Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 12 additions & 3 deletions eval_protocol/mcp/execution/manager.py
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,7 @@
from dataclasses import asdict
from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Union

import anyio
from openai.types import CompletionUsage

from vendor.tau2.data_model.message import AssistantMessage, UserMessage
Expand Down Expand Up @@ -464,11 +465,19 @@ async def _execute_rollout(
)

except asyncio.CancelledError:
logger.error(f"🚨 AsyncIO Cancel Error in roll out {rollout_idx}", exc_info=True)
failure_reason = "asyncio context cancelled"
logger.error(
f"🚨 Error in rollout {session.dataset_row.id} {rollout_idx}: {failure_reason}", exc_info=True
)
except (anyio.ClosedResourceError, anyio.BrokenResourceError):
failure_reason = "anyioconnection/resource error"
logger.error(
f"🚨 Error in rollout {session.dataset_row.id} {rollout_idx}: {failure_reason}", exc_info=True
)
except Exception as e:
logger.error(f"🚨 Error in rollout {rollout_idx}: {e}", exc_info=True)
failure_reason = str(e)
error_msg = str(e) if str(e) else f"{type(e).__name__}: Unexpected error"
logger.error(f"🚨 Error in rollout {session.dataset_row.id} {rollout_idx}: {error_msg}", exc_info=True)
failure_reason = error_msg
finally:
if failure_reason:
trajectory.terminated = True
Expand Down
Loading