open-swe/agent/tools/schedule_thread_wakeup.py
Johannes du Plessis 918641d3e2
fix: purge expired thread_wakeup crons (#1656)
* fix: purge expired thread_wakeup crons

One-shot wakeup crons set an end_time that stops re-firing but the cron
row is never deleted, so dead rows accumulate (86 in prod). Add a purge
that deletes thread_wakeup crons past their end_time, called
opportunistically before scheduling a new wakeup, plus a one-time
backfill script. Conservative: matches only kind=thread_wakeup with a
past end_time.

* chore: retrigger Open SWE review

---------

Co-authored-by: open-swe[bot] <open-swe@users.noreply.github.com>
(cherry picked from commit 9e5a1924ef306269322c31342a1831e57831cfee)
2026-07-02 19:23:27 -04:00

212 lines
7.2 KiB
Python

"""Tool that schedules a one-shot re-trigger of the current agent thread."""
from __future__ import annotations
import logging
from datetime import UTC, datetime, timedelta
from typing import Any
from langgraph.config import get_config
from langgraph_sdk import get_client
from ..utils.thread_ops import langgraph_url
logger = logging.getLogger(__name__)
_AGENT_ASSISTANT_ID = "agent"
_MIN_DELAY_SECONDS = 60
_MAX_DELAY_SECONDS = 86_400
_END_TIME_PADDING_SECONDS = 90
_WAKEUP_KIND = "thread_wakeup"
_PURGE_PAGE_SIZE = 100
_DEFAULT_WAKEUP_PROMPT = (
"This is an automated re-trigger of this thread. The agent scheduled this "
"wakeup to poll for updates. Check the current state of whatever you were "
"waiting on and continue from there."
)
def _ceil_to_next_minute(value: datetime) -> datetime:
"""Round a datetime up to the next whole minute."""
rounded = value.replace(second=0, microsecond=0)
if rounded == value:
return rounded
return rounded + timedelta(minutes=1)
def _build_one_shot_cron(fire_time: datetime) -> str:
"""Build a 5-field cron expression that fires at ``fire_time`` (UTC)."""
return " ".join(
[
str(fire_time.minute),
str(fire_time.hour),
str(fire_time.day),
str(fire_time.month),
"*",
]
)
def _parse_iso(value: Any) -> datetime | None:
if not isinstance(value, str) or not value:
return None
try:
return datetime.fromisoformat(value.replace("Z", "+00:00"))
except ValueError:
return None
async def find_expired_wakeup_cron_ids(client: Any, *, now: datetime) -> list[str]:
"""Return the ids of ``thread_wakeup`` crons whose ``end_time`` has passed.
Conservative: matches solely on ``metadata.kind == "thread_wakeup"`` AND a
past ``end_time``, so analyzer/dashboard crons are never selected. Paginates
fully before returning so the result is stable to delete afterwards.
"""
expired_ids: list[str] = []
offset = 0
while True:
page = await client.crons.search(
metadata={"kind": _WAKEUP_KIND},
limit=_PURGE_PAGE_SIZE,
offset=offset,
)
if not page:
break
for cron in page:
if not isinstance(cron, dict):
continue
end_time = _parse_iso(cron.get("end_time"))
cron_id = cron.get("cron_id")
if end_time is not None and end_time < now and isinstance(cron_id, str) and cron_id:
expired_ids.append(cron_id)
if len(page) < _PURGE_PAGE_SIZE:
break
offset += len(page)
return expired_ids
async def purge_expired_wakeup_crons(client: Any, *, now: datetime) -> int:
"""Delete ``thread_wakeup`` crons whose ``end_time`` has already passed.
Each wakeup is a thread-bound cron with an ``end_time`` (~90s past its fire)
that stops it re-firing, but the cron row itself is never removed, so dead
rows accumulate. This deletes only those dead rows. Returns the count deleted.
"""
expired_ids = await find_expired_wakeup_cron_ids(client, now=now)
deleted = 0
for cron_id in expired_ids:
await client.crons.delete(cron_id)
deleted += 1
return deleted
async def _purge_expired_wakeups_best_effort() -> None:
"""Opportunistically purge expired wakeup crons; never raises."""
try:
client = get_client(url=langgraph_url())
deleted = await purge_expired_wakeup_crons(client, now=datetime.now(UTC))
if deleted:
logger.info("Purged %d expired thread_wakeup cron(s)", deleted)
except Exception:
logger.warning("Failed to purge expired thread_wakeup crons", exc_info=True)
async def _create_wakeup_cron(
*,
thread_id: str,
fire_time: datetime,
prompt: str,
configurable: dict[str, Any],
) -> dict[str, Any]:
client = get_client(url=langgraph_url())
schedule = _build_one_shot_cron(fire_time)
end_time = fire_time + timedelta(seconds=_END_TIME_PADDING_SECONDS)
run_config: dict[str, Any] = {"configurable": configurable}
cron = await client.crons.create_for_thread(
thread_id,
_AGENT_ASSISTANT_ID,
schedule=schedule,
input={"messages": [{"role": "user", "content": prompt}]},
config=run_config,
end_time=end_time,
timezone="UTC",
metadata={
"kind": "thread_wakeup",
"thread_id": thread_id,
},
)
cron_id = cron.get("cron_id") if isinstance(cron, dict) else getattr(cron, "cron_id", None)
return {
"success": True,
"cron_id": cron_id,
"scheduled_for": fire_time.isoformat(),
"thread_id": thread_id,
}
async def schedule_thread_wakeup(delay_minutes: int, prompt: str | None = None) -> dict[str, Any]:
"""Schedule a one-shot re-trigger of the current thread after a delay.
Use this when you need to poll or check back on something later — e.g.
waiting for CI to finish, a deploy to complete, or an external process
to settle. The current thread will be re-invoked with the given prompt
(or a default wakeup message) after the specified delay.
Args:
delay_minutes: How many minutes from now to wait before re-triggering.
Minimum 1 minute, maximum 1440 (24 hours).
prompt: Optional message to send to the thread when it wakes up.
If omitted, a default polling prompt is used.
Returns a dict with ``success``, ``cron_id``, ``scheduled_for`` (ISO UTC),
and ``thread_id``.
"""
if not isinstance(delay_minutes, int) or delay_minutes < 1:
return {"success": False, "error": "delay_minutes must be a positive integer (>= 1)"}
delay_seconds = delay_minutes * 60
if delay_seconds < _MIN_DELAY_SECONDS:
return {"success": False, "error": "delay must be at least 1 minute"}
if delay_seconds > _MAX_DELAY_SECONDS:
return {"success": False, "error": "delay must be at most 1440 minutes (24 hours)"}
config = get_config()
configurable = config.get("configurable", {}) if isinstance(config, dict) else {}
thread_id = configurable.get("thread_id")
if not isinstance(thread_id, str) or not thread_id:
return {"success": False, "error": "No thread_id in current run config"}
fire_time = _ceil_to_next_minute(datetime.now(UTC) + timedelta(seconds=delay_seconds))
wakeup_prompt = (
prompt.strip() if isinstance(prompt, str) and prompt.strip() else _DEFAULT_WAKEUP_PROMPT
)
passthrough_keys = (
"repo",
"source",
"slack_thread",
"linear_issue",
"github_login",
"user_email",
"schedule_id",
)
wakeup_configurable: dict[str, Any] = {"thread_id": thread_id}
for key in passthrough_keys:
value = configurable.get(key)
if value is not None:
wakeup_configurable[key] = value
await _purge_expired_wakeups_best_effort()
try:
return await _create_wakeup_cron(
thread_id=thread_id,
fire_time=fire_time,
prompt=wakeup_prompt,
configurable=wakeup_configurable,
)
except Exception as exc:
logger.exception("Failed to schedule thread wakeup for %s", thread_id)
return {"success": False, "error": str(exc)}