蓝鲸:安装SaaS组件bk_monitor失败分析解决

使用./bk_install saas-o 安装发现bk_monitor(蓝鲸监控)组件报错“ERROR deploy failed: timeout”。

单独尝试安装各个组件:

#故障自愈
[root@rbtnode1 install]# ./bk_install saas-o bk_fta_solutions

#日志检索
[root@rbtnode1 install]# ./bk_install saas-o bk_log_search

#节点管理
[root@rbtnode1 install]# ./bk_install saas-o bk_nodeman

#标准运维
[root@rbtnode1 install]# ./bk_install saas-o bk_sops

#蓝鲸监控
[root@rbtnode1 install]# ./bk_install saas-o bk_monitor

发现前面几个bk_fta_solutions、bk_log_search、bk_nodeman、bk_sops都可以安装成功,唯独对bk_monitor安装,依然报错如下:

[root@rbtnode1 install]# ./bk_install saas-o bk_monitor
省略输出..
2020-03-09 13:27:36 125  INFO    check deploy result. retry 132
2020-03-09 13:27:39 125  INFO    check deploy result. retry 133
2020-03-09 13:27:39 134  ERROR  deploy failed: timeout
[192.168.1.6]20200309-132739 153   Deploy saas bk_monitor failed.
[192.168.1.6]20200309-132739 47   Abort

进一步查看agent日志(/data/bkce/logs/paas_agent/agent.log),最终因为部署任务timeout而终止,未见其他明显报错:

2020/03/09 13:24:57 job.go:279: Building wheels for collected packages: gevent, netifaces, arrow, msgpack-python, wrapt, itypes, backports.shutil-get-terminal-size, simplegeneric, scandir

2020/03/09 13:24:57 job.go:279:   Running setup.py bdist_wheel for gevent: started

2020/03/09 13:27:32 job.go:279:   Running setup.py bdist_wheel for gevent: still running...

2020/03/09 13:27:38 job.go:297: Deployment task execution timeout

查了些网上资料,说是因为机器配置不够,增加核数为6即可解决,但实际我测试无效,报错不变;
在蓝鲸官方群咨询,客服给出一个解决方案:
– https://bk.tencent.com/s-mart/community/question/787

但是实际这个Case和我这里遇到的情况并不一样,因为我这没有看到那个error。
晚上重新整理下思路,借鉴案例中清理环境的方式,然后重新部署,这次agent.log看到报错信息了:

2020/03/10 02:29:54 job.go:279:   File "/opt/py27_e/lib/python2.7/site-packages/pymysql/connections.py", line 906, in _read_packet

2020/03/10 02:29:54 job.go:279:     packet.check_error()

2020/03/10 02:29:54 job.go:279:   File "/opt/py27_e/lib/python2.7/site-packages/pymysql/connections.py", line 367, in check_error

2020/03/10 02:29:54 job.go:279:     err.raise_mysql_exception(self._data)

2020/03/10 02:29:54 job.go:279:   File "/opt/py27_e/lib/python2.7/site-packages/pymysql/err.py", line 120, in raise_mysql_exception

2020/03/10 02:29:54 job.go:279:     _check_mysql_exception(errinfo)

2020/03/10 02:29:54 job.go:279:   File "/opt/py27_e/lib/python2.7/site-packages/pymysql/err.py", line 115, in _check_mysql_exception

2020/03/10 02:29:54 job.go:279:     raise InternalError(errno, errorvalue)

2020/03/10 02:29:54 job.go:279: django.db.utils.InternalError: (1049, u"Unknown database 'bkdata_monitor_alert'")

2020/03/10 02:29:55 job.go:304: error waiting for Cmd exit status 1

这提示居然是没有这个名称为bkdata_monitor_alert的数据库??
结合之前的agent日志是确认有建表操作成功的,说明是环境清理操作很可能把对应组件的库也给删除了。

这里先不深究,直接查看当前的数据库列表:

MySQL [(none)]> show databases;
+--------------------+
| Database           |
+--------------------+
| information_schema |
| bk_fta_solutions   |
| bk_log_search      |
| bk_monitor         |
| bk_nodeman         |
| bk_sops            |
| bksuite_common     |
| job                |
| jobLog             |
| mysql              |
| open_paas          |
| performance_schema |
| sys                |
+--------------------+
13 rows in set (0.00 sec)

果然没有这个bkdata_monitor_alert库,这里先直接尝试创建一个空库试下:

MySQL [(none)]> create database bkdata_monitor_alert character set utf8;
Query OK, 1 row affected (0.01 sec)

再次尝试bk_monitor的安装:

# 再次安装bk_monitor
[root@rbtnode1 install]# ./bk_install saas-o bk_monitor

# 监控agent.log
[root@rbtnode1 paas_agent]# pwd
/data/bkce/logs/paas_agent
[root@rbtnode1 paas_agent]# tail -20f agent.log 

发现这次agent.log日志最终显示Job正常完成了:

省略部分日志..

2020/03/10 02:45:38 job.go:279:   Applying sessions.0001_initial... OK

2020/03/10 02:45:38 job.go:279: ------change db success------

2020/03/10 02:47:25 job.go:279: ------ start app server ------

2020/03/10 02:47:25 job.go:279: su: ignore --preserve-environment, it's mutually exclusive to --login.

2020/03/10 02:47:25 job.go:279: /etc/profile: line 77: ulimit: open files: cannot modify limit: Operation not permitted

2020/03/10 02:47:25 job.go:279: /etc/profile: line 78: ulimit: open files: cannot modify limit: Operation not permitted

2020/03/10 02:47:25 job.go:279: /etc/profile: line 79: ulimit: open files: cannot modify limit: Operation not permitted

2020/03/10 02:47:25 job.go:279: /etc/profile: line 80: ulimit: open files: cannot modify limit: Operation not permitted

2020/03/10 02:47:26 job.go:279: Last login: Mon Mar  9 14:01:54 CST 2020

2020/03/10 02:47:28 job.go:279: Job Done

2020/03/10 02:47:28 job.go:306: RunJob end ... ...

赶紧去看下安装的窗口,发现这次bk_monitor终于安装成功了:

[root@rbtnode1 install]# ./bk_install saas-o bk_monitor
省略部分日志..

2020-03-10 02:47:24 125  INFO    check deploy result. retry 107
2020-03-10 02:47:26 125  INFO    check deploy result. retry 108
2020-03-10 02:47:29 125  INFO    check deploy result. retry 109
2020-03-10 02:47:30 131  INFO   bk_monitor have been deployed successfully
[192.168.1.6]20200310-024730 151   SaaS application bk_monitor has been deployed successfully
[192.168.1.6]20200310-024730 56   install saas-o(bk_monitor) done

登陆蓝鲸的工作台,也确认这次蓝鲸监控组件已经安装成功,可以正常操作了。

This entry was posted in 蓝鲸智云 and tagged , , . Bookmark the permalink.