Mustafa Eyceoz
Mustafa Eyceoz
It would be nice for use in Kubernetes, OpenShift, and other cloud-based environments if the user had the option of setting hyperparameters via env vars rather than having to create...
Hi! I'm using this space to write about my current progress, updates and plans. The biggest piece right now is the addition of "collectors to the benchmark wrapper": COLLECTORS -...
### Name of Feature or Improvement Adding KubeRay direct support to the SDK ### Description of Problem the Feature Should Solve For users who wish to use the SDK as...
Add proper documentation for installing/using the CodeFlare stack without ODH (within the `project-codeflare` repo)
We've slowly been slipping on good docstring/type hint practice in the SDK (which is notably also how our docs are generated). It's time to do the somewhat-tedious work of fixing...
Currently, cluster.up() throws an exception when a cluster with the same name (in the same namespace) already exists. Investigate changing this behavior, and see if any issues would arise from...
A number of demos upcoming with specific requests: - Highlighting InstaScale cost saving - Highlighting MCAD queueing, quota, priority, and preemption - Highlighting more ray cluster features and more interactive...
Currently, when calling using the CLI's `status` or `details` commands, they will also write the cluster's AW back to current dir, since `get_cluster()` recreates the `Cluster` object. Perhaps we should...
Provide some documentation on how a developer can set up an environment to actually use/test/develop the CodeFlare components (like working on the SDK) without an existing available OpenShift cluster. Current...
Merge both `cluster.status()` and `cluster.is_ready()` into one one function (likely still called `status`). This will tell the user exactly where in the process of setup their cluster is currently (whether...