Skip to content

armgit / letta

Public Imported from GitHub
0

Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.

Add LLM client compaction errors to traces (#9474)

amysguan80a0d1a95f5f
diff --git a/letta/llm_api/llm_client_base.py b/letta/llm_api/llm_client_base.py
index b4b8d69d7..43fa9e27e 100644
--- a/letta/llm_api/llm_client_base.py
+++ b/letta/llm_api/llm_client_base.py
@@ -161,7 +161,11 @@ class LLMClientBase:
return
if response_json is None:
- return
+ if error_msg:
+ response_json = {"error": error_msg, "error_type": error_type}
+ else:
+ logger.warning(f"Skipping telemetry: no response_json or error_msg (call_type={self._telemetry_call_type})")
+ return
try:
pydantic_actor = self.actor.to_pydantic() if hasattr(self.actor, "to_pydantic") else self.actor
diff --git a/letta/services/summarizer/summarizer.py b/letta/services/summarizer/summarizer.py
index e7aabbae7..b5a11e020 100644
--- a/letta/services/summarizer/summarizer.py
+++ b/letta/services/summarizer/summarizer.py
@@ -542,27 +542,35 @@ async def simple_summary(
)
# AnthropicClient.stream_async sets request_data["stream"] = True internally.
- stream = await llm_client.stream_async(req_data, summarizer_llm_config)
- async for _chunk in interface.process(stream):
- # We don't emit anything; we just want the fully-accumulated content.
- pass
-
- content_parts = interface.get_content()
- text = "".join(part.text for part in content_parts if isinstance(part, TextContent)).strip()
-
- # Log telemetry after stream processing
- await llm_client.log_provider_trace_async(
- request_data=req_data,
- response_json={
- "content": text,
- "model": summarizer_llm_config.model,
- "usage": {
- "input_tokens": getattr(interface, "input_tokens", None),
- "output_tokens": getattr(interface, "output_tokens", None),
+ try:
+ stream = await llm_client.stream_async(req_data, summarizer_llm_config)
+ async for _chunk in interface.process(stream):
+ pass
+
+ content_parts = interface.get_content()
+ text = "".join(part.text for part in content_parts if isinstance(part, TextContent)).strip()
+
+ await llm_client.log_provider_trace_async(
+ request_data=req_data,
+ response_json={
+ "content": text,
+ "model": summarizer_llm_config.model,
+ "usage": {
+ "input_tokens": getattr(interface, "input_tokens", None),
+ "output_tokens": getattr(interface, "output_tokens", None),
+ },
},
- },
- llm_config=summarizer_llm_config,
- )
+ llm_config=summarizer_llm_config,
+ )
+ except Exception as e:
+ await llm_client.log_provider_trace_async(
+ request_data=req_data,
+ response_json=None,
+ llm_config=summarizer_llm_config,
+ error_msg=str(e),
+ error_type=type(e).__name__,
+ )
+ raise
if not text:
logger.warning("No content returned from summarizer (streaming path)")