fix: stop infinite self-update retry loop on failure

agent/internal/agent/agent.go

@@ -28,8 +28,10 @@ type Agent struct {
 	vectorVersion  string
 	deploymentMode string
 
-	mu            sync.Mutex
-	sampleResults []client.SampleResultMsg
+	mu                  sync.Mutex
+	sampleResults       []client.SampleResultMsg
+	failedUpdateVersion string // skip retries for this version
+	updateError         string // report failure to server
 }
 
 func New(cfg *config.Config) (*Agent, error) {
@@ -146,10 +148,16 @@ func (a *Agent) handlePendingAction(action *client.PendingAction) {
 		if a.deploymentMode == "DOCKER" {
 			slog.Warn("received update command but running in Docker — ignoring",
 				"targetVersion", action.TargetVersion)
+			a.updateError = "running in Docker"
 			return
 		}
+		if action.TargetVersion == a.failedUpdateVersion {
+			return // already failed for this version, don't retry
+		}
 		if err := a.handleSelfUpdate(action); err != nil {
 			slog.Error("self-update failed", "error", err)
+			a.failedUpdateVersion = action.TargetVersion
+			a.updateError = err.Error()
 		}
 	default:
 		slog.Warn("unknown pending action type", "type", action.Type)
@@ -164,6 +172,10 @@ func (a *Agent) sendHeartbeat() {
 	a.mu.Unlock()
 
 	hb := buildHeartbeat(a.supervisor, a.vectorVersion, a.deploymentMode, results)
+	updateErr := a.updateError
+	if updateErr != "" {
+		hb.UpdateError = updateErr
+	}
 	if err := a.client.SendHeartbeat(hb); err != nil {
 		slog.Warn("heartbeat error", "error", err)
 		// Put results back so they retry on the next heartbeat
@@ -173,6 +185,7 @@ func (a *Agent) sendHeartbeat() {
 			a.mu.Unlock()
 		}
 	} else {
+		a.updateError = "" // clear only after successful delivery
 		slog.Debug("heartbeat sent", "pipelines", len(hb.Pipelines), "sampleResults", len(results))
 	}
 }

agent/internal/client/client.go

@@ -202,6 +202,7 @@ type HeartbeatRequest struct {
 	VectorVersion  string            `json:"vectorVersion,omitempty"`
 	DeploymentMode string            `json:"deploymentMode,omitempty"`
 	SampleResults  []SampleResultMsg `json:"sampleResults,omitempty"`
+	UpdateError    string            `json:"updateError,omitempty"`
 }
 
 // SampleRequestMsg is received from the server via config poll.

src/app/api/agent/heartbeat/route.ts

@@ -66,6 +66,7 @@ const heartbeatSchema = z.object({
     schema: z.array(z.object({ path: z.string(), type: z.string(), sample: z.string() })).optional(),
     error: z.string().optional(),
   })).optional(),
+  updateError: z.string().max(500).optional(),
 });
 
 let lastCleanup = 0;
@@ -113,7 +114,7 @@ export async function POST(request: Request) {
       );
     }
 
-    const { pipelines: rawPipelines, hostMetrics, agentVersion, vectorVersion, deploymentMode } = parsed.data;
+    const { pipelines: rawPipelines, hostMetrics, agentVersion, vectorVersion, deploymentMode, updateError } = parsed.data;
 
     // Validate pipeline ownership: only accept pipelines belonging to this agent's environment
     const validPipelineIds = new Set(
@@ -126,9 +127,13 @@ export async function POST(request: Request) {
 
     const now = new Date();
 
-    // Check if pendingAction should be cleared (agent has updated to target version)
+    // Check if pendingAction should be cleared
     let clearPendingAction = false;
-    if (agentVersion) {
+    if (updateError) {
+      // Agent reported update failure — clear to stop retry loop
+      clearPendingAction = true;
+      console.warn("Agent update failed, clearing pending action:", agent.nodeId, updateError);
+    } else if (agentVersion) {
       const currentNode = await prisma.vectorNode.findUnique({
         where: { id: agent.nodeId },
         select: { pendingAction: true },