feat(filtering): filter pvc by their name and a regex to avoid replication

Signed-off-by: SkalaNetworks <contact@skala.network>

Author: SkalaNetworks

README.md

@@ -10,6 +10,7 @@ the actual replication itself.
 
 - **Automated Lifecycle**: Automatically creates `VolumeReplication` objects for PVCs with the appropriate annotation.
 - **Inheritance**: Can inherit the `VolumeReplicationClass` from the PVC or from the PVC's namespace (if not specified on the PVC).
+- **Exclusion by Name**: Supports excluding PVCs from replication using a global regular expression.
 - **VRC Selector**: Supports selecting a `VolumeReplicationClass` using a selector, allowing for more dynamic configuration based on `StorageClass` groups.
 - **Cleanup**: Automatically deletes `VolumeReplication` resources when their parent PVC is deleted or when the replication annotation is removed.
 - **Leader Election**: Supports high availability with leader election to ensure only one instance is active at a time.
@@ -147,6 +148,19 @@ If the annotation is modified, the `VolumeReplication` is updated accordingly wi
 
 If the annotation is deleted on both the PVC and the namespace, the VolumeReplication is deleted.
 
+### Excluding PVCs from replication
+
+It is possible to exclude some PVCs from being replicated, even if they have the correct annotations (or their namespace has them).
+This is done by providing a regular expression to the controller using the `--exclusion-regex` flag or the `EXCLUSION_REGEX` environment variable.
+
+Any PVC whose name matches the regular expression will be ignored by the controller.
+For example, if you set `EXCLUSION_REGEX` to `^prime-.*`, all PVCs starting with `prime-` will be excluded from replication.
+
+This feature is useful to avoid unnecessary replications of temporary PVCs (for example, for "prime" PVCs created by the [Container Data Importer](https://github.com/kubevirt/containerized-data-importer)).
+
+> [!NOTE]
+> If the regular expression is empty, no PVC will be excluded (unless it doesn't have the appropriate annotations).
+
 ## Configuration
 
 The controller can be configured using command-line flags or environment variables:
@@ -155,6 +169,7 @@ The controller can be configured using command-line flags or environment variabl
 |------|----------------------|---------|-------------|
 | `--kubeconfig` | - | - | Path to a kubeconfig file. If not provided, it assumes in-cluster configuration. |
 | `--namespace` | `NAMESPACE` | - | **Required**. The namespace where the controller is deployed (used for leader election). |
+| `--exclusion-regex` | `EXCLUSION_REGEX` | - | Optional regular expression to exclude PVCs from replication by name. |
 
 Standard `klog` flags are also supported for logging configuration.
 

cmd/main.go

@@ -3,12 +3,13 @@ package main
 import (
 	"context"
 	"flag"
+	"os"
+	"os/signal"
+
 	"github.com/skalanetworks/volume-replicator/internal/k8s"
 	"github.com/skalanetworks/volume-replicator/internal/replicator"
 	"k8s.io/client-go/tools/leaderelection"
 	"k8s.io/klog/v2"
-	"os"
-	"os/signal"
 )
 
 func main() {
@@ -18,6 +19,7 @@ func main() {
 	var kubeconfig, namespace string
 	flag.StringVar(&kubeconfig, "kubeconfig", "", "path to kubeconfig file")
 	flag.StringVar(&namespace, "namespace", os.Getenv("NAMESPACE"), "deployment namespace")
+	flag.StringVar(&replicator.ExclusionRegex, "exclusion-regex", os.Getenv("EXCLUSION_REGEX"), "regex to exclude PVCs from replication")
 	klog.InitFlags(nil)
 	flag.Parse()
 

internal/replicator/utils.go

@@ -3,6 +3,7 @@ package replicator
 import (
 	"context"
 	"fmt"
+	"regexp"
 
 	"github.com/skalanetworks/volume-replicator/internal/constants"
 	"github.com/skalanetworks/volume-replicator/internal/k8s"
@@ -13,6 +14,8 @@ import (
 	"k8s.io/klog/v2"
 )
 
+var ExclusionRegex string
+
 // isVolumeReplicationCorrect verifies if the definition of a VolumeReplication conforms to its originating PVC
 func isVolumeReplicationCorrect(pvc *corev1.PersistentVolumeClaim, vr *unstructured.Unstructured) bool {
 	key := fmt.Sprintf("%s/%s", vr.GetNamespace(), vr.GetName())
@@ -168,3 +171,21 @@ func getPvcProvisioner(pvc *corev1.PersistentVolumeClaim) string {
 	// Fallback to the deprecated annotation
 	return pvc.Annotations[constants.DeprecatedStorageProvisionerAnnotation]
 }
+
+// pvcNameMatchesExclusion returns whether a PVC has a name matching the exclusion regex
+func pvcNameMatchesExclusion(pvc *corev1.PersistentVolumeClaim) bool {
+	// If no regex is provided, return that it doesn't match
+	// This is to avoid Go matching "" as "everything matches"
+	if ExclusionRegex == "" {
+		return false
+	}
+
+	// Match the user-provided regex
+	match, err := regexp.MatchString(ExclusionRegex, pvc.Name)
+	if err != nil {
+		klog.Errorf("failed to parse exclusion regex: %s", err.Error())
+		return false
+	}
+
+	return match
+}

internal/replicator/utils_test.go

@@ -662,3 +662,62 @@ func TestGetPvcProvisioner(t *testing.T) {
 		})
 	}
 }
+
+func TestPvcNameMatchesExclusion(t *testing.T) {
+	tests := []struct {
+		name           string
+		exclusionRegex string
+		pvcName        string
+		expected       bool
+	}{
+		{
+			name:           "Empty regex matches nothing",
+			exclusionRegex: "",
+			pvcName:        "any-pvc",
+			expected:       false,
+		},
+		{
+			name:           "Exact match",
+			exclusionRegex: "^test-pvc$",
+			pvcName:        "test-pvc",
+			expected:       true,
+		},
+		{
+			name:           "Prefix match",
+			exclusionRegex: "^skip-.*",
+			pvcName:        "skip-this-pvc",
+			expected:       true,
+		},
+		{
+			name:           "No match",
+			exclusionRegex: "^skip-.*",
+			pvcName:        "dont-skip-this",
+			expected:       false,
+		},
+		{
+			name:           "Partial match (regexp.MatchString behavior)",
+			exclusionRegex: "exclude",
+			pvcName:        "pvc-to-exclude-here",
+			expected:       true,
+		},
+		{
+			name:           "Invalid regex",
+			exclusionRegex: "[invalid",
+			pvcName:        "any-pvc",
+			expected:       false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ExclusionRegex = tt.exclusionRegex
+			pvc := &corev1.PersistentVolumeClaim{
+				ObjectMeta: metav1.ObjectMeta{
+					Name: tt.pvcName,
+				},
+			}
+			result := pvcNameMatchesExclusion(pvc)
+			require.Equal(t, tt.expected, result)
+		})
+	}
+}

internal/replicator/vrc.go

@@ -15,6 +15,12 @@ import (
 // The VRC can be provided through annotations as a value or as a selector.
 // The annotations can be placed on the PVC or on its namespace.
 func getVolumeReplicationClass(pvc *corev1.PersistentVolumeClaim) string {
+	// If the PVC is to be excluded, return an empty replication class
+	if pvcNameMatchesExclusion(pvc) {
+		klog.Infof("PVC %s/%s matches exclusion pattern, no replication class to apply", pvc.Namespace, pvc.Name)
+		return ""
+	}
+
 	// Retrieve the literal VRC provided on the PVC
 	value := getVolumeReplicationClassValue(pvc)
 	if value != "" {

internal/replicator/vrc_test.go

@@ -191,12 +191,33 @@ func TestGetVolumeReplicationClass(t *testing.T) {
 			},
 			expectedResult: "",
 		},
+		{
+			name: "Excluded by name regex",
+			pvc: &corev1.PersistentVolumeClaim{
+				ObjectMeta: metav1.ObjectMeta{
+					Name:      "exclude-this-pvc",
+					Namespace: nsName,
+					Annotations: map[string]string{
+						constants.VrcValueAnnotation: vrcName,
+					},
+				},
+			},
+			expectedResult: "",
+		},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			clearNamespaceIndexer(t)
 
+			// Set ExclusionRegex specifically for each test case
+			if tt.name == "Excluded by name regex" {
+				ExclusionRegex = "exclude-.*"
+			} else {
+				ExclusionRegex = ""
+			}
+			defer func() { ExclusionRegex = "" }()
+
 			if tt.namespace != nil {
 				err := NamespaceInformer.Informer().GetIndexer().Add(tt.namespace)
 				require.NoError(t, err)