Forum Discussion

Tuckson's avatar
Tuckson
Icon for Altostratus rankAltostratus
Jun 03, 2021

script not working under v16

Hi,

To create an external monitoring script (internal does not give enough flexibility) I downloaded a template somewhere, also downloaded a trial of andd this thingy

#!/bin/sh
# (c) Copyright 1996-2007 F5 Networks, Inc.
#
# @(#) $Id: http_monitor_cURL+GET,v 1.0 2007/06/28 16:10:15 deb Exp $
# (based on sample_monitor,v 1.3 2005/02/04 18:47:17 saxon)
#
# these arguments supplied automatically for all external monitors:
# $1 = IP (IPv6 notation. IPv4 addresses are passed in the form
#                         ::ffff:w.x.y.z
#                         where "w.x.y.z" is the IPv4 address)
# $2 = port (decimal, host byte order)
#
# Additional command line arguments ($3 and higher) may be specified in the monitor template
# This example does not expect any additional command line arguments
#
# Name/Value pairs may also be specified in the monitor template
# This example expects the following Name/Vaule pairs:
#  URI  = the URI to request from the server
#  RECV = the expected response (not case sensitive)
#
# remove IPv6/IPv4 compatibility prefix (LTM passes addresses in IPv6 format)
#
 
 
 
IP=`echo ${1} | sed 's/::ffff://'`
PORT=${2}
 
 
PIDFILE="/var/run/$(basename ${0}).${IP}_${PORT}.pid"
# kill of the last instance of this monitor if hung and log current pid
if [ -f $PIDFILE ]
then
   echo "EAV exceeded runtime needed to kill ${IP}:${PORT}" | logger -p local0.error
   kill -9 $(cat $PIDFILE) > /dev/null 2>&1
fi
echo "$$" > $PIDFILE
 
 
# send request & check for expected response
curl -m 5 -fNs http://10.128.182.169:8000/$(basename ${NODE_NAME}) | grep "200" 2>&1 > /dev/null
 
 
# mark node UP if expected response was received
if [ $? -eq 0 ]
then
    rm -f $PIDFILE
    echo "UP"
else
    rm -f $PIDFILE
fi
exit

Unfortunately our production loadbalancers turn out to have an older version (v12) running than the one I downloaded (v16). And appaerantly The NODE_NAME variable, which worked so fine in my own setup, is not present there, hence the script does not work.

Is there anything I can do to make this script on F5 v12 go to a target without having to setup each healthmonitor by hand en giving the name/ip as extra fixed parameter? Is the NODE_NAME variable or the ipaddress available in a different variable name or something like that?

THIA

4 Replies

  • Well the NODE_NAME is one of the predefined variables i what i think. But if its not working in v12, then you have to supply it as argument or update your script to define that variable manually.

     

    Can you put logging of $NODE_NAME in your script to see if it gives you something in different format compared to v12 & v16 or does it not get defined at all.

  • Well, I do not have access to that F5 myself unfortunately, but as far as I can see it is empty. The script did run but the url to be tested failed because it was not complete. I could test if I could still download a trial of of vs 12 somewhere?

    • Yes, you can download the v12, spin it up. Create that monitor, create that pool, make sure its reachable. Then you can turn on logging to see what's happening... Good Luck :)

       

      Sorry dont have v12 currently to test it out.

  • Downloaded the V12. Go it to work :-) The NODE_NAME was available however, some other changes in the script were necessary.