CVP Server NIC is down automatically in peak time

manivannan.paramasivam · ‎09-01-2010

Hi All,

We have IPCC Setup with IPCC 7.5 and CVP 7.0.

We have two CVP physical servers (co-resident with Call Server, VXML Server, Media server)

And call flows work via H323 protocol (not using SIP).

Load balanced via CSS, Call Routing VIA gatekeeper (ras)

We have noticed that in peak time any one of the CVP server NIC card goes down and come back automatically after few minutes.

Any one knows about any bug in CVP 7.0 or any design issue.

please help us resolve this issue.

with Regards,

Manivannan

geoff · ‎09-01-2010

Two questions.

1. are you using LAN teaming?

2. do you have all the tcp offload, scaling etc turned off? command window - what does "netsh int ip show offload" say?

Is this an HP box? If so, the HP LAN Utility is the best way to check and fix.

Regards,

Geoff

manivannan.paramasivam · ‎09-01-2010

C:\Documents and Settings\Administrator.GRPIPCC>netsh int ip show offload

Offload Options for interface "Local Area Connection" with index: 20002:
TCP Transmit Checksum
IP Transmit Checksum
TCP Receive Checksum
IP Receive Checksum

C:\Documents and Settings\Administrator.GRPIPCC>

We are using MCS7845 Server

geoff · ‎09-01-2010

That command should return nothing - you have TCP offload enabled. You need to disable those things. This is the cause of instability. There are plenty of references on the Internet about disabling TCP Offload.

You can do it on the adapter out of the Windows configuration for the Network.

If it is an HP DL380 in disguise (MCS-7845-H-x) , you can also do it from the HP Network Utility, which is easier to use and you can ensure that the TCP Offload Engine is turned off. Is it an HP box?

If it is, fire up the utility and tell me if LAN Teaming is engaged across the two NICs. Maybe you know - do you have two CAT5 connections?

Regards,

Geoff