Need to clear arp and initiate ping to reach host

[dbuckley77]

Spoiler

We have an odd situation on our network.  We have several hosts that are on the far end of radio links that we cannot reach unless we clear the arp on our core switch and then initiate a continuous ping to.  If the ping stops we cannot reach the host again until we clear the core arp and re-initiate the ping.

Any idea what's going on and what we can do to resolve?

[MHM Cisco World]

can you share config of SW 

MHM

[M02@rt37]

Hello @dbuckley77 

This issue is quite specific and indicative of potential ARP-related problems in a network with radio links. This situation can occur in networks where there are intermittent or unstable links, such as those involving wireless or radio-based connections. 

Ensure consistent ARP cache timeout settings across all devices, monitor and adjuste the ARP cache size if necessary. Implement static ARP entries for critical hosts to maintain connectivity and optimize the radio links to ensure stable connections. Additionally, check for VLAN and subnet configuration issues, enable logging to monitor ARP anomalies, and if feasible, establish redundant paths to the hosts to improve network reliability.

Also, do you check if there is a bug (fixed I hope) on your Platform/Firmware version ?

[chrihussey]

If it is specific to the radio network, perhaps that is where the problem lies. If it is a point to multipoint network, perhaps the main or hub radio is losing the mac **bleep** of the remotes after periods of quiet and the switch is sending the traffic properly but it's dying at the radios. It is only when the switch initiates a broadcast arp after clearing the table will the far end devices respond.  With the continuous ping, the radios maintain the mac **bleep** information and that's why it stays up. Perhaps a radio setting will fix it? Just a guess.

[Georg Pauwen]

Hello,

setting up IP SLAs to have the switch continuously ping these hosts...could be a solution ?