minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

Annotate dead max_running_requests_under_SLO (#25441)

Browse Source
This commit is contained in:
fzyzcjy
2026-05-16 09:22:13 +08:00
committed by GitHub
parent c8e5cf5768
commit 72eb231d28
2 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/managers/scheduler.py
+1
View File
@@ -784,6 +784,7 @@ class Scheduler(
if self.enable_metrics: if self.enable_metrics:
self.metrics_collector.emit_constants( self.metrics_collector.emit_constants(
max_total_num_tokens=self.max_total_num_tokens, max_total_num_tokens=self.max_total_num_tokens,
# TODO: max_running_requests_under_SLO has no setter — dead chain.
max_running_requests_under_SLO=getattr( max_running_requests_under_SLO=getattr(
self, "max_running_requests_under_SLO", None self, "max_running_requests_under_SLO", None
), ),
python/sglang/srt/observability/scheduler_metrics_mixin.py
+1
View File
@@ -954,6 +954,7 @@ class SchedulerMetricsMixin:
if self.disaggregation_mode == DisaggregationMode.PREFILL: if self.disaggregation_mode == DisaggregationMode.PREFILL:
self.stats.utilization = -1 self.stats.utilization = -1
else: else:
# TODO: max_running_requests_under_SLO has no setter — sglang:utilization stuck at 0 (regressed #22713).
max_under_slo = getattr(self, "max_running_requests_under_SLO", None) max_under_slo = getattr(self, "max_running_requests_under_SLO", None)
if max_under_slo is not None and max_under_slo > 0: if max_under_slo is not None and max_under_slo > 0:
self.stats.utilization = max( self.stats.utilization = max(