openclaw - 💡(How to fix) Fix [Bug]: Isolated cron agentTurn fails with 'setup failed: sandbox' — intermittent, pre-runner, before LLM invocation

Bug type

Regression (intermittent — recovered after upgrade to 2026.5.12)

Summary

Isolated cron jobs intermittently fail with (cron status: error). The failure occurs before the LLM is invoked — the isolated child process cannot initialize. The next scheduled run typically succeeds, indicating a transient infrastructure-level issue rather than a persistent logic bug.

Environment

• OpenClaw version: 2026.5.12 (verified responding)
• OS: macOS 15.4.1 (Darwin 24.6.0, arm64)
• Node: v25.6.1
• Gateway: launchd, loopback 127.0.0.1:18789 (main), plus independent agent profiles on 18794–19001

Affected Cron Jobs

Error Details

Cron status:
Reported failure:
Variability: Some occurrences report instead of explicit sandbox failure. Both occur at the same lifecycle phase: isolated session initialization, before the runner process starts.

What Works (same gateway, same time)

• Interactive sessions on the same gateway: instant response ✅
• Unsandboxed (main session) task execution (e.g., via CLI from Cole/main): succeeds when isolated does not ✅
• Gateway health: OpenClaw status

Overview ┌──────────────────────┬───────────────────────────────────────────────────────────────────────────────────────────────┐ │ Item │ Value │ ├──────────────────────┼───────────────────────────────────────────────────────────────────────────────────────────────┤ │ OS │ macos 15.7.5 (arm64) · node 25.6.1 │ │ Dashboard │ http://127.0.0.1:18789/ │ │ Tailscale exposure │ serve · todds-mac-mini.taila9b7dc.ts.net · https://todds-mac-mini.taila9b7dc.ts.net │ │ Channel │ stable (config) │ │ Update │ pnpm · up to date · npm latest 2026.5.12 │ │ Gateway │ local · ws://127.0.0.1:18789 (local loopback) · reachable 61ms · auth token · Todds-Mac-mini. │ │ │ local (192.168.68.67) app 2026.5.12 macos 15.7.5 │ │ Gateway self │ Todds-Mac-mini.local (192.168.68.67) app 2026.5.12 macos 15.7.5 │ │ Gateway service │ LaunchAgent installed · loaded · running (pid 90674) │ │ Node service │ LaunchAgent installed · loaded · running (pid 627) │ │ Agents │ 9 · 1 bootstrap file present · sessions 93 · default main active 1m ago │ │ Memory │ enabled (plugin memory-core) · not checked │ │ Plugin compatibility │ none │ │ Probes │ skipped (use --deep) │ │ Events │ none │ │ Tasks │ 0 active · 0 queued · 0 running · 94 issues · audit 3 warn · 3988 tracked │ │ Heartbeat │ 60m (main), disabled (crash), disabled (dex), disabled (holt), disabled (knox), disabled │ │ │ (mercer), disabled (quinn), disabled (shaw), disabled (sloane) │ │ Sessions │ 93 active · default gpt-5.5 (200k ctx) · 9 stores │ └──────────────────────┴───────────────────────────────────────────────────────────────────────────────────────────────┘

Security audit Skipped in fast status. Full report: openclaw security audit Deep probe: openclaw status --deep

Channels ┌──────────┬─────────┬────────┬────────────────────────────────────────────────────────────────────────────────────────┐ │ Channel │ Enabled │ State │ Detail │ ├──────────┼─────────┼────────┼────────────────────────────────────────────────────────────────────────────────────────┤ │ Discord │ ON │ OK │ token config×6 (MTQ3…XkPU · len 72) · accounts 6/6 │ └──────────┴─────────┴────────┴────────────────────────────────────────────────────────────────────────────────────────┘

Sessions ┌────────────────────────────┬────────┬─────────┬──────────────┬─────────────────────┬─────────────────────────────────┐ │ Key │ Kind │ Age │ Model │ Runtime │ Tokens │ ├────────────────────────────┼────────┼─────────┼──────────────┼─────────────────────┼─────────────────────────────────┤ │ agent:main:discord:channel │ group │ 1m ago │ kimi-k2.6 │ OpenClaw Pi Default │ 99k/131k (76%) │ │ :1477… │ │ │ │ │ │ │ agent:quinn:cron:ae2a9f10- │ cron │ 4m ago │ gpt-5.5 │ OpenAI Codex │ 27k/400k (7%) · 🗄️ 99% cached │ │ 5907-… │ │ │ │ │ │ │ agent:main:cron:17f4421a- │ cron │ 14m ago │ kimi-k2.6 │ OpenClaw Pi Default │ 59k/200k (30%) │ │ 3392-4… │ │ │ │ │ │ │ agent:dex:cron:dce9b855- │ cron │ 20m ago │ gpt-5.5 │ OpenAI Codex │ 44k/400k (11%) · 🗄️ 99% cached │ │ 2f38-4a… │ │ │ │ │ │ │ agent:main:main:heartbeat │ direct │ 22m ago │ gpt-5.5 │ OpenAI Codex │ 50k/400k (12%) · 🗄️ 99% cached │ │ agent:sloane:cron:7b5cc983 │ cron │ 40m ago │ gpt-5.5 │ OpenAI Codex │ 41k/400k (10%) · 🗄️ 100% cached │ │ -314a… │ │ │ │ │ │ │ agent:shaw:cron:92d31051- │ cron │ 44m ago │ gpt-5.4 │ OpenAI Codex │ 87k/1050k (8%) · 🗄️ 99% cached │ │ 4234-4… │ │ │ │ │ │ │ agent:main:cron:36a4a208- │ cron │ 1h ago │ gpt-5.5 │ OpenAI Codex │ 83k/400k (21%) · 🗄️ 99% cached │ │ 8997-4… │ │ │ │ │ │ │ agent:main:cron:6c0b4c01- │ cron │ 2h ago │ gpt-5.5 │ OpenAI Codex │ 78k/400k (20%) · 🗄️ 99% cached │ │ deed-4… │ │ │ │ │ │ │ agent:main:cron:7a13fc4f- │ cron │ 3h ago │ gpt-5.5 │ OpenAI Codex │ 43k/400k (11%) · 🗄️ 97% cached │ │ 5821-4… │ │ │ │ │ │ └────────────────────────────┴────────┴─────────┴──────────────┴─────────────────────┴─────────────────────────────────┘

FAQ: https://docs.openclaw.ai/faq Troubleshooting: https://docs.openclaw.ai/troubleshooting Next steps: Need to share? openclaw status --all Need to debug live? openclaw logs --follow Need to test channels? openclaw status --deep reports healthy, all plugins loaded ✅

• Agent profile gateways: crash (18794), holt (18795), quinn (18850), dex (18820) all running and responding ✅

Observed Pattern

1. Cron fires on schedule → isolated session attempts to spawn
2. Setup fails within ~60 seconds → OR
3. All fallback models exhaust with same setup-phase failure (not inference timeout)
4. Next scheduled run (30m–1h later) → succeeds normally
5. No manual intervention required for recovery

Related Issues

• #82662 — Andy283 reported on 2026.5.12, macOS x64, with 6 fallback models exhausted. Same version, same failure phase (pre-runner), same recovery pattern. Our error wording differs ( vs ) but symptoms are nearly identical.

Distinctions from #82662

• Error wording: vs
• Platform: macOS arm64 vs x64
• Primary model: GPT-5.5 vs DeepSeek
• Scope: Multi-agent fleet (dex, quinn, main) vs single memory-core managed cron

Suggested Investigation

1. Is there a race condition or resource exhaustion during isolated child process initialization that explains intermittent failure + self-recovery?
2. Does the sandbox setup path have a separate internal timeout distinct from the cron ?
3. Could the arm64 vs x64 difference matter for sandbox spawn reliability?
4. Is there a way to surface more detailed sandbox setup diagnostics (errno, spawn trace) instead of the generic message?

Logs Available

• (structured JSON)
• Gateway stdout/stderr captured by launchd
• Agent-specific logs at per profile

Happy to provide full logs or test patches.