If you are seeing the neighbor table overflow message in the dmesg log it means that the ARP cache of your server has run out of space. This can happen if you host multiple cloud servers and or massively scale Kubernetes, Proxmox or other virtualization solutions on one physical server, as every VM / Pod / Container potentially has its own IP address which consumes ARP memory.
This can be easily solved by adding the following lines to /etc/sysctl.conf
:
net.ipv4.neigh.default.gc_thresh1 = 80000
net.ipv4.neigh.default.gc_thresh2 = 90000
net.ipv4.neigh.default.gc_thresh3 = 100000
So if you are running many VMs or want to use your servers for HPC you need to tune those parameters also.