[Feature][Docs] Adjust prefill release & expose load metrics (#6884)

2026-04-24 01:29:57 +08:00 · 2026-03-17 15:23:13 +08:00
parent daaf498213
commit b61731bb96
20 changed files with 576 additions and 59 deletions
@@ -14,9 +14,8 @@ scheduler:
  hit-ratio-weight: 1.0
  load-balance-weight: 0.05
  cache-block-size: 4
-  tokenizer-url: "http://0.0.0.0:8098" # optional tokenizer service endpoint
-  tokenizer-timeout-secs: 2
  waiting-weight: 10
+  stats-interval-secs: 5 # interval in seconds for periodic stats logging (running requests, cache hit rate)

 manager:
  health-failure-threshold: 3