fix: fix the traceid-evaluationid mismatch

Author: AAgnihotry

src/uipath/_cli/_evals/_runtime.py

@@ -210,7 +210,12 @@ def _upsert():
     def on_start(
         self, span: Span, parent_context: context_api.Context | None = None
     ) -> None:
-        """Called when span starts - upsert with RUNNING status (non-blocking)."""
+        """Called when span starts - upsert with RUNNING status (non-blocking).
+
+        Parent spans (eval_set_run, evaluation) are created before their children
+        in the code flow, so their async upsert tasks are submitted first, ensuring
+        correct ordering at the backend without blocking evaluation execution.
+        """
         # Only track evaluation-related spans
         if span.attributes and self._is_eval_span(span):
             self._upsert_span_async(span, status_override=self.span_status.RUNNING)

src/uipath/_cli/cli_eval.py

@@ -207,6 +207,10 @@ async def execute_eval():
                     LlmOpsHttpExporter() if should_register_progress_reporter else None
                 )
 
+                # Set trace_id early if eval_set_run_id is already known
+                if live_tracking_exporter and eval_context.eval_set_run_id:
+                    live_tracking_exporter.trace_id = eval_context.eval_set_run_id
+
                 if should_register_progress_reporter:
                     progress_reporter = StudioWebProgressReporter(
                         live_tracking_exporter