nnictl experiment delete -all fail for stopped expriment · microsoft/nni#4543

(2 comments) (1 reaction) (1 assignee)Python (13,504 stars) (1,830 forks)batch import

good first issue

説明

Describe the issue: nnictl experiment delete -all fail for stop expriment. comment out https://github.com/microsoft/nni/blob/master/nni/tools/nnictl/nnictl_utils.py#L509 may solve this problem.

Environment: linux

NNI version: 2.6
Training service (local|remote|pai|aml|etc): local
Client OS: centos
Server OS (for remote mode only):
Python version: 3.7
PyTorch/TensorFlow version: NA
Is conda/virtualenv/venv used?: yes
Is running in Docker?: no

Configuration:

Experiment config (remember to remove secrets!):
Search space:

Log message: bash output: INFO: This action will delete experiment 74ujrqym, and it's not recoverable.
INFO: do you want to continue?[y/N]:y ERROR: There is no experiment running. ERROR: Please set correct experiment id.

How to reproduce it?:

コントリビューターガイド

技術スタック: python
領域: backend
Issue 種別: bug
難度: 2
推定時間: under 1 hour
活動状況: stale
明確さ: mostly clear
前提条件: basic Pythonfamiliarity with nnictl
初心者向け度: 80
調査方針: The issue suggests that nnictl experiment delete all fails for stopped experiments. The recommended fix is to comment out line 509 in nni/tools/nnictl/nnictl utils.py which likely contains a guard that checks for a running experiment. A contributor should examine the logic around that line to ensure that deleting all experiments, including stopped ones, works correctly. There are no linked pull requests or recent comments, so this is an open task with a straightforward starting point.