zabbix的监控主机数量将近300，且运行了一年时间了，最近zabbix server服务监控历史数据等服务不断自身告警、查询性能也变得很低 关于历史数据的两个参数，在zabbix server的配置文件中 可以选择关闭housekeeper禁止自动定期清除历史记录数据，因为对于大数据的删除会直接影响zabbix的性能、或者调整相应参数 HousekeepingFrequency 取值范围：0-24 默认值：1 说明：housekeep执行频率，默认每小时回去删除一些过期数据。如果server重启，那么30分钟之后才执行一次，接下来，每隔一小时在执行一次。 MaxHousekeeperDelete 取值范围： 0-1000000 默认值：5000 housekeeping一次删除的数据不能大于MaxHousekeeperDelete 数据库优化 一、设置独立表空间（innodb_file_per_table=1） # 5.6版本以上自动开启 以上版本跳过这一段 1、清空history数据 [url=] [/url][root@Zabbix-Server ~]# mysql -u zabbix -pMariaDB [(none)]> use zabbix;Reading table information for completion of table and column namesYou can turn off this feature to get a quicker startup with -ADatabase changedMariaDB [zabbix]> MariaDB [zabbix]> truncate table history;Query OK, 0 rows affected (0.19 sec)MariaDB [zabbix]> optimize table history;+----------------+----------+----------+-------------------------------------------------------------------+| Table | Op | Msg_type | Msg_text |+----------------+----------+----------+-------------------------------------------------------------------+| zabbix.history | optimize | note | Table does not support optimize, doing recreate + analyze instead || zabbix.history | optimize | status | OK |+----------------+----------+----------+-------------------------------------------------------------------+2 rows in set (0.81 sec)MariaDB [zabbix]> truncate table history_str;Query OK, 0 rows affected (0.05 sec) MariaDB [zabbix]> truncate table history_uint;Query OK, 0 rows affected (6.32 sec)[url=] [/url]2、修改表结构 [url=][/url]MariaDB [(none)]> use zabbix;Reading table information for completion of table and column namesYou can turn off this feature to get a quicker startup with -ADatabase changedMariaDB [zabbix]> Alter table history_text drop primary key, add index (id), drop index history_text_2, add index history_text_2 (itemid, id);Query OK, 0 rows affected (1.11 sec) Records: 0 Duplicates: 0 Warnings: 0MariaDB [zabbix]> Alter table history_log drop primary key, add index (id), drop index history_log_2, add index history_log_2 (itemid, id);Query OK, 0 rows affected (0.14 sec) Records: 0 Duplicates: 0 Warnings: 0[url=] [/url]修改完之后再按照官网上的过程创建四个存储过程： 3、将官方的四个分散代码拷贝至一个文件保存为sql，导入数据库；  View Code[url=] [/url][root@Zabbix-Server ~]# mysql -u zabbix -p zabbixEnter password: Reading table information for completion of table and column namesYou can turn off this feature to get a quicker startup with -AWelcome to the MariaDB monitor. Commands end with ; or \g.Your MariaDB connection id is 48790Server version: 5.5.52-MariaDB MariaDB ServerCopyright (c) 2000, 2016, Oracle, MariaDB Corporation Ab and others.Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.MariaDB [zabbix]> use zabbix;Database changedMariaDB [zabbix]> source /root/zabbix-partition.sql;Query OK, 0 rows affected (0.04 sec)Query OK, 0 rows affected (0.00 sec)Query OK, 0 rows affected (0.00 sec)Query OK, 0 rows affected (0.00 sec)MariaDB [zabbix]> CALL partition_maintenance('zabbix', 'history_log', 28, 24, 14);+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801100000,1515600000) |+---------------------------------------------------------------+1 row in set (0.18 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801110000,1515686400) |+---------------------------------------------------------------+1 row in set (0.48 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801120000,1515772800) |+---------------------------------------------------------------+1 row in set (0.67 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801130000,1515859200) |+---------------------------------------------------------------+1 row in set (1.02 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801140000,1515945600) |+---------------------------------------------------------------+1 row in set (1.22 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801150000,1516032000) |+---------------------------------------------------------------+1 row in set (1.44 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801160000,1516118400) |+---------------------------------------------------------------+1 row in set (1.64 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801170000,1516204800) |+---------------------------------------------------------------+1 row in set (1.85 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801180000,1516291200) |+---------------------------------------------------------------+1 row in set (2.04 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801190000,1516377600) |+---------------------------------------------------------------+1 row in set (2.23 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801200000,1516464000) |+---------------------------------------------------------------+1 row in set (2.42 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801210000,1516550400) |+---------------------------------------------------------------+1 row in set (2.62 sec)+---------------------------------------------------------------+| msg |+---------------------------------------------------------------+| partition_create(zabbix,history_log,p201801220000,1516636800) |+---------------------------------------------------------------+1 row in set (2.85 sec)+--------------------+--------------------+| table | partitions_deleted |+--------------------+--------------------+| zabbix.history_log | N/A |+--------------------+--------------------+1 row in set (3.10 sec)Query OK, 0 rows affected, 1 warning (3.10 sec)[url=] [/url]4、对想要分区的表进行表分区 [url=][/url]DELIMITER $$CREATE PROCEDURE `partition_maintenance_all`(SCHEMA_NAME VARCHAR(32))BEGIN CALL partition_maintenance(SCHEMA_NAME, 'history', 7, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'history_log', 7, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'history_str', 7, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'history_text', 7, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'history_uint', 7, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'trends', 365, 24, 14); CALL partition_maintenance(SCHEMA_NAME, 'trends_uint', 365, 24, 14);END$$DELIMITER ; 以上代码部分的含义为(库名,表名,保存多少天的数据,每隔多久生成一个分区,本次生成多少分区） [url=] [/url]mysql> source /root/partition_maintenance_all.sql;Query OK, 0 rows affected (0.00 sec)mysql> CALL partition_maintenance_all('zabbix'); 5、Housekeeper 设置 Zabbix用户界面中的 "Administration" -> "部分提供了所有选项。确保在右上角的下拉列表中选择"Housekeeping" 您应该看到类似于以下的屏幕：  
6、加入计划任务 不要让数据库用完你的分区，上面示例是如何创建14天额外分区的，在第15天，数据库将无法粘贴历史/趋势数据，因此会发生数据丢失。 所以每隔一段时间（通过cron或其他方法）重新运行这些存储过程。通过这样做，分区将始终存在，可以插入数据。 #Q-2018-1/9 30 4 * * 1 /usr/bin/mysql -uzabbix -pzabbix -e "use zabbix;" -e "CALL partition_maintenance_all('zabbix');" 实际在生产环境中上述操作运行一段时间后，Zabbix server的log文件会报如下错误，events表主键重复、主键不能自动增长、导致zabbix不能告警 2581:20180208:213930.461 [Z3005] query failed: [1062] Duplicate entry '8703' for key 'PRIMARY' [insert into events (eventid,source,object,objectid,clock,ns,value) values (8703,0,0,19518,1518097170,457297996,1);可以使用如下命令删除events记录 [root@Zabbix-Server zabbix]# mysql -u zabbix -pzabbix -e "use zabbix;" -e 'delete from events';如果想要删除表的所有数据，truncate语句要比 delete 语句快。 因为 truncate 删除了表，然后根据表结构重新建立它，而 delete 删除的是记录，并没有尝试去修改表。 不过truncate命令虽然快，却不像delete命令那样对事务处理是安全的。 另外注意的是mysql数据库清空表默认是不回收空间的（对应步骤1） 回收表空间的命令 optimize table historyoptimize table history_uint针对MySQL的不同数据库存储引擎，在optimize使用清除碎片，回收闲置的数据库空间，把分散存储（fragmented）的数据和索引重新挪到一起(defragmentation)，对I/O速度有好处。 当然optimize在对表进行操作的时候，会加锁，所以不宜经常在程序中调用。 |
