openclaw - 💡(How to fix) Fix Cron jobs timeout after ~150s despite timeoutSeconds=300 - Bailian/Qwen3.5-plus [1 comments, 2 participants]

Description

Cron jobs are timing out after approximately 150 seconds (2.5 minutes), even though gents.defaults.timeoutSeconds is set to 300 seconds (5 minutes). The error message suggests the timeout is coming from the model API side, not OpenClaw's internal timeout.

Affected Tasks

Configuration

OpenClaw Version: 2026.4.2 (d74a122)

agents.defaults.timeoutSeconds: 300 (now increased to 600 for testing)

Model: bailian/qwen3.5-plus

Error Details

Error message from cron runs: Request timed out before a response was generated. Please try again, or increase gents.defaults.timeoutSeconds in your config.

Observations:

1. All timeouts occur around 120-160 seconds, not at the configured 300 seconds
2. The error is delivered successfully via QQ Bot (push notification works)
3. Tasks that complete under 60 seconds work fine (e.g., simple Heartbeat checks)
4. Complex tasks (morning report with weather/news/GitHub queries, Ace project status reports) consistently timeout

Possible Causes

1. pi-agent-core internal timeout: Documentation shows imeoutMs: 120_000 (2 minutes) in pi.md examples. Is this hardcoded?

2. Bailian/Qwen API timeout: Does the Bailian provider have a server-side timeout limit?

3. Model thinking/reasoning mode: Could thinking mode add extra latency that pushes requests over the limit?

Questions for Maintainers

1. Is there a hardcoded timeout in pi-agent-core that overrides gents.defaults.timeoutSeconds?

2. Does the Bailian provider configuration support a custom timeout setting?

3. Should we configure timeout at the provider level instead of (or in addition to) the agent level?

4. Is there a recommended timeout value for complex agentic tasks with Bailian/Qwen models?

Workaround Attempted

Increased gents.defaults.timeoutSeconds from 300 to 600 seconds. Will monitor if this resolves the issue or if the timeout is enforced at a different layer.

Environment

• OS: Windows NT 10.0.26200 (x64)
• Node: v24.14.0
• OpenClaw: 2026.4.2
• Channel: QQ Bot
• Model Provider: Bailian (Dashscope)
• Model: qwen3.5-plus

Related Documentation

From gent-loop.md: `

• Agent runtime: gents.defaults.timeoutSeconds default 172800s (48 hours); enforced in unEmbeddedPiAgent abort timer. `

From pi.md example: ypescript timeoutMs: 120_000, // 2 minutes

This suggests there may be multiple timeout layers that need to be configured consistently.

Priority: Medium - affects scheduled tasks but system remains functional
Impact: Cron jobs for Heartbeat, Morning Report, and Tieba interaction fail intermittently