Replies: 1 comment
|
从当前
建议抓一组最小证据再定位具体层:每次线路请求记录时间、
尤其请补充精确的 DSH commit/version、所用 profile 中的 retry mode 和 max delay。若相邻两次请求间隔小于 更新:我把这套定位方法整理成了完整的四时间线 runbook,包括 wire-attempt ↔ durable-event 对照表、 https://sandbaseai.github.io/deepseek-harness-handbook/responses-overload-retry.html |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
使用 dsh-tui 连接 OpenAI Codex 兼容服务时,gpt-5.6 系列模型请求持续高频失败,上游典型报错:
服务端观测到 dsh-tui 的请求疑似存在以下行为,想确认是预期设计还是 bug:
走的是 POST /v1/responses。需要更多信息我可以提供。
All reactions