Cisco 2960X getting ARP Entry for device on different VLAN, breaks communication

I've been battling this issue and can't seem to find a cause.

I have (2) Cisco 2960X devices (out of 40) that are populating the ARP cache with a device on a different VLAN.

The switches are on VLAN42, end device is on VLAN 30.

This ARP entry makes the switches unable to communicate with the end device (Cisco Prime Server)

If I clear the ARP entry on the switch, communication is restored.

About a week later, the ARP entry comes back and communication is broken once again.

If I understand correctly, since the devices are on separate VLANs, the core devices responsible for inter-VLAN routing should handle the ARP table, the switch should be sending packets to the default gateway, and not be populating the ARP table with devices on different L2 networks.

Does anyone have any ideas on this?

I've researched extensively, I've seen mention of IOS bugs but this problem just recently surfaced after moving all Switches from VLAN 40 to 42.

The ARP tables on downstream devices (HSRP pairs) all appears to be in order.

My assumption is that since there is an ARP entry for a non-local device, it's not hitting the gateway for routing and therefore not reaching the intended endpoint.

Any help is appreciated.