3S_COMMIT=$COMMIT INSTALL_K3S_EXEC=server ./install-k3s.sh
$ kg no,po -A
$ vim pvc.yaml
$ vim pv-pod.yaml
$ k apply -f pvc.yaml
$ kg pvc -A
$ k apply -f pv-pod.yaml
$ kg po,pv,pvc -A
$ k delete -f pv-pod.yaml
$ k delete -f pvc.yaml
$ kg pv -A
$ kubectl get deployment -n kube-system local-path-provisioner -o=jsonpath='{$.spec.template.spec.containers[:1].image}'
$ k apply -f pvc.yaml
$ k apply -f pv-pod.yaml
$ kg pv,pod,pvc -A
$ k delete -f pvc.yaml; sleep 40; k delete -f pv-pod.yaml
$ kg pv,pvc -A
$ k apply -f pvc.yaml
$ k apply -f pv-pod.yaml
$ k delete -f pvc.yaml
$ k delete -f pv-pod.yaml
$ kg pvc,pv -A

Results:

No dangling leftover volume claims, they're reclaimed appropriately.

$ cat pvc.yaml

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: test-pvc
  namespace: kube-system
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 10Mi

$ cat pv-pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: test-pod
  namespace: kube-system
spec:
  containers:
    - name: debian
      image: digitalocean/doks-debug
      command: ["sleep", "infinity"]
      volumeMounts:
        - name: data
          mountPath: /data
  volumes:
    - name: data
      persistentVolumeClaim:
        claimName: test-pvc

[zc-devs]

First change may be introducing new opt:

  setup: |-
    #!/bin/sh
    while getopts "m:s:p:a:" opt
    do
        case $opt in
            p)
            absolutePath=$OPTARG
            ;;
            s)
            sizeInBytes=$OPTARG
            ;;
            m)
            volMode=$OPTARG
            ;;
            a)
            action=$OPTARG
            ;;
        esac
    done
    if [ "$action" = "create" ]
    then
      mkdir -m 0777 -p ${absolutePath}
      chmod 700 ${absolutePath}/..
    fi
  teardown: |-
    #!/bin/sh
    set -x
    while getopts "m:s:p:a:" opt
    do
        case $opt in
            p)
            absolutePath=$OPTARG
            ;;
            s)
            sizeInBytes=$OPTARG
            ;;
            m)
            volMode=$OPTARG
            ;;
            a)
            action=$OPTARG
            ;;
        esac
    done
    if [ "$action" = "delete" ]
    then
      rm -rf ${absolutePath}
    fi

In that case I don't get Illegal option -a error, but helper pod fails anyways.
Then I tried to debug:

  teardown: |-
    #!/bin/sh
    sleep infinity

/ # ls -lah /var/lib/rancher/k3s/storage/local/ssd/pvc-a9809075-743a-4a86-ba28-38ca3d8256cd_kube-system_test-pvc
ls: can't open '/var/lib/rancher/k3s/storage/local/ssd/pvc-a9809075-743a-4a86-ba28-38ca3d8256cd_kube-system_test-pvc': Permission denied
total 0

/ # ls -lah /var/lib/rancher/k3s/storage/local/ssd
total 3G
drwx------   14 root     root        4.0K Mar 29 22:12 .
drwxr-xr-x    3 root     root        4.0K Mar 29 22:33 ..
drwxrwxrwx    2 root     root        4.0K Mar 29 22:13 pvc-a9809075-743a-4a86-ba28-38ca3d8256cd_kube-system_test-pvc

/ # id
uid=0(root) gid=0(root) groups=0(root),10(wheel)

Then I compared Pod definitions between v0.24 and v0.26:

[VestigeJ]

TLDR from above link where I encountered this issue while testing the latest COMMIT_ID for v1.28 branch

$ kg pv -A

NAME            CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS   CLAIM              STORAGECLASS   REASON   AGE
checking-path   5Gi        RWO            Recycle          Failed   default/test-pvc   local-path              50m

[0xMALVEE]

is this resolved already?

[brandond]

nope. issue is still open and PR is not merged. Waiting for end of code freeze.

[zc-devs]

Hi, I've just tested #9964 with Local path config: local-storage.yaml.

While PV creates, it cannot be deleted, the helper pod fails.

helper-pod-create-pvc-v0.24.yaml
helper-pod-create-pvc-v0.26.yaml
helper-pod-delete-pvc-v0.24.yaml
helper-pod-delete-pvc-v0.26.yaml

I think, the main difference is v0.26 doesn't use privileged security context flag, as I've noticed in #9833 (comment). Have to mention also, that I use Oracle Linux 9 with enabled SELinux. Directory permissions are:

# ls -lanZ /var/lib/rancher/k3s/storage/
total 48
drwx------. 12    0    0 system_u:object_r:container_file_t:s0           4096 May 21 19:24 .
drwxr-xr-x.  6    0    0 system_u:object_r:container_var_lib_t:s0        4096 Jan 18 18:32 ..

Volume permissions (v0.24):

# ls -lanZ pvc-9391085e-05dc-42f4-9c0d-25a1b4e6fe4a_kube-system_test-pvc/
total 12
drwxrwxrwx.  2 0 0 system_u:object_r:container_file_t:s0:c131,c199 4096 May 21 19:26 .
drwx------. 12 0 0 system_u:object_r:container_file_t:s0           4096 May 21 19:24 ..
-rw-r--r--.  1 0 0 system_u:object_r:container_file_t:s0:c131,c199    5 May 21 19:29 test.txt

Volume permissions (v0.26):

# ls -lanZ pvc-1fdbed66-a718-463d-908e-21bd731934c4_kube-system_test-pvc/
total 12
drwxrwxrwx.  2 0 0 system_u:object_r:container_file_t:s0:c143,c741 4096 May 21 19:34 .
drwx------. 11 0 0 system_u:object_r:container_file_t:s0           4096 May 21 19:33 ..
-rw-r--r--.  1 0 0 system_u:object_r:container_file_t:s0:c143,c741    5 May 21 19:34 test.txt

Should I file a separate issue?