[f-nsp] Interoperability issue between Windows NLB cluster with Foundry SX800

[Jared Valentine]

Sounds like you need to map a static arp entry to multiple Ethernet ports?
I believe it's possible with the SX800, but you'll have to use a hairpin in
order to get the functionality to work.  These instructions assume the
following:

 

SX800 is routing

Servers use the SX800 as their default gateway

Servers are in VLAN 100

SX800 VLAN100 VE address is 10.0.0.1

 

Here's something to try:

 

1.)    Remove VE100 from VLAN100.

2.)    Add another port to VLAN100 untagged (example 2/1)

3.)    Create new VLAN (101) 

4.)    Add single port, untagged, into VLAN101 (example eth 2/2)

5.)    Create VE101 in VLAN101

6.)    Assign VE101 with 10.0.0.1 

7.)    "Loop" 2/1 to 2/2

8.)    Create the static-arp entry on port 2/2

 

You can tweak these instructions for other scenarios (SX800 not routing,
just switching), but the concept remains the same.  If the SX800 isn't
routing, then you'll create VLAN101, move the physical uplink from VLAN100
to VLAN101, loop VLANs 100/101, add your static-arp entry to the loop port
in VLAN101.  

 

Good luck,

 

-Jared

 

 

 

From: foundry-nsp-bounces at puck.nether.net
[mailto:foundry-nsp-bounces at puck.nether.net] On Behalf Of Sujith
Sent: Wednesday, June 02, 2010 7:27 PM
To: foundry-nsp
Subject: [f-nsp] Interoperability issue between Windows NLB cluster with
Foundry SX800

 

Hello,

 

Thank you for your valuable suggestions and time.

 

The client is having a Windows NLB cluster configured for an application and
is terminated to the Foundry switch.  The Windows cluster is receiving
request constantly on one server and not on other server.

 

We tried configuring static MAC address on both the Ethernet ports as
suggested by Foundry guide for Windows NLB setup and followed some
guidelines from Cisco guide also.  Both the guidance were failing to deliver
the request on both the servers. 

 

The client is running a Hyper-V on two separate physical servers and each
Hyper-V is holding 3 Nos. of virtual machines.  The Windows NLB cluster is
configured only for 1 Virtual machine in each physical server.

 

The servers are terminated in the Foundry SX800.  The windows NLB cluster's
are configured to work in Unicast mode and also tested with Multicast mode
also.  On both the testing, we are not getting the results on both the
servers.  The Windows NLB cluster is working perfectly with Cisco switch
without any issue.

 

The client tested the Windows NLB cluster configured separately on the
physical servers (test bench) which is connected to the Foundry switch and
we get the results as expected.  But when tested with Windows NLB cluster
configured on a Virtual machine, it fails.

 

Do someone have come across this issue and is there any workaround to
resolve the same?

 

Thanks

-Sujith

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20100613/6b51ad92/attachment.html>