CUCM Network Teaming Issue...

[Virstra001]

Hello,

In our environment we have two subscriber servers connected to two different core switches. Every subscriber server is dual NIC and we have performed teaming on them such that one NIC is connected to one core switch and other NIC to other core switch.

Recently we had an activity wherein we wanted to test that if incase one core switch goes down, the traffice flowis hrough the other core switch or not.

Accordingly we switched off the preimary core switch. The CUCM subscriber servers (through nw teaming) got connected to the other core switch. This was expected and it worked too but after some time when we switched on the primary core switch and then switched down the secondary core switch, the two CUCM servers did neither get connected to the primary core switch nor to the secondary switch. Those two servers became unreachable and at last we had to reboot the servers and then they became reachable.

If anyone has faced similar issues please guide us with the solution.CUCM version is 7.1.5.

Regards,

Sargam Patne

[Anas Abueideh]

Hi,

the network cards are configured as failover not teaming. that means one of them are working at one time.

from the command line run command "set network status" to check the status of network failover. if it is not enabled run command "set network failover ena"

HTH

Anas

please rate if it is helpful

[Virstra001]

Hello buddy,

Thanks for the mail.

Just wanted to let you know that we have already configured the failover part. Thats the reason when the primary core switch went down the CUCM servers failovered to the secondary core switch. The problem begaun when the primary core switch was brought up and then the secondary core switch was taken down. It was expected that the CUCM servers should have falledback to the primary core switch. But this did not happen.

Instead both the CUCM servers bacame unreacable.Then we brough the secondary core switch up but still the servers were unreachable. Then we had to reboot the servers and then they became reachable.

Regards,

Sargam Patne

[Anas Abueideh]

Hi,

can you test without rebooting the core switches, try to disconnect the cables and shut down the interfaces from core switch.

It should work.

HTH

Anas

please rate if it is helpful

[Virstra001]

Hi Friend,

Thanks for your reply.

For your information there are two interface eth0 & eth1 both are team with bond 0. where as eth0 connected to RSH01 & eth1 connected to RSH02.

Before the failover activity eth0 interface in active mode. Traffic routed through RSH01.

after shutdown RSH01 eth0 gets disabled & eth1 come up, traffic routed through RSH02.

upto this all are working fine.

After shuddown RSH02 eth1 gets disabled & eth0 come up, but CUCM server not came up. Traffic routed through RSH01 because other system works fine except CUCM.

Just wanted to know exact reason for CUCM server not came up is it due to interface hang, teming not worked properly, link betweeen eth0 to RSH01 got stuck.

We have checked system logs but not able to track the error excatly. Can u help us to know excat cause of this trouble.

Regards,

Sargam Patne

[Anas Abueideh]

Hi,

just to confirm that I understand well, after shutting RSH02 down. RSH01 is working and traffic routed to eth0, but CUCM service is down. that means if you do a ping you will get reply, right ?

Anas

[Virstra001]

Hi,

No!!!!! We are not able to ping the server.

Regards,

Prashant Dhotre

[Aman Soi]

Hi Prashant,

Are u getting some CDP related messages on switch where CUCM is connected?Just have a look : there is a bug related to NIC teaming as below:

CSCsy08872 Bug Details

CUCM CDP reports wrong duplex after reboot with Network Failover enabled

Symptom: Connected switch erroneously reports a duplex mismatch:Mar 19 15:00:54.073 EST: %CDP-4-DUPLEX_MISMATCH: duplex mismatch discovered on FastEthernet0/24 (not half duplex), with cm701.cisco.com bond0 (half duplex).Conditions: Using network failover / NIC teaming on a Cisco Unified Communications Manager 7.x server. Server has been rebooted (or rebooted after upgrading) since enabling network failover.Workaround: Disable CDP on the switch ports the CUCM server is connected to: no cdp enableORDo the following workaround on the CUCM server: This workaround is only effective until the next reboot of the system. It must be applied after every reboot. **Note that this workaround interrupts network connectivity briefly to the server and will likely be service impacting***-Disable network failover: execute "set network failover dis" from the CLI -Restart the "Cisco CDP" service: Go to Tools -> Control Center - Network Services Select the server in question Select the radio button next to the "Cisco CDP" service located in the "System Services" section Click the "Restart" button -Enable network failover: execute "set network failover ena" from the CLI

regds,

aman

[Virstra001]

Hi Aman,

Thanks for your reply.

We will check error logs on switch. Will try workaround as you said.

One more question is it CUCM server hardware bug or something else.

Regards,

Prashant Dhotre

[Virstra001]

Hello Aman,

For your referance I am pasting the interface status of call manager as well as the switchport to which the callmanager is connected:

Switchport:

interface GigabitEthernet5/3

switchport mode access

switchport access vlan 600

speed 100

duplex full

Call Manager eth0:

Ethernet 0

DHCP         : disabled           Status     : up

IP Address   :                    IP Mask    :

Link Detected: yes                Mode       : Auto disabled, Full, 100MB/s

Queue Length : 1000               MTU        : 1500

MAC Address  : 5c:f3:fc:0b:d0:b0

RX stats:

bytes   :   1083386797   packets :     13260392   errors :            0

dropped :            0   overrun :            0   mcast  :         1681

TX stats:

bytes   :   1244344397   packets :      8219458   errors :            0

dropped :            0   carrier :            0   colsns :            0

Please suggest if the duplex setting is ok.....if not then how should it be.

Regards,

Sargam Patne

[Aman Soi]

Hi Sargam,

Can u try making the switch port and call manager NIC port configured to Auto for Duplex and Speed?

regds,

aman