kubeasz/docs/guide/helm.md

148 lines
4.1 KiB
Markdown
Raw Normal View History

2018-05-23 23:58:11 +08:00
# Helm
`Helm`致力于成为k8s集群的应用包管理工具希望像linux 系统的`RPM` `DPKG`那样成功确实在k8s上部署复杂一点的应用很麻烦需要管理很多yaml文件configmap,controller,service,rbac,pv,pvc等等而helm能够整齐管理这些文档版本控制参数化安装方便的打包与分享等。
- 建议积累一定k8s经验以后再去使用helm对于初学者来说手工去配置那些yaml文件对于快速学习k8s的设计理念和运行原理非常有帮助而不是直接去使用helm面对又一层封装与复杂度。
- 本文基于helm 3建议版本helm 2 文档[请看这里](helm2.md)
2018-05-23 23:58:11 +08:00
## 安装 helm
2018-05-23 23:58:11 +08:00
在官方repo下载[release版本](https://github.com/helm/helm/releases)中自带的二进制文件即可以Linux amd64为例
```
wget https://get.helm.sh/helm-v3.2.1-linux-amd64.tar.gz
mv ./linux-amd64/helm /usr/bin
```
- 启用官方 charts 仓库
```
helm repo add stable https://kubernetes-charts.storage.googleapis.com/
```
## 使用 helm 安装应用
helm3 安装命令与 helm2 稍有变化个人习惯先下载对应charts到本地然后按照固定目录格式安装以创建一个redis集群举例
- 创建 redis-cluster 目录
2018-05-23 23:58:11 +08:00
``` bash
mkdir -p /opt/charts/redis-cluster
cd /opt/charts/redis-cluster
```
- 下载最新stalbe/redis-ha
```
helm repo update
helm pull stable/redis-ha
```
- 解压 charts复制 values.yaml设置
```
tar zxvf redis-ha-*.tgz
cp redis-ha/values.yaml .
```
- 创建 start.sh 脚本记录启动命令
```
cat > start.sh << EOF
#!/bin/sh
set -x
ROOT=$(cd `dirname $0`; pwd)
cd $ROOT
helm install redis \
--create-namespace \
--namespace dependency \
-f ./values.yaml \
./redis-ha
EOF
```
2018-05-23 23:58:11 +08:00
- 查看当前目录结构如下
```
tree .
.
├── redis-ha # redis-ha 原始charts目录
├── start.sh # 启动命名脚本
└── values.yaml # 个性化参数配置
```
- 修改当前目录的 values.yaml 为你的个性化配置
``` bash
#举例values.yaml 配置如下没有启用PV
#cat values.yaml
image:
repository: redis
tag: 5.0.6-alpine
replicas: 2
## Redis specific configuration options
redis:
port: 6379
masterGroupName: "mymaster" # must match ^[\\w-\\.]+$) and can be templated
config:
## For all available options see http://download.redis.io/redis-stable/redis.conf
min-replicas-to-write: 1
min-replicas-max-lag: 5 # Value in seconds
maxmemory: "4g" # Max memory to use for each redis instance. Default is unlimited.
maxmemory-policy: "allkeys-lru" # Max memory policy to use for each redis instance. Default is volatile-lru.
repl-diskless-sync: "yes"
rdbcompression: "yes"
rdbchecksum: "yes"
resources:
requests:
memory: 200Mi
cpu: 100m
limits:
memory: 4000Mi
## Sentinel specific configuration options
sentinel:
port: 26379
quorum: 1
resources:
requests:
memory: 200Mi
cpu: 100m
limits:
memory: 200Mi
hardAntiAffinity: true
## Configures redis with AUTH (requirepass & masterauth conf params)
auth: false
persistentVolume:
enabled: false
hostPath:
path: "/data/mcs-redis/{{ .Release.Name }}"
```
- 执行安装
```
bash ./start.sh
```
- 查看安装
```
helm ls -A
NAME NAMESPACE REVISION UPDATED STATUS CHART APP VERSION
redis dependency 1 2020-05-28 20:57:31.166002853 +0800 CST deployed redis-ha-4.4.4 5.0.6
# 查看k8s上资源
kubectl get pod,svc -n dependency
NAME READY STATUS RESTARTS AGE
pod/redis-redis-ha-server-0 2/2 Running 0 119s
pod/redis-redis-ha-server-1 2/2 Running 0 104s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/redis-redis-ha ClusterIP None <none> 6379/TCP,26379/TCP 119s
service/redis-redis-ha-announce-0 ClusterIP 10.68.41.65 <none> 6379/TCP,26379/TCP 119s
service/redis-redis-ha-announce-1 ClusterIP 10.68.64.49 <none> 6379/TCP,26379/TCP 119s
```