openclaw - 💡(How to fix) Fix [Bug]: expired lost tasks survive past cleanupAfter; add tasks reap-orphans CLI [1 comments, 2 participants]

Summary

Expired lost task records can remain in the durable task store after cleanupAfter passes.

At 2026-05-06T01:06:29Z, openclaw tasks list --status lost --json still showed 3 cron tasks whose cleanupAfter had already expired.

Environment

• OpenClaw CLI: 2026.4.23
• CLI warning on every run: Config was last written by a newer OpenClaw (2026.4.24); current version is 2026.4.23.

Repro

1. Run openclaw tasks audit.
2. Wait until a known batch of lost cron tasks passes cleanupAfter.
3. Run:
   • openclaw tasks list --status lost --json
   • openclaw tasks maintenance --json

Actual

tasks list --status lost --json still showed these expired records:

[
  {
    "taskId": "0828e588-39e9-497c-af9a-17a413ed9ab6",
    "label": "verify:cron-fix-stability-20260429",
    "cleanupAfter": "2026-05-06T01:06:04.460Z"
  },
  {
    "taskId": "4b06d99e-3915-4bdf-a675-448a64cd3e88",
    "label": "📮 PTT 多版面監控",
    "cleanupAfter": "2026-05-06T01:06:04.460Z"
  },
  {
    "taskId": "333b5537-94ac-4651-9054-30ac6c6f5e9a",
    "label": "usage-guard:model-routing",
    "cleanupAfter": "2026-05-06T01:06:04.460Z"
  }
]

At the same time, openclaw tasks maintenance --json reported candidates that look prunable:

{
  "maintenance": {
    "tasks": {
      "reconciled": 1,
      "recovered": 0,
      "cleanupStamped": 0,
      "pruned": 3
    }
  }
}

So the system appears able to identify expired orphan tasks, but they were still present after cleanupAfter elapsed.

Expected

One of these should exist:

1. Expired lost tasks are auto-pruned once cleanupAfter passes.
2. A dedicated operator command exists, e.g. openclaw tasks reap-orphans, to sweep expired orphan / lost records deterministically.

Request

Please add an explicit CLI for this path, such as:

openclaw tasks reap-orphans

Suggested behavior:

• prune lost tasks whose cleanupAfter <= now
• print counts by status / runtime
• support --json
• optionally support --dry-run

This would make cron cleanup observable and give operators a targeted recovery path when auto-prune misses records.