Conn/Ping Test/Graph to Secondary (Backup) IP 🔗 link

Good afternoon,

Having just upgraded from Xymon 4.2.3 on CentOS 5.11, our environment is now Xymon 4.3.28-1.el7.terabithia on CentOS 7.4.

We have a group of systems defined that monitors (via conn/ping) the internal interface of the primary router at each of our remote offices to tell us whether or not the connection is up.  Each office also has a secondary/backup network connection that is used for load balancing, VoIP, and failover.

In order to see each site as a single row on the web page we wrote a client-side extension to ping the backup router's internal interface.  The monitoring, display, and alerting all work like we'd hoped (and has for a long time).  Unlike the primary connection's monitor, however, the backup connection is not creating a RRD file so can't be graphed.  I am trying to fix that, and have been looking at the help page for setting up a custom graph but I can't seem to connect the dots properly.

The key logic from /etc/xymon-client/ext/voiptest.sh are:


##
## Logic to build remote site list happens first
##

foreach my $voipsvr (@rmt_names) {

     $color = GREEN;
     $status = $bbtest . " ok";
     $DATA = "";

     ##
     ## Logic to determine secondary IP from primary IP happens here
     ##

     my ($voipip, $svrname, $junk) = split(/:/, $voipsvr, 3);
     my $pingres = `/usr/sbin/fping -Ae $voipip`;
     $DATA .= $pingres;

     if ($pingres =~ "unreachable") {
        $color = RED;
        $status = $bbtest . " NOT ok"
     }

     ## Send to Hobbit
     #############################################################################
    my $report_date = `/bin/date`;
     chomp($report_date);
     system("$ENV{XYMON} $ENV{XYMSRV} 'status $svrname.voip $color $report_date - $status\n\n$DATA'\n");
}


Could anybody point me in the right direction to get the RRD graph for each secondary TCP/Ping test to work?

Thanks,
Scott

P.S. -- The log/graph wasn't working prior to the upgrade either.  We just upgraded first to make sure our environment was current...
P.P.S. -- We may have set this test up incorrectly, so please feel free to teach us if there is a better way to have done it...

Is there anything in the xymon rrd log?  Have you set up the NCV pieces in
the configs?  For example, I have a test that looks at the thread count for
an internal service...I have a file under /etc/xymon/xymonserver.cfg.d
named after the test (probably not required but helps keep the associations
clear)...I have the following in the file:

GRAPHS_threads="threads"

TEST2RRD+=",threads=ncv"

SPLITNCV_threads="*:NONE,total:GAUGE,consumed:GAUGE"

I prefer splitncv to make xymon split the rrd data into multiple individual
files.  The output in the test that drives the graph detail is:

# Tell Xymon about it


$XYMON $XYMSRV "status $MACHINE.$COLUMN $COLOR $TIMESTAMP


${MSG}
<!--
pctutil: ${THREAD_UTIL}
total: ${TOTAL_THREADS}
consumed: ${THREADS_CONSUMED}
-->

Summary
${COMPONENT} - $THREAD_UTIL% threads used

Details
${STATS_nocolon}
"

the content with the 'var:  is what contains the data to be tracked in
rrd.  It's not clear from the snippet you provided that your message
contains this or if it's being sent across a different channel (I'm
guessing not).  Basically, Xymon may not even know it has data to collect
since it's not being put in a format it expects.  IIRC, it should parse
either "var: value" and "var=value".

=G=

note: there is also a graph definition in /etcxymon/graphs.d
PPS: I'm using the terabithia rpms as well
PPPS: getting the user defined graphs working the way you want can be
tricky...I almost always have to go rooting through the archives.

On Mon, Dec 18, 2017 at 1:32 PM, Scott Driemeier-Showers <

user-8d275c0639b6@xymon.invalid> wrote:

I have been attempting to use the “standard” tcp (conn) tests and pre-defined graphs, rather than building something “custom”.

In /etc/xymon/xymonserver.cfg I added:

TEST2RRD=” … ,voip=tcp”
GRAPHS=” … ,voip”

I did not configure it to use NCV as I understood, based on my reading online, that was only necessary with output formatted as colon-separated “columnar” data.  Since this test runs a simple fping the output looks like this:

10.10.16.251 is alive (25.8 ms)

I rewrote the script yesterday as a shell script (no perl).  Here’s the entire thing:


#!/bin/sh

COLUMN=voip
for RMT in `cat /etc/xymon/hosts.cfg | egrep "vya|tal" | tr -s ' ' | cut -d" " -f1,2 --output-delimiter=,`
do
   ADDR=`echo $RMT | awk -F, '{print $1}'`
   SITE=`echo $RMT | awk -F, '{print $2}'`
   RSLT=ok
   STAT=green
   VOIP=${ADDR/192.168/10.10}
   VOIP=${VOIP/.5./.6.}
   SRVR=${SITE//./,}
   DATA=`/usr/sbin/fping -Ae $VOIP`

   if [[ $DATA == *"unreachable"* ]]
   then
      RSLT="NOT ok"
      STAT=red
   fi
   MESG="Service $COLUMN on $SITE is $RSLT"

   # Results to Xymon
   #echo $ADDR $SRVR $DATA $STAT $MESG
   $XYMON $XYMSRV "status $SRVR.$COLUMN $STAT `date +"%a %b %d %H:%M:%S %Y"` $COLUMN $RSLT

${MESG}


${DATA}
"
done

exit 0


Unfortunately no, the RRD log files are not being created either – so it looks like I’ve missed some critical steps there too.

Thanks,
Scott

I expect you're going to have to use NCV to generate the rrds.  I don't
think Xymon will look at a number and just decide to track it.  I expect
someone will correct me if this is not the case.  I'd focus on getting
Xymon to create and populate the rrd files, then worry about the graphing.

=G=

On Tue, Dec 19, 2017 at 5:42 PM, Scott Driemeier-Showers <

user-8d275c0639b6@xymon.invalid> wrote: