- Hide errors from lower level libraries as debug information.
- Log debug info to a file.
- Fix top memory usage parsing bug.
- Make timeouts configurable.
- Use a fallback port if desired local tunnel port is taken.
- Improve Markdown documentation.
- Add new tutorial notebooks.
- Use JupyterLab by default.
- Release on PyPI.
More Dask stability improvements and new features:
- Synchronization to and from cluster for allocation and Jupyter deployments.
- Quick Jupyter deployment app: `idact-notebook`, or `python -m idact.notebook`.
Multiple stability and debugability improvements, among them:
- More task stage info is available for the user.
- More actions are retried on failure.
- Conda dev environment and pip install and are now tested on Travis.
- Deployment of Jupyter Notebook on allocated nodes.
- Deployment of Dask on allocated nodes.
- Support for public key authentication.
- A way to push and pull the environment from local machine to cluster.
- A way to examine CPU and memory usage of a node.
- A notebook for testing functionality on the Prometheus cluster (ProSandbox.ipynb).
- Tunneling of remote node ports to local ports.
- A way to set log level.
- Support for Python 3.5.
- Added basic node allocation with password authentication.
- Added a way to save and load the environment.