1. Deploy GPU sharing plug-in in k8s Before deployment, you need to ensure that nvidia-driver and nvidia-docker are installed on the k8s node, and the default runtime of docker is set to nvidia Copy 1 2 3 4 5 6 7 8 9 10 # cat /etc/docker/daemon.json { "runtimes": { "nvidia": { "path": "/usr/bin/nvidia-container-runtime", "runtimeArgs": [] } }, "default-runtime": "nvidia", } 1. Install gpushare-device-plugin in helm Copy 1 2 3

In order to verify the private deployment later , the intranet environment needs to quickly build a k8s cluster. Because I usually use kubeasz and kubespray to handle large-scale clusters before, this time for small environment clusters, it will be more efficient to directly use kubeadm to deploy. The following records the kubeadm deployment process: Cluster nodes: Copy 1 2 3 4 192.168.1.206 sd-cluster-206 node 192.168.1.207 sd-cluster-207 master,etcd 192.168.1.208 sd-cluster-208

Requirements: Alibaba Cloud cluster can resolve internal domain names Office network resolves internal domain names + office network Internet resolution Solution: For the first problem, use Alibaba Cloud PrivateZone for resolution For the second problem, configure the internal domain name zone in PrivateZone, and then synchronize it to the office network bind9 server through Alibaba Cloud synchronization tool; For the office network DNS resolution entry, use Dnsmasq to process, and

Earlier, we introduced how to install and trigger tasks in Argo Workflow. This article mainly introduces a new tool: What is ArgoEvents? Argo Events is an event-driven Kubernetes workflow automation framework. It supports more than 20 different events (such as webhooks, S3 drops, cronjobs, message queues - such as Kafka, GCP PubSub, SNS, SQS, etc.) Features: Supports events from 20+ event sources and more than 10 triggers. Ability to customize

Introduction & Architecture Argo Workflows is an open source container-level workflow engine for coordinating parallel jobs on Kubernetes. Argo Workflows implements the entire architecture functionality by abstracting Kubernetes CRDs (Custom Resource Definitions), such as Workflow Template, Workflow, and Cron Workflow. What can Argo workflow do? Define workflows, where each step in the workflow is a container. Model multi-step workflows as a series of tasks, or use directed acyclic graphs (DAGs)

When building Npm for front-end projects, it often takes a long time to pull the front-end library. In addition, reuse between different jobs is also a problem. Whether it is artifacts or cache, we need to persist the reused files. Here we take cache as an example Note: The Gitlab runner here is deployed to the k8s cluster using the helm chart method, and the runner deployment is ignored; the

Quickly create an Alibaba Cloud ECS host Specify the terraform version Here we specify the Alibaba Cloud provider version information and set the terraform version requirements Copy 1 2 3 4 5 6 7 8 9 10 11 12 13 # mkdir aliyun-ecs-one && cd aliyun-ecs-one # touch versions.tf # vim versions.tf terraform { required_providers { alicloud = { source = "aliyun/alicloud" version = "1.115.1" } } required_version = ">=

Install Terraform Mac system installation Copy 1 2 brew tap hashicorp/tap brew install hashicorp/tap/terraform Linux system installation Ubuntu installation Copy 1 2 3 curl -fsSL https://apt.releases.hashicorp.com/gpg | sudo apt-key add - sudo apt-add-repository "deb [arch=amd64] https://apt.releases.hashicorp.com $(lsb_release -cs) main" sudo apt-get update && sudo apt-get install terraform CentOS system Copy 1 2 3 sudo yum install -y yum-utils sudo yum-config-manager --add-repo https://rpm.releases.hashicorp.com/RHEL/hashicorp.repo sudo yum -y install terraform Verify installation Copy