Docker 1.12: Now with Built-in Orchestration! | Docker Blog
【参考訳】Docker 1.12: ついにオーケストレーションを組み込み！ · Pocketstudio Technology Log
［速報］Docker、最新バージョンでDocker Engineにオーケストレーション機能を内蔵。外部ツール不要でクラスタ運用を実現。DockerCon 16 － Publickey
Docker だけでオーケストレーションできるってよ
試してみた
Version
- Version 1.12.0-rc2-beta16 (build: 9493)
- OS X EI Capitan 10.11.5（15F34）
macで動いている Docker ホストを使おうと思ったが
$ docker node ls
ID NAME MEMBERSHIP STATUS AVAILABILITY MANAGER STATUS
5cqlhnd928r9pbklk3n7r9nb2 * moby Accepted Ready Active Leader
$
で返ってくるけど
$ docker-machine ip moby
Host does not exist: "moby"
$
でipが返ってこないのでVirtualBoxでやる
(どっちにしろMacがDHCPで設定されてるたのでswarmで連携するのには向いてない感じ)
複数台構成を試してみる
- Manager 3台
- Worker 3 台
構成をVirtualBoxを使って試してみる
VirtualBox上にdocker machine を作成
$ docker-machine create swarm-manager01 --driver virtualbox
$ docker-machine create swarm-manager02 --driver virtualbox
$ docker-machine create swarm-manager03 --driver virtualbox
$ docker-machine create swarm-worker01 --driver virtualbox
$ docker-machine create swarm-worker02 --driver virtualbox
$ docker-machine create swarm-worker03 --driver virtualbox
名前が manager/worker と付いているけどまだ各々バラバラに動いている状態
ホスト情報を確認する
$ docker-machine ls
NAME ACTIVE DRIVER STATE URL SWARM DOCKER ERRORS
swarm-manager01 * virtualbox Running tcp://192.168.99.103:2376 v1.12.0-rc2
swarm-manager02 - virtualbox Running tcp://192.168.99.104:2376 v1.12.0-rc2
swarm-manager03 - virtualbox Running tcp://192.168.99.105:2376 v1.12.0-rc2
swarm-worker01 - virtualbox Running tcp://192.168.99.106:2376 v1.12.0-rc2
swarm-worker02 - virtualbox Running tcp://192.168.99.107:2376 v1.12.0-rc2
swarm-worker03 - virtualbox Running tcp://192.168.99.108:2376 v1.12.0-rc2
$
- swarm-manager01 : 192.168.99.103:2376
- swarm-manager02 : 192.168.99.104:2376
- swarm-manager03 : 192.168.99.105:2376
- swarm-worker01 : 192.168.99.106:2376
- swarm-worker02 : 192.168.99.107:2376
- swarm-worker03 : 192.168.99.108:2376
Managerを構成する
swarm-manager01 で init
$ eval $(docker-machine env swarm-manager01)
$ docker swarm init --listen-addr 192.168.99.103:2377
$ #自動でjoinしないようにしたい場合は --auto-accept none とか付ける
swarm-manager02 を join --manager
$ eval $(docker-machine env swarm-manager02)
$ docker swarm join --manager --listen-addr 192.168.99.104:2377 192.168.99.103:2377
This node joined a Swarm as a manager.
$
swarm-manager03 を join --manager
$ eval $(docker-machine env swarm-manager03)
$ docker swarm join --manager --listen-addr 192.168.99.105:2377 192.168.99.103:2377
This node joined a Swarm as a manager.
$
Workerを構成する
join コマンドに --manager オプションをつけないとworker
swarm-worker01 を join
$ eval $(docker-machine env swarm-worker01)
$ docker swarm join --listen-addr 192.168.99.106:2377 192.168.99.103:2377
This node joined a Swarm as a worker.
$
swarm-worker02 を join
$ eval $(docker-machine env swarm-worker02)
$ docker swarm join --listen-addr 192.168.99.107:2377 192.168.99.103:2377
This node joined a Swarm as a worker.
$
swarm-worker03 を join
$ eval $(docker-machine env swarm-worker03)
$ docker swarm join --listen-addr 192.168.99.108:2377 192.168.99.103:2377
This node joined a Swarm as a worker.
$
Nginx を 起動してみる
- nginxを frontend というservice名で ５つ起動する
Manager node で操作する
$ eval $(docker-machine env swarm-manager01)
作成
$ docker service create --name frontend --replicas 5 -p 80:80/tcp nginx:latest
0ktteptq3v85p499fsd6hjeah
$
確認
manager は worker 兼務
$ docker service tasks frontend
ID NAME SERVICE IMAGE LAST STATE DESIRED STATE NODE
64ikgt2aogz4yvf2420087eeu frontend.1 frontend nginx:latest Running 3 minutes Running swarm-manager01
5pirlmlbm29tagjx7t7ty96o2 frontend.2 frontend nginx:latest Running 3 minutes Running swarm-manager03
8rebiu9bf74g2fe971z03d4m7 frontend.3 frontend nginx:latest Running 3 minutes Running swarm-worker01
3codqzxy7qui6t62hbu1axu8r frontend.4 frontend nginx:latest Running 3 minutes Running swarm-manager02
0zi5xg7qecxjhgrfhm1bh6q4t frontend.5 frontend nginx:latest Running 3 minutes Running swarm-worker03
5 > 2 にスケールダウン！
$ docker service scale frontend=2
frontend scaled to 2
$
$ docker service tasks frontend
ID NAME SERVICE IMAGE LAST STATE DESIRED STATE NODE
3codqzxy7qui6t62hbu1axu8r frontend.4 frontend nginx:latest Running 8 minutes Running swarm-manager02
0zi5xg7qecxjhgrfhm1bh6q4t frontend.5 frontend nginx:latest Running 8 minutes Running swarm-worker03
$
この状態で swarm-manager01のipでアクセスしても繋がる
worker のipでもつながる。バランシングできている
(反応は悪い。時間がかかる。VM上げ過ぎ？)
http://192.168.99.103/
http://192.168.99.104/
http://192.168.99.105/
http://192.168.99.106/
http://192.168.99.107/
http://192.168.99.108/
2 > 50 にスケールアップ
$ docker service scale frontend=50 ; while true; do date; docker service tasks frontend; sleep 1; done
frontend scaled to 50
2016年 6月21日 火曜日 18時08分32秒 JST
ID NAME SERVICE IMAGE LAST STATE DESIRED STATE NODE
53zq623h6pydgwz8gibbyg0ma frontend.1 frontend nginx:latest Assigned Less than a second Running swarm-worker02
b2342f9gr6ddzhgxkv4ehao4w frontend.2 frontend nginx:latest Assigned Less than a second Running swarm-worker03
5mofgkdru0g62y80g3yzr8po4 frontend.3 frontend nginx:latest Assigned Less than a second Running swarm-worker02
3codqzxy7qui6t62hbu1axu8r frontend.4 frontend nginx:latest Running 16 minutes Running swarm-manager02
0zi5xg7qecxjhgrfhm1bh6q4t frontend.5 frontend nginx:latest Running 16 minutes Running swarm-worker03
bpt3lqq15jh1rqjli7pejcafc frontend.6 frontend nginx:latest Assigned Less than a second Running swarm-worker03
892j9gm1zjfprktjd0h8v65zp frontend.7 frontend nginx:latest Assigned Less than a second Running swarm-worker02
93f0xh7jo99leq5qfcmv3fwg2 frontend.8 frontend nginx:latest Assigned Less than a second Running swarm-worker01
bd028u32w2dn0qra7qdougim2 frontend.9 frontend nginx:latest Assigned Less than a second Running swarm-manager01
3lu7qnvlpv7hcet94ex3mkktr frontend.10 frontend nginx:latest Assigned Less than a second Running swarm-worker02
704qhs8fafrkqu892bp1mfuvu frontend.11 frontend nginx:latest Assigned Less than a second Running swarm-worker02
60q9h6739t6quwhb0s3ry5vic frontend.12 frontend nginx:latest Assigned Less than a second Running swarm-manager03
2gn9go7e8vspa1tlfrs81bmtv frontend.13 frontend nginx:latest Assigned Less than a second Running swarm-worker01
edph6viy94ojhcup2k5hkj6wa frontend.14 frontend nginx:latest Assigned Less than a second Running swarm-worker01
4rfworr36efy3xdblknyt0afc frontend.15 frontend nginx:latest Assigned Less than a second Running swarm-manager03
1m22vxsnzq5ycs7thwo5bxqsf frontend.16 frontend nginx:latest Assigned Less than a second Running swarm-manager03
7pgdqys387811krx3hnkyeo10 frontend.17 frontend nginx:latest Assigned Less than a second Running swarm-manager01
b5wjgv6qhe0bq124xkrjjqj5p frontend.18 frontend nginx:latest Assigned Less than a second Running swarm-manager02
926l77zjfvoegwalopmxoto98 frontend.19 frontend nginx:latest Assigned Less than a second Running swarm-worker03
dyt78zxvpx5nefgipbpp2f6bt frontend.20 frontend nginx:latest Assigned Less than a second Running swarm-manager03
d67a68cysagkezm5s8uzjcv6c frontend.21 frontend nginx:latest Assigned Less than a second Running swarm-manager01
cmd6wqb5o6h7vhggr0oc1agyc frontend.22 frontend nginx:latest Assigned Less than a second Running swarm-manager01
7wdzv2opjk3vamx9vzz7b8jtb frontend.23 frontend nginx:latest Assigned Less than a second Running swarm-manager01
7fh7k7lp1p8g45x8giud05rsi frontend.24 frontend nginx:latest Assigned Less than a second Running swarm-worker01
9pgcixsz2i2j9wfhqnomu78bk frontend.25 frontend nginx:latest Assigned Less than a second Running swarm-manager02
aw0dlr1n4ncddwfnkkls3qpxu frontend.26 frontend nginx:latest Assigned Less than a second Running swarm-manager02
7swtukjp7cjxir3cj9kghwaca frontend.27 frontend nginx:latest Assigned Less than a second Running swarm-manager03
d344xub98wzp7mqfuvgrj9kgr frontend.28 frontend nginx:latest Assigned Less than a second Running swarm-manager01
4le0yrl7xzaijna82ffff72xp frontend.29 frontend nginx:latest Assigned Less than a second Running swarm-manager01
48vkny4d13u2fjskty1pfro1i frontend.30 frontend nginx:latest Assigned Less than a second Running swarm-worker01
6g03xa6lgfd8gqctpszpk3qey frontend.31 frontend nginx:latest Assigned Less than a second Running swarm-manager01
eoan69u29a6ah0vi05ihl8ccp frontend.32 frontend nginx:latest Assigned Less than a second Running swarm-worker01
9ie6s3b7viybtbyc7g32lsddj frontend.33 frontend nginx:latest Assigned Less than a second Running swarm-manager03
4t7n6owvqbd3wd2fyhb6p033t frontend.34 frontend nginx:latest Assigned Less than a second Running swarm-worker01
6adu64lap9xp8if0f32e6uvqd frontend.35 frontend nginx:latest Assigned Less than a second Running swarm-worker03
09v5jml347eor4a83xfh7ng6a frontend.36 frontend nginx:latest Assigned Less than a second Running swarm-manager02
aie6px9sg75ju3pbnh5vs8pyt frontend.37 frontend nginx:latest Assigned Less than a second Running swarm-worker03
3k3qtoszxtwcm3junsjuo9bxg frontend.38 frontend nginx:latest Assigned Less than a second Running swarm-worker02
8wikdzvj293fo0qe7q477pkcp frontend.39 frontend nginx:latest Assigned Less than a second Running swarm-manager03
2jbil2dexyjd6jh85182m5n15 frontend.40 frontend nginx:latest Assigned Less than a second Running swarm-worker01
29bphav3mp67yuimkqpdvmx18 frontend.41 frontend nginx:latest Assigned Less than a second Running swarm-manager02
0dt1tc8707uf519ua2gc07qcj frontend.42 frontend nginx:latest Assigned Less than a second Running swarm-worker01
6vtc5qkftqtpac4xegzs07qlp frontend.43 frontend nginx:latest Assigned Less than a second Running swarm-manager03
aq2ya45sturcn6ekr1ejeggj7 frontend.44 frontend nginx:latest Assigned Less than a second Running swarm-worker03
5e4mv8ap983bnd0e8vxz4o0p8 frontend.45 frontend nginx:latest Assigned Less than a second Running swarm-manager02
4e46ffegsx2j750hyehjdnlht frontend.46 frontend nginx:latest Assigned Less than a second Running swarm-worker03
7dpfq95k7o4qm9nyranb6r1yh frontend.47 frontend nginx:latest Assigned Less than a second Running swarm-worker02
8xvdmtxrd9ucprlusf7vn7f9j frontend.48 frontend nginx:latest Assigned Less than a second Running swarm-manager02
8d5t4d52kdx2btmzv8givx818 frontend.49 frontend nginx:latest Assigned Less than a second Running swarm-manager01
10wg1ljjavdem11u4vxsnv122 frontend.50 frontend nginx:latest Assigned Less than a second Running swarm-worker02
$
秒で起動！
均等に起動
$ docker service tasks frontend | grep -v NODE |awk '{print $NF}' | sort | uniq -c
9 swarm-manager01
8 swarm-manager02
8 swarm-manager03
9 swarm-worker01
8 swarm-worker02
8 swarm-worker03
$
Worker が止まったらどうなる?
さあOrchestrationしてみせろやー！ 
kill swarm-worker03
$ docker-machine kill swarm-worker03
２秒後くらい？
(1Mac上ですからね)
$ # DESIRED STATE が Accepted になって移動
$ docker service tasks frontend | grep -v NODE |awk '{print $NF}' | sort | uniq -c
10 swarm-manager01
10 swarm-manager02
10 swarm-manager03
10 swarm-worker01
10 swarm-worker02
$