Michele Zanotti

cluster-autoscaler cannot scale to Azure VMSS when running k8s on-prem

2

Hi everyone, I am trying to use the cluster-autoscaler for provisioning new nodes on Azure Virtual Machine Scale Sets, starting from a cluster run locally on my PC using [kind](https://kind.sigs.k8s.io/)....

Hello everyone, When creating an `ElasticQuota` resource, if the "max" field is not provided then its default value is zero. Is this the intended behavior? Quoting [KEP 9 - Capacity...

[Trimaran] Support GPU resource

3

Hello everyone, It would be nice if Trimaran supported load-aware scheduling based on GPU utilization. The [load-watcher](https://github.com/paypal/load-watcher) service supports Prometheus sources, so the metrics exposed by the NVIDIA DCGM Exporter...

## Description Invoking `optimize_model` with either `input_data=None` or `input_data=[]` generates the following output: ``` 2023-03-09 18:09:34 | INFO | Running Speedster on CPU ``` However, the model is not optimized...

Speedster `optimize_model` function should return an object of type [BaseInferenceLearner](https://github.com/nebuly-ai/nebullvm/blob/main/nebullvm/operations/inference_learners/base.py#L41). However when optimizing Stable Diffusion models, the function returns an object of type `StableDiffusionPipeline`. It would be useful if the...

## Description The Nebullvm auto-installer (`nebullvm.installers.auto_installer`) installs only the libraries available for the current hardware, even when `--all` options are specified. This means that if I run the installer on...

MPS server not serving any request after connecting with wrong user ID

1

## Problem description MPS Server requires the clients to run with the same user ID, which is 1000 by default. If a container requesting MPS resources runs with a different...

### Description Currently, when enabling Dynamic GPU Partitioning on a node, it is possible to choose only between MIG or MPS by adding one of the following labels: `nos.nebuly.com/gpu-partitioning: "mig"`...

## Problem description When changing the partitioning mode of a node from MPS to MIG, the nvidia-device-plugin crashes and therefore any new MIG device created by `nos` is never exposed...