auto push

content/posts/k8s/controller-runtime.md

@@ -0,0 +1,40 @@
+---
+title: "Controller Runtime"
+date: 2024-06-01T10:42:13+08:00
+draft: false
+toc: true
+tags: [informer,controller,workqueue]
+---
+
+> [controller-runtime](https://github.com/kubernetes-sigs/controller-runtime)是在[client-go/tools/cache](https://github.com/kubernetes/client-go/tree/master/tools/cache)和[client-go/util/workqueue](https://github.com/kubernetes/client-go/tree/master/util/workqueue)的基础上实现的, 了解`client-go/tools/cache`和`client-go/util/workqueue`对理解`controller-runtime`很有帮助
+
+## 介绍informer
+
+带着问题看
+
+## 开发CRD时想到的一些问题
+
+### 更新local store缓存和触发reconcile是否有先后顺序
+
+### 同一个crd object会不会同时被reconcile
+
+这个全靠Queue数据结构设计的精妙, 保证了正在执行的reconcile不会处理相同的object
+
+向queue中增加object之前会检查是否有次object存在于queue中，如果不存在则加入dirty set，如果也不存在于processing set才会加入queue中，当processing中的处理完成之后（调用Done），会将object从processing set种移除，如果次object在处理过程中加入到了dirty set，则将object再次加入到queue中
+https://www.cnblogs.com/daniel-hutao/p/18010835/k8s_clientgo_workqueue
+
+有几种队列，Queue，DelayingQueue，RateLimitingQueue
+
+### 如何解决进入reconcile之后读到的是旧数据的问题
+
+读到旧数据是否说明是先出发reconcile再更新local store的
+
+My cache might be stale if I read from a cache! How should I deal with that?
+
+在更新或patch status之后，通过wait.Pool(100ms, 2s, func()(bool, error))校验cache中的本object数据直至更新
+
+https://github.com/kubernetes-sigs/controller-runtime/blob/main/FAQ.md#q-my-cache-might-be-stale-if-i-read-from-a-cache-how-should-i-deal-with-that
+
+https://github.com/kubernetes/test-infra/blob/8f0f19a905a20ed6f76386e5e11343d4bc2446a7/prow/plank/reconciler.go#L516-L520
+
+

content/posts/k8s/rke2.md

@@ -0,0 +1,125 @@
+---
+title: "RKE2"
+date: 2024-07-01T21:24:49+08:00
+draft: false
+toc: true
+tags: [k8s,rke2]
+---
+
+> 通过RKE2快速搭建测试使用的k8s集群环境
+
+## 环境准备
+
+1. 准备bridge网络br0
+2. 准备ubuntu 22.04 server qcow2镜像
+3. 准备libvirt环境
+
+## 创建虚拟机
+
+### 准备 cloudinit 镜像
+
+```bash
+cat <<EOFALL > gen-cloudinit-iso.sh
+#!/bin/bash
+
+set -eux
+
+CLOUD_INIT_DIR="/var/lib/libvirt/disks/\${VM}/cloudinit"
+FILENAME="\${CLOUD_INIT_DIR}/init.iso"
+
+mkdir -p \${CLOUD_INIT_DIR}
+
+cat <<EOF > \${CLOUD_INIT_DIR}/meta-data
+instance-id: \${VM}
+local-hostname: \${VM}
+EOF
+
+# 更多配置参照 https://cloudinit.readthedocs.io/en/latest/explanation/format.html
+cat <<EOF > \${CLOUD_INIT_DIR}/user-data
+#cloud-config
+EOF
+
+# 参考 kubevirt /pkg/cloud-init/cloud-init.go:defaultIsoFunc
+xorrisofs -output \$FILENAME -volid cidata -joliet -rock -partition_cyl_align on \${CLOUD_INIT_DIR}/user-data \${CLOUD_INIT_DIR}/meta-data
+EOFALL
+
+VM=k8s-node01 bash gen-cloudinit-iso.sh
+```
+
+### 准备系统盘并创建虚拟机
+
+```bash
+VM=k8s-node01
+mkdir -p /var/lib/libvirt/disks/${VM}
+qemu-img create -f qcow2 -F qcow2 -b /var/lib/libvirt/images/ubuntu.qcow2 /var/lib/libvirt/disks/${VM}/sysdisk.qcow2 200G
+qemu-img create -f qcow2 /var/lib/libvirt/disks/${VM}/datadisk01.qcow2 500G
+qemu-img create -f qcow2 /var/lib/libvirt/disks/${VM}/datadisk02.qcow2 500G
+qemu-img create -f qcow2 /var/lib/libvirt/disks/${VM}/datadisk03.qcow2 500G
+
+virt-install \
+  --name ${VM} \
+  --memory 16384 \
+  --vcpus 8 \
+  --disk /var/lib/libvirt/disks/${VM}/sysdisk.qcow2,device=disk,bus=scsi \
+  --disk /var/lib/libvirt/disks/${VM}/datadisk01.qcow2,device=disk,bus=scsi \
+  --disk /var/lib/libvirt/disks/${VM}/datadisk02.qcow2,device=disk,bus=scsi \
+  --disk /var/lib/libvirt/disks/${VM}/datadisk03.qcow2,device=disk,bus=scsi \
+  --disk /var/lib/libvirt/disks/${VM}/cloudinit/init.iso,device=cdrom,bus=scsi \
+  --network bridge=br0 \
+  --import \
+  --os-variant ubuntu22.10 \
+  --noautoconsole
+```
+
+## 安装 RKE2
+
+### 脚本在线安装
+
+```bash
+# TODO 指定 cni
+curl -sfL https://rancher-mirror.rancher.cn/rke2/install.sh | INSTALL_RKE2_MIRROR=cn sh -
+systemctl enable rke2-server.service
+systemctl start rke2-server.service
+```
+
+### 离线安装
+
+TODO
+
+### 配置
+
+```bash
+# kubectl ctr crictl...
+CONFIG="PATH=\$PATH:/var/lib/rancher/rke2/bin/"
+grep "$CONFIG" ~/.bashrc || echo "$CONFIG" >> ~/.bashrc && source ~/.bashrc
+
+# command auto completiom
+CONFIG="source <(kubectl completion bash)"
+grep "$CONFIG" ~/.bashrc || echo "$CONFIG" >> ~/.bashrc && source ~/.bashrc
+
+# KUBECONFIG ENV
+CONFIG="export KUBECONFIG=/etc/rancher/rke2/rke2.yaml"
+grep "$CONFIG" ~/.bashrc || echo "$CONFIG" >> ~/.bashrc && source ~/.bashrc
+
+# CRI_CONFIG_FILE
+CONFIG="export CRI_CONFIG_FILE=/var/lib/rancher/rke2/agent/etc/crictl.yaml"
+grep "$CONFIG" ~/.bashrc || echo "$CONFIG" >> ~/.bashrc && source ~/.bashrc
+
+# install helm
+HELM_LATEST_VERSION=v3.15.2
+wget https://get.helm.sh/helm-${HELM_LATEST_VERSION}-linux-amd64.tar.gz
+tar -zxvf helm-${HELM_LATEST_VERSION}-linux-amd64.tar.gz
+mv linux-amd64/helm /usr/local/bin/helm
+rm -f helm-${HELM_LATEST_VERSION}-linux-amd64.tar.gz && rm -rf linux-amd64/
+```
+
+## 安装 rook ceph
+
+https://rook.io/docs/rook/latest-release/Getting-Started/quickstart/
+
+## 安装 kube-prometheus-stack
+
+## 参考
+
+- [[RKE2 docs] quickstart](https://docs.rke2.io/zh/install/quickstart)
+- [[RKE2 docs] CLI 工具](https://docs.rke2.io/zh/reference/cli_tools)

content/posts/linux/shell-script.md

@@ -493,6 +493,12 @@ docker save -o all.tar image-a:1.0.0 image-b:1.0.0 image-c:1.0.0
 docker load < all.tar
 ```
 
+## TODO
+
+- 看一下 [the art of command line](https://github.com/jlevy/the-art-of-command-line) 然后融入进这片文档
+- 增加 shell 常用的控制循环和条件判断语法
+
 ## 参考
 
-- [awk](https://www.ruanyifeng.com/blog/2018/11/awk.html)
+- [awk](https://www.ruanyifeng.com/blog/2018/11/awk.html)
+- [the art of command line](https://github.com/jlevy/the-art-of-command-line)

content/posts/virt/cloudinit.md

@@ -8,7 +8,7 @@ tags: [cloudinit]
 
 ## cloudinit 介绍
 
-> 用于初始化网络, 主机名, 根文件系统, 用户名密码
+> 用于在新建的虚拟机中进行时间设置、密码设置、扩展跟文件系统所在分区、设置主机名、运行脚本、安装软件包等初始化设置
 
 ## cloudinit iso 镜像制作