fix: add resource quota evaluation timed out to transient (#12536)

argoproj · May 7, 2024 · 574fd3a · 574fd3a
1 parent 93e981d
commit 574fd3a
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 1 deletion.
diff --git a/util/errors/errors.go b/util/errors/errors.go
@@ -28,7 +28,7 @@ func IsTransientErr(err error) bool {
 		return false
 	}
 	err = argoerrs.Cause(err)
-	isTransient := isExceededQuotaErr(err) || apierr.IsTooManyRequests(err) || isResourceQuotaConflictErr(err) || isTransientNetworkErr(err) || apierr.IsServerTimeout(err) || apierr.IsServiceUnavailable(err) || matchTransientErrPattern(err) ||
+	isTransient := isExceededQuotaErr(err) || apierr.IsTooManyRequests(err) || isResourceQuotaConflictErr(err) || isResourceQuotaTimeoutErr(err) || isTransientNetworkErr(err) || apierr.IsServerTimeout(err) || apierr.IsServiceUnavailable(err) || matchTransientErrPattern(err) ||
 		errors.Is(err, NewErrTransient(""))
 	if isTransient {
 		log.Infof("Transient error: %v", err)
@@ -57,6 +57,10 @@ func isResourceQuotaConflictErr(err error) bool {
 	return apierr.IsConflict(err) && strings.Contains(err.Error(), "Operation cannot be fulfilled on resourcequota")
 }
 
+func isResourceQuotaTimeoutErr(err error) bool {
+	return apierr.IsInternalError(err) && strings.Contains(err.Error(), "resource quota evaluation timed out")
+}
+
 func isTransientNetworkErr(err error) bool {
 	switch err.(type) {
 	case *net.DNSError, *net.OpError, net.UnknownNetworkError:

diff --git a/util/errors/errors_test.go b/util/errors/errors_test.go
@@ -57,6 +57,10 @@ func TestIsTransientErr(t *testing.T) {
 		assert.False(t, IsTransientErr(apierr.NewConflict(schema.GroupResource{}, "", nil)))
 		assert.True(t, IsTransientErr(apierr.NewConflict(schema.GroupResource{Group: "v1", Resource: "resourcequotas"}, "", nil)))
 	})
+	t.Run("ResourceQuotaTimeoutErr", func(t *testing.T) {
+		assert.False(t, IsTransientErr(apierr.NewInternalError(errors.New(""))))
+		assert.True(t, IsTransientErr(apierr.NewInternalError(errors.New("resource quota evaluation timed out"))))
+	})
 	t.Run("ExceededQuotaErr", func(t *testing.T) {
 		assert.False(t, IsTransientErr(apierr.NewForbidden(schema.GroupResource{}, "", nil)))
 		assert.True(t, IsTransientErr(apierr.NewForbidden(schema.GroupResource{Group: "v1", Resource: "pods"}, "", errors.New("exceeded quota"))))