Jiaxin Shan
Jiaxin Shan
It would be great to come up a scalable way to generate python SDK client. Currently, it's not scalable to add sdk for each framework. Is there some framework support...
@xyhuang I plan to add S3 reporter. Do you want me to Create a separate issue to track it just sync status here?
Could I take this task? I am recently doing benchmark using openmpi and like to add support in kubebench
Add PR #167 to add MPI support.
@PatrickXYS We do prefer json logging. No big preference. The challenge is different components use different solutions - kubebuilder - logrus Let's see if we can consolidate into one
Current library provides `reconcile` logics. It's kind of neutral and we did some refactor last year to make it work with both low level controller (tf, mxnet) or high level...
Let's wait for https://github.com/kubeflow/community/pull/522 finalized and then come back to implementation. @qiankunli If you can add more use case, that would be great
> Maybe more helper functions would resolve comparison issue? I think we can add some type name convention by defining ReplicaType type and add more util functions. @MartinForReal Yeah, I...
I think first option means we can not provide high level reconciler for users? If that's true, I prefer second option and probably give clear documents and let user know...
Things to figure out. 1. code repo process, project name -> confirm with Bobby. 2. tech stack? Kubebuilder version, Kubernetes version etc 3. integration environments - Prow or Github Actions,...