Xymon Mailing List Archive search

Setting up second network tester

list Eric Jacobs
Thu, 31 Jan 2008 21:28:24 -0500
Message-Id: <fnu04k$fbn$user-c17d2b339195@xymon.invalid>

"Henrik Stoerner" <user-ce4a2c883f75@xymon.invalid> wrote in message news:user-1200914237e6@xymon.invalid...
On Thu, Jan 31, 2008 at 03:29:43PM -0500, Eric Jacobs wrote:
My Hobbit server does the display, network testing and alerting. Using the
default 5 minute interval for network tests. Some have complained that
connectivity tests to certain routers should happen more frequently. But, as
far as I can see, network testing interval is an all or nothing variable. So
I'm thinking of setting up second network tester and using NET variable in
bb-hosts to have the particular routers in question tested by second network
tester, running its tests more frequently.

Preamble done - now, how do I go about setting up second network tester?
Configuring another server and disabling all unnecessary tasks in
hobbitlaunch.cfg? If so, which will I need for the tester to work properly?
You don't have to setup a different Hobbit server, if your current
server has enough capacity to run the network tests. E.g. use this
script "netchecks.sh":

 #!/bin/sh

 BBLOCATION=$1 bbtest-net --report=bbnet-$1 --ping --checkresponse
 exit 0

Then in your hobbitlaunch.cfg you have two (or more) entries:

 [net-std]
       ENVFILE /usr/lib/hobbit/server/etc/hobbitserver.cfg
CMD netchecks.sh standard
LOGFILE $BBSERVERLOGS/bb-network.log
INTERVAL 5m

 [net-frequent]
       ENVFILE /usr/lib/hobbit/server/etc/hobbitserver.cfg
CMD netchecks.sh frequent
LOGFILE $BBSERVERLOGS/bb-network.log
INTERVAL 1m

This runs two sets of network tests - one with 5 minute intervals,
the other with 1 minute intervals. Then in bb-hosts, flag each host with
"NET:standard" or "NET:frequent".


Regards,
Henrik

Thanks -

Looks like an excellent solution for now. But I would still like to learn how to set up a second network tester. Due to recent merger and other situations, it looks like we will be adding a large number of machines to be monitored and I worry about capacity

eric