How to make an IP phone failover to LSP after power outage

[rejackson]

I have read through several old posts about this. After the power outage in Amsterdam our WAN connection did not come up and we are still fighting that. Our office has power and but the phones will not connect to the LSP because they cannot connect to the main site first to get the LSP information. Really a silly set up not being able to put the LSP in the MCIPADD parameter or some other DHCP parameter. In my experience power outages are more common than WAN outages. After failing to connect for a while the phones reboot on their own anyway. The posts I read are pretty old. Is there a solution for this yet?

 

[kyle555]

I'm not sure it is an unsolved problem in and of itself.

I would see the phones, sunny day

-Get DHCP from core
-Register to core CM
-Get alternate gatekeeper list from CM - including the LSP

In WAN fail, the phones would happily register to the LSP
In power and WAN fail, I would think there's something you can do to make the phones come up with previous settings, including the AGK to register to the LSP. To be honest, I can't think of what that is off the top of my head, but I figure its gotta be in there somewhere.

 

[dialtonebone]

I'd place the LSP IP address in the MCIPADD field as the second option, with your CM core as the first. CM should also be distributing this information out from the survivable servers list in the NR. Like kyle said, it doesn't sound like a problem. I'd start double checking on the configs.

 

[kyle555]

I was thinking about how this could be a problem in a centralized DHCP configuration.

Power hit + WAN fail = phones needing to come up and get to the LSP without DHCP, and I think reuse time is a/the parameter that is key.

The setup in question having a reuse time of 0 would prevent the phones from ever reusing an IP. Also, I would think that a VLANTEST of 0 might be problematic in this case, as would strictly adhering to DHCP leases and never reusing addresses if DHCP is no longer available - which you would do with "DHCPSTD=1" - but neither of those settings for VLANTEST or DHCPSTD that could muck it up are defaults.

I think with a bit of testing you should be able to make it work the way you want.

## REUSETIME specifies the number of seconds that DHCP will be attempted with a VLAN ID of
##  zero (if the value of L2Q is 1) or with untagged frames (if the value of L2Q is 0 or 2)
##  before reusing the IP address (and associated address information) that it had the last
##  time it successfully registered with a call server, if such an address is available.
##  While reusing an address, DHCP will enter the extended rebinding state described above
##  for DHCPSTD.
##  Valid values are 0 and 20 through 999; the default value is 60.
##  A value of zero means that DHCP will try forever (i.e., no reuse).
##  This parameter is supported by:
##       96x1 H.323 R6.0 and later
##       96x1  SIP  R6.0 and later
##	 B189 H.323 R1.0 and later
##       96x0 H.323 R3.1 and later
##       96x0  SIP  R2.5 and later
## SET REUSETIME 90

 

[randyb19]

What model phones are you using? If it is the 46xx series, you should be putting the LSP IP in the DHCP option. If it is the 96xx series, you should be placing the node name of the LSP on page 3 of the network region under "BACKUP SERVERS".

 

[wpetilli]

We run DHCP from our hubs and in the voice and data 176/242 options we have the registration addresses with the LSP as the last in the list. When that GW loses its heartbeat after 3 mins or so, the 8300 kicks in and the phones register locally in an instant. Our remote office setup have an MPLS link and a backup VPN. I don't run the network, but we def block the voice subnets from communicating over the VPN. We only allow access for mngt, in the event I need to access that remote gw. If there was just a WAN outage and no power the phones have cached memory. when power goes to the phones I have to check if they still recover because DHCP is more than likely accessible over the backup VPN.