We use a monitoring system that does TCP based checks on our various systems and can alerts us based on criteria we define. So we right shell scripts that run locally and return certain values and tie those scripts to specific TCP ports using /etc/inetd.conf and /etc/services - This is the script we use to monitor replication on every machine (it's much shorter without my excessive comments):
#!/bin/sh # # # Bruce's MySQL Replication Verification Script # /usr/local/mysql/bin/mysql -e "show status like 'Slave_running';" This script is then tied to a port, so any web browser or our monitoring system hits http://mysqlserver:9999 (or whatever port you decide on) should get this: Variable_name Value Slave_running ON >From there our monitor takes that data and looks for the keyword "ON", if it's there it's happy, if it matches the keyword "OFF" it sends an alert page and marks the instance as in warning state, any response that doesn't include "ON" or "OFF" generates a service down state and also sends pages etc... (If MySQL is running then the slave status will either be ON or OFF... If mysql isn't running the mysql client returns it's own error saying it's unable to connect). Best Regards, Bruce On 12/28/04 1:44 PM, "Bruce Dembecki" <[EMAIL PROTECTED]> wrote: > Tucker, Gabriel wrote: >> Anil >> >> Write a script that does a slave status and check if either of the threads >> are >> running. You could >> further check for error numbers and descriptions. This is what we do. >> >> Gabe >> -----Original Message----- >> From: Anil Doppalapudi [mailto:[EMAIL PROTECTED] >> Sent: Tuesday, December 21, 2004 6:05 AM >> To: [EMAIL PROTECTED] >> Subject: Monitoring replication in mysql >> >> >> >> Hi, >> >> we have no of mysql replication setups in our setup. how to monitor those >> replication setups.my aim is if any slave goes down >> my script should immediately send an alert mail to me. if anybody having >> already developed scripts please let me know otherwise just give me an idea >> what to monitor in in replication setup. >> >> Thanks >> Anil >> DBA > > We have a script that monitors output from SHOW SLAVE STATUS, but > actually had one time when replication died, but output from above > command looked perfectly fine. It was due to massive table corruption, > which was in turn due to filesystem corruption. Now, we have the same > test running, but we also have a backup monitor which inserts a value in > the master and tries to read it from all replicants. We allow an > acceptable delay (5-10 minutes) before we page all admins with this > backup test. > > Greg > -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
