03-30-2016 11:42 AM
I am just asking for a good explaination to our users. I understand that ARP/MAC table lookup can be complicated.
Sometimes an end host record is showing it is connecting to an upstream switch. For example, a Ten-Gig switch port which we know it is not possible for a standalone Dell server. We are sure that:
1. All switches from core to access-layer switch the server is connected to are in NetMRI.
2. We have the networks in as "exclude from management".
3. This server is not VMWare for any virtualization inside a chassis with multiple interface macs.
4. All ARP are collected from Cisco FWSMs.
5. We did see this server was connected to the right access-layer switch in the past.
6. Switch Port Management data collection is configured in 12 hours cycles.
Again, I am not looking for a solution but an explaination (maybe several) to our users why this sometimes happens.
Solved! Go to Solution.
04-05-2016 08:05 AM
The 12 hour cycles may be the issue. The MAC address tables on the switches typically have a 5 minute timeout. If the MAC is not there on the access switch but remains on the upstream switch due to different polling times of the switches, it can cause this to happen.
I believe there are some updates to the heuristics in the latest version that may help with this situation.
04-07-2016 12:08 PM
Also make sure the "Automatic ARP Refresh before Switch Port Polling" in Settings > Collection and Groups. What that does is try to refresh the ARP/MAC tables by looking at the list of IPs that NetMRI has seen on that switch in the past, and sending a single packet to them. This works as long as there is no firewall blocking that packet.