my company manages and supports a number of switches (100's) for another UK wide company. We have had a number of issues in the past year where the NOC will lose management of the device via the management VLAN. The only way to get this back is to go on to the device and issue the clear arp-cache command. I have been asked to explain both a: why this happens and b: why clearing the arp cache solves it.
Looking for any help possible on this question.
There is not much information here about the network so we do not have much to work with. Let me suggest an approach that might help you to find the cause of the problem. The next time that it happens have the person who is dispatched to solve the problem do a show arp and record the results. Then clear arp. Then do another show arp and record the results. It is very likely that the MAC address for the management address in the table the first time is not the MAC address of the switch. Then look in the switch mac address table for this MAC of the other device. Once you have identified the switch port where it is connected you should be able to determine what is causing this device to send an arp response claiming the IP that is the management address of the switch.
Hi Rick, thanks for the reply. Definetly something I will try.
The network is linked via mpls and on various subnets per site eg site 1 10.44.10.x site 2 10.44.20.x and so on. Sites range in size from a couple of switches to a couple of dozen. Different types of switches and different levels of ios. So no common denominator there. Issue arrises on diferent sites as well with random switches. Doesn't happen often, maybe about 6-7 times in past year.