ling/hermes-agent
1
0
Fork 0
mirror of https://github.com/NousResearch/hermes-agent.git synced 2026-05-02 08:47:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix(agent): spawn OpenRouter pre-warm thread only once per process

Browse Source 
Each AIAgent.__init__() was unconditionally starting a daemon thread to
pre-warm the OpenRouter model metadata cache.  In gateway mode a new
AIAgent is created for every incoming message, so one OS thread leaked
per request.  After ~1 000 messages the process hit the Linux thread
limit and raised RuntimeError: can't start new thread for all subsequent
requests.

Add a module-level threading.Event (_openrouter_prewarm_done) that is
set before the thread is started.  Subsequent AIAgent instantiations
skip the spawn entirely; fetch_model_metadata() is cached for 1 hour so
the single background call is sufficient.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Vlad Ra
2026-04-29 15:47:27 +01:00
committed by Teknium
parent 502debed91
commit a7fb79efb2
1 changed files with 15 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
run_agent.py
View File

@@ -323,6 +323,12 @@ _PATH_SCOPED_TOOLS = frozenset({"read_file", "write_file", "patch"})
# Maximum number of concurrent worker threads for parallel tool execution.
_MAX_TOOL_WORKERS = 8
# Guard so the OpenRouter metadata pre-warm thread is only spawned once per
# process, not once per AIAgent instantiation. Without this, long-running
# gateway processes leak one OS thread per incoming message and eventually
# exhaust the system thread limit (RuntimeError: can't start new thread).
_openrouter_prewarm_done = threading.Event()
# Patterns that indicate a terminal command may modify/delete files.
_DESTRUCTIVE_PATTERNS = re.compile(
r"""(?:^|\s|&&|\|\||;|`)(?:
@@ -1107,10 +1113,17 @@ class AIAgent:
# Pre-warm OpenRouter model metadata cache in a background thread.
# fetch_model_metadata() is cached for 1 hour; this avoids a blocking
# HTTP request on the first API response when pricing is estimated.
if self.provider == "openrouter" or self._is_openrouter_url():
# Use a process-level Event so this thread is only spawned once — a new
# AIAgent is created for every gateway request, so without the guard
# each message leaks one OS thread and the process eventually exhausts
# the system thread limit (RuntimeError: can't start new thread).
if (self.provider == "openrouter" or self._is_openrouter_url()) and \
not _openrouter_prewarm_done.is_set():
_openrouter_prewarm_done.set()
threading.Thread(
target=lambda: fetch_model_metadata(),
target=fetch_model_metadata,
daemon=True,
name="openrouter-prewarm",
).start()
self.tool_progress_callback = tool_progress_callback