VK-CMD

From epsciwiki
Revision as of 21:21, 27 February 2024 by Tsai (talk | contribs) (Created page with "<span id="docker-image-for-virtual-kubelet-cmd"></span> = Docker Image for Virtual-Kubelet-Cmd = This repository contains the Docker image for [https://github.com/tsaie79/vir...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Docker Image for Virtual-Kubelet-Cmd

This repository contains the Docker image for virtual-kubelet-cmd, a BASH command provider for Virtual Kubelet. The image construction process is based on the KinD project guidelines.

Overview

This repository hosts the virtual-kubelet-cmd binary and its required files. When a container is created from this image, it generates a vk-cmd directory. The binary and its related files are subsequently transferred into this directory. To execute vk-cmd, users can use the scripts found in the run directory.

Docker Image Construction

The Dockerfile used to build the image for this project is located at docker/images/base/Dockerfile. To build the Docker image, run the make quick command in your terminal.

Executing vk-cmd

To execute vk-cmd, adhere to the following procedure: 1. Configure the required environment variables. 2. Establish SSH tunnels, if they are not already in place. 3. Execute the start.sh script located in the vk-cmd directory.

Note Scripts for ruunning vk-cmd are located in the run directory.

1. Configuring the Environment Variables

This section defines several environment variables essential for configuring the Virtual Kubelet command. Each variable serves a specific purpose:

Variable Name Description
NODENAME The name of the node in the Kubernetes cluster.
KUBECONFIG Points to the location of the Kubernetes configuration file, which is used to connect to the Kubernetes API server. By default, it’s located at $HOME/.kube/config.
VKUBELET_POD_IP The IP address of the VK that metrics server talks to. If the metrics server is running in a Docker container and VK is running on the same host, this is typically the IP address of the docker0 interface.
KUBELET_PORT The port on which the Kubelet service is running. The default port for Kubelet is 10250. This is for the metrics server and should be unique for each node.
JIRIAF_WALLTIME Sets a limit on the total time that a node can run. It should be a multiple of 60 and is measured in seconds. If it’s set to 0, there is no time limit.
JIRIAF_NODETYPE Specifies the type of node that the job will run on. This is just for labeling purposes and doesn’t affect the actual job.
JIRIAF_SITE Used to specify the site where the job will run. This is just for labeling purposes and doesn’t affect the actual job.

2. Building SSH Tunnels

If you are running vk-cmd on a remote machine, you will need to build an SSH tunnel to the control plane. Two tunnels are required: one for the API server and one for the metrics server. The following commands can be used to build these tunnels:

Communication Direction Environment Variables SSH Command
VK to API server APISERVER_PORT=6443
CONTROL_PLANE_IP="control-plane-ip"
ssh -NfL $APISERVER_PORT:localhost:$APISERVER_PORT $CONTROL_PLANE_IP
Metrics server to VK CONTROL_PLANE_IP="control-plane-ip" ssh -NfR *:$KUBELET_PORT:localhost:$KUBELET_PORT $CONTROL_PLANE_IP

Note: The * symbol in the SSH command for Metrics server to VK is used to bind the port to all network interfaces. To ensure the port is accessible from all interfaces, you must set GatewayPorts to yes in the /etc/ssh/sshd_config file. After making this change, restart the SSH service on the control plane using the command sudo service ssh restart.

3. Executing the start.sh Script

Once the environment variables are configured and the SSH tunnels are established, you can initiate the start.sh script located in the vk-cmd directory. This script launches the vk-cmd binary and establishes a connection with the Kubernetes API server.

./start.sh $KUBECONFIG $NODENAME $VKUBELET_POD_IP $KUBELET_PORT $JIRIAF_WALLTIME $JIRIAF_NODETYPE $JIRIAF_SITE

Example of Running vk-cmd

#!/bin/bash

## ssh tunnel

## name node
export NODENAME="vk"
export KUBECONFIG="$HOME/.kube/config" # the config file that is used to connect to the api-server. Usually it is located at $HOME/.kube/config
export VKUBELET_POD_IP="123.123.123.123" # the ip address of the node that is used to connect to the api-server. Usually it is the ip address of the docker0 interface if the api-server is running in a docker container 
export KUBELET_PORT="10260"

export JIRIAF_WALLTIME="0" # should be multiple of 60 and in seconds. if 0, then no walltime limit
export JIRIAF_NODETYPE="cpu"
export JIRIAF_SITE="mylin"

## run vk-cmd
### update the image to the latest version
export VK_CMD_IMAGE="jlabtsai/vk-cmd:main"
docker pull $VK_CMD_IMAGE

container_id=$(docker run -itd --rm --name vk-cmd $VK_CMD_IMAGE)
docker cp ${container_id}:/vk-cmd `pwd` && docker stop ${container_id}

cd `pwd`/vk-cmd

echo "=============================="

echo "api-server config: $KUBECONFIG; nodename: $NODENAME is runnning..."
echo "vk ip: $VKUBELET_POD_IP from view of metrics server; vk kubelet port: $KUBELET_PORT"
echo "walltime: $JIRIAF_WALLTIME; nodetype: $JIRIAF_NODETYPE; site: $JIRIAF_SITE"

./start.sh $KUBECONFIG $NODENAME $VKUBELET_POD_IP $KUBELET_PORT $JIRIAF_WALLTIME $JIRIAF_NODETYPE $JIRIAF_SITE

Environment Characteristics for vk-cmd Launch Scenarios

Environment Container Runtime Interface (CRI) Is Host Network Available in Container? Are Host Shell Environment Variables Available?
Jiriaf2301 Docker No No
ifarm Apptainer Yes Yes
Perlmutter (NERSC) Shifter Yes Yes

Use kubernetes API to get information about the cluster

Please see tools/kubernetes-api/readme.md for details.

Related Resources

The vk-cmd implementation is inspired by the mock provider in the Virtual Kubelet project. The source code for vk-cmd can be found in the virtual-kubelet-cmd repository. The Dockerfile used to build the vk-cmd image is based on the guidelines from the KinD project. The built image is hosted on Docker Hub.