[BUG] Dataset downloader reports success and replaces data after a failed download
维护者通常 1 天内回复
评估
调研方向
Start with datasets/get_test_data.sh and trace the download, extraction, and destination replacement steps for each selected dataset. Reproduce the failure using the temporary directory and fake wget command in the issue body, then check that failed downloads or extraction return a nonzero status and preserve existing data. Done means the same behavior holds for every selected dataset.
由索引模型根据 Issue 内容生成。
描述
Describe the bug
datasets/get_test_data.sh can return exit status 0 after a download fails, while removing and recreating an existing destination directory. A developer can therefore lose the prior dataset and receive a success status even though the replacement was never downloaded or extracted.
Steps to reproduce
From a source checkout, this uses a temporary directory and a fake wget; it does not touch an existing dataset or make a network request:
task_dir=$(mktemp -d)
mkdir -p "$task_dir/bin" "$task_dir/work/cvrp"
cp datasets/get_test_data.sh "$task_dir/work/"
printf 'existing data\n' > "$task_dir/work/cvrp/sentinel"
printf '#!/bin/sh\nexit 8\n' > "$task_dir/bin/wget"
chmod +x "$task_dir/bin/wget"
(
cd "$task_dir/work"
env -u CUOPT_S3_URI -u CUOPT_AWS_ACCESS_KEY_ID -u CUOPT_AWS_SECRET_ACCESS_KEY \
PATH="$task_dir/bin:$PATH" ./get_test_data.sh --cvrp
printf 'exit status: %s\n' "$?"
if test -f cvrp/sentinel; then echo 'original preserved'; else echo 'original removed'; fi
)
Observed: the script reports a failed download and missing archive, prints exit status: 0, and prints original removed.
Expected behavior
Report a nonzero status when a selected dataset cannot be downloaded or extracted, and preserve an existing destination unless a complete replacement is ready. The same distinction should hold for any selected dataset, not just CVRP.
Environment and scope
Reproduced on Linux against main at 52d14b4c97de5591a8054ef1971a92e10617aed7. The fake-download case exercises the HTTP fallback in a disposable directory. An isolated ordinary run also observed an HTTP 404 followed by status 0; the issue is the script's failure handling rather than the availability of any particular external URL.
- 主要语言
- Cuda
- 星标
- 1.1k
- 派生
- 255
- 平均合并
- 1 天 11 小时
- 30 天内合并 PR
- 123
环境准备
- 没有 Dockerfile 或 Docker Compose 文件
- 有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
NVIDIA/cuopt 的其他 Issue
-
[BUG] Constraint validation reports CSR offset count as the number of rows可能已有人在做 @0z5a 今天认领。 未关闭
难度 2/5 1-3 小时 新手友好度 78/100
维护者通常 1 天内回复
-
ci: pr-test-summary job shows red when the GitHub API returns persistent 5xx errors可能已有人在做 @0z5a 今天认领。 未关闭
难度 2/5 1-3 小时 新手友好度 72/100
维护者通常 1 天内回复
-
难度 1/5 1 小时以内 新手友好度 82/100
维护者通常 1 天内回复
-
难度 4/5 3-5 天 新手友好度 48/100
维护者通常 1 天内回复
-
难度 4/5 3-5 天 新手友好度 32/100
维护者通常 1 天内回复
相似的 Issue
-
bug
难度 2/5 1-3 小时 新手友好度 69/100
jejjohnson/gaussx#752 ·
维护者通常 1 天内回复
-
bug ready-for-agent
难度 2/5 1-3 小时 新手友好度 86/100
kerti/balances-v2#801 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 73/100
-
channels:edit check:passed
难度 1/5 1 小时以内 新手友好度 65/100
维护者通常 4 天内回复
-
难度 2/5 1-3 小时 新手友好度 87/100
MIT-LCP/croissant-baker#174 ·