VMXNET3 nic loses the ability to update ARP table after N hours
Posted
by
Peter
on Server Fault
See other posts from Server Fault
or by Peter
Published on 2014-04-09T19:27:36Z
Indexed on
2014/06/06
9:28 UTC
Read the original article
Hit count: 533
I have a fedora 18 VM that stops updating the arp table on eth1 after running for a number of hours to days. There are other VMs on the same hypervisor that can access all of the same networks without issue. A tcpdump of the offending NIC shows only ARP broadcasts but no responses. None of the other VMs on the vDS see the ARP broadcasts from the offending NIC.
The only way I can currently solve the problem is to reboot the VM and then everything works for a while. I've tried changing the port on the vDS and even flipping the network configurations after I lose eth1's ARP table, but the ARP problem follows eth1 but I can access the machines that were originally on eth1.
If I statically add the arp entries for machines on the same subnet I have no problems with connectivity. The Hypervisor is an HP BL49X series with flex-10 network modules. Has anyone seen anything like this before?
© Server Fault or respective owner