pt-heartbeat --update --daemonize 只执行一次秒退的问题

Posted 2021-01-25 dataart

使用pt-heartbeat帮助测试mysql从库delay时长-

第一次运行测试成功，运行顺畅。

过几天再次用--update测试时，发现pt-heartbeat只update一次就自己退出了，--daemonize参数似乎不起作用。

而且程序退出的exitcode还是0.

排查半天未发现问题。

后在从库上测试，发现出现同样问题，但这次注意到--stop的时候有提示：

[[email protected] mysql3306]# /usr/bin/pt-heartbeat --stop --database=pt --defaults-file=/data/mysql/mysql3306/conf/my3306.cnf --socket=/data/mysql/mysql3306/3306.sock --ask-pass
Successfully created file /tmp/pt-heartbeat-sentinel

查看pt-heartbeat代码发现该文件存在pt-heartbeat就会自动退出。

删除掉该文件后，主库上终于可以运行pt了。

 

重新仔细阅读pt-heartbeat的man手册才发现，手册里早已经进行了说明。

       --sentinel
           type: string; default: /tmp/pt-heartbeat-sentinel

           Exit if this file exists.

       --stop
           Stop running instances by creating the sentinel file.

           This should have the effect of stopping all running instances which are watching the same sentinel file.  If none of "--update", "--monitor" or "--check" is specified, "pt-heartbeat" will exit
           after creating the file.  If one of these is specified, "pt-heartbeat" will wait the interval given by "--interval", then remove the file and continue working.

           You might find this handy to stop cron jobs gracefully if necessary, or to replace one running instance with another.  For example, if you want to stop and restart "pt-heartbeat" every hour
           (just to make sure that it is restarted every hour, in case of a server crash or some other problem), you could use a "crontab" line like this:

            0 * * * * pt-heartbeat --update -D test --stop
              --sentinel /tmp/pt-heartbeat-hourly

           The non-default "--sentinel" will make sure the hourly "cron" job stops only instances previously started with the same options (that is, from the same "cron" job).

           See also "--sentinel".

[[email protected] mysql3306]# cat /tmp/pt-heartbeat-sentinel
Remove this file to permit pt-heartbeat to run