check_mysql_health : Return code of 25 is out of bounds

[vAJ]

Have a check of a MySQL database that is constantly throwing the error: (Return code of 25 is out of bounds)

Command:

Code: Select all

/usr/local/nagios/libexec/check_mysql_health --hostname=aus02nqhpmysq01.dev.*******.com --port=3306 --username=NagiosMon --password="**********" --database=information_schema --mode threadcache-hitrate --warning 90: --critical 80:

When I run this from he cl, I occasionally get this error:

Code: Select all

Illegal division by zero at /usr/local/nagios/libexec/check_mysql_health line 488.

Thoughts? I'm running the latest version of this wizard. check_mysql_health (2.1)

[scottwilkerson]

Does it return correct sometimes?

By chance does aus02nqhpmysq01.dev.*******.com point to a loadbalancer IP or something of that nature, or is it a single server?

[vAJ]

Single server and it responds well most of the time. One single poll will fail on occasion. Next poll corrects itself.

Looks like it's in the calculation of connections/sec, like python craps out on itself or something.

[scottwilkerson]

Could this server be reaching it max-connections (configurable in /etc/my.cnf)?

[vAJ]

It's a lightly used dev/test mysql database with 5000 max connections. Doubt it's getting close to that. I found a python junkie on our development team that is going to dig into this script.

[tmcdonald]

Did you mean to type a return of 255 or 25? If it is 25 I would suspect the script itself assigning the return value incorrectly (check the if/else clauses for the warn/crit values) but for 255 it could be anything really.

[vAJ]

It's a 25.

The illegal division error when running from CLI makes me think it's a problem with the python itself. Looking at stats of the target in real time show it's doing absolutely nothing and I can only reproduce when execute the check in rapid fire.

[abrist]

The illegal division error when running from CLI makes me think it's a problem with the python itself. Looking at stats of the target in real time show it's doing absolutely nothing and I can only reproduce when execute the check in rapid fire.

I would suspect that this is internal to the script itself. Just to clarify, the issue only happens when you run the same check multiple times near simultaneously?

[vAJ]

Well, it happens occasionally as a scheduled check. My assumption is when the script returns the "Illeal Division" to Nagios, Nagios then displays that as "Return Code 25..."

[abrist]

Are these checks running through multiple gearman servers? (One of the servers could have an issue)
It could also be that occasionally the check returns a bad value or unexpected character that is breaking the check or it could be occasionally timing out.