I have devices which don’t send gratuitous arp when they plugged into a L3 switch. A problem occurs when one of this devices fails - and is replaced by another one (with another mac -- IP the same). The l3 switch doesn't update his ARP Table with the new mac and so the ping fails. When I clear the arp cache -> the arp table is updated by a new arp - request and the ping works.
The next workaround was to modify the ARP TIMEOUT to 60 sec. So when is swap an failed device - this takes longer then 60 sec - I thought that the arp cache is cleared in the meanwhile for this interface BUT it wasn't
How affect's the arp timeout - timer on the arp cache. I modified it many different values but the arp entry does not disappear??
I believe that you are confusing the arp table and the cam table. You are correct that the associated entries in the cam for an interface are purged if the switch interface goes down. And in my experience the entries in the arp table are purged if the interface on which they are learned goes down. But frequently the device doing the arp is the layer 3 router and not the switch to which the PC or server was connected. So the old PC is removed, the replacement PC is booted up and configured, and the entry in the arp table does not change.
I am surprised to read your post. In my experience setting the arp timeout has been effective. Would it be possible for you to recreate the issue and to post some outputs of events during the test? In particular I would be interested to see the output of debug arp and the output of show ip interface for the interface where the test device is connected (to verify the arp setting).
I found my mistake - I set the arptimeout on the L2 Switchinterface - it should be set on the L3 SVI - now it works fine
thx to all
I am glad that you have resolved your issue and that it is now working correctly. Thank you for posting back to the forum to let us know that you have found the problem and have fixed it. This is an easy mistake to make and I am glad that you found it and fixed it.
We are experiencing an issue with ARP as well, I hope you will be able to help
We have 2 Exchange Mail Servers in vlan 1 and one Exchange mail server is vlan 2
From time to time mail servers in vlan 1 are not able to ping mail server in vlan 2. Only when I clear the arp manually the ping is possible
In the meantime my computer in vlan 1 is able to ping mail server in vlan 2
Moreover, I have set arp timeout on interface vlan 1 for 10 min, but this does not help..
Do you have any thoughts on this?
If we had more information we would be able to give you better answers. Are the two mail servers and your PC in vlan 1 and the mail server in vlan 2 all on the same switch? or on different switches? If they are in different vlans what is doing the layer 3 forwarding between vlans? When you manually clear the arp is this on the layer 3 device or on the mail server? Posting the configuration might help us to understand what is going on and to identify the issue.
Thank you for your answer
We found out that these 2 mail servers in vlan 1 were in a cluster with a VIP address and when this address switched from one server to another it did not notify the gateway about it (gratuitous arp was switched off). We switched it back on and now it should be working fine
Thanks for posting back and letting us know that you have solved the main issue by enabling gratuitous arp on the servers. As far as the arp timeout I wonder when the problem began did you wait at least 10 minutes before you cleared the arp cache?
Yes, it never helped. I removed the command and reapplied it but had no chance to test it. Initially it was set to 10 min and once we had this problem for couple of hours at night before I cleared it manually in the morning