docs: record PEND reason classification as unfinished work (pending unified analysis)
Adds the PEND reason taxonomy (docs/pend-reason-investigation-design.md) to left_scheduler_work 八.4 推迟/未决, with the three open decisions for unified analysis: code strategy (reuse 3 codes + subcodes vs expand PREASON_* to ~20), wiring priority (P0 main_reason + limit attach), and bjobs -p-equivalent MCP capability. Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
parent
f66a626bae
commit
6f520a677b
1 changed files with 1 additions and 0 deletions
|
|
@ -188,6 +188,7 @@ spec 声称 36/36 任务完成,但 re-audit(research.md)发现缺陷。来
|
|||
### 4. 推迟 / 未决
|
||||
- `scheduler_modify` 工具推迟——apiserver 当前无 modify 端点(control 仅 kill/stop/resume/checkpoint),需新增后端端点
|
||||
- 3 个未决决策:LLM 模型选型(Claude API/自建/混合)、多集群 URM 来源识别、air-gapped LLM 访问
|
||||
- **PEND reason 分类体系(未完成,2026-08-13 调查+设计稿已出)**:见 `docs/pend-reason-investigation-design.md`。数据链路已通(bug-1513,ac6131d),但**喂进去的 reason 只有 2 种**——已实现阻挡点 8 处里仅 3 处产生 reason,limit 插件/队列/时间窗/fairshare 等 8 类 LSF 必有原因全部静默或桩。**待统一决策**:(a) code 策略——复用 3 code + 明细子码 vs 扩展 PREASON_* 到 ~20 个;(b) 接线优先级——建议 P0=修 main_reason + limit 插件 attach reason,P1=队列状态/时间窗/worker 过滤,P2=依赖/license/保留/exclusive/fairshare;(c) 是否补 LSF 风格的 `bjobs -p` 等价能力到 MCP 工具。
|
||||
|
||||
### 5. 设计层加固(文档已写、代码零实现)
|
||||
- **Auto-pilot 模式**:用户显式授权范围内的自主行为
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue