fix: M2 on_failure assignee 从 tasks 表读取 + infrastructure 防递归(司马懿 Review #65)
CI / lint (pull_request) Successful in 7s
CI / test (pull_request) Failing after 9s
CI / notify-on-failure (pull_request) Successful in 1s

M2: on_failure 中 assignee = meta.get('from', '') 读到 'system' 而非实际 Agent
→ 改为 SELECT must_haves, assignee FROM tasks 直接读 tasks.assignee 字段

附带:infrastructure failure 改为直接 DB INSERT,不走 _send_toolchain_task 防递归
This commit is contained in:
cfdaily
2026-06-13 23:47:12 +08:00
parent a5d5d2d974
commit 3bca794902
2 changed files with 62 additions and 34 deletions
+38 -22
View File
@@ -297,22 +297,24 @@ class ToolchainHandler(BaseTaskHandler):
logger.info("Toolchain %s: verify failed (%s), marked failed", logger.info("Toolchain %s: verify failed (%s), marked failed",
task_id, verify.reason) task_id, verify.reason)
# 读取 must_haves 获取事件上下文 # 读取 must_hives 获取事件上下文 + assignee 从 tasks 表读取
meta = {} meta = {}
assignee = agent_id
try: try:
conn = get_connection(db_path) conn = get_connection(db_path)
row = conn.execute( row = conn.execute(
"SELECT must_haves FROM tasks WHERE id=?", (task_id,) "SELECT must_haves, assignee FROM tasks WHERE id=?", (task_id,)
).fetchone() ).fetchone()
if row and row["must_haves"]: if row:
meta = json.loads(row["must_haves"]) if row["must_haves"]:
meta = json.loads(row["must_haves"])
assignee = row["assignee"] or agent_id
conn.close() conn.close()
except Exception: except Exception:
pass pass
action_type = meta.get("action_type", "") action_type = meta.get("action_type", "")
context_data = meta.get("context", {}) context_data = meta.get("context", {})
assignee = meta.get("assignee", "") or meta.get("from", "")
# 三分路决策 # 三分路决策
route = self._classify_failure(verify) route = self._classify_failure(verify)
@@ -403,29 +405,43 @@ class ToolchainHandler(BaseTaskHandler):
self, task_id: str, agent_id: str, self, task_id: str, agent_id: str,
verify: VerifyResult, db_path: Path, verify: VerifyResult, db_path: Path,
) -> None: ) -> None:
"""基础设施失败 → _send_toolchain_task @jiangwei-infra(防递归)""" """基础设施失败 → 直接在 _toolchain DB 创建 task @jiangwei-infra防递归"""
# 直接在 _toolchain DB 创建 task(不走 Gitea webhook)
try: try:
from src.api.toolchain_routes import _send_toolchain_task from datetime import datetime
_send_toolchain_task( new_task_id = f"tc-{int(datetime.now().timestamp() * 1000)}"
to_agent="jiangwei-infra", must_hives = json.dumps({
title=f"[基础设施] Gitea API 不可用 - {task_id}", "event_type": "infrastructure_failure",
description=( "action_type": "infrastructure_failure",
f"Gitea API 不可用,原任务 {task_id} 无法通过正常路径处理。\n" "steps": [
f"请检查 Gitea 服务状态和网络连通性。"
),
event_type="infrastructure_failure",
action_type="infrastructure_failure",
steps=[
"检查 Gitea 服务状态(http://192.168.2.154:3000)", "检查 Gitea 服务状态(http://192.168.2.154:3000)",
"检查网络连通性", "检查网络连通性",
"恢复后提交 action report", "恢复后提交 action report",
], ],
context_data={"original_task_id": task_id, "verify_reason": verify.reason}, "context": {"original_task_id": task_id, "verify_reason": verify.reason},
source="toolchain_handler", "from": "system",
"source": "toolchain_handler_on_failure",
}, ensure_ascii=False)
conn = get_connection(db_path)
conn.execute(
"INSERT INTO tasks (id, title, description, assignee, assigned_by, "
"must_haves, task_type, status) VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
(
new_task_id,
f"[基础设施] Gitea API 不可用 - {task_id}",
f"Gitea API 不可用,原任务 {task_id} 无法通过正常路径处理。\n"
f"请检查 Gitea 服务状态和网络连通性。",
"jiangwei-infra",
"system",
must_hives,
"toolchain",
"pending",
)
) )
logger.info("Toolchain %s: infrastructure failure → task created for jiangwei-infra", conn.commit()
task_id) conn.close()
logger.info(
"Toolchain %s: infrastructure failure → task %s created for jiangwei-infra",
task_id, new_task_id)
except Exception as e: except Exception as e:
logger.error( logger.error(
"Toolchain %s: failed to create infrastructure_failure task: %s", "Toolchain %s: failed to create infrastructure_failure task: %s",
+24 -12
View File
@@ -398,12 +398,14 @@ class TestSendToolchainTask:
class TestOnFailureRouting: class TestOnFailureRouting:
def test_business_failure_creates_gitea_comment(self, handler, tmp_db): def test_business_failure_creates_gitea_comment(self, handler, tmp_db):
"""Business failure → Gitea PR comment""" """Business failure → Gitea PR comment @task assignee (not must_hives field)"""
# S4: must_hives does NOT contain assignee — production data doesn't have it
must_haves = json.dumps({ must_haves = json.dumps({
"action_type": "review_result", "action_type": "review_result",
"context": {"repo": "sanguo/test", "pr_number": 42}, "context": {"repo": "sanguo/test", "pr_number": 42},
"assignee": "zhangfei-dev", "from": "system",
}) })
# assignee is set on the tasks table row (as production code writes it)
_insert_task(tmp_db, "t-fail", must_haves) _insert_task(tmp_db, "t-fail", must_haves)
with patch.object(handler, "_create_gitea_comment") as mock_comment: with patch.object(handler, "_create_gitea_comment") as mock_comment:
@@ -414,9 +416,12 @@ class TestOnFailureRouting:
call_args = mock_comment.call_args call_args = mock_comment.call_args
assert call_args[0][0] == "sanguo/test" assert call_args[0][0] == "sanguo/test"
assert call_args[0][1] == 42 assert call_args[0][1] == 42
# M2: comment body should @ the task's assignee from tasks table
comment_body = call_args[0][2]
assert "@zhangfei-dev" in comment_body
def test_infrastructure_failure_creates_task(self, handler, tmp_db): def test_infrastructure_failure_creates_task(self, handler, tmp_db):
"""Infrastructure failure → _send_toolchain_task for jiangwei-infra""" """Infrastructure failure → direct DB task for jiangwei-infra (no reverse dep)"""
must_haves = json.dumps({ must_haves = json.dumps({
"action_type": "review_result", "action_type": "review_result",
"context": {"repo": "sanguo/test", "pr_number": 42}, "context": {"repo": "sanguo/test", "pr_number": 42},
@@ -427,15 +432,22 @@ class TestOnFailureRouting:
mock_comment.return_value = False # Gitea API down mock_comment.return_value = False # Gitea API down
with patch.object(handler, "_create_gitea_issue") as mock_issue: with patch.object(handler, "_create_gitea_issue") as mock_issue:
mock_issue.return_value = False # Gitea API still down mock_issue.return_value = False # Gitea API still down
with patch("src.api.toolchain_routes._send_toolchain_task") as mock_send: verify = VerifyResult(False, "no_action", "no action_report")
mock_send.return_value = "tc-infra" handler.on_failure("t-infra", "zhangfei-dev", tmp_db, verify)
verify = VerifyResult(False, "no_action", "no action_report")
handler.on_failure("t-infra", "zhangfei-dev", tmp_db, verify) # S3: should directly INSERT into DB, not call _send_toolchain_task
# Should eventually try to create infrastructure_failure task # Verify a new task was created in DB for jiangwei-infra
mock_send.assert_called() conn = get_connection(tmp_db)
call_kwargs = mock_send.call_args rows = conn.execute(
assert call_kwargs[1]["action_type"] == "infrastructure_failure" "SELECT * FROM tasks WHERE assignee=?",
assert call_kwargs[1]["to_agent"] == "jiangwei-infra" ("jiangwei-infra",)
).fetchall()
conn.close()
assert len(rows) >= 1, "No infrastructure_failure task created"
infra_task = rows[0]
assert infra_task["task_type"] == "toolchain"
meta = json.loads(infra_task["must_haves"])
assert meta["action_type"] == "infrastructure_failure"
# --------------------------------------------------------------------------- # ---------------------------------------------------------------------------