[cisco-voip] InterCluster Disconnects
Wes Sisk
wsisk at cisco.com
Thu Nov 19 10:50:03 EST 2009
H.245 interruptions are usually caused by one of several things:
1. a race condition inside ccm.exe processing of the required signals
2. firewalls attempting to inspect the h.245 session or tearing down
perceived inactive sessions. This defect:
CSCsq17141 CUCM - Allow TCP KeepAlives for H.323 should enable H.245
TCP KAs also
helps counter that issue from inside CM. make sure you have the fix and
then enable the service parameter. Aside from that if there is any
firewall involved it needs to not teardown idle sessions at least for
h.245. If firewall is choking on parsing or inspection of h.245 that is
a separate issue to be addressed on the firewall itself.
3. network congestion or dropped packets. Do you notice any delay in
call setup or only during transfers? CM traces will help to isolate
exactly what is happening to the h.245 session. Information from this
defect should make it amply clear:
CSCsm26355 need clear messages for h245 tcp session failure
View the bugs in Cisco BugToolkit to get complete details.
/Wes
On Thursday, November 19, 2009 10:29:35 AM, Jim Reed
<jreed at swiftnews.com> wrote:
> *Wes,
> Thanks for that little tidbit. Which, of course, leads to the logical
> question --- Any ideas on what is causing that to be interrupted?
> Latency across the MPLS? A misconfiguration in Call Manager or IPCC?
> Etc. Etc.
>
> Thank You...
> **--
> Jim Reed
> Technology Wrangler
> Swift Communications, Inc.
> 970-683-5646 (Direct)
> 775-772-7666 (Cell)
>
> /"Not only is it not right.
> It's not even wrong."
> / The Pauli Proverb
> Wolfgang Pauli
> *
>
> On 11/19/09 7:55 AM, "Wes Sisk" <wsisk at cisco.com> wrote:
>
> hold/unhold triggers h.245 communication over the ICT to redirect
> the media. If this is failing in your script it will likely fail
> for users performing hold, park, conference, and transfer as well.
> You may have worked around it in the script but the underlying
> problem is still present. Most likely h.245 TCP communication
> between the clusters is being interrupted.
>
> /Wes
>
> On Wednesday, November 18, 2009 7:17:17 PM, Jim Reed
> <_jreed at swiftnews.com_> <_mailto:jreed at swiftnews.com_> wrote:
>
> Re: InterCluster Disconnects *Just as an FYI in case this
> issue should affect someone else or they have seen it
> previously, in an exchange between Anthony Holloway and myself
> on this issue, he pointed me towards turning on the ENG trace
> in IPCC. This lead to the suspicion that it was the Call Hold
> or Call Unhold steps that were possibly causing the issue for
> the calls across the intercluster trunk. Instead of placing
> the calls in queue on hold, running a thirty (30) second delay
> during which hold music played, then taking the calls off hold
> and playing a prompt that allowed the caller to continue to
> hold or press 1 to leave a message, I replaced the Call Hold
> step with thirty (30) seconds of music via a prompt set to be
> interruptible and removed the Call Unhold step completely.
> Unfortunately, it was near the end of the day and I could not
> run as many tests as I would have liked. After hours calls go
> directly to voicemail if the caller doesn't know the extension
> of the person they're trying to contact. I will test in
> earnest tomorrow morning and if I can get forty (40) to fifty
> (50) consecutive calls without a failure, I think it will
> definitely point a finger at something to do with placing a
> call on hold that goes across an intercluster trunk. But,
> playing the devils advocate, I won't say that's the problem
> right now until I get those forty (40) or fifty (50)
> consecutive, successful calls. If anyone has any thoughts
> along this line, would greatly appreciate hearing them.
>
> Thank You...
> **--
> Jim Reed
> Technology Wrangler
> Swift Communications, Inc.
> 970-683-5646 (Direct)
> 775-772-7666 (Cell)
>
> /"Not only is it not right.
> It's not even wrong."
> / The Pauli Proverb
> Wolfgang Pauli
>
> *
> On 11/17/09 6:10 PM, "Alex Hannah" <_alex.hannah at gmail.com_>
> <_mailto:alex.hannah at gmail.com_> wrote:
>
>
>
> Gather and post the MIVRlogs located at C:\prog
> files\wfavvid\logs\MIVR on the cluster you are having
> problems with. MAKE certain you have a time date and pref
> an ext that called so myself or others can parse the logs.
> Out of curiosity where are your transcoders located?
> What codec goes across the trunk?
>
> Sent from my iPhone
>
> On Nov 17, 2009, at 7:19 PM, Jim Reed
> <_jreed at swiftnews.com_> wrote:
>
>
>
> *I will do my best to explain this as clearly as
> possible. Please advise if additional information is
> needed.
>
> We have two (2) separate VoIP systems that communicate
> via intercluster trunks across an MPLS WAN. For
> purposes of this discussion, we'll call them System
> GPC and System CMNM. System GPC is 4.5-Meg MPLS and
> system CMNM is 9-Meg MPLS.
>
> When the contact center number is dialed on System
> GPC, the calls are forwarded to an extension (aka
> route point) on System CMNM that puts the caller into
> the contact center script.
>
> The caller hears the initial recording with no
> problem. That recording instructs the caller to press
> 2 for option A; press 3 for option B; press 4 for
> option C; etc. etc.
>
> On somewhat frequent occasions, when the caller makes
> their selection, the call is disconnected.
>
> This does not happen on calls that are placed to
> numbers associated directly with System CMNM.
>
> System CMNM has a toll free number associated with the
> contact center. When I forward the calls from System
> GPC to that toll free number instead of the route
> point on System CMNM, there are no disconnects.
>
> I have reset the intercluster trunks between the two
> systems. When I look at the call statistics in the
> router(s), they show the code for normal call clearing.
>
> Anyone got any thoughts on why the selection of the
> CSQ option in the script is causing that disconnect?
>
> I have looked at the T1 controller stats and the
> serial interface stats associated with that PRI
> controller and there are no errors. System GPC has no
> calling issues locally for calls that come in on those
> same PRI channels.
>
> Thank You...
> **--
> Jim Reed
> Technology Wrangler
> Swift Communications, Inc.
> 970-683-5646 (Direct)
> 775-772-7666 (Cell)
>
> /"Not only is it not right.
> It's not even wrong."
> / The Pauli Proverb
> Wolfgang Pauli
>
> *
>
>
>
>
>
>
>
>
>
>
> _______________________________________________
> cisco-voip mailing list
> _cisco-voip at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-voip
> _
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20091119/babd2087/attachment.html>
More information about the cisco-voip
mailing list