We are running a 2-node OES2 SP1 Linux cluster (64bit) and configured 5 NSS volumes as a cluster resource. We see that after a while certain resources go comatose because of a failed ncpcon bind command.
This command does not always seem to fail, so it's perfectly possible that we can migrate the resource 1 or 2 times between the nodes but all the following attempts fail and make the resource go comatose. The error gives no further details about a possible reason for this failure.

In some rare cases we even see eDirectory crash/defunct when this occurs, the machine hangs and the only solution then is to reboot the machine.

Does anybody else see this unstable NCP behaviour with eDirectory 8.8 SP4 (on a 64bit machine)?

<wild idea>Wouldn't it be better to completely remove the NCP part from eDirectory and make it a seperate application/process</wild idea>