fix: bound task failures and release stalled workers

artemis/sdk/agent.py

@@ -32,6 +32,7 @@ import shutil
 from shutil import which
 import sys
 import threading
+import time
 from types import NoneType
 from typing import Any, TypeVar, overload
 import uuid
@@ -573,6 +574,13 @@ class Agent:
             )
             try:
                 if device_lock is not None and os.environ.get("ARTEMIS_CLOUD_MODE") != "1":
+                    from artemis.runtime.task_control import control_dir, _write_atomic
+
+                    if sess_id and os.environ.get("ARTEMIS_TASK_WORKER") == "1":
+                        _write_atomic(
+                            control_dir(str(sess_id)) / "task_control_state.json",
+                            {"status": "queued", "updated_at": time.time(), "pid": os.getpid()},
+                        )
                     queue_cancel_event = threading.Event()
                     acquire_task = asyncio.create_task(
                         asyncio.to_thread(
@@ -605,6 +613,7 @@ class Agent:
 
                     task_control = TaskControl(str(sess_id), device_lock)
                     context.task_control = task_control
+                    task_control._state("running")
                 self._prepare_output_files(task=task)
                 if os.environ.get("ARTEMIS_CLOUD_MODE") != "1":
                     if self._ui_adb_client is not None:
@@ -712,6 +721,14 @@ class Agent:
                                     state=last_state_snapshot,
                                     error=err,
                                 )
+                                from artemis.runtime.task_result import save_failure_result
+
+                                if sess_id:
+                                    save_failure_result(
+                                        str(sess_id),
+                                        err,
+                                        code=flash_result.get("code", "execution_failed"),
+                                    )
                                 if context.data_engine:
                                     context.data_engine.end_session("failed")
                             return output
@@ -862,6 +879,10 @@ class Agent:
                     state=last_state_snapshot,
                     error=err,
                 )
+                from artemis.runtime.task_result import save_failure_result
+
+                if sess_id:
+                    save_failure_result(str(sess_id), err)
                 if context.data_engine:
                     context.data_engine.end_session("failed")