做过Zabbix的同学都知道，Zabbix通过专用的Agent或者SNMP收集相关的监控数据，然后存储到数据库里面实时在前台展示。Zabbix监控数据主要分为以下两类：
历史数据：history相关表,从history_uint表里面可以查询到设备监控项目的最大，最小和平均值,即存储监控数据的原始数据。
趋势数据：trends相关表,趋势数据是经过Zabbix计算的数据，数据是从history_uint里面汇总的，从trends_uint可以查看到监控数据每小时最大，最小和平均值，即存储监控数据的汇总数据。
Zabbix可以通过两种方式获取历史数据：
1.通过Zabbix前台获取历史数据
通过Zabbix前台查看历史数据非常简单，可以通过Monitoring->Lastest data的方式查看。也可以点击右上角的As plain test按钮保存成文本文件。
2.通过前台获取的数据进行处理和二次查询有很多限制，因此可以通过SQL语句直接从后台DB查询数据。
首先大家应该熟悉SQL语句Select 常用用法：
|
1
2
3
4
5
6
7
|
SELECT [ALL | DISTINCT] Select_List [INTO [New_Table_name]
FROM { Table_name | View_name} [ [,{table2_name | view2_name}
[,...] ]
[ WHERE Serch_conditions ]
[ GROUP BY Group_by_list ]
[ HAVING Serch_conditions ]
[ ORDER BY Order_list [ASC| DEsC] ]
|
说明：
1)SELECT子句指定要查询的特定表中的列，它可以是*，表达式，列表等。
2)INTO子句指定要生成新的表。
3)FROM子句指定要查询的表或者视图。
4)WHERE子句用来限定查询的范围和条件。
5)GROUP BY子句指定分组查询子句。
6)HAVING子句用于指定分组子句的条件。
7)ORDER BY可以根据一个或者多个列来排序查询结果，在该子句中，既可以使用列名，也可以使用相对列号，ASC表示升序，DESC表示降序。
8）mysql聚合函数：sum()，count()，avg()，max()，avg()等都是聚合函数，当我们在用聚合函数的时候，一般都要用到GROUP BY 先进行分组，然后再进行聚合函数的运算。运算完后就要用到Having子句进行判断了,例如聚合函数的值是否大于某一个值等等。
从Zabbix数据库中查询监控项目方法,这里已查询主机的网卡流量为例子：
1）通过hosts表查找host的ID。
|
1
2
3
4
5
6
7
|
mysql> select host,hostid from hosts where host="WWW05";
+-------+--------+
| host | hostid |+-------+--------+
| WWW05 | 10534 |+-------+--------+
1 row in set (0.00 sec)
|
2）通过items表查找主的监控项和key以及itemid。
|
1
2
3
4
5
6
7
|
mysql> select itemid,name,key_ from items where hostid=10534 and key_="net.if.out[eth0]";
+--------+-----------------+------------------+
| itemid | name | key_ |
+--------+-----------------+------------------+
| 58860 | 发送流量： | net.if.out[eth0] |
+--------+-----------------+------------------+
1 row in set (0.00 sec)
|
3）通过itemid查询主机的监控项目（history_uint或者trends_uint)，单位为M。
主机流入流量：
|
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
|
mysql> select from_unixtime(clock) as DateTime,round(value/1024/1024,2) as Traffic_in from history_uint where itemid="58855" and from_unixtime(clock)>='2014-09-20' and from_unixtime(clock)<'2014-09-21' limit 20;
+---------------------+------------+
| DateTime | Traffic_in |+---------------------+------------+
| 2014-09-20 00:00:55 | 0.10 || 2014-09-20 00:01:55 | 0.09 || 2014-09-20 00:02:55 | 0.07 || 2014-09-20 00:03:55 | 0.05 || 2014-09-20 00:04:55 | 0.03 || 2014-09-20 00:05:55 | 0.06 || 2014-09-20 00:06:55 | 0.12 || 2014-09-20 00:07:55 | 0.05 || 2014-09-20 00:08:55 | 0.10 || 2014-09-20 00:09:55 | 0.10 || 2014-09-20 00:10:55 | 0.12 || 2014-09-20 00:11:55 | 0.12 || 2014-09-20 00:12:55 | 0.13 || 2014-09-20 00:13:55 | 3.16 || 2014-09-20 00:14:55 | 0.23 || 2014-09-20 00:15:55 | 0.24 || 2014-09-20 00:16:55 | 0.26 || 2014-09-20 00:17:55 | 0.23 || 2014-09-20 00:18:55 | 0.14 || 2014-09-20 00:19:55 | 0.16 |+---------------------+------------+
20 rows in set (0.82 sec)
|
主机流出流量：
|
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
|
mysql> select from_unixtime(clock) as DateTime,round(value/1024/1024,2) as Traffic_out from history_uint where itemid="58860" and from_unixtime(clock)>='2014-09-20' and from_unixtime(clock)<'2014-09-21' limit 20;
+---------------------+-------------+
| DateTime | Traffic_out |+---------------------+-------------+
| 2014-09-20 00:00:00 | 4.13 || 2014-09-20 00:01:00 | 3.21 || 2014-09-20 00:02:00 | 2.18 || 2014-09-20 00:03:01 | 1.61 || 2014-09-20 00:04:00 | 1.07 || 2014-09-20 00:05:00 | 0.92 || 2014-09-20 00:06:00 | 1.23 || 2014-09-20 00:07:00 | 2.76 || 2014-09-20 00:08:00 | 1.35 || 2014-09-20 00:09:00 | 3.11 || 2014-09-20 00:10:00 | 2.99 || 2014-09-20 00:11:00 | 2.68 || 2014-09-20 00:12:00 | 2.55 || 2014-09-20 00:13:00 | 2.89 || 2014-09-20 00:14:00 | 4.98 || 2014-09-20 00:15:00 | 6.56 || 2014-09-20 00:16:00 | 7.34 || 2014-09-20 00:17:00 | 6.81 || 2014-09-20 00:18:00 | 7.67 || 2014-09-20 00:19:00 | 4.11 |+---------------------+-------------+
20 rows in set (0.74 sec)
|
4）如果是两台设备，汇总流量，假如公司出口有两台设备，可以用下面的SQL语句汇总每天的流量。下面SQL语句是汇总上面主机网卡的进出流量的。
|
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
|
mysql> select from_unixtime(clock,"%Y-%m-%d %H:%i") as DateTime,sum(round(value/1024/1024,2)) as Traffic_total from history_uint where itemid in (58855,58860) and from_unixtime(clock)>='2014-09-20'and from_unixtime(clock)<'2014-09-21' group by from_unixtime(clock,"%Y-%m-%d %H:%i") limit 20;
+------------------+---------------+
| DateTime | Traffic_total |+------------------+---------------+
| 2014-09-20 00:00 | 4.23 || 2014-09-20 00:01 | 3.30 || 2014-09-20 00:02 | 2.25 || 2014-09-20 00:03 | 1.66 || 2014-09-20 00:04 | 1.10 || 2014-09-20 00:05 | 0.98 || 2014-09-20 00:06 | 1.35 || 2014-09-20 00:07 | 2.81 || 2014-09-20 00:08 | 1.45 || 2014-09-20 00:09 | 3.21 || 2014-09-20 00:10 | 3.11 || 2014-09-20 00:11 | 2.80 || 2014-09-20 00:12 | 2.68 || 2014-09-20 00:13 | 6.05 || 2014-09-20 00:14 | 5.21 || 2014-09-20 00:15 | 6.80 || 2014-09-20 00:16 | 7.60 || 2014-09-20 00:17 | 7.04 || 2014-09-20 00:18 | 7.81 || 2014-09-20 00:19 | 4.27 |+------------------+---------------+
20 rows in set (1.52 sec)
|
5)查询一天中主机流量的最大值，最小值和平均值。
|
1
2
3
4
5
6
7
|
mysql> select date as DateTime,round(min(traffic)/2014/1024,2) as TotalMinIN,round(avg(traffic)/1024/1024,2) as TotalAvgIN,round(max(traffic)/1024/1024,2) as TotalMaxIN from (select from_unixtime(clock,"%Y-%m-%d") as date,sum(value) as traffic from history_uint where itemid in (58855,58860) and from_unixtime(clock)>='2014-09-20' and from_unixtime(clock)<'2014-09-21' group by from_unixtime(clock,"%Y-%m-%d %H:%i") ) tmp;
+------------+------------+------------+------------+
| DateTime | TotalMinIN | TotalAvgIN | TotalMaxIN |+------------+------------+------------+------------+
| 2014-09-20 | 0.01 | 4.63 | 191.30 |+------------+------------+------------+------------+
1 row in set (1.74 sec)
|
6）查询主机组里面所有主机CPU Idle平均值（原始值）。
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
|
mysql> select from_unixtime(hi.clock,"%Y-%m-%d %H:%i") as DateTime,g.name as Group_Name,h.host as Host, hi.value as Cpu_Avg_Idle from hosts_groups as hg join groups g on g.groupid = hg.groupid join items i on hg.hostid = i.hostid join hosts h on h.hostid=i.hostid join history hi on i.itemid = hi.itemid where g.name='上海机房--项目测试' and i.key_='system.cpu.util[,idle]' and from_unixtime(clock)>='2014-09-24' and from_unixtime(clock)<'2014-09-25' group by h.host,from_unixtime(hi.clock,"%Y-%m-%d %H:%i") limit 10;
+------------------+----------------------------+----------+--------------+
| DateTime | Group_Name | Host | Cpu_Avg_Idle |+------------------+----------------------------+----------+--------------+
| 2014-09-24 00:02 | 上海机房--项目测试 | testwb01 | 94.3960 |
| 2014-09-24 00:07 | 上海机房--项目测试 | testwb01 | 95.2086 |
| 2014-09-24 00:12 | 上海机房--项目测试 | testwb01 | 95.4308 |
| 2014-09-24 00:17 | 上海机房--项目测试 | testwe01 | 95.4580 |
| 2014-09-24 00:22 | 上海机房--项目测试 | testwb01 | 95.4611 |
| 2014-09-24 00:27 | 上海机房--项目测试 | testwb01 | 95.2939 |
| 2014-09-24 00:32 | 上海机房--项目测试 | testwb01 | 96.0896 |
| 2014-09-24 00:37 | 上海机房--项目测试 | testwb01 | 96.5286 |
| 2014-09-24 00:42 | 上海机房--项目测试 | testwb01 | 96.8086 |
| 2014-09-24 00:47 | 上海机房--项目测试 | testwb01 | 96.6854 |
+------------------+----------------------------+----------+--------------+
10 rows in set (0.75 sec)
|
7)查询主机组里面所有主机CPU Idle平均值（汇总值）。
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
|
mysql> select from_unixtime(hi.clock,"%Y-%m-%d %H:%i") as Date,g.name as Group_Name,h.host as Host, hi.value_avg as Cpu_Avg_Idle from hosts_groups as hg join groups g on g.groupid = hg.groupid join items i on hg.hostid = i.hostid join hosts h on h.hostid=i.hostid join trends hi on i.itemid = hi.itemid where g.name='上海机房--项目测试' and i.key_='system.cpu.util[,idle]' and from_unixtime(clock)>='2014-09-10' and from_unixtime(clock)<'2014-09-11' group by h.host,from_unixtime(hi.clock,"%Y-%m-%d %H:%i") limit 10;
+------------------+----------------------------+----------+--------------+
| Date | Group_Name | Host | Cpu_Avg_Idle |
+------------------+----------------------------+----------+--------------+
| 2014-09-10 00:00 | 上海机房--项目测试 | testwb01 | 99.9826 |
| 2014-09-10 01:00 | 上海机房--项目测试 | testwb01 | 99.9826 |
| 2014-09-10 02:00 | 上海机房--项目测试 | testwb01 | 99.9825 |
| 2014-09-10 03:00 | 上海机房--项目测试 | testwb01 | 99.9751 |
| 2014-09-10 04:00 | 上海机房--项目测试 | testwb01 | 99.9843 |
| 2014-09-10 05:00 | 上海机房--项目测试 | testwb01 | 99.9831 |
| 2014-09-10 06:00 | 上海机房--项目测试 | testwb01 | 99.9829 |
| 2014-09-10 07:00 | 上海机房--项目测试 | testwb01 | 99.9843 |
| 2014-09-10 08:00 | 上海机房--项目测试 | testwb01 | 99.9849 |
| 2014-09-10 09:00 | 上海机房--项目测试 | testwb01 | 99.9849 |
+------------------+----------------------------+----------+--------------+
10 rows in set (0.01 sec)
|
8）其它与Zabbix相关的SQL语句。
查询主机已经添加但没有开启监控主机：
|
1
|
select host from hosts where status=1;
|
查询NVPS的值：
|
1
2
3
4
5
6
7
|
mysql> SELECT round(SUM(1.0/i.delay),2) AS qps FROM items i,hosts h WHERE i.status='0' AND i.hostid=h.hostid AND h.status='0' AND i.delay<>0;
+--------+
| qps |+--------+
| 503.40 |+--------+
1 row in set (0.11 sec)
|
查询IDC机房的资产信息：
|
1
2
3
4
5
6
7
8
|
mysql> select name,os,tag,hardware from host_inventory where hostid in (select hostid from hosts_groups where groupid=69) limit 2;
+-------+----------------------------+------+-------------------+
| name | os | tag | hardware |
+-------+----------------------------+------+-------------------+
| SHDBM | CentOS release 5.2 (Final) | i686 | ProLiant DL360 G5 || SHDBS | CentOS release 5.2 (Final) | i686 | ProLiant DL360 G5 |+-------+----------------------------+------+-------------------+
2 rows in set (0.00 sec)
|
查询Zabbix interval分布情况：
|
1
2
3
4
5
6
7
8
9
10
11
12
13
|
mysql> select delay,count(*),concat(round(count(*) / (select count(*) from items where status=0)*100,2),"%") as percent from items where status=0 group by delay order by 2 desc;
+-------+----------+---------+
| delay | count(*) | percent |
+-------+----------+---------+
| 3600 | 41168 | 38.92% || 300 | 35443 | 33.51% || 600 | 16035 | 15.16% || 60 | 12178 | 11.51% || 0 | 902 | 0.85% || 36000 | 46 | 0.04% || 30 | 1 | 0.00% |+-------+----------+---------+
7 rows in set (0.68 sec)
|
总结：通过SQL语句可以查询出任何监控项目的数据，并且在SQL语句的末尾通过into outfile '/tmp/zabbix_result.txt'直接把查询的结果保存到系统上面，在通过脚本发送查询结果到指定的用户，实现自动化查询的过程，网上很少有介绍Zabbix数据库查询的文章，希望对大家有所帮助。
本文转自 sfzhang 51CTO博客，原文链接：http://blog.51cto.com/sfzhang88/1558254，如需转载请自行联系原作者