ceph-csi/e2e/cephfs_helper.go
Rakshith R 43f753760b cleanup: resolve nlreturn linter issues
nlreturn linter requires a new line before return
and branch statements except when the return is alone
inside a statement group (such as an if statement) to
increase code clarity. This commit addresses such issues.

Updates: #1586

Signed-off-by: Rakshith R <rar@redhat.com>
2021-07-22 06:05:01 +00:00

255 lines
7.4 KiB
Go

package e2e
import (
"context"
"encoding/json"
"fmt"
"regexp"
"strings"
snapapi "github.com/kubernetes-csi/external-snapshotter/client/v4/apis/volumesnapshot/v1"
v1 "k8s.io/api/core/v1"
metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
"k8s.io/client-go/kubernetes"
"k8s.io/kubernetes/test/e2e/framework"
e2elog "k8s.io/kubernetes/test/e2e/framework/log"
)
const (
adminUser = "admin"
)
// validateSubvolumegroup validates whether subvolumegroup is present.
func validateSubvolumegroup(f *framework.Framework, subvolgrp string) error {
cmd := fmt.Sprintf("ceph fs subvolumegroup getpath myfs %s", subvolgrp)
stdOut, stdErr, err := execCommandInToolBoxPod(f, cmd, rookNamespace)
if err != nil {
return fmt.Errorf("failed to exec command in toolbox: %w", err)
}
if stdErr != "" {
return fmt.Errorf("failed to getpath for subvolumegroup %s with error %v", subvolgrp, stdErr)
}
expectedGrpPath := "/volumes/" + subvolgrp
stdOut = strings.TrimSpace(stdOut)
if stdOut != expectedGrpPath {
return fmt.Errorf("error unexpected group path. Found: %s", stdOut)
}
return nil
}
func createCephfsStorageClass(
c kubernetes.Interface,
f *framework.Framework,
enablePool bool,
params map[string]string) error {
scPath := fmt.Sprintf("%s/%s", cephfsExamplePath, "storageclass.yaml")
sc, err := getStorageClass(scPath)
if err != nil {
return err
}
sc.Parameters["fsName"] = "myfs"
sc.Parameters["csi.storage.k8s.io/provisioner-secret-namespace"] = cephCSINamespace
sc.Parameters["csi.storage.k8s.io/provisioner-secret-name"] = cephFSProvisionerSecretName
sc.Parameters["csi.storage.k8s.io/controller-expand-secret-namespace"] = cephCSINamespace
sc.Parameters["csi.storage.k8s.io/controller-expand-secret-name"] = cephFSProvisionerSecretName
sc.Parameters["csi.storage.k8s.io/node-stage-secret-namespace"] = cephCSINamespace
sc.Parameters["csi.storage.k8s.io/node-stage-secret-name"] = cephFSNodePluginSecretName
if enablePool {
sc.Parameters["pool"] = "myfs-data0"
}
// overload any parameters that were passed
if params == nil {
// create an empty params, so that params["clusterID"] below
// does not panic
params = map[string]string{}
}
for param, value := range params {
sc.Parameters[param] = value
}
// fetch and set fsID from the cluster if not set in params
if _, found := params["clusterID"]; !found {
fsID, stdErr, failErr := execCommandInToolBoxPod(f, "ceph fsid", rookNamespace)
if failErr != nil {
return failErr
}
if stdErr != "" {
return fmt.Errorf("error getting fsid %v", stdErr)
}
sc.Parameters["clusterID"] = strings.Trim(fsID, "\n")
}
sc.Namespace = cephCSINamespace
_, err = c.StorageV1().StorageClasses().Create(context.TODO(), &sc, metav1.CreateOptions{})
return err
}
func createCephfsSecret(f *framework.Framework, secretName, userName, userKey string) error {
scPath := fmt.Sprintf("%s/%s", cephfsExamplePath, "secret.yaml")
sc, err := getSecret(scPath)
if err != nil {
return err
}
if secretName != "" {
sc.Name = secretName
}
sc.StringData["adminID"] = userName
sc.StringData["adminKey"] = userKey
delete(sc.StringData, "userID")
delete(sc.StringData, "userKey")
sc.Namespace = cephCSINamespace
_, err = f.ClientSet.CoreV1().Secrets(cephCSINamespace).Create(context.TODO(), &sc, metav1.CreateOptions{})
return err
}
// unmountCephFSVolume unmounts a cephFS volume mounted on a pod.
func unmountCephFSVolume(f *framework.Framework, appName, pvcName string) error {
pod, err := f.ClientSet.CoreV1().Pods(f.UniqueName).Get(context.TODO(), appName, metav1.GetOptions{})
if err != nil {
e2elog.Logf("Error occurred getting pod %s in namespace %s", appName, f.UniqueName)
return fmt.Errorf("failed to get pod: %w", err)
}
pvc, err := f.ClientSet.CoreV1().
PersistentVolumeClaims(f.UniqueName).
Get(context.TODO(), pvcName, metav1.GetOptions{})
if err != nil {
e2elog.Logf("Error occurred getting PVC %s in namespace %s", pvcName, f.UniqueName)
return fmt.Errorf("failed to get pvc: %w", err)
}
cmd := fmt.Sprintf(
"umount /var/lib/kubelet/pods/%s/volumes/kubernetes.io~csi/%s/mount",
pod.UID,
pvc.Spec.VolumeName)
_, stdErr, err := execCommandInDaemonsetPod(
f,
cmd,
cephfsDeamonSetName,
pod.Spec.NodeName,
cephfsContainerName,
cephCSINamespace)
if stdErr != "" {
e2elog.Logf("StdErr occurred: %s", stdErr)
}
return err
}
func deleteBackingCephFSVolume(f *framework.Framework, pvc *v1.PersistentVolumeClaim) error {
imageData, err := getImageInfoFromPVC(pvc.Namespace, pvc.Name, f)
if err != nil {
return err
}
cmd := fmt.Sprintf("ceph fs subvolume rm %s %s %s", fileSystemName, imageData.imageName, subvolumegroup)
_, stdErr, err := execCommandInToolBoxPod(f, cmd, rookNamespace)
if err != nil {
return err
}
if stdErr != "" {
return fmt.Errorf("error deleting backing volume %s %v", imageData.imageName, stdErr)
}
return nil
}
type cephfsSubVolume struct {
Name string `json:"name"`
}
func listCephFSSubVolumes(f *framework.Framework, filesystem, groupname string) ([]cephfsSubVolume, error) {
var subVols []cephfsSubVolume
stdout, stdErr, err := execCommandInToolBoxPod(
f,
fmt.Sprintf("ceph fs subvolume ls %s --group_name=%s --format=json", filesystem, groupname),
rookNamespace)
if err != nil {
return subVols, err
}
if stdErr != "" {
return subVols, fmt.Errorf("error listing subolumes %v", stdErr)
}
err = json.Unmarshal([]byte(stdout), &subVols)
if err != nil {
return subVols, err
}
return subVols, nil
}
// getSubvolumepath validates whether subvolumegroup is present.
func getSubvolumePath(f *framework.Framework, filesystem, subvolgrp, subvolume string) (string, error) {
cmd := fmt.Sprintf("ceph fs subvolume getpath %s %s --group_name=%s", filesystem, subvolume, subvolgrp)
stdOut, stdErr, err := execCommandInToolBoxPod(f, cmd, rookNamespace)
if err != nil {
return "", err
}
if stdErr != "" {
return "", fmt.Errorf("failed to getpath for subvolume %s with error %s", subvolume, stdErr)
}
return strings.TrimSpace(stdOut), nil
}
func getSnapName(snapNamespace, snapName string) (string, error) {
sclient, err := newSnapshotClient()
if err != nil {
return "", err
}
snap, err := sclient.
VolumeSnapshots(snapNamespace).
Get(context.TODO(), snapName, metav1.GetOptions{})
if err != nil {
return "", fmt.Errorf("failed to get volumesnapshot: %w", err)
}
sc, err := sclient.
VolumeSnapshotContents().
Get(context.TODO(), *snap.Status.BoundVolumeSnapshotContentName, metav1.GetOptions{})
if err != nil {
return "", fmt.Errorf("failed to get volumesnapshotcontent: %w", err)
}
snapIDRegex := regexp.MustCompile(`(\w+\-?){5}$`)
snapID := snapIDRegex.FindString(*sc.Status.SnapshotHandle)
snapshotName := fmt.Sprintf("csi-snap-%s", snapID)
e2elog.Logf("snapshotName= %s", snapshotName)
return snapshotName, nil
}
func deleteBackingCephFSSubvolumeSnapshot(
f *framework.Framework,
pvc *v1.PersistentVolumeClaim,
snap *snapapi.VolumeSnapshot) error {
snapshotName, err := getSnapName(snap.Namespace, snap.Name)
if err != nil {
return err
}
imageData, err := getImageInfoFromPVC(pvc.Namespace, pvc.Name, f)
if err != nil {
return err
}
cmd := fmt.Sprintf(
"ceph fs subvolume snapshot rm %s %s %s %s",
fileSystemName,
imageData.imageName,
snapshotName,
subvolumegroup)
_, stdErr, err := execCommandInToolBoxPod(f, cmd, rookNamespace)
if err != nil {
return err
}
if stdErr != "" {
return fmt.Errorf("error deleting backing snapshot %s %v", snapshotName, stdErr)
}
return nil
}