We are replicating from one set of esx 4.1 hosts to another set of esx 4.1 hosts. We never had any problems with this until we upgraded to 4.1. Now about once a week one of the hosts will go down and need to be rebooted.

vmware tech support say the don't much in the logs except errors communicating with the LUNs.
What I think might have been the problem is that we were keeping the target LUNs in the vmware storage group and the esx host could see them but while replication is running it couldn't talk to them, recoverpoint had control. I think this might have caused the host to go into panic mode.
I have removed the target LUNs from the storage group and will now only add them when failing over.