David Young wrote: > Do people think that it works well enough to replace carp(4)? By how much a packet can be delayed without causing a failover? Is it 1/256 of a second? Can the kernel guarantee that a process can process a packet within that limit regardless of how busy the box is? Alex