Collector for generic metadata of HPC execution environment focussed on benchmarking.
Currently the tool is not publicly available on PyPi and is installed directly
from the repository. It is recommended to install the package into an
environment. For example using Python venv
could be done as follows:
python -m venv venv
source venv/bin/activate
pip install -U pip
pip install git+ssh://[email protected]/INM-6/gather-metadata.git
If SSH is not available, use
git+https://[email protected]/INM-6/gather-metadata.git
together with a
GitHub Personal Access Token (PAT).
Running the tool with a directory name will create that directory and gather all information into it. You can nicely zip it and store it together with the other result data. The intermediate folder is not needed anymore, but may be useful to already extract rudimentary first metadata to annotate the run.
gathermetadata about;
tar -czf some-run-id-or-so.tgz about && rm -rf about
If you know about additional metadata sources please do not just customize your
fork but create a pull request to add the gathering globally in recordables
(here),
even if your command is extremely specific to your case or environment! The
tool will always try to gather the maximum of avilable data and skip any source
not found automatically. It has been taken great care that broken entries will
not annoy anyone.
To install in development mode add the dev
feature flag to the pip install
line of the normal installation procedure:
pip install -e .[dev]
To run the test suite type
pytest gathermetadata
This project is licensed under GNU General Public License v3.0 only. See LICENSE for details.
SPDX-License-Identifier: GPL-3.0-only
SPDX-Copyright: 2023, Forschungszentrum Jülich GmbH, Jülich, Germany
SPDX-Author: Dennis Terhorst <[email protected]>