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

Demote per-request load-back log to debug

Browse Source 
init_load_back logged 'loading back N tokens for node M' at INFO for
every request triggering an L2->L1 HiCache load-back (hi_mamba already
had it at debug). Demote and switch both to lazy %-formatting so the
f-string is not built when debug is off.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
leavelet
2026-06-10 10:22:35 +00:00
co-authored by Claude Fable 5
parent 3eef989739
commit afdd1d0992
2 changed files with 7 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/mem_cache/hi_mamba_radix_cache.py
+3 -1
View File
@@ -265,7 +265,9 @@ class HiMambaRadixCache(MambaRadixCache):
loading_values = self.load_back(last_node, mem_quota) loading_values = self.load_back(last_node, mem_quota)
if loading_values is not None: if loading_values is not None:
logger.debug( logger.debug(
f"loading back {len(loading_values)} tokens for node {last_node.id}" "loading back %d tokens for node %d",
len(loading_values),
last_node.id,
) )
return loading_values, last_node return loading_values, last_node
python/sglang/srt/mem_cache/hiradix_cache.py
+4 -2
View File
@@ -3737,8 +3737,10 @@ class HiRadixCache(RadixCache):
if last_node.evicted: if last_node.evicted:
loading_values = self.load_back(last_node, mem_quota) loading_values = self.load_back(last_node, mem_quota)
if loading_values is not None: if loading_values is not None:
logger.info( logger.debug(
f"loading back {len(loading_values)} tokens for node {last_node.id}" "loading back %d tokens for node %d",
len(loading_values),
last_node.id,
) )
return loading_values, last_node return loading_values, last_node