cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
584
Views
0
Helpful
1
Replies

Monitoring ISDN - event manager applet

1977bjorn
Level 1
Level 1

Hi all!

 

I have successfully set up some event manager applets on Cisco routers for triggering actions based on the local log in the router, for instance cpu or routing issues (see bottom). I would like to set up some monitoring for isdn as well. We have a bunch of gateways and they are all monitored using snmp-walk and presented in cacti. This is the bottom line. Now to the problem.

There is a inherent problem with monitoring isdn with snmp. Some gw:s run MGCP and are backhauled to cucm (and some are not). The cucm will fill up the gw before moving on to another one and so on. The problem is that under high load and short communication (a call could be as short as 20 seconds as we have equipment dialing in as well as people) - the statistics "lie" in some extent. For instance - there can be more calls than what is possible as the polling period is too long. So for instance - a gw with 60 channels can report more than 60 channels as peak usage / per minute - which is actually correct. But this is hard to understand for some people (those who read the final BI reports) - and I can agree to this as well. What is also a problem is that the load can peak during short periods - and this really does´nt show in a graph. The load has to be way too high for an extended period of time for it to "flatline" in a graph. I think that this is not really a problem with either monitoring software or the setup but rather the function of polling. An example:

- Hey there, could you take out some statistics on the isdn usage we have on site x ?

Well, if you look at the graph from a 24/7 perspective - it does not really seem to be that utilized. What I would rather have (or at least I think I do) is a local trigger in the router like event manager applet that can trigger when all the lines fill up. One site can have many gw:s so that one is filled up might not be a problem, but at least we can consolidate the fillups and have exact statistics on when we went out of channels. I really like the idea that the monitoring processing is done locally in the router and that it just ships it of with use of some script or so.

To shrink the polling perios is not a feasable solution as it will introduce too much traffic and there is also some common bugs related to "too short polling periods" that will report wrong values

There is also another perspective to this problem. We can alway aquire more channels - but in some cases we will always be out of lines (like exceptional call volumes due to things we cannot control - like telco problems - force majure and so on). We are implementing an emergency prompt system that teamleaders and so on can manage themselves to cut the inbound load in extreme cases.

 

So, any ideas?

 

I would rather not have a debug running all the time in the gateway to be able to trigger this.

I had a look at the "snmp-server enable traps ds0-busyout" but I´m not sure that this will help me.

 

regards

Björn

 

heres an example of a current event app which works really well:

event manager applet cputrigger
event syslog pattern "SYS-1-CPURISINGTHRESHOLD"
action 0 cli command "enable"
action 1 cli command "sh proc cpu sorted | append flash:cpuoutput.txt"
action 2 cli command "sh voice call summ | append flash:cpuoutput.txt"

So, do you know of any good trigger for ISDN ?

and by the way - we use 2800 and 2900 routers mainly for voice gw:s equipped with multiple PRI:s

 

1 Reply 1

1977bjorn
Level 1
Level 1

Bump!

 

Really - no one?

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco