Increase size of timeout variable to resolve error relating to large …

…timeouts
triton-inference-server · Dec 1, 2023 · a008a7a · a008a7a
1 parent 34915be
commit a008a7a
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 6 deletions.
diff --git a/src/infer_request.cc b/src/infer_request.cc
@@ -42,7 +42,7 @@ InferRequest::InferRequest(
     const std::vector<std::shared_ptr<PbTensor>>& inputs,
     const std::set<std::string>& requested_output_names,
     const std::string& model_name, const int64_t model_version,
-    const std::string& parameters, const uint32_t flags, const int32_t timeout,
+    const std::string& parameters, const uint32_t flags, const uint64_t timeout,
     const intptr_t response_factory_address, const intptr_t request_address,
     const PreferredMemory& preferred_memory, const InferenceTrace& trace)
     : request_id_(request_id), correlation_id_(correlation_id), inputs_(inputs),
@@ -145,7 +145,7 @@ InferRequest::ShmHandle()
   return shm_handle_;
 }
 
-int32_t
+uint64_t
 InferRequest::Timeout()
 {
   return timeout_;

diff --git a/src/infer_request.h b/src/infer_request.h
@@ -84,7 +84,7 @@ class InferRequest {
       const std::set<std::string>& requested_output_names,
       const std::string& model_name, const int64_t model_version,
       const std::string& parameters, const uint32_t flags = 0,
-      const int32_t timeout = 0, const intptr_t response_factory_address = 0,
+      const uint64_t timeout = 0, const intptr_t response_factory_address = 0,
       const intptr_t request_address = 0,
       const PreferredMemory& preferred_memory =
           PreferredMemory(PreferredMemory::DEFAULT, 0),
@@ -100,7 +100,7 @@ class InferRequest {
   void SetFlags(uint32_t flags);
   const std::set<std::string>& RequestedOutputNames();
   bi::managed_external_buffer::handle_t ShmHandle();
-  int32_t Timeout();
+  uint64_t Timeout();
   bool IsDecoupled();
   void SetIsDecoupled(const bool is_decoupled);
   PreferredMemory& GetPreferredMemory();
@@ -158,7 +158,7 @@ class InferRequest {
   int64_t model_version_;
   std::string parameters_;
   uint32_t flags_;
-  int32_t timeout_;
+  uint64_t timeout_;
   intptr_t response_factory_address_;
   intptr_t request_address_;
   bool is_decoupled_;

diff --git a/src/pb_stub.cc b/src/pb_stub.cc
@@ -1581,7 +1581,7 @@ PYBIND11_EMBEDDED_MODULE(c_python_backend_utils, module)
                       const std::vector<std::string>& requested_output_names,
                       const std::string& model_name,
                       const int64_t model_version, const uint32_t flags,
-                      const int32_t timeout,
+                      const uint64_t timeout,
                       const PreferredMemory& preferred_memory,
                       const InferenceTrace& trace,
                       const py::object& parameters_) {