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

Jump to bottom

refactor: Remove explicit callings to garbage collect #55

Merged
merged 3 commits into from
Aug 6, 2024
Merged

refactor: Remove explicit callings to garbage collect #55

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
2302f2f
Remove explicit calling garbage collect
kthui Aug 2, 2024
5718785
Collect garbage once at the end of finalize()
kthui Aug 2, 2024
b4981a0
Apply logging suggestions from review
kthui Aug 2, 2024
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
13 changes: 6 additions & 7 deletions src/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -288,7 +288,6 @@ def response_loop(self):
if item is None:
break
response_sender, response, response_flag = item
del item
try:
response_sender.send(response, response_flag)
except Exception as e:
Expand All @@ -298,9 +297,6 @@ def response_loop(self):
finally:
if response_flag == pb_utils.TRITONSERVER_RESPONSE_COMPLETE_FINAL:
self.ongoing_request_count -= 1
del response_sender
if self.ongoing_request_count == 0:
gc.collect()

def create_response(self, vllm_output, prepend_input):
"""
Expand Down Expand Up @@ -447,9 +443,6 @@ async def generate(self, request):
finally:
if decrement_ongoing_request_count:
self.ongoing_request_count -= 1
del response_sender
if self.ongoing_request_count == 0:
gc.collect()

def verify_loras(self, request):
# We will check if the requested lora exists here, if not we will send a
Expand Down Expand Up @@ -527,3 +520,9 @@ def finalize(self):
if self._response_thread is not None:
self._response_thread.join()
self._response_thread = None

# When using parallel tensors, the stub process may not shutdown due to
# unreleased references, so manually run the garbage collector once.
self.logger.log_info("[vllm] Running Garbage Collector on finalize...")
gc.collect()
self.logger.log_info("[vllm] Garbage Collector on finalize... done")
Loading