postStart hook commands timeout

apis/controller/v1alpha1/devworkspaceoperatorconfig_types.go

@@ -189,6 +189,13 @@ type WorkspaceConfig struct {
 	RuntimeClassName *string `json:"runtimeClassName,omitempty"`
 	// CleanupCronJobConfig defines configuration options for a cron job that automatically cleans up stale DevWorkspaces.
 	CleanupCronJob *CleanupCronJobConfig `json:"cleanupCronJob,omitempty"`
+	// PostStartTimeout defines the maximum duration the PostStart hook can run
+	// before it is automatically failed. This timeout is used for the postStart lifecycle hook
+	// that is used to run commands in the workspace container. The timeout is specified in seconds.
+	// Duration should be specified in a format parseable by Go's time package, e.g. "20s", "2m".
+	// If not specified or "0", the timeout is disabled.
+	// +kubebuilder:validation:Optional
+	PostStartTimeout string `json:"postStartTimeout,omitempty"`
 }
 
 type WebhookConfig struct {

controllers/workspace/devworkspace_controller.go

@@ -327,7 +327,9 @@ func (r *DevWorkspaceReconciler) Reconcile(ctx context.Context, req ctrl.Request
 		&workspace.Spec.Template,
 		workspace.Config.Workspace.ContainerSecurityContext,
 		workspace.Config.Workspace.ImagePullPolicy,
-		workspace.Config.Workspace.DefaultContainerResources)
+		workspace.Config.Workspace.DefaultContainerResources,
+		workspace.Config.Workspace.PostStartTimeout,
+	)
 	if err != nil {
 		return r.failWorkspace(workspace, fmt.Sprintf("Error processing devfile: %s", err), metrics.ReasonBadRequest, reqLogger, &reconcileStatus), nil
 	}

pkg/config/sync.go

@@ -431,6 +431,10 @@ func mergeConfig(from, to *controller.OperatorConfiguration) {
 				to.Workspace.CleanupCronJob.Schedule = from.Workspace.CleanupCronJob.Schedule
 			}
 		}
+
+		if from.Workspace.PostStartTimeout != "" {
+			to.Workspace.PostStartTimeout = from.Workspace.PostStartTimeout
+		}
 	}
 }
 
@@ -601,6 +605,9 @@ func GetCurrentConfigString(currConfig *controller.OperatorConfiguration) string
 		if workspace.IdleTimeout != defaultConfig.Workspace.IdleTimeout {
 			config = append(config, fmt.Sprintf("workspace.idleTimeout=%s", workspace.IdleTimeout))
 		}
+		if workspace.PostStartTimeout != defaultConfig.Workspace.PostStartTimeout {
+			config = append(config, fmt.Sprintf("workspace.postStartTimeout=%s", workspace.PostStartTimeout))
+		}
 		if workspace.ProgressTimeout != "" && workspace.ProgressTimeout != defaultConfig.Workspace.ProgressTimeout {
 			config = append(config, fmt.Sprintf("workspace.progressTimeout=%s", workspace.ProgressTimeout))
 		}

pkg/library/container/container.go

@@ -45,7 +45,7 @@ import (
 // rewritten as Volumes are added to PodAdditions, in order to support e.g. using one PVC to hold all volumes
 //
 // Note: Requires DevWorkspace to be flattened (i.e. the DevWorkspace contains no Parent or Components of type Plugin)
-func GetKubeContainersFromDevfile(workspace *dw.DevWorkspaceTemplateSpec, securityContext *corev1.SecurityContext, pullPolicy string, defaultResources *corev1.ResourceRequirements) (*v1alpha1.PodAdditions, error) {
+func GetKubeContainersFromDevfile(workspace *dw.DevWorkspaceTemplateSpec, securityContext *corev1.SecurityContext, pullPolicy string, defaultResources *corev1.ResourceRequirements, postStartTimeout string) (*v1alpha1.PodAdditions, error) {
 	if !flatten.DevWorkspaceIsFlattened(workspace, nil) {
 		return nil, fmt.Errorf("devfile is not flattened")
 	}
@@ -77,7 +77,7 @@ func GetKubeContainersFromDevfile(workspace *dw.DevWorkspaceTemplateSpec, securi
 		podAdditions.Containers = append(podAdditions.Containers, *k8sContainer)
 	}
 
-	if err := lifecycle.AddPostStartLifecycleHooks(workspace, podAdditions.Containers); err != nil {
+	if err := lifecycle.AddPostStartLifecycleHooks(workspace, podAdditions.Containers, postStartTimeout); err != nil {
 		return nil, err
 	}
 

pkg/library/container/container_test.go

@@ -87,7 +87,7 @@ func TestGetKubeContainersFromDevfile(t *testing.T) {
 		t.Run(tt.Name, func(t *testing.T) {
 			// sanity check that file is read correctly.
 			assert.True(t, len(tt.Input.Components) > 0, "Input defines no components")
-			gotPodAdditions, err := GetKubeContainersFromDevfile(tt.Input, nil, testImagePullPolicy, defaultResources)
+			gotPodAdditions, err := GetKubeContainersFromDevfile(tt.Input, nil, testImagePullPolicy, defaultResources, "")
 			if tt.Output.ErrRegexp != nil && assert.Error(t, err) {
 				assert.Regexp(t, *tt.Output.ErrRegexp, err.Error(), "Error message should match")
 			} else {

pkg/library/lifecycle/poststart.go

@@ -16,17 +16,32 @@ package lifecycle
 import (
 	"fmt"
 	"strings"
+	"time"
 
 	dw "github.com/devfile/api/v2/pkg/apis/workspaces/v1alpha2"
 	corev1 "k8s.io/api/core/v1"
+	"sigs.k8s.io/controller-runtime/pkg/log"
 )
 
-const redirectOutputFmt = `{
+const (
+	// `tee` both stdout and stderr to files and to the main output streams.
+	redirectOutputFmt = `{
+  # This script block ensures its exit code is preserved
+  # while its stdout and stderr are tee'd.
+  _script_to_run() {
+    %s # This will be replaced by scriptWithTimeout
+  }
+  _script_to_run
+} 1> >(tee -a "/tmp/poststart-stdout.txt") 2> >(tee -a "/tmp/poststart-stderr.txt" >&2)
+`
+
+	noTimeoutRedirectOutputFmt = `{
 %s
 } 1>/tmp/poststart-stdout.txt 2>/tmp/poststart-stderr.txt
 `
+)
 
-func AddPostStartLifecycleHooks(wksp *dw.DevWorkspaceTemplateSpec, containers []corev1.Container) error {
+func AddPostStartLifecycleHooks(wksp *dw.DevWorkspaceTemplateSpec, containers []corev1.Container, postStartTimeout string) error {
 	if wksp.Events == nil || len(wksp.Events.PostStart) == 0 {
 		return nil
 	}
@@ -54,7 +69,7 @@ func AddPostStartLifecycleHooks(wksp *dw.DevWorkspaceTemplateSpec, containers []
 			return fmt.Errorf("failed to process postStart event %s: %w", commands[0].Id, err)
 		}
 
-		postStartHandler, err := processCommandsForPostStart(commands)
+		postStartHandler, err := processCommandsForPostStart(commands, postStartTimeout)
 		if err != nil {
 			return fmt.Errorf("failed to process postStart event %s: %w", commands[0].Id, err)
 		}
@@ -68,7 +83,41 @@ func AddPostStartLifecycleHooks(wksp *dw.DevWorkspaceTemplateSpec, containers []
 	return nil
 }
 
-// processCommandsForPostStart builds a lifecycle handler that runs the provided command(s)
+// processCommandsForPostStart processes a list of DevWorkspace commands
+// and generates a corev1.LifecycleHandler for the PostStart lifecycle hook.
+func processCommandsForPostStart(commands []dw.Command, postStartTimeout string) (*corev1.LifecycleHandler, error) {
+	if postStartTimeout == "" {
+		// use the fallback if no timeout propagated
+		return processCommandsWithoutTimeoutFallback(commands)
+	}
+
+	originalUserScript, err := buildUserScript(commands)
+	if err != nil {
+		return nil, fmt.Errorf("failed to build aggregated user script: %w", err)
+	}
+
+	// The user script needs 'set -e' to ensure it exits on error.
+	// This script is then passed to `sh -c '...'`, so single quotes within it must be escaped.
+	scriptToExecute := "set -e\n" + originalUserScript
+	escapedUserScriptForTimeoutWrapper := strings.ReplaceAll(scriptToExecute, "'", `'\''`)
+
+	fullScriptWithTimeout := generateScriptWithTimeout(escapedUserScriptForTimeoutWrapper, postStartTimeout)
+
+	finalScriptForHook := fmt.Sprintf(redirectOutputFmt, fullScriptWithTimeout)
+
+	handler := &corev1.LifecycleHandler{
+		Exec: &corev1.ExecAction{
+			Command: []string{
+				"/bin/sh",
+				"-c",
+				finalScriptForHook,
+			},
+		},
+	}
+	return handler, nil
+}
+
+// processCommandsWithoutTimeoutFallback builds a lifecycle handler that runs the provided command(s)
 // The command has the format
 //
 // exec:
@@ -79,7 +128,7 @@ func AddPostStartLifecycleHooks(wksp *dw.DevWorkspaceTemplateSpec, containers []
 //	  - |
 //	    cd <workingDir>
 //	    <commandline>
-func processCommandsForPostStart(commands []dw.Command) (*corev1.LifecycleHandler, error) {
+func processCommandsWithoutTimeoutFallback(commands []dw.Command) (*corev1.LifecycleHandler, error) {
 	var dwCommands []string
 	for _, command := range commands {
 		execCmd := command.Exec
@@ -99,9 +148,96 @@ func processCommandsForPostStart(commands []dw.Command) (*corev1.LifecycleHandle
 			Command: []string{
 				"/bin/sh",
 				"-c",
-				fmt.Sprintf(redirectOutputFmt, joinedCommands),
+				fmt.Sprintf(noTimeoutRedirectOutputFmt, joinedCommands),
 			},
 		},
 	}
 	return handler, nil
 }
+
+// buildUserScript takes a list of DevWorkspace commands and constructs a single
+// shell script string that executes them sequentially.
+func buildUserScript(commands []dw.Command) (string, error) {
+	var commandScriptLines []string
+	for _, command := range commands {
+		execCmd := command.Exec
+		if execCmd == nil {
+			// Should be caught by earlier validation, but good to be safe
+			return "", fmt.Errorf("exec command is nil for command ID %s", command.Id)
+		}
+		var singleCommandParts []string
+		for _, envVar := range execCmd.Env {
+			singleCommandParts = append(singleCommandParts, fmt.Sprintf("export %s=%q", envVar.Name, envVar.Value))
+		}
+
+		if execCmd.WorkingDir != "" {
+			singleCommandParts = append(singleCommandParts, fmt.Sprintf("cd %q", execCmd.WorkingDir))
+		}
+		if execCmd.CommandLine != "" {
+			singleCommandParts = append(singleCommandParts, execCmd.CommandLine)
+		}
+		if len(singleCommandParts) > 0 {
+			commandScriptLines = append(commandScriptLines, strings.Join(singleCommandParts, " && "))
+		}
+	}
+	return strings.Join(commandScriptLines, "\n"), nil
+}
+
+// generateScriptWithTimeout wraps a given user script with timeout logic,
+// environment variable exports, and specific exit code handling.
+// The killAfterDurationSeconds is hardcoded to 5s within this generated script.
+// It conditionally prefixes the user script with the timeout command if available.
+func generateScriptWithTimeout(escapedUserScript string, postStartTimeout string) string {
+	// Convert `postStartTimeout` into the `timeout` format
+	var timeoutSeconds int64
+	if postStartTimeout != "" && postStartTimeout != "0" {
+		duration, err := time.ParseDuration(postStartTimeout)
+		if err != nil {
+			log.Log.Error(err, "Could not parse post-start timeout, disabling timeout", "value", postStartTimeout)
+			timeoutSeconds = 0
+		} else {
+			timeoutSeconds = int64(duration.Seconds())
+		}
+	}
+
+	return fmt.Sprintf(`
+export POSTSTART_TIMEOUT_DURATION="%d"
+export POSTSTART_KILL_AFTER_DURATION="5"
+
+_TIMEOUT_COMMAND_PART=""
+_WAS_TIMEOUT_USED="false" # Use strings "true" or "false" for shell boolean
+
+if command -v timeout >/dev/null 2>&1; then
+  echo "[postStart hook] Executing commands with timeout: ${POSTSTART_TIMEOUT_DURATION} seconds, kill after: ${POSTSTART_KILL_AFTER_DURATION} seconds" >&2
+  _TIMEOUT_COMMAND_PART="timeout --preserve-status --kill-after=${POSTSTART_KILL_AFTER_DURATION} ${POSTSTART_TIMEOUT_DURATION}"
+  _WAS_TIMEOUT_USED="true"
+else
+  echo "[postStart hook] WARNING: 'timeout' utility not found. Executing commands without timeout." >&2
+fi
+
+# Execute the user's script
+${_TIMEOUT_COMMAND_PART} /bin/sh -c '%s'
+exit_code=$?
+
+# Check the exit code based on whether timeout was attempted
+if [ "$_WAS_TIMEOUT_USED" = "true" ]; then
+  if [ $exit_code -eq 143 ]; then # 128 + 15 (SIGTERM)
+    echo "[postStart hook] Commands terminated by SIGTERM (likely timed out after ${POSTSTART_TIMEOUT_DURATION}s). Exit code 143." >&2
+  elif [ $exit_code -eq 137 ]; then # 128 + 9 (SIGKILL)
+    echo "[postStart hook] Commands forcefully killed by SIGKILL (likely after --kill-after ${POSTSTART_KILL_AFTER_DURATION}s expired). Exit code 137." >&2
+  elif [ $exit_code -ne 0 ]; then # Catches any other non-zero exit code
+    echo "[postStart hook] Commands failed with exit code $exit_code." >&2
+  else
+    echo "[postStart hook] Commands completed successfully within the time limit." >&2
+  fi
+else
+  if [ $exit_code -ne 0 ]; then
+    echo "[postStart hook] Commands failed with exit code $exit_code (no timeout)." >&2
+  else
+    echo "[postStart hook] Commands completed successfully (no timeout)." >&2
+  fi
+fi
+
+exit $exit_code
+`, timeoutSeconds, escapedUserScript)
+}