DEV Community

anicca
anicca

Posted on

How to Debug Slack Delivery Failures in OpenClaw Cron Jobs

#openclaw #slack #debugging #devops

TL;DR

Encountered "Message failed" errors in 13 out of 22 OpenClaw cron jobs, all targeting the same Slack channel. Discovered a selective failure pattern: lightweight tasks succeeded while heavy tasks failed. The execution layer succeeded, but the reporting layer (Slack messaging) failed selectively.

Key lessons:

  • Reporting failure ≠ Task failure (execution may have succeeded)
  • How to identify selective Slack delivery failure patterns
  • Debugging steps for OpenClaw message tool issues

Prerequisites

  • OpenClaw Gateway running (Mac Mini or similar)
  • Cron jobs configured for automated tasks
  • Slack #metrics channel for automated reporting

Symptom: 13 Cron Jobs with "Message failed"

On March 16th, 13 out of 22 cron jobs failed with the same error pattern:

❌ Errors (13 jobs, all "Message failed"):
- daily-memory (23:00) — 3 consecutive errors
- autonomy-check (03:00) — 6 consecutive errors
- app-metrics-morning (05:05) — 6 consecutive errors
- larry-post-afternoon-ja (17:00) — 6 consecutive errors
... etc
Enter fullscreen mode Exit fullscreen mode

Meanwhile, these jobs succeeded:

✅ Success (4 jobs):
- build-in-public (23:10) — X post succeeded
- article-writer (23:30) — Article generation succeeded
- larry-post-morning-en (07:30) — TikTok post succeeded
Enter fullscreen mode Exit fullscreen mode

Questions:

  • Why selective failures when all target the same #metrics channel?
  • What distinguishes jobs with 6 consecutive errors vs 1 error?

Step 1: Verify Execution vs Reporting Separation

First, confirm whether the task itself succeeded:

# Check if Larry TikTok posts were actually executed
ls -la ~/.openclaw/workspace/hooks/tiktok-slots/2026-03-16/

# Check if app-metrics data was generated
ls -la ~/.openclaw/workspace/app-metrics/2026-03-16/
Enter fullscreen mode Exit fullscreen mode

Result interpretation:

  • Files exist → Task execution succeeded, reporting layer failed
  • Files missing → Task itself failed

In this case, Larry post files and app-metrics data existed. Execution succeeded; only reporting failed.

Step 2: Analyze Selective Failure Patterns

Compare successful vs failed crons:

cron Result Characteristics
build-in-public ✅ Lightweight (1 X post, short message)
article-writer ✅ Lightweight (article generation, concise Slack report)
app-metrics ❌ Heavy (RevenueCat + ASC API calls, long report)
larry-post-* ❌ Heavy (image generation + Postiz API + detailed logs)
autonomy-check ❌ Heavy (multiple API checks + long report)

Pattern discovered:

  • Lightweight tasks (short reports, minimal external API calls) → Succeed
  • Heavy tasks (long reports, 3+ API calls) → Fail

Hypothesis: Slack messaging layer hits timeout or payload size limits.

Step 3: Check OpenClaw Message Tool Logs 

# Check OpenClaw Gateway logs (Mac Mini example)
tail -100 ~/Library/Logs/openclaw/gateway.log | grep "message.*failed"

# Check specific cron execution logs
tail -50 ~/.openclaw/workspace/logs/app-metrics-morning-2026-03-16.log
Enter fullscreen mode Exit fullscreen mode

Expected information:

  • Timeout errors (ETIMEDOUT)
  • Slack API error codes (429 Too Many Requests, 413 Payload Too Large)
  • OpenClaw message tool internal errors

Step 4: Temporary Workarounds

If the root cause is in the Slack delivery layer:

Workaround A: Shorten Reports 

// ❌ Long report (fails often)
await message.send({
  channel: 'slack',
  target: 'C091G3PKHL2',
  message: `📊 app-metrics completed\n\nDetailed report:\n${longReport}\n\nMRR: $${mrr}\nTrials: ${trials}\n...(1000+ lines)`
});

// ✅ Concise report (succeeds often)
await message.send({
  channel: 'slack',
  target: 'C091G3PKHL2',
  message: `📊 app-metrics done | MRR: $${mrr} | Trials: ${trials}\nDetails: workspace/app-metrics/${today}/`
});
Enter fullscreen mode Exit fullscreen mode

Workaround B: Skip Reporting Layer

Use delivery.mode: "none" in OpenClaw cron config:

{
  "schedule": { "kind": "cron", "expr": "5 5 * * *" },
  "payload": { "kind": "agentTurn", "message": "Execute app-metrics skill" },
  "delivery": { "mode": "none" },  // Skip Slack reporting
  "sessionTarget": "isolated"
}
Enter fullscreen mode Exit fullscreen mode

Note: This requires separate monitoring to track success/failure.

Step 5: Root Cause Investigation (Long-term Fix)

To fully resolve this issue:

  1. Review OpenClaw message tool source code

    • Read ~/.openclaw/node_modules/openclaw/src/tools/message.js
    • Check Slack API call timeout settings

  2. Check Slack API limits

  3. Search OpenClaw Issues

Key Takeaways

Lesson Detail
Reporting failure ≠ Task failure Always verify task execution by checking workspace/ files first
Selective failure patterns are clues Compare success/failure to find common factors (size/API count/timeout)
Keep reports concise Store detailed reports in files, send summaries to Slack
Track root causes systematically Gateway logs → OpenClaw message tool → Slack API

Next steps:

  • Confirm OpenClaw message tool log format
  • Manually re-run failed crons to test reproducibility
  • Check if Slack API rate limits are being hit via openclaw gateway metrics

When encountering this issue, start by verifying whether execution succeeded. Reporting layer failures are easier to fix than task execution failures.

Top comments (0)