this intrigued me as well as your setup should’ve worked, even if the dhcpoffer 
coming back through the backup vrrp member.  actually, vrrp doesn’t even come 
into play during initial dhcp.  the client broadcasts the dhcpdiscover msg and 
either vrrp member can relay that msg to the dhcp server as both switches 
participate in the same user vlan.  the unicast back from the dhcp server would 
have a dip of either interface ip, not the vrrp ip.

seems something else was broken.  what did you do to get it working?

mike

From: bounce-29047156-33657...@listserv.unc.edu 
[mailto:bounce-29047156-33657...@listserv.unc.edu] On Behalf Of Stephen Wilson
Sent: Wednesday, April 20, 2011 10:50 AM
To: Enterasys Customer Mailing List
Subject: RE: [enterasys] General VRRP question

 

Walt,

I wish I could give you a definitive answer, but we never determined exactly 
why it wouldn’t work.  We spent around 4 hours with an Enterasys engineer 
troubleshooting the issue but never nailed down the exact cause.  My best guess 
is this: 

VRRP only answers for the virtual address, it originates any traffic as the 
real address on the interface.  So in the case of a DHCP relay, if .1 is your 
VRRP address and .2 and .3 are your real interface addresses, the Relay IP 
Address in the DHCP packets will be either .2 or .3, depending on which router 
is master at that time.  However, when the DHCP server sends traffic to the 
client it could go to either the .2 or .3 router, depending on which of the 
equal cost routes was chosen by the upstream router.  If the receiving router 
is not the router that originally relayed the packet, the Relay IP address 
field doesn’t match any interface address, and is dropped.  

Again, I never proved  this theory but it is the best explanation I’ve come up 
with.  If my assumption is incorrect hopefully someone on the list will correct 
me.

Stephen Wilson

Network Manager

WCU Networking and Telecommunications

828-227-3215

 

 

From: Walter Witkowski [mailto:wwitk...@dccc.edu] 
Sent: Tuesday, April 19, 2011 4:01 PM
To: Enterasys Customer Mailing List
Subject: RE: [enterasys] General VRRP question

 

Stephen,

 

Thanks for the response.  I not sure I understand why you're situation would 
not work.  I would think that it didn't matter which path was used back to the 
user with the DHCP response as long as their gateway was programmed with the 
VRRP address that was being used.  Can you expand  further as I am working in 
an educational environment with student and adminstrative subnets and DHCP 
scopes for each.

 

thanks in advance 

walt

 

>>> Stephen Wilson <swil...@email.wcu.edu> 4/19/2011 3:46 PM >>>

Walt,

I haven’t had the opportunity to test but based off past experience your first 
assumption is correct.  Any traffic that originates within the subnet will exit 
through the Master VRRP router.  When traffic originates from outside the 
subnet it could enter the subnet using either router, but all responses to that 
traffic will exit using the Master router.  You could have asymmetric routing 
of traffic occur if you have equal cost paths from outside of the subnet and 
don’t do anything to make traffic prefer the path through the master.  We had a 
situation similar to this when we first implemented VRRP…  A client would send 
a DHCP request to a remote subnet, but the response would come back through the 
other VRRP router, which resulted in DHCP not working for those clients.  Hope 
this helps.

Stephen Wilson

Network Manager

WCU Networking and Telecommunications

828-227-3215

 

From: Walter Witkowski [mailto:wwitk...@dccc.edu] 
Sent: Tuesday, April 19, 2011 3:38 PM
To: Enterasys Customer Mailing List
Subject: [enterasys] General VRRP question

 

Hi all!  Just need some clarification here.  Here's an example of a subnet 
connected to 2 Cores.

Core 1 Vlan 717 interface  x.x.x.3

Core 2 Vlan 717 interface x.x.x.2

Vrrp address x.x.x.1 (GW address for users)

 

1)  Is the vrrp address only used when a subnet user originates a connection to 
another network?  Which means all communications originated from this subnet 
flows through the core that holds the VRRP address?  I believe this is correct. 

2)  When configured with VRRP, a connection can originate and enter the subnet 
from either core.  Will the response always exit via the core that was the 
origination point?

 

thanks in advance

ww

 

 

Walt Witkowski Primary Network Specialist - Sungard Higher Education
Office of Information Technology - Delaware County Community College

610-359-5017(phone) / 610-359-4123 (fax)/856-217-4430(cell)

wwitkow...@dccc.edu

*       --To unsubscribe from enterasys, send email to lists...@unc.edu with 
the body: unsubscribe enterasys swil...@email.wcu.edu 




*       --To unsubscribe from enterasys, send email to lists...@unc.edu with 
the body: unsubscribe enterasys wwitk...@dccc.edu 

*       --To unsubscribe from enterasys, send email to lists...@unc.edu with 
the body: unsubscribe enterasys swil...@email.wcu.edu 




*       --To unsubscribe from enterasys, send email to lists...@unc.edu with 
the body: unsubscribe enterasys Michael.D'estie...@dhs.gov 


---
To unsubscribe from enterasys, send email to lists...@unc.edu with the body: 
unsubscribe enterasys arch...@mail-archive.com

Reply via email to