On Thu, 25 Mar 2010, gregwm wrote: >> I thought about NFS but the problem is there's no failover with that, >> the share goes down and every server is rendered useless, also I >> anticipate a lot more reads than writes and having the data available >> locally is going to be much faster. A minute of two delay is ok. I'm >> going to take a look at DRDB and incron, > > > so true re nfs. please share as your experiences unfold with drbd and > incron. I'm interested in hearing about the other options, but my NFS advice was not to do NFS mounting to a single point. It was to do NFS mounting from every machine to every other machine and do every write to all of them. That would keep everything synced, always. It might be tricky to arrange, though. Mike