Docker Compose部署RocketMQ和控制中心

>最近单位有个应用需要上`RocketMQ`，考虑到资源不是特别充足且该应用重要性分级不高，打算采用`Docker Compose`方式单机部署。让我们一起来看看该过程中有哪些坑💥💥

# 创建RocketMQ相关目录和文件
创建持久化日志和数据目录
```
#namesrv的日志目录
mkdir -p /data/rocketmq/namesrv/logs
#broker的日志、数据、配置文件目录
mkdir -p /data/rocketmq/broker/{conf,logs,store}
```
创建broker.conf配置文件
```
brokerClusterName = DefaultCluster
brokerName = broker-a
brokerId = 0
deleteWhen = 04
fileReservedTime = 48
brokerRole = ASYNC_MASTER
flushDiskType = ASYNC_FLUSH
# IP设置为主机IP
brokerIP1=192.168.124.80
autoCreateTopicEnable=true
autoCreateSubscriptionGroup=true
waitTimeMillsInSendQueue=1000
#设置namesrv地址，建议不做修改
namesrvAddr=rmqnamesrv:9876
traceTopicEnable=true
```

# 编排docker
编排docker-compose.yaml
```yml
version: '3.8'
services:
rmqnamesrv:
image: apache/rocketmq:4.7.1
container_name: rmqnamesrv
ports:
- 9876:9876
restart: always
volumes:
- /data/rocketmq/namesrv/logs:/home/rocketmq/logs
environment:
JAVA_OPT: -server -Xms256m -Xmx512m -Xmn128m -XX:MetaspaceSize=128m -XX:MaxMetaspaceSize=320m
command: ["sh","mqnamesrv"]
#看环境需求，是使用默认网络模式还是host模式
#network_mode: "host"

broker:
image: apache/rocketmq:4.7.1
container_name: rmqbroker
ports:
- 10909:10909
- 10911:10911
- 10912:10912
restart: always
volumes:
- /data/rocketmq/broker/logs:/home/rocketmq/logs
- /data/rocketmq/broker/store:/home/rocketmq/store
- /data/rocketmq/broker/conf/broker.conf:/home/rocketmq/rocketmq-4.7.1/conf/broker.conf
depends_on:
- 'rmqnamesrv'
environment:
JAVA_OPT: -server -Xms256m -Xmx512m -Xmn128m -XX:MetaspaceSize=128m -XX:MaxMetaspaceSize=320m
command: ["sh","mqbroker","-c","/home/rocketmq/rocketmq-4.7.1/conf/broker.conf"]
#network_mode: "host"

rmqdashboard:
image: apacherocketmq/rocketmq-dashboard:latest
container_name: rocketmq-dashboard
ports:
- 8081:8080
restart: always
depends_on:
- 'rmqnamesrv'
environment:
- JAVA_OPTS= -Xms256M -Xms512M -Xmn128M -Drocketmq.namesrv.addr=rmqnamesrv:9876 -Dcom.rocketmq.sendMessageWithVIPChannel=false
#network_mode: "host"
```
> 如果使用了host网络模式，`broker.conf`配置文件的`namesrvAddr=rmqnamesrv:9876`可以设置为`namesrvAddr=主机IP:9876`

# 启动RocketMQ
```
docker compose up
```
> docker compose分为V1和V2版本，独立安装的V1和V2版本使用`docker-compose`方式启动，插件安装的V2版本使用'docker compose'方式启动

# 报错
好了，现在坑💥来了

可以看到namesrv启动成功，但是broker异常退出，并且没有任何日志,导致后面有依赖关系的dashboard连接broker也出现异常。

## 落坑
**docker compose启动指定container**</br>
因为该docker-compose.yaml文件有3个container且有依赖关系，为方便排错，暂时去掉依赖关系，并只启动broker container
```yml
version: '3.8'
services:
rmqnamesrv:
...
broker:
...
# depends_on:
# - 'rmqnamesrv'
environment:
JAVA_OPT: -server -Xms256m -Xmx512m -Xmn128m -XX:MetaspaceSize=128m -XX:MaxMetaspaceSize=320m
command: ["sh","mqbroker","-c","/home/rocketmq/rocketmq-4.7.1/conf/broker.conf"]

rmqdashboard:
...
```
启动后错误同上。
```
docker compose up broker
```

**直接使用docker run方式启动**
```shell
docker run --name rmqbroker --restart always -p 10909:10909 -p 10911:10911 -p 10912:10912 -e "JAVA_OPT=-server -Xms256m -Xmx512m -Xmn128m -XX:MetaspaceSize=128m -XX:MaxMetaspaceSize=320m" -v /data/rocketmq/broker/logs:/home/rocketmq/logs -v /data/rocketmq/broker/store:/home/rocketmq/store -v /data/rocketmq/broker/conf/broker.conf:/home/rocketmq/rocketmq-4.7.1/conf/broker.conf apache/rocketmq:4.7.1 sh mqbroker -c /home/rocketmq/rocketmq-4.7.1/conf/broker.conf
```
果然，还是一样的错误且没有日志。在求助搜索引擎后也没有找到合适的解决办法。

**上大招**💘，求助`deepseek`！
1. 配置 Docker Compose 日志记录
2. 强制显示实时日志
3. 捕获容器退出时的状态
4. 启用 Docker 调试日志
5. 手动调试容器启动过程

逐一按照`最强AI`的思路，在最后的手动调试启动过程中发现了端倪（*填坑耗时6️⃣小时*）

## 爬坑
1. **启动容器，但不启动broker**
```shell
docker run -it \
--name rmqbroker-debug \
-v /data/rocketmq/broker/logs:/home/rocketmq/logs \
-v /data/rocketmq/broker/store:/home/rocketmq/store \
-v /data/rocketmq/broker/conf/broker.conf:/home/rocketmq/rocketmq-4.7.1/conf/broker.conf \
apache/rocketmq:4.7.1 \
sh -c "tail -f /dev/null"
```
>这里我们在dockr启动后，不执行启动mqbroker的命令，只保持容器运行

成功启动，没有之前的异常退出和反复重启等故障，见到曙光啦🌥️！从目前来看，应该是之前执行启动mqbroker时出错了

2. **进入容器手动执行命令**
```
#进入容器
docker exec -it rmqbroker-debug bash
# 手动执行 Broker 启动命令
sh mqbroker -c /home/rocketmq/rocketmq-4.7.1/conf/broker.conf
```
可以正常执行但是无任何输出，几个意思？检查日志吧
```
[root@docker logs]# pwd
/data/rocketmq/broker/logs
[root@docker logs]# ls -l
total 0

#容器内
[rocketmq@ff756bdbe93e logs]$ pwd
/home/rocketmq/logs
[rocketmq@ff756bdbe93e logs]$ ls -l
total 0
```
明明挂载了本地路径，为什么容器内外都无法生成日志呢，权限问题❓
```
[rocketmq@ff756bdbe93e ~]$ ls -lh
total 0
drwxr-xr-x 2 root root 6 Feb 8 06:43 logs
drwxr-xr-x 1 rocketmq rocketmq 17 Jul 31 2021 rocketmq-4.7.1
drwxr-xr-x 2 root root 6 Feb 8 06:43 store

[rocketmq@ff756bdbe93e conf]$ pwd
/home/rocketmq/rocketmq-4.7.1/conf
[rocketmq@ff756bdbe93e conf]$ ls -l
total 36
drwxr-xr-x 1 rocketmq rocketmq 118 Jun 2 2020 2m-2s-async
drwxr-xr-x 1 rocketmq rocketmq 118 Jun 2 2020 2m-2s-sync
drwxr-xr-x 1 rocketmq rocketmq 91 Jun 2 2020 2m-noslave
-rw-r--r-- 1 root root 361 Jan 24 01:47 broker.conf
drwxr-xr-x 1 rocketmq rocketmq 72 Jun 2 2020 dledger
-rw-r--r-- 1 rocketmq rocketmq 14978 Jun 2 2020 logback_broker.xml
-rw-r--r-- 1 rocketmq rocketmq 3836 Jun 2 2020 logback_namesrv.xml
-rw-r--r-- 1 rocketmq rocketmq 3761 Jun 2 2020 logback_tools.xml
-rw-r--r-- 1 rocketmq rocketmq 1305 Jun 2 2020 plain_acl.yml
-rw-r--r-- 1 rocketmq rocketmq 834 Jun 2 2020 tools.yml

```
rocketmq的家目录是rocketmq属主/组的，并且配置目录下除了挂载的`broker.conf`其他文件也都是rocketmq属主/组。

已经很明确，就是权限问题导致的之前启动异常退出，以及手动调试时无报错情况。
3. **查看dockerfile确认权限问题**</br>
[完整文件请看这里](https://hub.docker.com/layers/apache/rocketmq/4.7.1/images/sha256-10a5abbc735bde5c34d0e52fc47e857c36fccb108a9692e20b45269f10f4325e)
```
...
#添加用户和组
7 ARG uid=3000
8 ARG gid=3000
9 |4 gid=3000 group=rocketmq uid=3000 user=rocketmq /bin/sh -c groupadd -g ${gid} ${group} && useradd -u ${uid} -g ${gid} -m -s /bin/bash ${user}
...
#设置工作目录
12 ENV ROCKETMQ_HOME=/home/rocketmq/rocketmq-4.7.1
13 WORKDIR /home/rocketmq/rocketmq-4.7.1
...
#更改家目录的属主/组
16 |5 gid=3000 group=rocketmq uid=3000 user=rocketmq version=4.7.1 /bin/sh -c chown -R ${uid}:${gid} ${ROCKETMQ_HOME}
...
```
真相大白❗❗❗官方镜像设置了容器内运行用户为`rocketmq`,并修改了相关目录的权限。
## 修正错误
修改权限
```
[root@docker data]# chown -R 3000:3000 rocketmq/
[root@docker data]# ls -lh
total 12K
drwxr-xr-x 4 3000 3000 61 Feb 6 16:47 rocketmq
```
启动最初完整的docker compose
```
docker compose up
```
成功启动


查看dashboard

## 测试功能
写个python测试
```python
import json
from rocketmq.client import Producer, Message

producer = Producer('testGroup')
producer.set_name_server_address('192.168.124.80:9876')
producer.start()

msg = Message('testTopic')
msg_body = {"id":"test_id","name":"test_name","message":"test_message"}
ss = json.dumps(msg_body).encode('utf-8')
msg.set_body(ss)
producer.send_sync(msg)

producer.shutdown()
~
```
查看dashboard