Bharath Appali

Current EM design doesn't have a mechanism to release the memory of it's internal components when it's idle, the components include process queues, stage processors, completed ETD's (which takes up...

First EM Patch #278 has been added to autotune. To make it easier for review and also for the followers of autotune I would like to make an incremental progress...

Make Autotune start in EM only mode

1

For testing and other purposes we expect Autotune to be started in EM only mode where only EM comes up when Autotune is started.

Shutsdown kubernetes client after it's usage

1

This PR adds a trivial change to shutdown the kubernetes client after it's usage Signed-off-by: bharathappali

The Kubernetes service was not closed after initialization in these functions [matchPodsToAutotuneObject](https://github.com/kruize/autotune/blob/mvp_demo/src/main/java/com/autotune/analyzer/deployment/AutotuneDeployment.java#L257) , [getAutotuneConfig](https://github.com/kruize/autotune/blob/mvp_demo/src/main/java/com/autotune/analyzer/deployment/AutotuneDeployment.java#L451).

Current input JSON has the queries which consists of query variables, queries need to be evaluated and checked if there are any unknown variables and known variables need to be...

The current implementation of `MetricCollectionHandler` is not scheduled and also it needs to be cycled for number of warmup and measurement cycles

To extract the pod name, instead of selecting the pod based on searching container name and deployment name it would be easier to have the key and value of the...

Adds the implementation to check for pod metrics to detect the load. Currently Work in progress

Pytorch Job not getting detected by the training operator when launched via katib

5

**What steps did you take and what happened:** I'm a new katib user and I have installed katib and training operator in standalone installation on my local minikube cluster, and...