feat: include container name in error message. Fixes #10007

docs/upgrading.md

@@ -104,6 +104,10 @@ To disable this set `metricsConfig.secure` to `false`.
 When returning a map or array in an expression, you would get a Golang representation.
 This now returns plain JSON.
 
+### Container name in error messages
+
+Error messages are prefixed with container name, you may need to adjust your `lastRetry.message` expressions or `TRANSIENT_ERROR_PATTERN` variable. See [Conditional retries](retries.md)
+
 ### `ARGO_TEMPLATE` removed from main container
 
 The environment variable `ARGO_TEMPLATE` which is an internal implementation detail is no longer available inside the `main` container of your workflow pods.

test/e2e/cli_test.go

@@ -894,7 +894,7 @@ func (s *CLISuite) TestWorkflowRetryWithRecreatedPVC() {
 			assert.Equal(t, wfv1.NodeFailed, status.Nodes.FindByDisplayName("print").Phase)
 			// This step is failed intentionally to allow retry. The error message is not related to PVC that is deleted
 			// previously since it is re-created during retry.
-			assert.Equal(t, "Error (exit code 1)", status.Nodes.FindByDisplayName("print").Message)
+			assert.Equal(t, "main: Error (exit code 1)", status.Nodes.FindByDisplayName("print").Message)
 		})
 }
 

workflow/controller/operator.go

@@ -1406,7 +1406,7 @@ func (woc *wfOperationCtx) assessNodeStatus(ctx context.Context, pod *apiv1.Pod,
 			woc.markNodePhase(ctrNodeName, wfv1.NodeRunning)
 		case c.State.Terminated != nil:
 			exitCode := int(c.State.Terminated.ExitCode)
-			message := fmt.Sprintf("%s (exit code %d): %s", c.State.Terminated.Reason, exitCode, c.State.Terminated.Message)
+			message := fmt.Sprintf("%s: %s (exit code %d): %s", c.Name, c.State.Terminated.Reason, exitCode, c.State.Terminated.Message)
 			switch exitCode {
 			case 0:
 				woc.markNodePhase(ctrNodeName, wfv1.NodeSucceeded)
@@ -1629,6 +1629,7 @@ func (woc *wfOperationCtx) inferFailedReason(pod *apiv1.Pod, tmpl *wfv1.Template
 		if t.Message != "" {
 			msg = fmt.Sprintf("%s: %s", msg, t.Message)
 		}
+		msg = fmt.Sprintf("%s: %s", ctr.Name, msg)
 
 		switch {
 		case ctr.Name == common.InitContainerName: