记一次FreeBSD系统中mysql服务异常的排查过程

随着监控助理突然提示很多数据库连接错误:
记一次FreeBSD系统中mysql服务异常的排查过程
文章图片

排查数据库错误便随之提上了日程。
重启大法 不得不说,有时候重启大法还是挺好使的。所以我们上来也尝试重启mysql

$ /usr/local/etc/rc.d/mysql-server stop $ /usr/local/etc/rc.d/mysql-server start

再次连接,数据数据直接就连不上了。此时便需要来到正确的轨迹上:看报错内容,根据报错内容来排查原因,解决问题。
错误日志 很遗憾的是,mysql在启动过程中,即使启动失败,也不会报什么的错误信息。我们查看mysql是否成功启动则需要使用mysql-server status命令:
root@YunzhiTest:/usr/home/panjie # /usr/local/etc/rc.d/mysql-server status mysql is not running.

而是否打印日志,以及日志的位置放在哪,则需要我们进行手动配置。在mysql服务成功启动的前提下,我们其实是可以使用mysql的相关命令来查看当前的配置文件位置的,无奈当前mysql并没有成功启动,所以此时则需要借助一些查询软件或是当初安装mysql使用的工具(比如FreeBSD的ports)来查找mysql的配置文件位置了。在FreeBSD中,mysql的配置文件位于/usr/local/etc/mysql中:
root@YunzhiTest:/usr/home/panjie # cd /usr/local/etc/mysql/ root@YunzhiTest:/usr/local/etc/mysql # ls keyringmy.cnfmy.cnf.sample

然后我们备份一个配置文件cp my.cnf my.cnf.bak后再对其进行编辑:
[mysqld] log= /var/log/mysql/mysqld.log log-error= /var/log/mysql/error.log user= mysql port= 3306

在 mysqld 下增加两项:log及log-error,分别存一般日志及错误日志。同时由于当前mysql启动的用户是mysql,还需要保证mysql用户对相关日志路径拥有绝对权限:
$ mdkir /usr/log/mysql $ chown mysql:mysql /usr/log/mysql

查看日志 此时我们再次启动mysql 服务,则可以查看在/var/log/mysql/下生成的error.log文件了:
$ /usr/local/etc/rc.d/mysql-server start

其比较重要的错误信息如下:
2022-07-11T14:22:25.946391Z 0 [Note] InnoDB: Rollback of non-prepared transactions completed 2022-07-11T14:22:25.946435Z 0 [Note] InnoDB: Setting file '/var/db/mysql/ibtmp1' size to 128 MB. Physically writing the file full; Please wait ... 2022-07-11T14:22:25.947132Z 0 [Note] InnoDB: Progress in MB: 1002022-07-11T14:22:26.085805Z 0 [Warning] InnoDB: Retry attempts for writing partial data failed. 2022-07-11T14:22:26.085855Z 0 [ERROR] InnoDB: Write to file /var/db/mysql/ibtmp1failed at offset 133169152, 1048576 bytes should have been written, only 0 were written. Operating system error number 28. Check that your OS and file system support files of this size. Check also that the disk is not full or a disk quota exceeded. 2022-07-11T14:22:26.085940Z 0 [ERROR] InnoDB: Error number 28 means 'No space left on device' 2022-07-11T14:22:26.085951Z 0 [Note] InnoDB: Some operating system error numbers are described at http://dev.mysql.com/doc/refman/5.7/en/operating-system-error-codes.html 2022-07-11T14:22:26.085968Z 0 [ERROR] InnoDB: Could not set the file size of '/var/db/mysql/ibtmp1'. Probably out of disk space

上述错误大概就是在说一个问题:磁盘空间满了,此问题导致mysql无法启动。
整理数据 问题的根本原因找到了,解决问题便成了最轻松的事情。
root@YunzhiTest:/usr/local/etc/mysql # df -h FilesystemSizeUsedAvail CapacityMounted on /dev/ufsid/59a7effe7885633c39G36G124M100%/ devfs1.0K1.0K0B100%/dev zroot/mengyunzhi48G40G8.4G82%/mengyunzhi zroot8.4G23K8.4G0%/zroot

首先我们查看my.cnf中的数据库文件配置路径:
datadir= /var/db/mysql tmpdir= /var/db/mysql_tmpdir slave-load-tmpdir= /var/db/mysql_tmpdir secure-file-priv= /var/db/mysql_secure

然后依次查看其占用空间:
root@YunzhiTest:/var/db # du -h -d 1 180M./portsnap 3.1M./etcupdate 8.0K./zfsd 36K./entropy 4.0K./ipf 4.0K./hyperv 87M./pkg 688K./ports 1.5G./freebsd-update 12K./ntp 148K./fontconfig 8.0K./sudo 18G./mysql 4.0K./mysql_secure 4.0K./mysql_tmpdir 8.0K./redis 8.0K./colord 20G.

发现mysql占用了18G,但实际上并没有这么多数据。进入mysql文件夹后继续查看:
root@YunzhiTest:/var/db/mysql # du -ah | sort -h 104M./log/log.ibd 105M./log 130M./mysql/slow_log.CSV 131M./mysql-bin.000108 136M./measurement/instrument.ibd 142M./mysql-bin.000113 145M./mysql-bin.000104 150M./mysql 190M./mysql-bin.000114 214M./mysql-bin.000111 224M./mysql-bin.000109 230M./mysql-bin.000103 256M./ib_logfile0 256M./ib_logfile1 256M./mysql-bin.000106 274M./mysql-bin.000107 287M./mysql-bin.000110 344M./mysql-bin.000102 346M./instrument 380M./mysql-bin.000112 404M./measurement/instrument_check_info_mandatory_instrument_check_ability_list.ibd 502M./mysql-bin.000120 658M./mysql-bin.000121 678M./mysql-bin.000125 786M./mysql-bin.000116 813M./mysql-bin.000123 900M./mysql-bin.000118 1.0G./measurement 1.0G./mysql-bin.000115 1.0G./mysql-bin.000117 1.0G./mysql-bin.000119 1.0G./mysql-bin.000122 1.0G./mysql-bin.000124 1.2G./switchgear1 1.2G./switchgear1/record_value.ibd 2.3G./ibdata1

最终发现空间大户如上,我们发现系统中的.mysql-bin文件占据了较大的空间,而mysql-bin文件大体有两个作用:1是用来进行数据恢复;2是在主从数据库的时保障高可用性。
虽然可以删除相应的mysql-bin文件,但是保留该文档还是有一定的必要性的。但我们可以将其保留的日期缩短一些,比如我们只保留一周的。查看文件的生成日期:
root@YunzhiTest:/var/db/mysql # ls -alh -rw-r-----1 mysqlmysql344M Jun 13 17:02 mysql-bin.000102 -rw-r-----1 mysqlmysql229M Jun 14 13:53 mysql-bin.000103 -rw-r-----1 mysqlmysql145M Jun 14 20:44 mysql-bin.000104 -rw-r-----1 mysqlmysql56M Jun 15 00:11 mysql-bin.000105 -rw-r-----1 mysqlmysql256M Jun 15 22:34 mysql-bin.000106 -rw-r-----1 mysqlmysql274M Jun 16 11:29 mysql-bin.000107 -rw-r-----1 mysqlmysql131M Jun 16 17:38 mysql-bin.000108 -rw-r-----1 mysqlmysql224M Jun 17 04:00 mysql-bin.000109 -rw-r-----1 mysqlmysql287M Jun 17 17:26 mysql-bin.000110 -rw-r-----1 mysqlmysql214M Jun 18 03:29 mysql-bin.000111 -rw-r-----1 mysqlmysql380M Jun 18 21:19 mysql-bin.000112 -rw-r-----1 mysqlmysql142M Jun 20 17:02 mysql-bin.000113 -rw-r-----1 mysqlmysql189M Jun 21 00:09 mysql-bin.000114 -rw-r-----1 mysqlmysql1.0G Jun 22 19:35 mysql-bin.000115 -rw-r-----1 mysqlmysql785M Jun 24 00:16 mysql-bin.000116 -rw-r-----1 mysqlmysql1.0G Jun 25 19:06 mysql-bin.000117 -rw-r-----1 mysqlmysql900M Jun 27 08:14 mysql-bin.000118 -rw-r-----1 mysqlmysql1.0G Jun 29 11:30 mysql-bin.000119 -rw-r-----1 mysqlmysql502M Jul1 13:09 mysql-bin.000120 -rw-r-----1 mysqlmysql657M Jul5 01:38 mysql-bin.000121 -rw-r-----1 mysqlmysql1.0G Jul6 21:05 mysql-bin.000122 -rw-r-----1 mysqlmysql813M Jul8 09:05 mysql-bin.000123 -rw-r-----1 mysqlmysql1.0G Jul 10 10:36 mysql-bin.000124 -rw-r-----1 mysqlmysql677M Jul 11 21:28 mysql-bin.000125

发现该文件当前保存了近1个月,此时我们先删除两个稍大的历史文件,把空间释放一些出来,然后再去修改一下my.cnf中的保留日期将其缩短为10天。
root@YunzhiTest:/var/db/mysql # rm mysql-bin.000115 root@YunzhiTest:/var/db/mysql # rm mysql-bin.000124 root@YunzhiTest:/var/db/mysql # df -h FilesystemSizeUsedAvail CapacityMounted on /dev/ufsid/59a7effe7885633c39G34G2.1G94%/ devfs1.0K1.0K0B100%/dev zroot/mengyunzhi48G40G8.4G82%/mengyunzhi zroot8.4G23K8.4G0%/zroot

将bin文件的保留天数据设置为10:
binlog_cache_size= 16M expire_logs_days= 10

最后尝试启动mysql
root@YunzhiTest:/usr/local/etc/mysql # /usr/local/etc/rc.d/mysql-server start Starting mysql. root@YunzhiTest:/usr/local/etc/mysql # /usr/local/etc/rc.d/mysql-server status mysql is not running.

其实除此方法外,如果你的第二硬盘空间够用,还可以直接把mysql的数据文件迁移到第二块硬盘上,我只所以没有这么做是由于我第二块硬盘的剩余空间也仅有8.4G,而这个值小于当前mysql的占用空间18G。所以即便是我想进行迁移,也迁移不过去。其根本原因是由于当下有个系统需要上传大量的较大的文件,而我并没有使用存储来处理这些文件,是时候使用存储来专门存放资源文件了。
【记一次FreeBSD系统中mysql服务异常的排查过程】追踪:虽然将expire_logs_days的值设置成了10,但mysql在启动的时候并没有自动删除历史的日志,可能还需要在某个时间节点上触发吧,待后续进行追踪。

    推荐阅读