fix(pod-replacement): exclude container requirement (#1004)

Pod replacement operates at the pod level and should
not require container information. The feature was
failing because the injector expected target containers
to be specified, but pod replacement works directly
with pods identified by their IP address.

This fix adds logic to detect when running the
pod-replacement command and skips the container
validation, creating a single config without container
dependencies instead.

Jira: CHAOSPLT-1330

Author: aymericDD

api/v1beta1/pod_replacement.go

@@ -5,7 +5,11 @@
 
 package v1beta1
 
-import "strconv"
+import (
+	"strconv"
+
+	chaostypes "github.com/DataDog/chaos-controller/types"
+)
 
 // PodReplacementSpec represents a pod replacement disruption
 type PodReplacementSpec struct {
@@ -27,7 +31,7 @@ func (s *PodReplacementSpec) Validate() error {
 // GenerateArgs generates injection or cleanup pod arguments for the given spec
 func (s *PodReplacementSpec) GenerateArgs() []string {
 	args := []string{
-		"pod-replacement",
+		chaostypes.DisruptionKindPodReplacement,
 		"inject",
 	}
 

cli/injector/main.go

@@ -222,6 +222,16 @@ func initConfig() {
 	// assign to the pointer to level the new value to persist it after this method
 	disruptionArgs.Level = chaostypes.DisruptionLevel(disruptionLevelRaw)
 
+	// check if we're running pod-replacement command which doesn't need containers
+	isPodReplacement := false
+
+	for _, arg := range os.Args {
+		if arg == chaostypes.DisruptionKindPodReplacement {
+			isPodReplacement = true
+			break
+		}
+	}
+
 	switch disruptionArgs.Level {
 	case chaostypes.DisruptionLevelPod:
 		// check for container ID flag
@@ -231,33 +241,39 @@ func initConfig() {
 			return
 		}
 
-		for containerName, containerID := range disruptionArgs.TargetContainers {
-			// retrieve container info
-			ctn, err := container.New(containerID, containerName)
-			if err != nil {
-				log.Fatalw("can't create container object", tags.ErrorKey, err)
+		if !isPodReplacement {
+			// Pod replacement operates at the pod level and doesn't need container information
+			for containerName, containerID := range disruptionArgs.TargetContainers {
+				// retrieve container info
+				ctn, err := container.New(containerID, containerName)
+				if err != nil {
+					log.Fatalw("can't create container object", tags.ErrorKey, err)
 
-				return
-			}
+					return
+				}
 
-			log.Infow("injector targeting container", tags.ContainerIDKey, containerID, tags.ContainerNameKey, containerName)
+				log.Infow("injector targeting container", tags.ContainerIDKey, containerID, tags.ContainerNameKey, containerName)
 
-			pid := ctn.PID()
+				pid := ctn.PID()
 
-			// keep pid for later if this is a chaos handler container
-			if disruptionArgs.OnInit && ctn.Name() == chaosInitContName {
-				handlerPID = pid
-			}
+				// keep pid for later if this is a chaos handler container
+				if disruptionArgs.OnInit && ctn.Name() == chaosInitContName {
+					handlerPID = pid
+				}
 
-			ctns = append(ctns, ctn)
-			pids = append(pids, pid)
-		}
+				ctns = append(ctns, ctn)
+				pids = append(pids, pid)
+			}
+		} else {
+			// check for pod IP flag
+			if disruptionArgs.TargetPodIP == "" {
+				log.Fatal("--target-pod-ip flag must be passed when --level=pod")
 
-		// check for pod IP flag
-		if disruptionArgs.TargetPodIP == "" {
-			log.Fatal("--target-pod-ip flag must be passed when --level=pod")
+				return
+			}
 
-			return
+			pids = []uint32{1}
+			ctns = []container.Container{nil}
 		}
 	case chaostypes.DisruptionLevelNode:
 		pids = []uint32{1}

cli/injector/pod_replacement.go

@@ -8,14 +8,16 @@ package main
 import (
 	"strconv"
 
+	"github.com/spf13/cobra"
+
 	"github.com/DataDog/chaos-controller/api/v1beta1"
 	"github.com/DataDog/chaos-controller/injector"
 	"github.com/DataDog/chaos-controller/o11y/tags"
-	"github.com/spf13/cobra"
+	chaostypes "github.com/DataDog/chaos-controller/types"
 )
 
 var podReplacementCmd = &cobra.Command{
-	Use:   "pod-replacement",
+	Use:   chaostypes.DisruptionKindPodReplacement,
 	Short: "Pod replacement subcommands",
 	Run:   injectAndWait,
 	PreRun: func(cmd *cobra.Command, args []string) {

examples/demo.yaml

@@ -217,6 +217,30 @@ spec:
             requests:
               memory: 16Mi
               cpu: 10m
+        ## Create a second container to ensure the pod replace is working as expected
+        - name: storage-monitor-2
+          image: alpine:latest
+          command: ["/bin/sh"]
+          args:
+            - -c
+            - |
+              echo "Storage monitoring container started"
+              # Create a test file to demonstrate persistence
+              echo "demo-storage-$(date)" > /mnt/shared/storage-info.txt
+              # Keep container running and periodically update the file
+              while true; do
+                echo "$(date): Storage is available" >> /mnt/shared/heartbeat.log
+                sleep 30
+              done
+          volumeMounts:
+            - mountPath: /mnt/shared
+              name: shared-storage
+          resources:
+            limits:
+              memory: 16Mi
+              cpu: 10m
+            requests:
+              memory: 16Mi
   volumeClaimTemplates:
     - metadata:
         name: shared-storage