Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add cancellation into response statistics #17

Merged
merged 1 commit into from
Feb 23, 2024
Merged

Add cancellation into response statistics #17

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
55 changes: 44 additions & 11 deletions src/square.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,12 +98,13 @@ namespace triton { namespace backend { namespace square {
//
class ModelParameters {
public:
enum DelayType { INFER, OUTPUT };
enum DelayType { INFER, OUTPUT, CANCEL };
enum InferResultType { SUCCESS, FAIL, EMPTY };

ModelParameters()
: custom_infer_delay_ns_(0), custom_output_delay_ns_(0),
custom_fail_count_(0), custom_empty_count_(0)
custom_cancel_delay_ns_(0), custom_fail_count_(0),
custom_empty_count_(0)
{
}
ModelParameters(common::TritonJson::Value& model_config_);
Expand All @@ -120,6 +121,7 @@ class ModelParameters {

size_t custom_infer_delay_ns_;
size_t custom_output_delay_ns_;
size_t custom_cancel_delay_ns_;
size_t custom_fail_count_;
size_t custom_empty_count_;
};
Expand All @@ -135,6 +137,8 @@ ModelParameters::ModelParameters(common::TritonJson::Value& model_config_)
parameters_json, "CUSTOM_INFER_DELAY_NS", &custom_infer_delay_ns_);
ReadParameter(
parameters_json, "CUSTOM_OUTPUT_DELAY_NS", &custom_output_delay_ns_);
ReadParameter(
parameters_json, "CUSTOM_CANCEL_DELAY_NS", &custom_cancel_delay_ns_);
ReadParameter(parameters_json, "CUSTOM_FAIL_COUNT", &custom_fail_count_);
ReadParameter(parameters_json, "CUSTOM_EMPTY_COUNT", &custom_empty_count_);
}
Expand All @@ -148,6 +152,8 @@ ModelParameters::Sleep(DelayType delay_type) const
Sleep(custom_infer_delay_ns_);
} else if (delay_type == DelayType::OUTPUT) {
Sleep(custom_output_delay_ns_);
} else if (delay_type == DelayType::CANCEL) {
Sleep(custom_cancel_delay_ns_);
}
}

Expand Down Expand Up @@ -526,18 +532,33 @@ ModelInstanceState::RequestThread(
uint64_t response_start_ns;
SET_TIMESTAMP(response_start_ns);

// Simulate compute delay, if provided.
model_state_->get_model_parameters().Sleep(
ModelParameters::DelayType::INFER);
// Check if the request is cancelled.
bool is_cancelled;
RESPOND_FACTORY_AND_RETURN_IF_ERROR(
factory.get(),
TRITONBACKEND_ResponseFactoryIsCancelled(factory.get(), &is_cancelled));

// Simulate compute or cancellation.
if (is_cancelled) {
// Simulate cancel clean-up delay, if provided.
model_state_->get_model_parameters().Sleep(
ModelParameters::DelayType::CANCEL);
} else {
// Simulate compute delay, if provided.
model_state_->get_model_parameters().Sleep(
ModelParameters::DelayType::INFER);
}

// Result type of the simulated inference.
// Result type of the simulated inference, ignore if cancelled.
ModelParameters::InferResultType result_type =
model_state_->get_model_parameters().InferResult(e, element_count);

// Populate 'compute_output_start_ns' and 'response' if not empty result.
// Simulate compute output based on result type and cancellation, and
// populate 'compute_output_start_ns' and 'response'.
uint64_t compute_output_start_ns = 0;
TRITONBACKEND_Response* response = nullptr;
if (result_type != ModelParameters::InferResultType::EMPTY) {
if (!is_cancelled &&
result_type != ModelParameters::InferResultType::EMPTY) {
// Timestamp at start of outputting compute tensors.
SET_TIMESTAMP(compute_output_start_ns);

Expand Down Expand Up @@ -584,13 +605,15 @@ ModelInstanceState::RequestThread(

// Set error for simulated failure.
TRITONSERVER_Error* error = nullptr;
if (result_type == ModelParameters::InferResultType::FAIL) {
if (is_cancelled) {
error = TRITONSERVER_ErrorNew(TRITONSERVER_ERROR_CANCELLED, "cancelled");
} else if (result_type == ModelParameters::InferResultType::FAIL) {
error = TRITONSERVER_ErrorNew(
TRITONSERVER_ERROR_UNKNOWN, "simulated failure");
}

// Send response if not empty.
if (result_type != ModelParameters::InferResultType::EMPTY) {
// Send response, if any.
if (response != nullptr) {
LOG_IF_ERROR(
TRITONBACKEND_ResponseSend(response, 0 /* flags */, error),
"failed sending response");
Expand All @@ -612,6 +635,16 @@ ModelInstanceState::RequestThread(
(std::string("sent response ") + std::to_string(e + 1) + " of " +
std::to_string(element_count))
.c_str());

// If cancelled, stop sending remaining responses, if any.
if (is_cancelled) {
LOG_MESSAGE(
TRITONSERVER_LOG_INFO,
(std::string("cancelled at response ") + std::to_string(e + 1) +
" of " + std::to_string(element_count))
.c_str());
break;
}
}

// Add some logging for the case where IN was size 0 and so no
Expand Down
Loading