ecce-launch: a submit whose answer is lost leaves the calculation Ready although the job is in the queue

未關閉
#887 1 則留言 0 個 reaction 已指派 1 人 在 GitHub 檢視

@ohlincha 已經在處理了。

開始於 2026年11月7日。

評估

這個 Issue 還沒有評估資料。

描述

bug

ecce-launch of 12 ORCA single points to a Slurm machine whose controller was slow (slurm_load_partitions: Socket timed out): two ended "not launched: timed out after 75 s" and never reached Slurm (correct). Five others printed nothing and stayed Ready in ECCE, but were in the queue (job names submit__). ECCE does not monitor them, and launching them again runs them twice.

Wanted:

  • When the submit command times out or its answer is lost, look the job up in the queue by its job name (submit__) before reporting; if it is there, store it as Submitted with that job id; if this cannot be decided, report "submitted? check the queue" and mark the calculation so it is not launched a second time without a warning.
  • ecce-launch prints one line for every calculation it was given, never nothing.
  • A way to attach an existing queue job to a Ready calculation: ecce-reconnect <calc> jobid=N (and the same in Reconnect Selected).
主要語言
C
星號
10
分支
8
平均合併
21 分鐘
30 天內合併 PR
1

環境準備

這個專案沒有提供開發容器、Dockerfile 或貢獻指南,環境需要你自己搭建:先看它的 README,通用步驟見我們的新手貢獻指南。

從這裡開始

  1. 先讀完整個 Issue,再讀專案的貢獻指南。
  2. 在 Issue 下留言說明你要接手 —— 這能避免兩個人做同樣的事。
  3. Fork 儲存庫,在一個分支上完成修改。
  4. 送出 Pull Request,並在描述裡引用這個 Issue 編號。

FriendsofECCE/ECCE 的其他 Issue

查看 FriendsofECCE/ECCE 的全部 Issue

相似的 Issue

更多 C Issue

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。