关于linux:容器化应用系统上生产的最佳实践

前言

最近忙的要死, 👻👻👻. 上一周来了一次比 996 更猛的 907. 这周二终于有点遭不住了, 调休一天, 略微休憩一下.

同时手痒的不行, 把筹备了良久的重磅文章发上来哈哈. 😆😆😆

不过工夫还是有点仓促, 所以这次就先开个头, 前面有工夫再细化.

容器化利用零碎上生产的最佳实际

1. 查看镜像、容器是否是用 root 启动以及配置其余特权. 如无必要, 一律应用普通用户.
2. 查看镜像 LANG 配置: LANG = en_US.UTF-8. 目标: 防止生产呈现 乱码等问题
3. 查看镜像时区配置: TZ=Asia/Shanghai 目标: 防止生产呈现时区不统一的问题
4. 配置内部化. 内部化伎俩有多种:
5. 同一个镜像, 从测试流转到生产. 给镜像打 ${version} 或${gitCommitId}这一类的标签. 目标: 通过版本号或 commit id, 保障正确地的版本流转到生产
6. 探讨每个组件的
7. 日志输入优化:
8. (可选) 依据须要, 装置 redis/kafka/rabbitmq 集群(并配置 exporter 监控)
9. 微服务参数优化:
10. 制作 DEV, TEST, UAT, Pre-PROD, PROD 的 DevOps pipeline.
11. 配置 Readiness 和 Liveness 探针.
12. 减少 JMX-exporter 监控和 Tracing 监控.
13. NGINX conf 倡议减少: worker_processes 1; 而后按需调节正本数.
14. (可选)配置 PDB, 指定降级或重启过程中:maxUnavailable 或minAvailable(特地实用于: 有状态利用. 典型如: redis, kafka, zookeeper 等)
15. 配置反亲和性podAntiAffinity. 保障同一组微服务 / 利用 / 组件尽可能打散在不同 node 上.

5-6 操作步骤:

      nodeSelector:
        zone: internet

11 步示例如下:

          livenessProbe:
            tcpSocket:
              port: 8080
            initialDelaySeconds: 60
          readinessProbe:
            httpGet:
              path: /myapp/services/
              port: 8080
              scheme: HTTP
            initialDelaySeconds: 60

14 步示例如下: (留神关键词: maxUnavailable 和 minAvailable)

kind: PodDisruptionBudget
apiVersion: policy/v1beta1
metadata:
  name: kafka-prod-kafka
  labels:
    app.kubernetes.io/instance: kafka-prod
    app.kubernetes.io/managed-by: strimzi-cluster-operator
    app.kubernetes.io/name: strimzi
    strimzi.io/cluster: kafka-prod
    strimzi.io/kind: Kafka
    strimzi.io/name: kafka-prod-kafka
spec:
  selector:
    matchLabels:
      strimzi.io/cluster: kafka-prod
      strimzi.io/kind: Kafka
      strimzi.io/name: kafka-prod-kafka
  maxUnavailable: 1

kind: PodDisruptionBudget
apiVersion: policy/v1beta1
metadata:
  name: redis-cluster-redis
  namespace: myapp
  labels:
    app.kubernetes.io/component: redis
    app.kubernetes.io/managed-by: redis-operator
    app.kubernetes.io/name: redis
    app.kubernetes.io/part-of: redis-cluster
    redis.kun/v1beta1: myapp_redis
spec:
  minAvailable: 2
  selector:
    matchLabels:
      app.kubernetes.io/component: redis
      app.kubernetes.io/managed-by: redis-operator
      app.kubernetes.io/name: redis
      app.kubernetes.io/part-of: redis-cluster
      redis.kun/v1beta1: myapp_redis

15 步骤示例如下: (留神关键词: podAntiAffinity)

kind: StatefulSet
apiVersion: apps/v1
metadata:
  name: redis-cluster-redis
  labels:
    app.kubernetes.io/component: redis
    app.kubernetes.io/managed-by: redis-operator
    app.kubernetes.io/name: redis
    app.kubernetes.io/part-of: redis-cluster
    redis.kun/v1beta1: myapp_redis
spec:
  replicas: 3
  selector:
    matchLabels:
      app.kubernetes.io/component: redis
      app.kubernetes.io/managed-by: redis-operator
      app.kubernetes.io/name: redis
      app.kubernetes.io/part-of: redis-cluster
      redis.kun/v1beta1: myapp_redis
  template:
    metadata:
      creationTimestamp: null
      labels:
        app.kubernetes.io/component: redis
        app.kubernetes.io/managed-by: redis-operator
        app.kubernetes.io/name: redis
        app.kubernetes.io/part-of: redis-cluster
        redis.kun/v1beta1: myapp_redis
    spec:
      containers:
        <...>
      affinity:
        podAntiAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
            - weight: 100
              podAffinityTerm:
                labelSelector:
                  matchLabels:
                    app.kubernetes.io/component: redis
                    app.kubernetes.io/managed-by: redis-operator
                    app.kubernetes.io/name: redis
                    app.kubernetes.io/part-of: redis-cluster
                    redis.kun/v1beta1: myapp_redis
                topologyKey: kubernetes.io/hostname
...

完

🎉🎉🎉

本文由博客一文多发平台 OpenWrite 公布！