ICM 7.0: ICMROUTER Connectivity with Duplex partner failed.

Unanswered Question
May 31st, 2007
User Badges:

We are using ICM 7.0 SR1, we are continuously facing an error of Connectivity with Duplex partner has been lost due to failure of the private network , or duplex partner is out of service?. on ICMROUTER.


We have two sites, connected over the WAN.ICMROUTERA and ICMROUTERB, whenever WAN goes down, ICMROUTERB processes restart. Due to which we are facing discrepancy on Web View Reports on the two sites. To resolve the issue we installed the Microsoft patch as stated in the link below.


http://www.cisco.com/en/US/products/sw/custcosw/ps1844/products_field_notice09186a0080715763.shtml



But after installing this patch does not seem that the issue is resolved. We are still facing problems in Web View Reports like Agent Half difference and Agent Missing on both the sites.


Where till my concept and my experience goes, WAN is process that can?t be stable for 365 days continuously, we are on duplex mode, so there must be a mechanism in ICM to control all these problems.


If you guys have any clue, do let me know, how can it be sort out.


Regards,

Humayun..

+923218287609

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
jeff.marshall Fri, 06/01/2007 - 02:44
User Badges:

Focus on the core problem first: Why is the Call Router disconnecting?


Do you have two diverged networks (1) Visible or Public and then (2) a Private network or is this collapsed onto a single network?


How is your QoS scheme designed and is it built to protect the high, medium and low priority traffic streams between the Loggers and Call Routers on the private network?


Describe your design a bit and we'll try to help further.


ICM can manage WAN outages by still running simplex at one side but under no circumstances is "continuously facing an error" good for the duplex design.


/Jeff

humayun.sami Fri, 06/01/2007 - 03:12
User Badges:

Thanks for the reply.

Call Router is disconnecting when the WAN link flaps and comes back after 3-4 sec or shifts to the backup link.


We have a visible link and a private link as well, ICM of both sides are in duplex mode - synchronized.


We have not defined any QoS on the ICM, but we are using QoS on the WAN Router's.


About the design as i told you that we are runinng in a duplex mode.

Call Enters to the Voice gateway from GSM>VOice Gateways>CVP>ICM.


let me know if you need anything else.


Regards,

Humayun.



bullett Fri, 06/01/2007 - 04:35
User Badges:

I'm not actually clear on the problem you have.


I think your wan goes down and side B restarts (as side A has majority of PG's visible)this would be normal as side B Node Manager is attempting to establish the connection by restarting the Router Process.


When the WAN is back up the system will resynch.


You then talk about webview and a reporting discrepancy? what is this difference where are you seeing it. Can you clarify the issue here?


How many WV servers do you have and how are they configured?

jeff.marshall Fri, 06/01/2007 - 04:54
User Badges:

You have to fix that WAN link then. No matter what else you do, each time that the WAN link flaps (when you are experiencing many failures as you describe) you are negatively impacting the system. If you cannot fix the link (say the provider isn't reliable) then I would change the design and not split the system in this manner. ICM can manage this once in a while as it takes time to resync the databases and then sync the HDS dbs for reporting ? I suspect that the quantity of your WAN failures never really allow the ICM system to fully sync, impacting reporting. Your WAN uptime goal should be single-digit failures per month ? or better. The far majority of North American customers experience near-zero of these types of failures over a calendar year, obviously this is dependant upon your provider.


Can you describe the quantity of sites and PGs? I?m assuming that this is a Cisco Call Manager based system (rather than TDM ACD)? We?d be happy to help discuss alternatives to geographically splitting the central controller (Loggers/Call Router pairs) to get your system more stable.


/Jeff


P.S. Is there some reason that you are not QoS marking packets from the application level and instead doing it only on the WAN routers?

humayun.sami Fri, 06/01/2007 - 05:17
User Badges:

Boss we are using ICM 7.0 with SR4, we are trying to get the WAN link stable, but on the other side i am concerned with the ICM, as ICM says that it is in synvhronised mode, if one goes down the other will complete the role and when the processes are back they will get sych with all the information they contain ICM should be having such an mechanism that it should get all back to normal.


We have 2 sites, we have 3 Pg's on each side(total 6). Why are you splitting these two sites, these are already splitted geographically..


e.g.

SITE A:

ICMROUTER A

LOGGER A

HDS A


SITEB:

ICMROUTER B

LOGGER B

HDS B


No there is no reason, the cisco advance services team did all this, and i guess i have read somewhere that if you have more than 50 Pg's, than we should be implementing QoS on Application level.



Regards.

Humayun.


Actions

This Discussion