分享知识，分享快乐

0%

starRocks部署

发表于

[clickhouse]
172.20.85.[111:113]
172.20.85.[138:142]

[fe]
172.20.85.[111:113]

[be]
172.20.85.[138:142]

1
2
ansible clickhouse -m copy -a "src=/root/.bashrc dest=/root/.bashrc "
ansible clickhouse -m shell -a "java -version  "
1
2
3
ansible clickhouse -m shell -a "yum install -y rsync  "
ansible clickhouse  -m synchronize  -a "src=/opt/module/StarRocks dest=/opt/module/"  
ansible clickhouse -m shell -a " ls /opt/module/StarRocks "

fe部署

第一次启动需指定–helper参数，后续再启动无需指定此参数
sh /opt/module/StarRocks/fe/bin/start_fe.sh --helper 172.20.85.111:9010 --daemon

1
2
ansible fe -m shell -a " sh /opt/module/StarRocks/fe/bin/stop_fe.sh "
ansible fe -m shell -a " sh /opt/module/StarRocks/fe/bin/start_fe.sh --daemon "
1
2
3
4
5
6
mysql -h 127.0.0.1 -P9030 -uroot -p
mysql> SHOW PROC '/frontends'\G
ALTER SYSTEM ADD FOLLOWER "172.20.85.112:9010";
ALTER SYSTEM ADD FOLLOWER "172.20.85.142:9010";
# ALTER SYSTEM DROP FOLLOWER "172.20.85.142:9010";
ALTER SYSTEM ADD OBSERVER "172.20.85.113:9010";

be 部署

1
2
3
4
5
6
mysql> SHOW PROC '/backends'\G
ALTER SYSTEM ADD BACKEND "cdh85-138:9050";
ALTER SYSTEM ADD BACKEND "cdh85-139:9050";
ALTER SYSTEM ADD BACKEND "cdh85-140:9050";
ALTER SYSTEM ADD BACKEND "cdh85-141:9050";
ALTER SYSTEM ADD BACKEND "cdh85-142:9050";
1
2
ansible be -m shell -a " /opt/module/StarRocks/be/bin/stop_be.sh "
ansible be -m shell -a " /opt/module/StarRocks/be/bin/start_be.sh --daemon "

如添加过程出现错误，需要通过以下命令将该 BE 节点从集群移除。

1
mysql> ALTER SYSTEM decommission BACKEND "host:port";

监控

wget https://github.com/prometheus/prometheus/releases/download/v2.29.1/prometheus-2.29.1.linux-amd64.tar.gz
tar -xf prometheus-2.29.1.linux-amd64.tar.gz -C /usr/local/

1
2
3
4
5
6
7
8
9
10
11
- job_name: 'StarRocks_Cluster01' # 每一个集群称之为一个job，可以自定义名字作为StarRocks集群名
  metrics_path: '/metrics'    # 指定获取监控项目的Restful Api

  static_configs:
    - targets: ['cdh85-111:8030','cdh85-112:8030','cdh85-113:8030']
      labels:
        group: fe # 这里配置了 fe 的 group，该 group 中包含了 3 个 Frontends

    - targets: ['cdh85-138:8040', 'cdh85-139:8040', 'cdh85-140:8040','cdh85-141:8040','cdh85-142:8040']
      labels:
        group: be # 这里配置了 be 的 group，该 group 中包含了 5 个 Backends

权限

1
2
3
4
5
6
7
8
9
10
11
12
13
-- 创建用户

CREATE USER dba_starRocks@'%' IDENTIFIED BY 'dba_starRocks';

给权限

GRANT SELECT_PRIV,LOAD_PRIV,ALTER_PRIV,CREATE_PRIV,DROP_PRIV ON *.*  TO dba_starRocks;

取消权限
REVOKE LOAD_PRIV,ALTER_PRIV,CREATE_PRIV,DROP_PRIV ON *.*  FROM  dba_starRocks;

查看权限
SHOW GRANTS FOR dba_starRocks

测试环境

[sr]
10.0.19.[156:158]

fe 重启

1
2
ansible sr -m shell -a " sh /opt/module/StarRocks/fe/bin/stop_fe.sh "
ansible sr -m shell -a " sh /opt/module/StarRocks/fe/bin/start_fe.sh --daemon "

be重启

1
2
ansible sr -m shell -a "sh  /opt/module/StarRocks/be/bin/stop_be.sh "
ansible sr -m shell -a "sh  /opt/module/StarRocks/be/bin/start_be.sh --daemon "

hive外部表

需要把hdfs-site.xml文件 放到be和fe的conf目录里 并重启

遇到权限问题 修改 hadoop_env.sh

1
export HADOOP_USER_NAME=hdfs

手动升级

1
\cp -r /opt/module/StarRocks-2.3.0-rc01/apache_hdfs_broker /opt/module/StarRocks/apache_hdfs_broker/../

监控

1
ansible be -m shell -a " netstat -nap |grep 9050 "

supervisord守护进程

/etc/supervisord.d/StartRocks.ini

1
2
3
4
5
6
7
8
9
10
11
12
[program:be]
process_name=%(program_name)s                        ;进程名称
directory=/opt/module/StarRocks/be                   ;工作目录be所在路径
command=sh /opt/module/StarRocks/be/bin/start_be.sh  ;运行的命令be的启动sh命令
autostart=true                                       ;是否随supervisor自动开启
autorestart=true                                     ;是否在挂了之后重启，意外关闭后会重启，比如kill掉
user=root                                            ;用户
numprocs=1                                           ;进程数 如果需要同时启动多个进程，进程名称需要不一样
startretries=3                                       ;启动重试次数
stopasgroup=true                                     ;是否停止子进程
killasgroup=true                                     ;是否杀死子进程
startsecs=5                                          ;启动5秒后，如果还是运行状态才认为进程已经启动
1
2
3
4
5
6
supervisorctl status        //查看所有进程的状态
supervisorctl stop be       //停止
supervisorctl start be      //启动
supervisorctl restart be    //重启
supervisorctl update be     //配置文件修改后使用该命令加载新的配置
supervisorctl reload        //重新启动配置中的所有程序

官网升级

升级 BE 前的准备

1
2
3
4
5
6
7
8
9
10
11
12
13

#为了避免 BE 重启期间不必要的 Tablet 修复，进而影响升级后的集群性能，建议在升级前先在 FE Leader 上执行如下命令以禁用 Tablet 调度功能，
> mysql -h 127.0.0.1 -P9030 -uroot -p

admin set frontend config ("max_scheduling_tablets"="0");
admin set frontend config ("disable_balance"="true");
admin set frontend config ("disable_colocate_balance"="true");

# 在所有 BE 重启升级完成后，通过 show backends 命令确认所有 BE 的 Alive 状态为 true 后，启用 Tablet 调度功能，

admin set frontend config ("max_scheduling_tablets"="2000");
admin set frontend config ("disable_balance"="false");
admin set frontend config ("disable_colocate_balance"="false");

升级

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
scp StarRocks-2.5.0-rc03.tar.gz usdp05:/opt/module/
#ansible clickhouse  -m copy  -a "src=/opt/module/StarRocks-2.5.0-rc03.tar.gz dest=/opt/module/"

cd /opt/module
tar -zxvf StarRocks-2.5.0-rc03.tar.gz 

# 升级 BE 节点
cd /opt/module/StarRocks/be
rm -rf lib.bak && mv lib lib.bak 
rm -rf bin.bak && mv bin bin.bak
cp -r /opt/module/StarRocks-2.5.0-rc03/be/lib .
cp -r /opt/module/StarRocks-2.5.0-rc03/be/bin .
sh bin.bak/stop_be.sh
sh bin/start_be.sh --daemon
ps aux | grep starrocks_be

# 升级 FE 节点
cd /opt/module/StarRocks/fe
rm -rf lib.bak && mv lib lib.bak 
rm -rf bin.bak && mv bin bin.bak
cp -r /opt/module/StarRocks-2.5.0-rc03/fe/lib .
cp -r /opt/module/StarRocks-2.5.0-rc03/fe/bin .
sh bin.bak/stop_fe.sh
sh bin/start_fe.sh --daemon
ps aux | grep StarRocksFE

# 升级 Broker
cd /opt/module/StarRocks/apache_hdfs_broker
rm -rf lib.bak && mv lib lib.bak 
rm -rf bin.bak && mv bin bin.bak
cp -r /opt/module/StarRocks-2.5.0-rc03/apache_hdfs_broker/lib  .   
cp -r /opt/module/StarRocks-2.5.0-rc03/apache_hdfs_broker/bin  .
sh bin.bak/stop_broker.sh
sh bin/start_broker.sh --daemon

批量操作

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
ansible sr -m shell -a " sh /opt/module/StarRocks/be/bin/stop_be.sh "
ansible sr -m shell -a " sh /opt/module/StarRocks/be/bin/start_be.sh --daemon "

ansible sr -m shell -a "supervisorctl status "
ansible sr -m shell -a "supervisorctl start starrocks_be"


ansible sr -m shell -a " sh /opt/module/StarRocks/fe/bin/stop_fe.sh "
ansible sr -m shell -a " sh /opt/module/StarRocks/fe/bin/start_fe.sh --daemon "

ansible sr -m shell -a "supervisorctl status "
ansible sr -m shell -a "supervisorctl start starrocks_fe"


supervisorctl start starrocks_broker