GoldenDB 启停命令

2026-09-20 09:27 圣彼得堡的村长阅读(23) 评论(0) 收藏举报

GoldenDB 启停命令

前面提到过，GoldenDB 在部署时会为每个组件创建一个操作系统用户，组件的相关操作，需要切换到相应的用户下执行。

组件	用户名规则
Manager	`zxmanager`
GTM	`zxgtm1`、`zxgtm2` ...
CN	`zxdbproxy1`、`zxdbproxy2` ...
DN	`zxdb1`、`zxdb2` ...
LoadServer	`zxload1`、`zxload2` ...
InsightServer/Agent	`insight`
安装检查和巡检	`zxgdbcheck`

因此，GoldenDB 提供了一套通用的组件管理命令：dbmoni -start 、dbmoni -stop 以及 dbstate ，实现组件的启动、关闭和状态查看。

1. 命令操作

1.1 Manager 组件

切换到 zxmanager 用户，一条命令同时管理 4 个主要进程：MetadataServer 、ProxyManager 、ClusterManager 以及 ommagent 代理。

[root@golden1 goldendb]# su - zxmanager

[zxmanager@golden1 ~]$ dbstate 
[metadataserver]The metadataserver process is running, PID:5216
[proxymanager]The proxymanager process is running, PID:5483
[clustermanager]The clustermanager process is running, PID:5710
[ommagent]The ommagent process is running, PID:5795

[zxmanager@golden1 ~]$ dbmoni -stop
[dbmoni -stop]The script dbmoni metadataserver process is stopped
[metadataserver]The metadataserver process is stopped
[dbmoni -stop]The script dbmoni proxymanager process is stopped
[proxymanager]The proxymanager process is stopped
[dbmoni -stop]The script dbmoni clustermanager process is stopped
[clustermanager]The clustermanager process is stopped
[dbmoni -stop]The script dbmoni ommagent process is stopped
[ommagent]The ommagent process is stopped

[zxmanager@golden1 ~]$  dbmoni -start
[dbmoni -start]dbmoni metadataserver ......starting......
[dbmoni -start]dbmoni proxymanager ......starting......
[dbmoni -start]dbmoni clustermanager ......starting......
[dbmoni -start]dbmoni ommagent ......starting......

[zxmanager@golden1 ~]$ dbstate
[metadataserver]The metadataserver process is running, PID:92029
[proxymanager]The proxymanager process is running, PID:92106
[clustermanager]The clustermanager process is running, PID:92176
[ommagent]The ommagent process is running, PID:92251

1.2 GTM 组件

切换到 zxgtm1 用户，一条命令同时操作 3 个主要进程： gtm、gtm_aide 和 ommagent 。

[root@golden3 DB]# su - zxgtm1

[zxgtm1@golden3 ~]$ dbmoni -stop
[dbmoni -stop]The script dbmoni gtm process is stopped
[gtm]The gtm_aide process is stopped
[gtm]The gtm process is stopped
[dbmoni -stop]The script dbmoni ommagent process is stopped
[ommagent]The ommagent process is stopped

[zxgtm1@golden3 ~]$ dbmoni -start
[dbmoni -start]dbmoni gtm ......starting......
[dbmoni -start]dbmoni ommagent ......starting......

gtm_aide 是 gtm 的守护进程，它监控本地 GTM 进程的健康状况，并与管理节点保持实时的心跳通信。如果 GTM 自身因为内存溢出或高并发压力导致进程 Hang 住，它自己无法发出告警或主备切换请求。这时候 gtm_aide 作为一个轻量级的辅助进程，可以迅速发起高可用仲裁流程，通知 CN 切换路由（跨多个分片的写操作时，CN 要从 GTM 获取 GTID）。

1.3 Proxy 组件

切换到 zxdbproxy1 用户，一条命令同时操作 dbproxy 和 ommagent 两个主进程。

[root@golden2 goldendb]# su - zxdbproxy1

[zxdbproxy1@golden2 ~]$ dbstate
[dbproxy]The dbproxy process is running, PID:23573
[ommagent]The ommagent process is running, PID:23519


[zxdbproxy1@golden2 ~]$ dbmoni -stop
[dbmoni -stop]The script dbmoni dbproxy process is stopped
[dbproxy]The dbproxy process is stopped
[dbmoni -stop]The script dbmoni ommagent process is stopped
[ommagent]The ommagent process is stopped

[zxdbproxy1@golden2 ~]$ dbmoni -start
[dbmoni -start]dbmoni dbproxy ......starting......
[dbmoni -start]dbmoni ommagent ......starting......

1.4 mysqld 数据库

切换到 zxdbN 用户，一条命令同时操作 dbproxy 、loadserver 和 ommagent 3 个主进程。

[root@golden1 ~]# su - zxdb2

[zxdb2@golden1 ~]$ dbstate
[dbagent]The dbagent process is running, PID:119185
[loadserver]The loadserver process is running, PID:119241
[ommagent]The ommagent process is running, PID:119580

--insight 用户下的mysqld 是 RDB，zxdb1 用户下的 mysqld 是系统租户，zxdb2 下面的是业务租户
[zxdb2@golden1 ~]$ ps -ef |grep mysqld
insight    1956      1 99 00:37 ?        14:47:49 /home/goldendb/insight/bin/mysqld --defaults-file=/home/goldendb/insight/etc/my.cnf --datadir=/home/goldendb/insight/data/data --pid-file=/home/goldendb/insight/bin/mysqld1.pid
zxdb1      6338      1 99 00:38 ?        15:46:04 /home/goldendb/zxdb1/bin/mysqld --defaults-file=/home/goldendb/zxdb1/etc/my.cnf --datadir=/home/goldendb/zxdb1/data/data --pid-file=/home/goldendb/zxdb1/bin/mysqld1.pid
zxdb2     29985  26712  0 15:26 pts/1    00:00:00 grep --color=auto mysqld
zxdb2    117979      1 99 02:10 ?        13:46:21 /zxdb2/bin/mysqld --defaults-file=/zxdb2/etc/my.cnf --datadir=/zxdb2/data/data --pid-file=/zxdb2/bin/mysqld1.pid

zxdbN 用户下， dbstate 只检查 GoldenDB 的 dbagent 、loadserver 和 ommagent 组件的状态，不直接显示 mysqld 的进程，mysqld 由 dbagent 托管，只要 dbagent 状态正常，且 mysqld 进程存在，就说明 DB 实例正常。

我们先看一下业务租户的一主两从同步状态：dbtool -cm -query-latest-slave -clusterID=2 -groupID=1

[zxmanager@golden1 ~]$  dbtool -cm -query-latest-slave -clusterID=2 -groupID=1
Send message to module[ClusterManager,127.0.0.1:6016] localport[5600] pid[115994] successfully!
The response message: RSP Code[0].{0:success; other: fail.}
[09-18 15:19:43:820]Successful response: 
ClusterID: 2, GroupID: 1
MasterDB[6]: [192.168.182.41:5502] cityID:1 roomID:1 mysql-bin.000001:564550
sync binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000001:564550
    DB[5]: [192.168.182.42:5501] cityID:1 roomID:1 mysql-bin.000001:564550
relay binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000001:564550
    DB[5]: [192.168.182.42:5501] cityID:1 roomID:1 mysql-bin.000001:564550
 
~success~

由于业务租户的 ClusterID 为 2，集中式集群，只有一个分片组，因此：-clusterID=2 -groupID=1 。

MasterDB 的信息： [192.168.182.41:5502] ；
同步备库：
- standby： [192.168.182.43:5501]
- standby： [192.168.182.42:5501]

我们停掉 MasterDB 验证一下：

[zxdb2@golden1 ~]$ dbmoni -stop
[dbmoni -stop]The script dbmoni dbagent process is stopped
Shutting down MySQL ....... SUCCESS! 
[dbagent]The dbagent process is stopped
[dbmoni -stop]The script dbmoni loadserver process is stopped
[loadserver]The loadserver process is stopped
[dbmoni -stop]The script dbmoni ommagent process is stopped
[ommagent]The ommagent process is stopped

[zxdb2@golden1 ~]$ ps -ef |grep zxdb2 |grep mysqld
zxdb2     45346  26712  0 15:28 pts/1    00:00:00 grep --color=auto mysqld

[zxdb2@golden1 ~]$ dbstate
[dbagent]The dbagent process is not running
[loadserver]The loadserver process is not running
[ommagent]The ommagent process is not running

可以看到，最先停掉该用户下的 mysqld 进程，随后停掉了 dbagent 、loadserver 和 ommagent 3 个主进程。再次查看业务租户的同步状态：

[zxmanager@golden1 ~]$  dbtool -cm -query-latest-slave -clusterID=2 -groupID=1
Send message to module[ClusterManager,127.0.0.1:6016] localport[5600] pid[61989] successfully!
The response message: RSP Code[0].{0:success; other: fail.}
[09-18 15:30:34:210]Successful response: 
ClusterID: 2, GroupID: 1
MasterDB[5]: [192.168.182.42:5501] cityID:1 roomID:1 mysql-bin.000002:867
sync binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000002:867
relay binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000002:867
 
~success~

MasterDB ： [192.168.182.42:5501]
备库是 [192.168.182.43:5501]
原先的主库 [192.168.182.41:5502] 则不在列表中了。

现在我们启动一下：

[zxdb2@golden1 ~]$ dbmoni -start
[dbmoni -start]dbmoni dbagent ......starting......
[dbmoni -start]dbmoni loadserver ......starting......
[dbmoni -start]dbmoni ommagent ......starting......

-- 数据库进程启动了
[zxdb2@golden1 ~]$ ps -ef |grep zxdb2 |grep mysqld
zxdb2     87011      1 97 15:50 ?        00:00:03 /zxdb2/bin/mysqld --defaults-file=/zxdb2/etc/my.cnf --datadir=/zxdb2/data/data --pid-file=/zxdb2/bin/mysqld1.pid
zxdb2     87567  26712  0 15:50 pts/1    00:00:00 grep --color=auto mysqld
[zxdb2@golden1 ~]$ dbstate
[dbagent]The dbagent process is running, PID:86234
[loadserver]The loadserver process is running, PID:86216
[ommagent]The ommagent process is running, PID:86275

再次查看主备关系：

[zxmanager@golden1 ~]$  dbtool -cm -query-latest-slave -clusterID=2 -groupID=1
Send message to module[ClusterManager,127.0.0.1:6016] localport[5600] pid[93443] successfully!
The response message: RSP Code[0].{0:success; other: fail.}
[09-18 15:51:38:320]Successful response: 
ClusterID: 2, GroupID: 1
MasterDB[5]: [192.168.182.42:5501] cityID:1 roomID:1 mysql-bin.000002:867
sync binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000002:867
    DB[6]: [192.168.182.41:5502] cityID:1 roomID:1 mysql-bin.000002:867
relay binlog DBs:
   *DB[4]: [192.168.182.43:5501] cityID:1 roomID:1 mysql-bin.000002:867
    DB[6]: [192.168.182.41:5502] cityID:1 roomID:1 mysql-bin.000002:867
 
~success~

原先的主库 [192.168.182.41:5502] 作为备库加入到复制关系中了。

1.5 Insight 组件

Insight 的管理命令是 serverdbmoni，稍微特殊一些。

[insight@golden1 ~]$ netstat -ntlp |grep 8444
(Not all processes could be identified, non-owned process info
 will not be shown, you would have to be root to see it all.)
tcp6       0      0 :::8444                 :::*                    LISTEN      68540/java          

[insight@golden1 ~]$ serverdbmoni -stop
[serverdbmoni -stop] serverdbmonidamon will stop, kill -9 68378
[serverdbmoni -stop] serverdbmonidamon is not running, stop cmd is ignored
[serverdbmoni -stop] insightserver will stop, kill -9 68540
[serverdbmoni -stop] insightserver is not running, stop cmd is ignored
[serverdbmoni -stop] /home/goldendb/insight/bin/.serverdbmonidamon_lock will stop, kill -9 128631
[serverdbmoni -stop] /home/goldendb/insight/bin/.serverdbmonidamon_lock is not running, stop cmd is ignored

状态查看：8444 端口是 Insight 的 WEB 端口。

[insight@golden1 ~]$ serverdbmoni -status
[serverdbmoni -status] serverdbmonidamon pid:
[serverdbmoni -status] insightserver pid:

[insight@golden1 ~]$ netstat -ntlp |grep 8444
(Not all processes could be identified, non-owned process info
 will not be shown, you would have to be root to see it all.)

启动 Insight ，再次查看状态，8444 端口已经被占用了。

[insight@golden1 ~]$ serverdbmoni -start
[serverdbmoni -start] serverdbmonidamon ......starting......Done


[insight@golden1 ~]$ serverdbmoni -status
[serverdbmoni -status] serverdbmonidamon pid:15874
[serverdbmoni -status] insightserver pid:16016

[insight@golden1 ~]$ netstat -ntlp |grep 8444
(Not all processes could be identified, non-owned process info
 will not be shown, you would have to be root to see it all.)
tcp6       0      0 :::8444                 :::*                    LISTEN      16016/java

2. Insight 界面操作

2.1 Manager 等组件的管理

2.2 租户的管理

总体来说，GoldenDB 的组件管理比较方便，既可以在 Insight 可视化界面进行集中管控，也可以在命令行中执行操作命令，虽然需要切换到组件对应的用户下，但是能够显著降低误操作风险。

这里是《实战派K8S&DB》，更多干货敬请关注公众号。如果本文对你有所帮助，欢迎点赞、推荐和转发，也欢迎关注后续文章，一起考证、一起学习 GoldenDB 分布式数据库。

刷新页面返回顶部

圣彼得堡的村长