Skip to content

[V1][Bugfix] Do not reset prefix caching metrics #14235

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

Merged
merged 1 commit into from
Mar 5, 2025
Merged

[V1][Bugfix] Do not reset prefix caching metrics #14235

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
6 changes: 3 additions & 3 deletions vllm/v1/metrics/loggers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,6 +34,9 @@ class LoggingStatLogger(StatLoggerBase):
def __init__(self):
self._reset(time.monotonic())
self.last_scheduler_stats = SchedulerStats()
# Prefix cache metrics. This cannot be reset.
# TODO: Make the interval configurable.
self.prefix_caching_metrics = PrefixCachingMetrics()

def _reset(self, now):
self.last_log_time = now
Expand All @@ -42,9 +45,6 @@ def _reset(self, now):
self.num_prompt_tokens: list[int] = []
self.num_generation_tokens: list[int] = []

# Prefix cache metrics. TODO: Make the interval configurable.
self.prefix_caching_metrics = PrefixCachingMetrics()

def _track_iteration_stats(self, iteration_stats: IterationStats):
# Save tracked stats for token counters.
self.num_prompt_tokens.append(iteration_stats.num_prompt_tokens)
Expand Down