Rogger Failover

krishnamohan.kp · ‎09-30-2009

Hi,

We do have an IPCC-7.5.5 with CVP 7.0.2 set up with Side-A and Side-B geographically separated with IP WAN.The agents are logging to CAD.

Side-A- Rogger-A, PG1A, PG2A (PG's locally failover) and CVP.

Side-B- Rogger-B, PG1B, PG2B(PG's locally failover) and CVP.

When the WAN link is fluctuating the agents of Side-B are getting logged out from CAD, but Side-A agents are unaffected.

Help Pls.

Thanks

Krishna Mohan

Side

geoff · ‎10-01-2009

I don't understand your notation - regarding "PG's locally failover". Are you saying you have PG2A and PG2B at one site for the CVPs there, and PG3A and PG3B at the other site for the CVPs there? That would be good.

Your Roggers and Agent PGs are split across the WAN. Are the private and public link are over different physical links?

Is the WAN fluctuation on one of these links, or both?

I have seen similar behaviour you mention regarding logout during testing when we brought the link down, and it probably does work that way, although I have never worked through a full analysis of the parameters.

What can you do to get stability of the links? Splitting sides geographically when the service is unreliable brings more problems than it tries to solve (continuity when one physical location is clobbered.)

Regards,

Geoff

krishnamohan.kp · ‎10-01-2009

Yes Geoff. precisely telling each site has its own Call manager cluster and a CVP server each.The Public link and private link are separate and fluctuation occurs on both.

Even when the Cenral controllers loose connectivity during fluctaution, the PG of Side -B is still having connectivity to Rogger -B and disconnects should not happen right.

Thanks

Krishna Mohan

david.macias · ‎10-01-2009

Ok, I'm still confused. You have two UCM clusters? If so, where are the PG pairs located local or remote to each other?

Maybe I'm dense today.

david

Blog

krishnamohan.kp · ‎10-01-2009

Yes David, I have 2 Call Manager Clusters.Each site has a pair of PGs.

SideA- PG1A- CCMPG, PG2A-VRUPG, PG1B-CCMPG,PG2B-VRUPG

SideB- PG3A-CCMPG, PG4A-VRUPG,PG3B-CCMPG,PG4B-VRUPG

Hope it clears.

Thanks

KrishnaMohan

robb.allred · ‎10-01-2009

Ok - one more for clarification, because this may be the crux of the issue:

Side A: PG1A and PG1B for UCM; PG2A and PG2B for VRU; the VRU pg's are connected via cross connect

Side B: PG3A and PG3B for UCM; PG4A and PG4B for VRU

You mention two UCM clusters - do they share PG Servers? So you have a Server with a PG for UCM cluster A, another PG for UCM cluster B, and another PG for the VRU - all on the same server? Correct?

krishnamohan.kp · ‎10-01-2009

Ok I shall explain:-

Side-A-

PGA Server- PIM1A- CCMPIM ( ClusterA), PIM2A-VRUPIM (CVP-A)

PGB Server-PIM1B-CCMPIM(ClusterA), PIM2B-VRUPIM(CVP-A)

The servers are connected back to back with cross over cable

Side B

PGA Server- PIM3A- CCMPIM ( ClusterB), PIM4A-VRUPIM(CVP-B)

PGB Server-PIM3B-CCMPIM(ClusterB), PIM4B-VRUPIM(CVP-B)

The servers are connected back to back with cross over cable

Thanks

Krishna Mohan

jimbrittcisco · ‎10-01-2009

I'd pull the logs from the CAD agents and CTI and work back through the components from there. In your search, you might find error codes you can look up that will point you to the piece(s) that are broken or ill configured.

geoff · ‎10-01-2009

OK, now I have a better picture. At each site you have a CUCM cluster, a CVP (1 or 2) and a pair of PGs with a crossover cable for the private link.

In addition, at each site is one of the Roggers. These are connected with a public and a private WAN. Are these disparate?

Fluctuations in the WAN affect the Roggers ONLY. The Call Router will go into simplex mode and back to duplex mode as the link comes and goes. I am totally surprised that this affects the peripherals.

Regards,

Geoff

krishnamohan.kp · ‎10-01-2009

Yes Geoff , Public link are private link are separate. Public is 6mb and private is 2mb dedicated

Thanks

KrishnaMohan

robb.allred · ‎10-01-2009

So, with two different UCM clusters - I'm assuming your side B agents are registered with Side B cluster and Side A agents with side A cluster?

When the WAN is fluctuating, which processes on your UCM PG(s) are failing? And is it just the public link that's fluctuating?

Robb

krishnamohan.kp · ‎10-07-2009

Hi Geoff,

Have one doubt.For the Central controllers to function independently when the Wan goes down , the Total PG's connected to it should be greater than n/2 right? Right now we have 2 PG's connected to each Rogger.

Rogger A- PG's 1 and 2

Rogger B- Pg's 3 and 4

If we add 5th PG should at both sides should it resolve the problem?

Please correct me if I am wrong.

Thanks

Krishna Mohan

csshenoy80 · ‎10-07-2009

how is your ccmpg configured? On the PG installaltion, how are the Callrouters selected for the Preference for connectivity. Also, for the sides, is the connectivity for call router selected as local or over wan?