The ARP packet rate(89pps) exceeded the rate limit (50pps) on interface...

[Apachez-]

That would depending on how far away this topology change occurs. If this particular device is part of this change and suddently hosts are seen through gi 0/2 instead of previous gi 0/1 then of course you will get a mac/arp "flap".

There is also a setting regarding if you want to log mac-flapping or not:

mac-flapping notification enable

[ciobis]

The access switches only keep ARP entries of the other switches' management IPs; I dont understand why, when adding a new switch to the topology, the whole ARP table needs to be refreshed; and why not only add the entry for the new switch.

Also, I still cannot explain how do ARP exceeds the rate-linit on a BAGG that is down.

This is very annoying.

[ciobis]

Something interesting..... we used to get the ARP packet rate exceeds logs from  three 10G switches in the network; recently, we have upgraded the firmware on one of them (the one that was reporting the most) to Comware Software, Version 5.20.105, Release 1808P27. After the upgrade, there are no more ARP exceeds reported from this switch (HP A5800-24G-SFP), but still getting them from the other 2, different platform (HP A5820AF-24XG)  which run on Comware Software, Version 5.20.105, Feature 1805P02-US. 

I dont know exactly what changed in the new software to stop these; we did not make changes to the configuration.

[Apachez-]

Read the release notes and you shall see...

[3comold]

ARP rate limit is not needed if the switch does not serve as  the access device for the users. By default its packet rate limit is 100 pps.

50 pps if not configured by the administrator, then it is an error un the software that the later upgrade has reverted to 100 pps. Therefore it is not anymore 50 pps and 89 pps is not a pic.

A server and/or a gateway can send a huge number of ARP packets. This attribute - and the ARP gateway protection - is usefull only at the access of the internetwork block.

[Kjartan]

I came across this problem couple of weeks ago, my loggs were flooded with ARP packet rate messages and also had a high CPU load. Turned out to be due to Tc flooding and after going for RSTP in instead of PVST the problem went away. I now have the occasional message. This was in a mixed Cisco, HP enviroment.

Kjartan

[DavidYerger]

I saw ARP packet rate exceeding the default rate on all the link agg'ed interfaces as well as TCs and duplicate IP addresses logged yesterday where the core switch side of a 20G linkagg had one of the members taken out of the link aggregation group to an IDF by a consultant.   

Simply adding 

port link-aggregation group 5

back into the interface configuration got everything OK again.