# Services with Docker and Compose

## 章節重點

### 理解什麼是 service
> Any processes, functionality, or data that must be discoverable and available over a
network is called a service.

現今的軟體大多都包含了許多子系統，而這些子系統會彼此交互依賴溝通
利用抽象的Service可以幫助我們用更簡單的角度去理解各個系統的職責

Docker也有提供service的功能 (Swarm)，幫助我們去管理service的生命週期，
以及協作和調度

### 介紹 Docker Swarm

* 名詞和定義介紹
* Automated resurrection and replication
* Automated rollout
* Service health and rollback

### 利用 Docker Compose 來管理 service

- 介紹YAML格式
- replicas
- volumes
- networks

## Docker Swarm 名詞

[官方概念介紹](https://docs.docker.com/engine/swarm/key-concepts/)
[Manager vs Worker nodes](https://docs.docker.com/engine/swarm/how-swarm-mode-works/nodes/)

* Swarm

> A swarm consists of multiple Docker hosts which run in swarm mode (managers or workers)

* Nodes

> A node is an instance of the Docker engine participating in the swarm

* Service

> A service is the definition of the tasks to execute on the manager or worker nodes

* Task (roughly interchangeable with container)

> A task carries a Docker container and the commands to run inside the container. It is the atomic scheduling unit of swarm.

### Service 和 Task(container) 的區別

> service的功能會由一群containers來負責實作


## Swarm 的功能介紹

### 前置步驟

// 把自己的機器變成一個 node
docker swarm init

// 啟動範例 service
docker service create \
 --publish 8080:80 \
 --name hello-world \
 dockerinaction/ch11_service_hw:v1

### Automated resurrection and replication

> swarm 可以自動保持 service 的 DESIRED STATE
> swarm 可以實現 service 的水平擴展
> 偵測維持 replicas 的設定，自動啟動或關閉 tasks(containers)

#### 查看 service

// list service
docker service ls

// list the container associated with a specific service
docker service ps hello-world

ID             NAME                IMAGE                               NODE             DESIRED STATE   CURRENT STATE           ERROR                         PORTS
bm217mfvfeqa   hello-world.1       dockerinaction/ch11_service_hw:v1   docker-desktop   Running         Running 2 minutes ago
lnnk5tc3l768    \_ hello-world.1   dockerinaction/ch11_service_hw:v1   docker-desktop   Shutdown        Failed 2 minutes ago    "task: non-zero exit (137)"

- DESIRED STATE: 希望的狀態
- CURRENT STATE: 系統當前狀態

#### 查看 service 的設定

docker service inspect --pretty hello-world

docker service inspect hello-world

https://docs.docker.com/engine/swarm/swarm-tutorial/rolling-update/
> Rollout parameters => UpdateConfig

ID:		oqo4kuy9zq6p7ys41mhyg2sb6
Name:		hello-world
Service Mode:	Replicated
 Replicas:	1
Placement:
UpdateConfig:
 Parallelism:	1
 On failure:	pause
 Monitoring Period: 5s
 Max failure ratio: 0
 Update order:      stop-first
RollbackConfig:
 Parallelism:	1
 On failure:	pause
 Monitoring Period: 5s
 Max failure ratio: 0
 Rollback order:    stop-first
ContainerSpec:
 Image:		dockerinaction/ch11_service_hw:v1@sha256:4f92ec42207773b4ea97afe03469ff30598d08dd2ec4a28e05550b3ba4bc5fd5
 Init:		false
Resources:
Endpoint Mode:	vip
Ports:
 PublishedPort = 8080
  Protocol = tcp
  TargetPort = 80
  PublishMode = ingress

#### 水平擴展 service

[service_scale](https://docs.docker.com/engine/reference/commandline/service_scale/)
[replicated vs global](https://docs.docker.com/engine/swarm/how-swarm-mode-works/services/)
[swarm ingress](https://docs.docker.com/engine/swarm/ingress/)

- Two types of service deployments
- replicated (specific replica number)
- global (one replica on each node)

docker service scale hello-world=3

docker service ps hello-world

### Automated rollout

> swarm 可以自動更新 service

[service_update](https://docs.docker.com/engine/reference/commandline/service_update/)


docker service update \
 --image dockerinaction/ch11_service_hw:v2 \
 --update-order stop-first \
 --update-parallelism 1 \
 --update-delay 30s \
 hello-world 

- update-parallelism 1

Service Mode:	Replicated
 Replicas:	3
UpdateStatus:
 State:		completed
 Started:	2 minutes ago
 Completed:	35 seconds ago
 Message:	update completed
Placement:
UpdateConfig:
 Parallelism:	1
 Delay:		30s
 On failure:	pause
 Monitoring Period: 5s
 Max failure ratio: 0
 Update order:      stop-first

- update-parallelism 3

Service Mode:	Replicated
 Replicas:	3
UpdateStatus:
 State:		completed
 Started:	About a minute ago
 Completed:	18 seconds ago
 Message:	update completed
Placement:
UpdateConfig:
 Parallelism:	3
 Delay:		30s
 On failure:	pause
 Monitoring Period: 5s
 Max failure ratio: 0
 Update order:      stop-first

### Service health and rollback

>A successful partnership with an orchestrator means clearly communicating the
expected requirements and behavior of the workload you’re asking it to orchestrate

- update 成會失敗的版本

docker service update \
--image dockerinaction/ch11_service_hw:start-failure \
hello-world

overall progress: 0 out of 3 tasks
1/3: starting  [============================================>      ]
2/3:
3/3: starting container failed: OCI runtime create failed: container_linux.go:3…
service update paused: update paused due to failure or early termination of task fjk5wq3ntlby27w7i5wlpdoy6

- 手動 rollback 到上一版

docker service update \
--rollback \
hello-world

- 在 update 的時候指定 update-failure-action
- update-failure-action: ("pause"|"continue"|"rollback")

docker service update \
--update-failure-action rollback \
--update-max-failure-ratio 0.6 \
--image dockerinaction/ch11_service_hw:start-failure \
hello-world

overall progress: rolling back update: 3 out of 3 tasks
1/3: running   [>                                                  ]
2/3: running   [>                                                  ]
3/3: running   [>                                                  ]
rollback: update rolled back due to failure or early termination of task y0ltb7g7a7q7ons8kjb9gjw2r
verify: Service converged

#### Docker 如何檢查 task 的 health

>Docker is application agnostic. Rather than making assumptions about how to
determine whether a specific task is healthy, it lets you specify a health check command

- **HEALTHCHECK** in Dockerfile

https://hub.docker.com/r/dockerinaction/ch11_service_hw/tags

HEALTHCHECK --interval=10s CMD ["/bin/httpping"]

- 也可以透過 update 指定 health check 的設定

docker service update \
 --health-cmd /bin/httpping \
 --health-interval 10s \
 hello-world

[update 的 options](https://docs.docker.com/engine/reference/commandline/service_update/#options)

## Declarative service environments with Compose V3

>Setting all of these parameters when you’re managing a service from the command
line is a mess


- Docker Compose File 的優點
- 避免人為的下指令減少犯錯機會
- 更有系統的去設定 service 的參數
- 可以做版本控制 (git)
- Docker Compose File 的缺點
- 相較於指令會不夠靈活

> When you need to model a whole environment of services, you should use a Docker
stack. A stack describes collections of services, volumes, networks, and other configuration abstractions.

### A YAML primer

https://zh.wikipedia.org/wiki/YAML

### Collections of services with Compose V3

https://docs.docker.com/compose/compose-file/compose-file-v3/

version: "3.7"
services:
  postgres:
    image: dockerinaction/postgres:11-alpine
    environment:
      POSTGRES_PASSWORD: example
  mariadb:
    image: dockerinaction/mariadb:10-bionic
    environment:
      MYSQL_ROOT_PASSWORD: example
  adminer:
    image: dockerinaction/adminer:4
    ports:
      - 8080:8080

### Creating and updating stack

- create

docker stack deploy -c databases.yml my-databases

- update

  adminer:
    image: dockerinaction/adminer:4
    ports:
      - 8080:8080
    deploy:
      replicas: 3

### Scaling down and removing services

- scale down

  adminer:
    image: dockerinaction/adminer:4
    ports:
      - 8080:8080
    deploy:
      replicas: 2

- remove service

把 mariadb 的 definition 刪掉

# 比較不好的方式 但可以成功移除
docker service remove my-databases_mariadb

# 比較正確但會刪掉沒有在yaml定義的resources
docker stack deploy -c databases.yml --prune my-databases

### Stateful services and preserving data

- 增加volume的設定

version: "3.7"
volumes:
  pgdata:
services:
  postgres:
    image: dockerinaction/postgres:11-alpine
    volumes:
      - type: volume
        source: pgdata
        target: /var/lib/postgresql/data
    environment:
      POSTGRES_PASSWORD: example
  adminer:
    image: dockerinaction/adminer:4
    ports:
      - 8080:8080
    deploy:
      replicas: 1

- 重新啟動 service

docker service remove my-databases_postgres

docker stack deploy -c databases.yml my-databases

### Load balancing, service discovery, and networks with Compose

>Port publishing for a service is different from publishing a
port on a container. Whereas containers directly map the port on the host interface to an
interface for a specific container, services might be made up of many replica containers.

>Docker accommodates services by creating virtual IP (VIP) addresses and balancing requests for a specific service between all of the associated replicas.


#### ingress network

- handles all port forwarding from the host interface to services

#### my-databases_default

- Docker will create a network for your stack named 'default', if not define network config in compose file

#### 可以定義自己的 network

- 讓不同的 stack 共用同個 network
- 讓 services 不共用同個 network

https://docs.docker.com/compose/compose-file/compose-file-v3/#network-configuration-reference

version: "3.7"
networks:
  foo:
    driver: overlay
volumes:
  pgdata:
services:
  postgres:
    image: dockerinaction/postgres:11-alpine
    volumes:
      - type: volume
        source: pgdata
        target: /var/lib/postgresql/data
    networks:
      - foo
    environment:
      POSTGRES_PASSWORD: example
  adminer:
    image: dockerinaction/adminer:4
    networks:
      - foo
    ports:
      - 8080:8080
    deploy:
      replicas: 1