Extreme VRRP/ARP Issue - default gateway does not respond

[Jasp80]

Hi

Hope someone can help.

We have two Extreme core Blackdiamonds with various flavours of Extreme edge switches. Each core has two mlagg'd connections to edge stack.

Recently the network was reconfigured to put routing on the cores, rather than a FW, and to set up the cores with VRRP. Fairly standard design, one core is master with the other as backup, VIP set as default gateway on the vlans.

Now, since this work has been done an odd issue has appeared. At some point during the day Windows server 2003, Linux and Macs devices can only ping within their own vlan but not outside of it, i.e. another VLAN. Pinging the default gateway (BD1) does NOT respond. However, pinging the second backup BD, not holding the VIP does get a response.

Clearing the arp cache on the devices in question brings the response back. Now, with 2003 it is not a problem. Changed the ARP settings in the registry and messed around a bit and that seems to have sorted it. However, this is not so easy on Linux or Macs.

This is starting to cause a few issues as you can imagine. Has anyone seen this behaviour before and managed to resolve it?

Any help gratefully received.

[Paul_Russo]

Hello Jasp80

Curious why you are using Master/Backup on the VRRP. Have you looked into Master/Master? If you are staying with Master/Backup are are all VLANs on the same Master if they are split on the two switches then we will need to route between them.

Thanks
P

[Drew_C]

Hi Jasp80,
Can you let us know which version of EXOS you're running on your cores? Also, please share the output of the following commands with us (you can sanitize the IPs if you choose):

code:

show switch

code:

show vlan

code:

show vrrp

Are there any log messages associated with the problem? Are you sure there are no duplicate gateway IPs?

Thanks,
-Drew