scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-10-03 02:22:24 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 80

feat(agent_session_endpoints): add session_status helper for busy/ready oscillation

Browse source 
Pure-function logic for the session status state machine already lives
in state_machine.derive_session_status_from_runs. This module is the
I/O wrapper that reads the session row, counts active runs, and
persists the new status only when the helper says it should change.

Used by every code path that flips a run's status:
  * POST /v2/sessions/{sid}/runs (queued -> session busy)
  * POST /v2/sessions/{sid}/followup (new run -> same)
  * GET  /v2/sessions/{sid}/runs/next/internal/poll (queued -> running)
  * POST /v2/sessions/{sid}/runs/{rid}/cancel (terminal -> ready)
  * POST /v2/sessions/{sid}/runs/{rid}/events:append (terminal -> ready)

Greptile P1 (review #PRR_kwDOKALCgc78u9En).
This commit is contained in:
Ishaan Jaffer 2026-05-06 15:51:16 -07:00
parent c2e8539f1e
commit f68869071b
No known key found for this signature in database
1 changed files with 59 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

59
litellm/proxy/agent_session_endpoints/session_status.py Normal file
View file

@ -0,0 +1,59 @@
"""
Session status driver — flips ``ready`` <-> ``busy`` based on the runs table.
Pure-function logic lives in ``state_machine.derive_session_status_from_runs``.
This module is the I/O wrapper that reads the session row, counts active
runs, and persists the new status only when the helper says it should change.
Called from every code path that flips a run's status:
* ``POST /v2/sessions/{sid}/runs`` (create — queued -> session busy)
* ``POST /v2/sessions/{sid}/followup`` (create new run — same)
* ``GET /v2/sessions/{sid}/runs/next/internal/poll`` (claim — queued -> running)
* ``POST /v2/sessions/{sid}/runs/{rid}/events:append`` (terminal -> ready)
Without these hooks, sessions that started ``ready`` would stay ``busy``
forever after the first run started — see Greptile P1 review on
PR #27330.
"""
from datetime import datetime, timezone
from litellm.proxy.agent_session_endpoints.constants import RUN_ACTIVE_STATUSES
from litellm.proxy.agent_session_endpoints.state_machine import (
derive_session_status_from_runs,
)
async def refresh_session_status_from_runs(prisma_client, session_id: str) -> None:
"""Look up the session, decide if it should flip, persist if yes.
Idempotent: if no transition is needed, no DB write happens.
Best-effort: callers should treat failures as non-fatal — the
cleanup sweeper is the safety net that catches sessions stuck in
``busy`` after the run rows have all moved to terminal.
"""
session = await prisma_client.db.litellm_agentsession.find_unique(
where={"id": session_id}
)
if session is None:
return
active = await prisma_client.db.litellm_agentrun.find_first(
where={
"session_id": session_id,
"status": {"in": list(RUN_ACTIVE_STATUSES)},
}
)
new_status = derive_session_status_from_runs(
current_session_status=session.status,
has_active_run=active is not None,
)
if new_status is None:
return
await prisma_client.db.litellm_agentsession.update(
where={"id": session_id},
data={
"status": new_status,
"updated_at": datetime.now(timezone.utc),
},
)