@onlineapps/conn-orch-orchestrator 1.0.45 → 1.0.47

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@onlineapps/conn-orch-orchestrator",
-  "version": "1.0.45",
+  "version": "1.0.47",
   "description": "Workflow orchestration connector for OA Drive - handles message routing and workflow execution",
   "main": "src/index.js",
   "scripts": {

package/src/WorkflowOrchestrator.js

@@ -253,33 +253,9 @@ class WorkflowOrchestrator {
         // Ignore - use default 0
       }
 
-      // IMPORTANT: Publish progress event for FAILED step (so it appears in trace)
-      // This ensures every cookbook step is visible in monitoring, even failed ones
-      const { publishToMonitoringWorkflow } = require('@onlineapps/mq-client-core').monitoring;
-      try {
-        await publishToMonitoringWorkflow(this.mqClient, {
-          event_type: 'progress',
-          workflow_id: workflow_id,
-          service_name: serviceName,
-          step_index: failedStepIndex,
-          step_id: current_step,
-          cookbook: cookbookDef,
-          context: context,
-          output: null,
-          error: {
-            message: error.message,
-            attempt: attemptNumber
-          },
-          status: 'failed',
-          timestamp: new Date().toISOString()
-        }, this.logger, { workflow_id, step_id: current_step });
-      } catch (monitoringError) {
-        // Don't fail workflow if monitoring publish fails
-        this.logger.warn('Failed to publish failed step progress to monitoring', {
-          workflow_id,
-          error: monitoringError.message
-        });
-      }
+      // NOTE: We do NOT publish workflow_progress on every retry attempt.
+      // Instead, we publish ONE progress event at the end (either success or final failure with DLQ entry).
+      // This keeps the trace clean - one entry per step, with complete _dlq_history in context.
 
       // Add attempt to DLQ history
       const attemptRecord = {
@@ -368,6 +344,7 @@ class WorkflowOrchestrator {
           error: error.stack
         });
         
+        // Publish to workflow.failed (DLQ queue - message stays until manual requeue/discard)
         await this.mqClient.publish('workflow.failed', {
           workflow_id,
           current_step,
@@ -382,6 +359,29 @@ class WorkflowOrchestrator {
           failed_at: new Date().toISOString()
         });
         
+        // ALSO publish to monitoring.workflow so dashboard shows the DLQ entry
+        // This is separate from the DLQ message - it's just for visibility
+        try {
+          await publishToMonitoringWorkflow(this.mqClient, {
+            event_type: 'failed',
+            workflow_id: workflow_id,
+            service_name: serviceName,
+            step_index: failedStepIndex,
+            step_id: current_step,
+            cookbook: cookbookDef,
+            context: finalContext,
+            error: { message: error.message },
+            status: 'failed',
+            timestamp: new Date().toISOString()
+          }, this.logger, { workflow_id, step_id: current_step });
+        } catch (monitoringError) {
+          // Don't fail if monitoring publish fails
+          this.logger.warn('[WorkflowOrchestrator] Failed to publish DLQ entry to monitoring', {
+            workflow_id,
+            error: monitoringError.message
+          });
+        }
+        
         console.log(`[WorkflowOrchestrator] [DLQ] ✓ Successfully published workflow.failed for ${workflow_id}`);
         this.logger.info(`[WorkflowOrchestrator] [DLQ] ✓ Published workflow.failed: ${workflow_id}`);
       } catch (publishError) {