Use better value in timeout test L0_sequence_batcher (#5716)

* Use better value in timeout test L0_sequence_batcher * Format
triton-inference-server · Apr 28, 2023 · b795c01 · b795c01
1 parent e6eda20
commit b795c01
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 5 deletions.
diff --git a/qa/L0_sequence_batcher/request_timeout_models/custom_sequence_int32_timeout/config.pbtxt b/qa/L0_sequence_batcher/request_timeout_models/custom_sequence_int32_timeout/config.pbtxt
@@ -57,6 +57,6 @@ sequence_batching {
 parameters [
   {
     key: "execute_delay_ms"
-    value: { string_value: "3000" }
+    value: { string_value: "5000" }
   }
-]
+]
diff --git a/qa/L0_sequence_batcher/sequence_batcher_test.py b/qa/L0_sequence_batcher/sequence_batcher_test.py
@@ -2874,7 +2874,7 @@ def setUp(self):
     def send_sequence_with_timeout(self,
                                    seq_id,
                                    callback,
-                                   timeout_us=2000000,
+                                   timeout_us=3000000,
                                    request_pause_sec=0):
         with grpcclient.InferenceServerClient(
                 self.server_address_) as triton_client:
@@ -2903,8 +2903,8 @@ def test_request_timeout(self):
         # expect the timeout will only be expired on backlog sequence and reject
         # all requests of the sequence once expired.
         # Sending two sequences while the model can only process one sequence
-        # at a time. Each model execution takes 3 second and all requests have
-        # 2 second timeout, so the second sequence will be rejected.
+        # at a time. Each model execution takes 5 second and all requests have
+        # 3 second timeout, so the second sequence will be rejected.
 
         # correlation ID is 1-index
         seq1_res = []