Files
nexus/wiki/sources/yang-xia-ri-ji-4.md
2026-04-14 16:02:50 +08:00

43 lines
1.9 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
title: "养虾日记4一次「Context Limit Exceeded」错误排查"
type: source
tags: [AI, Agent, OpenClaw, Debug, Context-Window]
date: 2026-04-10
source_file: raw/微信公众号/养虾日记4 一次「Context Limit Exceeded」错误排查我以为是小问题结果踩了大坑.md
---
## Summary
- 核心主题OpenClaw Agent Context Limit问题的排查过程
- 问题域星枢Telegram频道报Context Limit Exceeded修改compaction配置无效
- 方法/机制:日志分析 → 发现模型被切换到deepseek-reasoner16K context→ 还原正确配置
- 结论/价值:错误信息可能是表面意思,真正问题藏在配置层级深处
## Key Claims
- 星枢的Telegram channel绑定到了只有16K context的deepseek-reasoner模型
- OpenClaw配置分层Global Configopenclaw.jsonvs Agent/Channel Specific Config
- 解决后建议让星枢Telegram用回MiniMax-M2.7200K context
## Key Quotes
> "不要默认认为错误信息就是表面意思。" — 比利哥
> "日志真的有用Gateway日志把问题写得明明白白只是我自己没仔细看。" — 比利哥
## Key Concepts
- [[Context-Window]]:模型的上下文窗口大小,影响对话长度
- [[Compaction]]对话压缩机制预留一半token给compactionsafeguard模式
- [[Fallback机制]]模型回退触发条件503/429/Timeout/配置错误)
- [[配置分层]]Global Config vs Agent/Channel Specific Config
## Key Entities
- [[OpenClaw]]AI Agent操作系统
- [[星枢]]xingshuOpenClaw主Agent
- [[deepseek-reasoner]]16K context的推理模型问题根源
- [[MiniMax-M2.7]]200K context的主力模型
## Connections
- [[yang-xia-ri-ji-2]] ← 问题关联 ← [[yang-xia-ri-ji-4]]Context Limit与记忆系统紧密相关
- [[养龙虾5天血泪史]] ← 深化 ← [[yang-xia-ri-ji-4]](更全面的内存问题调试经验)
## Contradictions
- 尚未发现