3548 switching its arp table around

So my PowerConnect 3548, is messing with its own ARP table I noticed.

 

I found this out to random times I am unable to access one of our servers or ping a host on the network. Sometimes, pinging a host results in 0/4 success rate.. or 3/4 success rate... RANDOM times throughout the day. Other times its fine. I decided to investigate farther and noticed it has something to do with the 3548's ARP table.

Attached below is my arp table Between when I could NOT ping host 192.168.2.90 and when I COULD ping host 192.168.2.90

(I cut stuff out to save formatting)

------------------------

console# show arp

vlan 102   e24        192.168.2.90    00:00:85:bb:c7:7e   dynamic

--------------------------

console# show arp

vlan 102   e9         192.168.2.90    00:00:85:bb:c7:7e   dynamic

-------------------

 

Notice how show arp shows 192.168.2.90 on TWO different ports? The host is physically on port e9, which is when you can successfully ping it.. but some times.. randomly..it will stop responding, and it shows itself on port e24. I watched this happen within seconds.

Interesting enough, on port e24, theres a consumer grade DLINK switch attached to it with two other hosts on that little switch... those two hosts have static IP address so its not a "two hosts have the same IP" problem... I temporarily removed that DLINK little switch to see if thats whats screwing with the arp table.

 

What could cause the switches arp table to swap ports around like this? This causes brief outage for network hosts. Although not EVERY host has this issue. Just a few.