<html> <head> <META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=us-ascii"> <meta name=Generator content="Microsoft Word 10 (filtered)"> <style>  </style> </head> <body lang=EN-US link=blue vlink=purple> <div class=Section1> Hi All, I wrote a useful Perl script 2 years ago to execute Nagios handlers which I forgot to share. I perused the Nagios source and various emails around the lists, but didn't see anything with a similar function. If there is something out there already, please let me know! I'm always interested in going more "standard" The script performs two functions: 1) Abstracts the Nagios "states" (service state, type, attempts) into a command line configuration option. This removes the need (in general) to write functionality into the other handlers to determine whether they should run during "OK" state, or just "CRITICAL", hard vs. soft, etc. I got tired of writing these into our many handler scripts. 2) Allow you to chain multiple scripts together with one handler. This is very important to us at our company. We want to, for example, run 2 debugging scripts when a service goes into a CRITICAL SOFT state, but then when it's CRITICAL HARD, restart something. Can you do this with a custom handler script? Of course, but that's a lot of work. (relatively speaking) Especially if you have many handlers and options, like we do. Let me give you two examples here using the script: define command { command_name db_snapshot_handler command_line $USER10$/handler.pl $SERVICESTATE$ $STATETYPE$ $SERVICEATTEMPT$ ,C,S,4 $USER10$/db_snapshot.pl $HOSTNAME$ ,C,S,4 $USER12$/logevent.pl source=Nagios severity=NOTIFICATION type=DBCRISISLOAD targethost=$HOSTNAME$ ,C,S,4 $USER10$/db_poke.pl -h $HOSTNAME$ -s DB_Crisis_Load ,C,H,+ $USER10$/db_snapshot.pl $HOSTNAME$ } (** I've taken the liberty of breaking out the lines to make them more clear. In the actual configuration, they are all in one line**) The first part calls the "handler.pl" script itself and gives it the current State, Type, and Attempt number. Past that, we have three triggering events: **** ,C,S,4 $USER10$/db_snapshot.pl $HOSTNAME$ The first ',' is the delimiter for a new command. This says, on a CRITICAL, SOFT state on the 4th attempt, run USER10$/db_snapshot.pl $HOSTNAME$ **** ,C,S,4 $USER12$/logevent.pl source=Nagios severity ..... On CRITICAL, SOFT on the 4th Attempt, run this command. In the case of two or more commands that run, they run in serial. The script does not currently check exit status or that kind of thing. (It was just a simple thing, really) **** ,C,S,4 $USER10$/db_poke.pl -h $HOSTNAME$ -s DB_Crisis ..... Again, same states as before, run a "db_poke.pl" script with options. So on this state, it will run three things in succession. Of course, all three have to complete before Nagios whacks them based on its own timeouts. Here is another example: define command { command_name oc4j_pokerestart command_line $USER10$/handler.pl $SERVICESTATE$ $STATETYPE$ $SERVICEATTEMPT$ ,C,H,+ $USER10$/oc4j_restart.pl -s Unknown $HOSTNAME$ ,CW,S,+ $USER10$/oc4j_poke.pl -s Unknown -h $HOSTNAME$ } Let's examine the criteria: **** ,C,H,+ $USER10$/oc4j_restart.pl -s Unknown $HOSTNAME$ This says on CRITICAL, HARD and ANY Attempt #, run the restart script. Of course, there is only ONE 'hard' event handler called, so this is safe. **** ,CW,S,+ $USER10$/oc4j_poke.pl -s Unknown -h $HOSTNAME$ On CRITICAL or WARNING, SOFT, and ANY attempt number, run this script. Obviously, it will run "N - 1" iterations of this, where N is your number of checks that fail before entering a hard state. (In our configuration, N = 3, so oc4j_poke.pl will execute twice) The script is not meant to be complicated or have a lot of features. There are some shortcomings: 1) Uses "system()" in perl - This may cause sub-shell spawns depending on what you call. 2) Is written in Perl - If you have lots of service state changes, or have performance issues, this may not be speedy enough for you. We've had no performance issues, though, in our configuration. 3) Long command lines/complicated chains/etc. - These sorts of things are probably still handled by custom scripts. Is there a better or standard way to do this? I could not find any. You can find a copy of this script here: <a href="http://www.xney.com/handler.txt">http://www.xney.com/handler.txt</a> -Karl </div> </body> </html>