Cache task events on disk per server; fix performance audit findings

This commit is contained in:
Ian Gulliver
2026-10-23 12:49:43 -07:00
parent dc2fcc6075
commit d6be98f9d8
13 changed files with 448 additions and 250 deletions
+2 -1
View File
@@ -2436,9 +2436,10 @@ components:
Status:
type: object
required: [started_at, db_bytes, last_event_seq, projects, active_tasks, sse_clients, clients, live_agents, running_turns, rate_limits, agent_processes, zombies, stuck_after_seconds, protocol_errors]
required: [started_at, db_bytes, last_event_seq, schema, projects, active_tasks, sse_clients, clients, live_agents, running_turns, rate_limits, agent_processes, zombies, stuck_after_seconds, protocol_errors]
properties:
live_agents: { type: integer, description: Agent processes currently running. }
schema: { type: string, description: "Fingerprint of the database's applied migrations. It changes when a migration lands or the database is recreated; the app then drops the events it cached from this server." }
agent_processes:
type: array
description: Every agent process byre is tracking, idle sessions included.