DUPLICATE REGISTERATIONS IN A VCS CLUSTER

[Chet Cronin]

We have a two VCS Control cluster configuration here and they are running X8.2.2

Problem is there are a number of duplicate registrations showing up on both servers.

I try to unregister one or both and it doesn't get rid of the duplicate.  I do know in the past to clear it the end point had to be restarted.

Is there anything else to try?   I have thought about putting the servers in maintenance mode until all the registrations are gone and then turn the registration mode off?

Any ideas ???

VR

Chet

[Patrick Sparkman]

Are you trying to unregister the endpoint from the correct VCS peer?  You have to directly to the peer the registration is located to unregister it.

It could be a network issue, if something breaks the link between the VCS and endpoint, and the endpoint sends another registration request before the previous one either timed out or expired, you can sometimes get duplicate registrations.

[Chet Cronin]

Thank you for the update.  Looks like CISCO put a function on the VCS that doesn't work either ...

I select Unregister and the system just immediately registers .. Doesn't make sense but it is what it is with CISCO.

Do you know if the same thing happens with the Polycom Gatekeepers??? Not that familiar with them.

[Martin Koch]

It would be handy to know a bit more about your deployment and also a bit more info about

the registration you see, like which protocol, a screenshot/copy paste of the registration info,

type and software version of these systems, info about the endpoint and vcs configuration, 

dns srv setup, how your network looks like and if there are any issues and so on, ...

Also if you see have any issues calling from or to these systems 

and if you have any suspicions what might be not ok in your network.

As Patrick said, such kind of symptoms can be caused by network issues.

That said, if its SIP and you are using SIP outbound, thats its intention to have registrations

on all peers of a cluster, so just seeing the same endpoint being registered to both is not

necessary wrong.

So without knowing more its hard to say.

[Chet Cronin]

It would be handy to know a bit more about your deployment and also a bit more info about

the registration you see, like which protocol, a screenshot/copy paste of the registration info,

type and software version of these systems, info about the endpoint and vcs configuration, 

dns srv setup, how your network looks like and if there are any issues and so on, ...

------------Unable to provide screenshots as the systems are on a classified network. Protocol=H323, address is the IP : 1719.  Normal information observed.  DNS/SRV records are set up properly. 

Also if you see have any issues calling from or to these systems 

and if you have any suspicions what might be not ok in your network.

----- systems are fully operational with no issues making IP or Alias calls.  No network issues observed.

As Patrick said, such kind of symptoms can be caused by network issues.

----- What kind of network issues is Patrick referring too.  Our network is fiber based HUB spoke.  CISCO Routers/Switches most Switches are GIG ports and is very stable.

---------SIP is not deployed .. Only H.323.  (Good to know about SIP registrations thank you)

---------Duplicate registrations will not negatively impact operations.. thank you.  That was my major concern. 

---------Using clustered VCS Control X8.2.2

99 percent of end points are MXP's and some C & EX series.  Most MXPs run F9.3.1 and C & EX running TC6.0 and TC7.3.

------------------------------------------------------------------------------------------------------------------------

Thank you.

[Martin Koch]

I did not say that the duplicate registrations will not impact operations, I only said

there might be reasons that you see them, but thats most likely only on sip.

You registration history might be interesting to check, also how long a registration lasts.

Do you have any issues with disconnected calls after some time?

I did not try how duplicate h323 registrations over a cluster should behave, on a single vcs there can only be one registration and you have a setting if a second one will be rejected or replace the first one.

Double check that you do not have a firewall, NAT helper, ALG, inspection, or something else L3 and up which tries to mess with the h323 signaling or timers.

Often Enterprise grade firewalls have some "hidden features" regards that which are

enabled by default, so you might not even see it in the base config as long as you do

not disable it.

Besides that on h323 there is more than the srv records, the endpoint gets the alternate gatekeeper from the VCS which are the other peers of the cluster.

It would be interesting to see on the endpoint which vcs it says its registered to and see if that flips

Btw, if you look at the registrations on the vcs web site you will find the registration

for all systems in the cluster, but it is marked with "this system" or the peer number.

You can also filter the registrations so you only see the ones on the system you are

on, but you have to do this manually.

Not that you got confused by that, but that should show all systems, not only some :-)

If its really a hanging h323 registration you need to check your network

[Chet Cronin]

Martin,

Thank you and appreciate the input on what to also look for ... picked up much information on your last comment.  Will check the firewall out as we are operating in an enterprise configuration. I think you are on to something there.

VR

Chet