1

Is it possible to monitor mysql like:

  • if some queries run more then 300 seconds - it will be warning
  • if some queries run more then 500 seconds - it will be critical

I tried:

/usr/lib/nagios/plugins/check_mysql_health --hostname localhost --username icinga --password XXX --mode slow-queries --warning 300

But this only shows the rate slow_queries/per second. How can I monitor mysql as I describe?

Thank you for your help.

Br,

3

2 Answers 2

1

Check mysql scripts by percona. There's some very good utilities: http://www.percona.com/software/percona-toolkit http://www.percona.com/software/percona-monitoring-plugins

0

Using Nagios to monitor MySQL service availability is OK but certainly not slow queries.

I use a simple bash script which run every x seconds in crontab and scan processlist to capture queries running for more than 180 seconds. Hope it fits into your case aswell:

    #!/bin/bash


[email protected] 

count=0

# capture all running queries
echo "<TABLE BORDER=1><TR><TH>Queries running more than 180 seconds</TH></TR></TABLE>" > /tmp/long_running_queries.htm
mysql -uroot -ppassword -s -e "SELECT  now(), ID, USER, HOST, DB, COMMAND, TIME, STATE, INFO FROM  information_schema.PROCESSLIST WHERE HOST not in ('%','localhost') AND COMMAND not in ('Sleep','Binlog Dump') AND TIME > 120;" > /tmp/all_running_queries.txt
echo "<TABLE BORDER=1>" >> /tmp/long_running_queries.htm

# store the output
while IFS= read -r ROW
do
    count=$(($count + 1))
    echo "<TR>" >> /tmp/long_running_queries.htm
    echo "$ROW"  >> /tmp/long_running_queries.htm
    echo "</TR>" >> /tmp/long_running_queries.htm
done < /tmp/all_running_queries.txt 

# if there are more than 2 long running queries then send the output from while loop above into mail
# else pruge the output
if (("$count" > "2"));
then
    echo "</TABLE>" >> /tmp/long_running_queries.htm
    Subject="$count SQL queries running for more than 180 Seconds"
    Body="Some text"
    echo $Body | mutt -a /tmp/long_running_queries.htm -s "$Subject" $Notify
else
    echo "" > /tmp/long_running_queries.htm
fi

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .