[cisco-voip] [cisco-VoIP] Odd CUCM IP Phone Registration behavior during Migration

Sat Jul 27 15:08:47 EDT 2013

Have you tried resetting one of the problematic 7940 pones to factory
default and let it try to register?

From: cisco-voip [mailto:cisco-voip-bounces at puck.nether.net] On Behalf Of
Dana Tong
Sent: Saturday, July 27, 2013 10:37 AM
To: cisco-voip at puck.nether.net (cisco-voip at puck.nether.net)
Subject: [cisco-voip] Odd CUCM IP Phone Registration behaviour during
Migration

Good morning all,

Weird one tonight.

Customer has a CUCM version 7.1(5) with approx. 900 handsets of 7912's,
7940's etc.

So I built up a new CUCM version 9.1(1) on new UCS hardware (vmware) and
everything looks fine. I cleaned up all the data an implemented some of the
intelligent call routing features in newer CUCM etc. Tested with some brand
new 7942's out of the box and all is good.

Tonight I get onsite to perform the migration. I start with the first site.
~110 phones. Update the DHCP option 150 and reset the phones.

Now about 40 of the ~110 phones come across and register to the new CUCM ok.
This leaves some 70 phones reporting configuring IP. The DHCP is on the
router, and the router connected to the LAN via a dot1q trunk. 

The registered phones are all of the 7912's and some 7940's. 

Two of the phones close by are 7940's. One is registered and one is not.
They both have the same firmware. They are both on the same access switch.
The unregistered phone does not have an IP address. However the router's ARP
and DHCP has an IP address in its database. I cleared the ARP, and cleared
the DHCP bindings. Restart the phone. I see the DHCP request in the debugs
and an address get assigned but the phone still doesn't get an IP.

So I manually set an IP address on the phone. It still won't register. I set
a continuous ping to the phone and I get maybe 2% ping response. With some
response times up in the 190ms range. Weird. This is all LAN based. Check
spanning tree and that's ok. The TAC have investigated and they are stumped.
They ask me to log another fault for the switching guys. A PC behind the
phone in the voice VLAN also has perfect pings to the new CUCM. So it seems
phone related.

Anyways so I do some more troubleshooting. Roll back the DHCP option 150 and
reset all the phones. They all re-register with the old cluster no problems.

I configure a new VLAN on the switches and new sub-interface on the router.
Configure a new DHCP scope for the new range and configure this phone to use
that new VLAN. At this point I haven't added the new subnet to BGP to this
subnet cannot see the Data Centre VLANs.

The phone comes up in the test VLAN and gets an IP address ok. I can ping
the phone and have no loss.

I configure the BGP routing and now the new subnet is advertised into the
network and I can reach the CUCM. 

Immediately, the phone starts dropping pings and is showing the same
sporadic response and packet loss.

I remove the subnet from BGP and now the phone is 100% responsive again.
What the???

So it seems there is an issue when the phone is trying to register to the
new CUCM. I've just checked the old cluster and noticed that this system is
running 7.1(5) UNRESTRICTED. So the last upgrade which was done for this
customer, it seems someone had installed the UNRESTRICTED version. Is this
affecting my migration? The new cluster is a fresh build of UCM 9.1 straight
on to new UCS hardware. So it's a standard restricted ISO.

I know that once you go UNRESTRICTED you cannot go back. However this was
not an upgrade or backup/ restore. Remember this is a fresh build.

Anyways I'm stumped. Can you guys think of anything??? It seems to be
related the 7940 series phones. We tried another site and all of the 7912's
migrated fine but the 7940's didn't. But then again there were "some" 7940's
in the HQ site which did work. 

Thanks in advance.

Cheers

Dana

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20130727/bc8fd398/attachment.html>