I'm having a weird problem with a CentOS 6.6 machine. It is a server running KVM and a few guest VMs using a bridge interface. Basically, my issue is this: The VMs on this host have no network connectivity issues at all, I can ssh to them and ping them just fine. The host, however, can pretty much only communicate with devices in its own subnet (including the VMs running on it) and a few switches...
If I run a tcpdump on it and try pinging from a device that can't communicate with it, I can see the ICMP echo requests coming in, but no echo reply is being sent back out. I've checked everything I can think of: restart networking, restart server, iptables is totally empty, selinux turned off, icmp_echo_ignore_all = 0, all routes are correct and identical to other servers in the same subnet that have no issues.
I'm pretty much out of ideas, any help would be greatly appreciated!
Thanks,
It seems like this could be a routing issue. Since the VMs inside KVM can route independently of the host, the issue is not going to be your physical network connectivity, but rather your logical network connectivity. Check your default gateway and static routes, make sure those are configured properly. If your host cannot find a route back to the devices pinging it, it cannot respond to icmp (ping) requests.