[cisco-voip] Call counters high on intercluster trunk

[Ryan Ratliff]

You can use the 'show status' CLI command to see how much free space  
on the log partition is available.  This can tell you how much you can  
increase the log files and not run out of room (remember to do SDL as  
well as SDI traces).   You can also use trace archiving through RTMT  
to automatically upload the log files to an SFTP server every few  
hours.  That way you don't have to keep tons of data on the server  
itself.

A call leak happens when a call is terminated in such a way that some  
part of ccm doesn't think the call got torn down.  Aside from the  
perfmon counters getting off most of the time a small amount of memory  
is leaked as well so in large numbers call leaks will cause problems.

-Ryan

On Sep 9, 2009, at 11:08 AM, Nate VanMaren wrote:

So the main cluster has grown quite a bit, and we're doing a new 2MB  
log file every 6 seconds on the main subsriber. So 375 log files don't  
last that long.

How big/how many is it safe to go and get 24 hrs?

What defines a call leak?

-Nate
-----Original Message-----
From: Ryan Ratliff [mailto:rratliff at cisco.com]
Sent: Wednesday, September 09, 2009 8:30 AM
To: Nate VanMaren
Cc: cisco-voip at puck.nether.net
Subject: Re: [cisco-voip] Call counters high on intercluster trunk

First make sure your CCM SDI and SDL traces cover 24 hours worth of
data (on both clusters).  Next start monitoring the active calls
counter at a time of day when you know there are no calls active.  If
you see the counter increment from one data point to the next (say
midnight Wednesday to midnight Thursday and there should be no calls
active at this time) then you collect 24 hours of CCM traces from all
servers on both clusters and open a TAC SR to have them analyzed for
call leaks.

-Ryan

On Sep 9, 2009, at 10:17 AM, Nate VanMaren wrote:


We have two UCM Clusters that are currently at 7.1.2.  The main one
has most of the phones and the other has about 80 phones.  There are
route patterns to the individual extensions that are on the 2nd
cluster because the extensions overlap.  There is also a 2XXXX route
pattern and PSTN access route patterns on the 2nd cluster that point
to the main cluster.

The strange behavior is the longer the clusters are up the higher and
higher the active call counters go.  After a few weeks we get about
2000 active calls both clusters, and if we reboot a node on either
cluster the calls in-between clear on both sides.

I cannot find any loops in the routing, the CSSs of the ICT on the 2nd
server don't allow access back to the main cluster.

Anyone have some ideas?

Thanks
-Nate


NOTICE: This email message is for the sole use of the intended
recipient(s) and may contain confidential and privileged information.
Any unauthorized review, use, disclosure or distribution is
prohibited. If you are not the intended recipient, please contact the
sender by reply email and destroy all copies of the original message.

_______________________________________________
cisco-voip mailing list
cisco-voip at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-voip


NOTICE: This email message is for the sole use of the intended  
recipient(s) and may contain confidential and privileged information.  
Any unauthorized review, use, disclosure or distribution is  
prohibited. If you are not the intended recipient, please contact the  
sender by reply email and destroy all copies of the original message.