手动配置 swarm
1. 手动运行
- 编译后手动运行，参考：Docker Swarm discovery
- 需要自定义配置服务发现服务，代码中默认是使用 docker Hub 的开放注册服务。生产环境需要配置自己的 distributed key/value store 注册服务。
- 所以本地采用 etcd
启动etcd
james@james-CW65:/media/james/linux/yangz/software/etcd/etcd-v3.3.0-rc.1-linux-amd64 > ./etcd --config-file etcd_default.conf
2018-03-06 11:51:20.890647 I | etcdmain: Loading server configuration from "etcd_default.conf"
2018-03-06 11:51:20.893130 I | etcdmain: etcd Version: 3.3.0-rc.1
2018-03-06 11:51:20.893167 I | etcdmain: Git SHA: d3c2acf
2018-03-06 11:51:20.893185 I | etcdmain: Go Version: go1.9.2
2018-03-06 11:51:20.893202 I | etcdmain: Go OS/Arch: linux/amd64
2018-03-06 11:51:20.893219 I | etcdmain: setting maximum number of CPUs to 4, total number of available CPUs is 4
2018-03-06 11:51:20.893242 W | etcdmain: no data-dir provided, using default data-dir ./default.etcd
2018-03-06 11:51:20.893342 N | etcdmain: the server is already initialized as member before, starting as etcd member...
2018-03-06 11:51:20.893452 I | embed: listening for peers on http://172.19.128.93:2380
2018-03-06 11:51:20.893481 I | embed: pprof is enabled under /debug/pprof
2018-03-06 11:51:20.893526 I | embed: listening for client requests on 172.19.128.93:2379
加入集群，在每个机器运行swarm agent。可以看到默认每分钟，agent去注册一次。
--advertise参数是Address of the Docker Engine joining the cluster. Swarm manager(s) MUST be able to reach the Docker Engine at this address. [$SWARM_ADVERTISE]
代码中的限制是：--advertise should be of the form ip:port or hostname:port
所以使用swarm必须通过tcp方式与docker daemon通信，不能用sock文件了。
MacBook-Air:src air$ swarm join --advertise=172.19.128.148:2375 etcd://172.19.128.93:2379
INFO[0000] Initializing discovery without TLS
INFO[0000] Registering on the discovery service every 1m0s... addr=172.19.128.148:2375 discovery=etcd://172.19.128.93:2379
INFO[0060] Registering on the discovery service every 1m0s... addr=172.19.128.148:2375 discovery=etcd://172.19.128.93:2379
INFO[0120] Registering on the discovery service every 1m0s... addr=172.19.128.148:2
开启swarm管理节点：
--host, -H [--host option --host option] ip/socket to listen on [$SWARM_HOST]
MacBook-Air:ac-agent air$ swarm manage -H tcp://172.19.128.148:2377 etcd://172.19.128.93:2379
INFO[0000] Initializing discovery without TLS
INFO[0000] Listening for HTTP addr=172.19.128.148:2377 proto=tcp
开始使用swarm:
可以看出在该节点上使用swarm的管理节点来获取集群信息，由于swarm是安装在一个没装docker的机器上，节点也是。所以没有获取到节点的容器任何信息。
james@james-CW65:/media/james/linux/yangz/software/etcd/etcd-v3.3.0-rc.1-linux-amd64 > docker -H tcp://172.19.128.148:2377 info
Containers: 0
Running: 0
Paused: 0
Stopped: 0
Images: 0
Server Version: swarm/1.2.8
Role: primary
Strategy: spread
Filters: health, port, containerslots, dependency, affinity, constraint, whitelist
Nodes: 1
(unknown): 172.19.128.148:2375
└ ID:
└ Status: Pending
└ Containers: 0
└ Reserved CPUs: 0 / 0
└ Reserved Memory: 0 B / 0 B
└ Labels:
└ Error: Cannot connect to the Docker daemon at tcp://172.19.128.148:2375. Is the docker daemon running?
└ UpdatedAt: 2018-03-06T04:11:37Z
└ ServerVersion:
Plugins:
Volume:
Network:
Log:
Swarm:
NodeID:
Is Manager: false
Node Address:
Kernel Version: 16.7.0
Operating System: darwin
Architecture: amd64
CPUs: 0
Total Memory: 0B
Name: MacBook-Air.local
Docker Root Dir:
Debug Mode (client): false
Debug Mode (server): false
Experimental: false
Live Restore Enabled: false
WARNING: No kernel memory limit support
james@james-CW65:/media/james/linux/yangz/software/etcd/etcd-v3.3.0-rc.1-linux-amd64 > ./etcdctl --endpoint http://172.19.128.93:2379 get /docker/swarm/nodes/172.19.128.148:2375
172.19.128.148:2375
加入集群，指定注册地址和服务发现服务地址：
james@james-CW65:/media/james/linux/yangz/software/docker > ./swarm_linux join --advertise=172.19.128.93:2375 etcd://172.19.128.93:2379
INFO[0000] Initializing discovery without TLS
INFO[0000] Registering on the discovery service every 1m0s... addr=172.19.128.93:2375 discovery=etcd://172.19.128.93:2379
会显示2个节点，但是由于docker没有开放tcp连接，没办法获取docker信息
james@james-CW65:/media/james/linux/yangz/software/etcd/etcd-v3.3.0-rc.1-linux-amd64 > docker -H tcp://172.19.128.148:2377 info
Containers: 0
Running: 0
Paused: 0
Stopped: 0
Images: 0
Server Version: swarm/1.2.8
Role: primary
Strategy: spread
Filters: health, port, containerslots, dependency, affinity, constraint, whitelist
Nodes: 2
(unknown): 172.19.128.148:2375
└ ID:
└ Status: Pending
└ Containers: 0
└ Reserved CPUs: 0 / 0
└ Reserved Memory: 0 B / 0 B
└ Labels:
└ Error: Cannot connect to the Docker daemon at tcp://172.19.128.148:2375. Is the docker daemon running?
└ UpdatedAt: 2018-03-06T05:18:00Z
└ ServerVersion:
(unknown): 172.19.128.93:2375
└ ID:
└ Status: Pending
└ Containers: 0
└ Reserved CPUs: 0 / 0
└ Reserved Memory: 0 B / 0 B
└ Labels:
└ Error: Cannot connect to the Docker daemon at tcp://172.19.128.93:2375. Is the docker daemon running?
└ UpdatedAt: 2018-03-06T06:16:52Z
└ ServerVersion:
Plugins:
Volume:
Network:
Log:
Swarm:
NodeID:
Is Manager: false
Node Address:
Kernel Version: 16.7.0
Operating System: darwin
Architecture: amd64
CPUs: 0
Total Memory: 0B
Name: MacBook-Air.local
Docker Root Dir:
Debug Mode (client): false
Debug Mode (server): false
Experimental: false
Live Restore Enabled: false
WARNING: No kernel memory limit support
MacBook-Air:src air$ swarm list etcd://172.19.128.93:2379
INFO[0000] Initializing discovery without TLS
172.19.128.148:2375
172.19.128.93:2375
MacBook-Air:src air$
- 注意这里 swarm 仅仅提供一个集群功能。将 docker 连成了一个整体。但是在该集群上吗并不能使用新版 docker 的命令行或 API client 来操作服务，会报错 404。
- 不要混淆了独立 swarm 和 swarm mode 的区别。这里我们创建的是 swarm 集群，并不提供服务级别的功能。参考：使用 Docker Toolbox 创建 Swarm 集群的问题-概念混淆导致
- 实际运行这样也是不能创建服务的。发现只有使用 docker swarm init 来从 docker Hub 上获取的 token 才能使用 swarm mode。为什么不在 docekr swarm init 的时候提供 kv 数据存储的参数，来本地注册呢。看 swarm 源代码也可以看出再创建集群的时候，使用的是默认的 docker hub 的数据库，真不知道这个为什么不让用户用自己的。
欢迎来到这里！
我们正在构建一个小众社区，大家在这里相互信任，以平等 • 自由 • 奔放的价值观进行分享交流。最终，希望大家能够找到与自己志同道合的伙伴，共同成长。
注册 关于