环境准备

方案1

  • 购买至少4 台阿里云服务器

方案2

  • 电脑性能够强大, 开4 台虚拟机

image.png
image.png
我的电脑性能如下:
image.png

工作模式

Docker Engine 1.12 introduces swarm mode that enables you to create a cluster of one or more Docker Engines called a swarm. A swarm consists of one or more nodes: physical or virtual machines running Docker Engine 1.12 or later in swarm mode.
There are two types of nodes: managers and workers.
Docker Swarm - 图4
If you haven’t already, read through the swarm mode overview and key concepts.

节点

  • 管理节点 (上图Manager)
    • 管理节点可以与管理节点互相通信
    • 管理节点可以操作工作节点, 反之不行
  • 工作节点 (上图Worker)

    搭建集群

    docker swarm 命令用法

    ```shell docker swarm —help

Usage: docker swarm COMMAND

Manage Swarm

Options: —help Print usage

Commands:

初始化集群

init Initialize a swarm

加入集群

join Join a swarm as a node and/or manager

创建集群token

join-token Manage join tokens

离开集群

leave Leave the swarm unlock Unlock swarm unlock-key Manage the unlock key

更新集群

update Update the swarm

Run ‘docker swarm COMMAND —help’ for more information on a command.

  2. <a name="IzzIU"></a>
  3. ## docker swarm init 命令用法
  4. ```shell
  5. docker swarm init --help
  7. Usage:  docker swarm init [OPTIONS]
  9. Initialize a swarm
  11. Options:
  12.       # 广播地址
  13.       --advertise-addr string           Advertised address (format: <ip|interface>[:port])
  14.       --autolock                        Enable manager autolocking (requiring an unlock key to start a stopped manager)
  15.       --cert-expiry duration            Validity period for node certificates (ns|us|ms|s|m|h) (default 2160h0m0s)
  16.       --dispatcher-heartbeat duration   Dispatcher heartbeat period (ns|us|ms|s|m|h) (default 5s)
  17.       --external-ca external-ca         Specifications of one or more certificate signing endpoints
  18.       --force-new-cluster               Force create a new cluster from current state
  19.       --help                            Print usage
  20.       --listen-addr node-addr           Listen address (format: <ip|interface>[:port]) (default 0.0.0.0:2377)
  21.       --max-snapshots uint              Number of additional Raft snapshots to retain
  22.       --snapshot-interval uint          Number of log entries between Raft snapshots (default 10000)
  23.       --task-history-limit int          Task history retention limit (default 5)

实例

  1. 初始化集群, 并将指定的机器设置为主节点 (Manager 角色)

image.png
(上图在192.168.101.67 服务器上操作)

  1. 使用上图中的命令将另外机器加入到1. 步骤中初始化完毕的集群中 (Work 角色)

image.png
(上图在192.168.101.68 服务器上操作)
image.png
(上图在192.168.101.67 服务器上操作)

  1. 可以使用以下命令生成在1. 步骤中的加入命令 ```shell

    以manager 角色加入集群

    docker swarm join-token manager

以worker 角色加入集群

docker swarm join-token worker

  1. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644219814198-902ba92a-2192-4d41-9637-8656d01f4612.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=121&id=uce5f0a64&margin=%5Bobject%20Object%5D&name=image.png&originHeight=212&originWidth=1604&originalType=binary&ratio=1&rotation=0&showTitle=false&size=23305&status=done&style=none&taskId=u03199486-d7dc-4ffe-9db3-15dba0bd1e6&title=&width=916.5714285714286)(上图在192.168.101.67 服务器上操作)
  3. 4. 复制3. 步骤中的命令到另外未加入集群的服务中, 使其以work 角色加入集群
  5. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644219918125-4a31c77f-9fa1-4c7d-885c-4016d437a480.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=70&id=uc5fe641b&margin=%5Bobject%20Object%5D&name=image.png&originHeight=123&originWidth=1637&originalType=binary&ratio=1&rotation=0&showTitle=false&size=19446&status=done&style=none&taskId=u82c236e5-993e-4c92-b09b-28c18f8b8e6&title=&width=935.4285714285714)<br />(上图在192.168.101.70 服务器上操作)<br />![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644219956513-b1ff6210-d6f5-49b1-98c9-1d8acdb9dc56.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=90&id=u2baf133e&margin=%5Bobject%20Object%5D&name=image.png&originHeight=157&originWidth=1487&originalType=binary&ratio=1&rotation=0&showTitle=false&size=26539&status=done&style=none&taskId=u1d68bc8f-2f00-4d41-bf8a-7f5de0c8b42&title=&width=849.7142857142857)<br />(上图在192.168.101.67 服务器上操作)
  7. 5. 使用docker swarm join-token manager 命令生成以manager 角色加入集群的命令, 与4. 步骤一样, 使另外一台服务器以manager 角色加入集群, 再返回到67 上查看集群成员信息
  9. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644220092512-836cdec0-79e4-4582-aa37-7ca59085ac82.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=105&id=ub5d30f11&margin=%5Bobject%20Object%5D&name=image.png&originHeight=184&originWidth=1446&originalType=binary&ratio=1&rotation=0&showTitle=false&size=32544&status=done&style=none&taskId=uc82b07d9-0413-48b5-a48a-ef441a74fa8&title=&width=826.2857142857143)
  11. <a name="LGtTM"></a>
  12. ## 搭建swarm 集群的流程
  14. 1. 生成主节点 (使用init 命令)
  15. 1. 加入 (manager/worker)
  17. <a name="foBGR"></a>
  18. # Raft 一致性算法
  19. <a name="aNPNR"></a>
  20. ## 抛出问题
  22. - 集群中(现在是双主双从的模式), 假设一个节点挂了, 其他节点是否可以用?
  24. <a name="dwvhu"></a>
  25. ## 定义
  27. - 保证大多数节点存活才可以用, 至少 > 1, 集群至少大于3 台
  29. <a name="ufzs7"></a>
  30. ## 实验
  32. 1. 将67 主节点停止 (模拟宕机), 可以发现在双主双从模式下的集群中, 另外一个节点也不能使用了
  34. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644220559001-5754a06d-027d-4507-a32d-0239255e28ac.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=48&id=u04ee40af&margin=%5Bobject%20Object%5D&name=image.png&originHeight=84&originWidth=1329&originalType=binary&ratio=1&rotation=0&showTitle=false&size=10878&status=done&style=none&taskId=u8028e613-4d45-47ae-bd33-4e5f26f1769&title=&width=759.4285714285714)<br />(上图在192.168.101.72 服务器上操作)
  36. 2. 再次上线67 主节点, 可以发现leader 从67 变为了72
  38. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644220780332-968ae7b3-1c4b-4c8c-a0d3-7533e0e61421.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=110&id=u769f80c9&margin=%5Bobject%20Object%5D&name=image.png&originHeight=192&originWidth=1500&originalType=binary&ratio=1&rotation=0&showTitle=false&size=33200&status=done&style=none&taskId=u0a1bfbdf-c519-43d2-8652-fdd9d436a42&title=&width=857.1428571428571)<br />(上图在192.168.101.67 服务器上操作)
  40. 3. 将70 (worker 节点) 离开集群, 可以发现在主节点的管理列表中, 该node 的status 变为down
  42. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644220880414-e07bd69d-6a4e-4b3d-84be-46ecacb77006.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=40&id=u93ef9338&margin=%5Bobject%20Object%5D&name=image.png&originHeight=70&originWidth=658&originalType=binary&ratio=1&rotation=0&showTitle=false&size=6687&status=done&style=none&taskId=u9e8bd028-0ce1-4cf2-a22d-7d2c0b7078f&title=&width=376)<br />(上图在192.168.101.70 服务器上操作)<br />![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644220903702-c9d2e860-ea2d-4ff6-a19e-064fbcad8312.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=105&id=u6fed4f5b&margin=%5Bobject%20Object%5D&name=image.png&originHeight=184&originWidth=1475&originalType=binary&ratio=1&rotation=0&showTitle=false&size=32773&status=done&style=none&taskId=ud6a86ea3-1f27-47dc-ae6c-a93cde98d95&title=&width=842.8571428571429)<br />(上图在192.168.101.72 服务器上操作)
  44. 4. 在任意manager 角色的节点上生成以manager 角色加入集群的命令, 然后在3. 步骤中离开集群的节点再次以manager 角色加入集群中, 可以发现该节点可以使用node ls 命令查看集群成员信息
  46. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644221130910-d95fe3f1-b18b-4a1a-8685-5ea474924323.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=214&id=u1de743b7&margin=%5Bobject%20Object%5D&name=image.png&originHeight=375&originWidth=1667&originalType=binary&ratio=1&rotation=0&showTitle=false&size=66653&status=done&style=none&taskId=u801f96a8-f348-47d5-a482-1769eff916b&title=&width=952.5714285714286)
  48. 5. 在67 和70 两个manager 服务器上在重复1. 步骤, 只剩下72 主节点存活, 发现72 主节点无法使用node ls 命令, 整个集群down
  50. ![image.png](https://cdn.nlark.com/yuque/0/2022/png/2516625/1644221692694-1d9d9aa3-4587-4c8b-ad29-a048f974ef98.png#clientId=uac7adfdf-922f-4&crop=0&crop=0&crop=1&crop=1&from=paste&height=41&id=ue75da0d4&margin=%5Bobject%20Object%5D&name=image.png&originHeight=72&originWidth=1327&originalType=binary&ratio=1&rotation=0&showTitle=false&size=9382&status=done&style=none&taskId=u53a0e4f0-0bca-44c4-9ce3-7094e093fad&title=&width=758.2857142857143)
  52. <a name="wdKhR"></a>
  53. ## 总结
  55. - 集群需要可用状态, 必须保证有3 个主节点, 其中: 
  56.    - 2 个主节点down, 整个集群down
  57.    - 1 个主节点down, 整个集群可以继续使用
  58.    - 总结成一句话: 保证大多数存活, 才可以使用
  60. <a name="ROi7G"></a>
  61. # 动态扩缩容
  62. <a name="eKUej"></a>
  63. ## 定义
  65. - 将容器变为了服务 (e.g. redis 服务可能有10 个副本组成, 也就是同时开启了10 个redis 容器)
  67. <a name="FGM4L"></a>
  68. ## docker service 命令用法
  69. ```shell
  70. docker service --help
  72. Usage:  docker service COMMAND
  74. Manage services
  76. Options:
  77.       --help   Print usage
  79. Commands:
  80.   create      Create a new service
  81.   inspect     Display detailed information on one or more services
  82.   ls          List services
  83.   ps          List the tasks of a service
  84.   rm          Remove one or more services
  85.   scale       Scale one or multiple replicated services
  86.   update      Update a service
  88. Run 'docker service COMMAND --help' for more information on a command.

体验

docker service create 命令用法

  1. docker service create --help
  3. Usage:  docker service create [OPTIONS] IMAGE [COMMAND] [ARG...]
  5. Create a new service
  7. Options:
  8.       --constraint list                  Placement constraints (default [])
  9.       --container-label list             Container labels (default [])
  10.       --dns list                         Set custom DNS servers (default [])
  11.       --dns-option list                  Set DNS options (default [])
  12.       --dns-search list                  Set custom DNS search domains (default [])
  13.       --endpoint-mode string             Endpoint mode (vip or dnsrr)
  14.   #环境配置
  15.   -e, --env list                         Set environment variables (default [])
  16.       --env-file list                    Read in a file of environment variables (default [])
  17.       --group list                       Set one or more supplementary user groups for the container (default [])
  18.       --health-cmd string                Command to run to check health
  19.       --health-interval duration         Time between running the check (ns|us|ms|s|m|h)
  20.       --health-retries int               Consecutive failures needed to report unhealthy
  21.       --health-timeout duration          Maximum time to allow one check to run (ns|us|ms|s|m|h)
  22.       --help                             Print usage
  23.       --host list                        Set one or more custom host-to-IP mappings (host:ip) (default [])
  24.       --hostname string                  Container hostname
  25.   #查看基本节点列表    
  26.   -l, --label list                       Service labels (default [])
  27.       --limit-cpu decimal                Limit CPUs (default 0.000)
  28.       --limit-memory bytes               Limit Memory (default 0 B)
  29.       --log-driver string                Logging driver for service
  30.       --log-opt list                     Logging driver options (default [])
  31.       --mode string                      Service mode (replicated or global) (default "replicated")
  32.       --mount mount                      Attach a filesystem mount to the service
  33.       --name string                      Service name
  34.       --network list                     Network attachments (default [])
  35.       --no-healthcheck                   Disable any container-specified HEALTHCHECK
  36.   #查看暴露的节点端口
  37.   -p, --publish port                     Publish a port as a node port
  38.       --replicas uint                    Number of tasks
  39.       --reserve-cpu decimal              Reserve CPUs (default 0.000)
  40.       --reserve-memory bytes             Reserve Memory (default 0 B)
  41.       --restart-condition string         Restart when condition is met (none, on-failure, or any)
  42.       --restart-delay duration           Delay between restart attempts (ns|us|ms|s|m|h)
  43.       --restart-max-attempts uint        Maximum number of restarts before giving up
  44.       --restart-window duration          Window used to evaluate the restart policy (ns|us|ms|s|m|h)
  45.       --secret secret                    Specify secrets to expose to the service
  46.       --stop-grace-period duration       Time to wait before force killing a container (ns|us|ms|s|m|h)
  47.   -t, --tty                              Allocate a pseudo-TTY
  48.       --update-delay duration            Delay between updates (ns|us|ms|s|m|h) (default 0s)
  49.       --update-failure-action string     Action on update failure (pause|continue) (default "pause")
  50.       --update-max-failure-ratio float   Failure rate to tolerate during an update
  51.       --update-monitor duration          Duration after each task update to monitor for failure (ns|us|ms|s|m|h) (default 0s)
  52.       --update-parallelism uint          Maximum number of tasks updated simultaneously (0 to update all at once) (default 1)
  53.   #配置用户信息
  54.   -u, --user string                      Username or UID (format: <name|uid>[:<group|gid>])
  55.       --with-registry-auth               Send registry authentication details to swarm agents
  56.   #配置工作目录
  57.   -w, --workdir string                   Working directory inside the container
  • 实例:
    • 启动自定义的nginx 服务

image.png

  • 根据服务名查看正在运行的nginx 服务

image.png

  • 查看正在运行的nginx 服务的具体信息
    1. docker service inspect mynginx
    2. [
    3. {
    4.    "ID": "ljlbfk4122sy3cfczmzbjv6k7",
    5.    "Version": {
    6.        "Index": 29
    7.    },
    8.    "CreatedAt": "2022-02-07T11:52:41.695689971Z",
    9.    "UpdatedAt": "2022-02-07T11:52:41.699705732Z",
    10.    "Spec": {
    11.        "Name": "mynginx",
    12.        "TaskTemplate": {
    13.            "ContainerSpec": {
    14.                "Image": "nginx:latest@sha256:0d17b565c37bcbd895e9d92315a05c1c3c9a29f762b011a10c54a66cd53c9b31",
    15.                "DNSConfig": {}
    16.            },
    17.            "Resources": {
    18.                "Limits": {},
    19.                "Reservations": {}
    20.            },
    21.            "RestartPolicy": {
    22.                "Condition": "any",
    23.                "MaxAttempts": 0
    24.            },
    25.            "Placement": {},
    26.            "ForceUpdate": 0
    27.        },
    28.        "Mode": {
    29.            "Replicated": {
    30.                "Replicas": 1
    31.            }
    32.        },
    33.        "UpdateConfig": {
    34.            "Parallelism": 1,
    35.            "FailureAction": "pause",
    36.            "MaxFailureRatio": 0
    37.        },
    38.        "EndpointSpec": {
    39.            "Mode": "vip",
    40.            "Ports": [
    41.                {
    42.                    "Protocol": "tcp",
    43.                    "TargetPort": 80,
    44.                    "PublishedPort": 8888,
    45.                    "PublishMode": "ingress"
    46.                }
    47.            ]
    48.        }
    49.    },
    50.    "Endpoint": {
    51.        "Spec": {
    52.            "Mode": "vip",
    53.            "Ports": [
    54.                {
    55.                    "Protocol": "tcp",
    56.                    "TargetPort": 80,
    57.                    "PublishedPort": 8888,
    58.                    "PublishMode": "ingress"
    59.                }
    60.            ]
    61.        },
    62.        "Ports": [
    63.            {
    64.                "Protocol": "tcp",
    65.                "TargetPort": 80,
    66.                "PublishedPort": 8888,
    67.                "PublishMode": "ingress"
    68.            }
    69.        ],
    70.        "VirtualIPs": [
    71.            {
    72.                "NetworkID": "r6t241jfgja5jndllate9em8s",
    73.                "Addr": "10.255.0.6/16"
    74.            }
    75.        ]
    76.    },
    77.    "UpdateStatus": {
    78.        "StartedAt": "0001-01-01T00:00:00Z",
    79.        "CompletedAt": "0001-01-01T00:00:00Z"
    80.    }
    81. }
    82. ]

  • 查看正在运行的所有服务

    1. docker service ls
    2. ID            NAME     MODE        REPLICAS  IMAGE
    3. ljlbfk4122sy  mynginx  replicated  1/1       nginx:latest

  • 使用update 命令增加nginx 副本数

image.png

  • 在任意的manager 节点上查看nginx 的分布情况

image.png
(主节点1)
image.png
(主节点2)

  • 访问测试:

image.png
(上图访问了没有在该机器上运行自定义nginx service 的8888 端口)
image.png
(上图可以看出该机器没有运行nginx 容器, 但是由于在同一个swarm 集群中, 同样可以访问到)

  • 再次动态扩缩容:

    1. docker service update --replicas 10 mynginx

  • 再次查看每个主节点中的容器使用情况

image.png

  • 缩容

image.png

docker service scale 命令用法

  1. docker service scale --help
  3. Usage:  docker service scale SERVICE=REPLICAS [SERVICE=REPLICAS...]
  5. Scale one or multiple replicated services
  7. Options:
  8.       --help   Print usage
  • 实例:
    • 将mynginx 副本数扩容到5

image.png

docker service rm 命令用法

  1. docker service rm --help
  3. Usage:  docker service rm SERVICE [SERVICE...]
  5. Remove one or more services
  7. Aliases:
  8.   rm, remove
  10. Options:
  11.       --help   Print usage
  • 实例:
    • 将mynginx 服务移除

image.png

与docker run 的区别

  • docker run 无扩缩容功能
  • docker service 具有扩缩容功能, 滚动更新, 灰度发布功能

概念总结

  • swarm
    • 集群的管理和编号, docker 可以初始化一个swarm 集群, 其他节点可以加入 (manager, worker)
  • node
    • 就是一个docker 节点, 多个节点就组成了一个网络集群 (manager, worker)
  • service
    • 任务, 可以在管理节点或者工作节点来运行, 核心
  • Task
    • 容器内的命令, 细节任务

image.png
image.png