Reproduction for jdx/mise#5665
Using mise
+ uv
in GitHub workflows with automatic Python virtualenv
activation doesn't work. Running python
causes the workflow hangs and memory
leak.
This is just a simple reproduction by running python
, in the actual use case
other uv
related commands were run and failed.
The step python --version
in .github/workflows/mise.yaml
hangs and got
killed eventually.
When running remotely on GitHub, the step failed with exit code 143 (run).
When running locally with act (through GitHub Local Actions in VSCode), the log has a bit more information as shown below, and the Docker container reached the memory limit and got OOMKilled.
⭐ Run Main python --version
🐳 docker exec cmd=[bash -e /var/run/act/workflow/4] user= workdir=
mise ERROR /opt/hostedtoolcache/uv/0.7.21/x86_64/uv failed
mise WARN uv venv failed: /opt/hostedtoolcache/uv/0.7.21/x86_64/uv exited with non-zero status: no exit status
Python 3.13.5
✅ Success - Main python --version [4m5.44525175s]
mise
automatically activates uv
virtualenv without any issues.