Random "DHCP server unreachable" on i2002 with BCM200 3.6

[shaun070]

We have a number of BCM200 units (running BCM Communications Manager Release 3.6 Build 2.2c) and when we reboot the BCMs (on a monthly basis after backup) quite often a number of i2002 sets will not succesfully reconnect. The message displayed on the set is "DHCP server unreachable".

The DHCP server in this case is a Windows 2000 server on a separate VLAN; a telco-managed router is responsible for forwarding DHCP packets from the set VLAN to the VLAN on which the DHCP server resides.

The problem has been observed with phones connected to Nortel Baystack 460-24T switches as well as Cisco Catalyst 3560 switches.

Not all of the sets fail to reconnect; the sets that fail to reconnect seem to be a random mix of Phase 1 (0602B59 and 603B59) and Phase 2 (0604D9C) sets of various vintages. We have seen it both on ancient Phase 1 sets and brand-new Phase 2 sets, so it doesn't seem to be common to a particular batch of sets.

We are often able to bring a number of the sets back online by unplugging the sets themselves; when that doesn't work restarting the DHCP service sometimes helps. However, sometimes neither unplugging the set nor restarting the DHCP service is able to bring a set back online.

We don't have the freedom to restart the DHCP forwarder, unfortunately.

Has anyone experienced or heard of similar symptoms?

I look forward to your suggestions.

 

[KnobTwidler]

Hey Shaun ... On the IP Phones there is a server 1 (or BCM1) and a server 2 (or BCM2) option ... After a while the the IP Phones switch over to the server 2 option .. so you should have the same IP Address for the S1 as well as the S2 .. Hope this helps

 

[MagnaRGP]

I don't think you have a set issue and the evidence is that it is random as to which set gets affected. I would tend to believe that it is related to switch or router config.

Do you have spanning tree enabled on the switches? Try giving half of your sets (choose a few that have been affected already) static IP's and see if they survive rebooting the BCM. If they always survive the reboots, then I think you can safely conclude that it's network related.

 

[shaun070]

KnobTwidler: I neglected to mention that the sets are configured via full DHCP, which has been setup as follows on the Windows 2000 Server:

Option Name: 128 Nortel-i2004-A
Vendor: Standard
String Value: Nortel-i2004-A,10.230.82.18:7000,1,10;10.230.82.18:7000,1,10.
Class: None

For the DHCP scope in which the DHCP server itself resides, the following is configured:

Option Name: 191 VLAN-A
Vendor: Standard
String Value: VLAN-A:82.
Class: None

So, thank you for your suggestion, but unfortunately we've already got both S1 and S2 configured.

 

[shaun070]

MagnaRGP: Have you seen issues with STP in this sort of scenario? What is it about STP that would cause any sort of issue in this situation?

None of our workstations experience problems such as this, and they are also configured via DHCP.

I expect that configuring the sets via static addressing would eliminate the problem entirely, but we will be testing this on our next visit to the site, just to be sure.

Thank you for your input -- do you have any more thoughts on this?