Hacktoberfest 2026: as issues que os mantenedores marcaram para outubro, abertas e boas para iniciantes. Ver issues do Hacktoberfest

Failed uploads are reported as success: Version.deploy() / Workspace.deploy_model() don't raise, and roboflow model upload / roboflow image upload <dir> exit 0

Aberta
  • Algo errado?
#545 0 comentários 0 reações 0 responsáveis Ver no GitHub

Mantenedores costumam responder em até 1 dia

Ninguém assumiu esta issue ainda.

  • #554 de @tonylampada — fechado sem integrar

Avaliação

Dificuldade
4/5
Tempo estimado
3-5 dias
Facilidade para iniciantes
55/100
Tipo de issue
Bug
Clareza
Claramente especificada
Status de atividade
Ativa
Stack de tecnologia
python
Domínio
api, cli

Direção de pesquisa

Start with Version._upload_zip() and Workspace._upload_zip() for swallowed upload failures, then trace _upload_model in roboflow/cli/handlers/model.py and _handle_upload_directory in roboflow/cli/handlers/image.py. Read the related issue #497 and inspect tests for model and image uploads before deciding how to represent partial failures. Done means failed uploads propagate as errors and the CLI exits non-zero; add tests for both paths.

Escrita pelo modelo de indexação a partir do texto da issue.

Descrição

Several upload paths catch errors, print them and continue, so scripts and the CLI cannot detect a failed upload (roboflow 1.7.0, also current main).

  1. Version._upload_zip() and Workspace._upload_zip() wrap both the request for the signed upload URL and the PUT in try/except Exception: print(...). Version.deploy() and Workspace.deploy_model() therefore return normally when the upload failed, and roboflow model upload (_upload_model in roboflow/cli/handlers/model.py) prints {"status": "uploaded"} and exits 0. The 429 branch raises a RuntimeError inside the same try, so it is swallowed too; since a version can own several models (#497), its message may also be outdated.
  2. Workspace.upload_dataset() (per-image flow) prints [ERR] <path> (...) for each failed image and returns None. roboflow image upload <dir> (_handle_upload_directory in roboflow/cli/handlers/image.py) then reports {"status": "imported", "count": N}, where N is the number of image files on disk, and exits 0, even if every upload failed. In the zip flow, a final status failed is passed through as "status": "failed", also with exit code 0.

Reproducer for 1, run from a checkout (offline, mocked HTTP):

from types import SimpleNamespace
from unittest.mock import patch
import responses
from tests.helpers import get_version

bundle = SimpleNamespace(model_type="yolov8n", archive_path=SimpleNamespace(name="roboflow_deploy.zip"))
with responses.RequestsMock() as rsps, patch(
    "roboflow.core.version.package_custom_weights_interactive", return_value=bundle
):
    rsps.add(responses.GET, "https://api.roboflow.com/Test Workspace Name/Test Project Name/1/uploadModel", status=500)
    print(get_version().deploy("yolov8n", "/tmp/weights", "weights/best.pt"))  # prints the error, then None

With the same mock and roboflow.Roboflow patched to return that version, roboflow --json model upload -p proj -v 1 -t yolov8n -m <dir> exits 0, and stdout is the printed error followed by {"status": "uploaded"}. For 2: with Project.upload_image raising ImageUploadError for every image, roboflow --json image upload <dir with 3 images> -p proj exits 0 and prints three [ERR] lines followed by {"status": "imported", "count": 3}.

Suggested behavior: raise RoboflowError from both _upload_zip methods (keeping the printed hints as the message) and let the CLI report it through output_error. For upload_dataset, return a summary (uploaded, duplicates, failed with reasons) and let the CLI exit non-zero when anything failed; treat a zip status failed as an error.

I'm happy to send a PR with tests for part 1. Part 2 changes a return value, so I'd like to hear your preference first. Related: #528 (annotation retries), #302.

cc @leeclemnet @digaobarbosa

Linguagem predominante
Python
Estrelas
634
Forks
144
Merge médio
1d 19h
PRs com merge (30d)
18

Preparar o ambiente

Abrir no Codespaces

Inicia o contêiner de desenvolvimento do projeto no navegador, com a sua própria conta do GitHub.

Primeiros passos

  1. Leia a issue inteira e depois o guia de contribuição do projeto.
  2. Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
  3. Faça um fork do repositório e trabalhe em uma branch.
  4. Abra um pull request que referencie o número da issue.

Mais de roboflow/roboflow-python

Todas as issues de roboflow/roboflow-python

Issues semelhantes

Mais issues de Python

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.