Your server is a cluster node of an HACMP cluster using heartbeat over disk as non-IP network. Because of excessive I/O disk heartbeats get lost and you receive a warning via errpt that the cluster might not work as expected because of that. Given that the heartbeat disk devices still exist and that they are in a working condition the following can be said: As long as
- you don't receive that message regularly,
- your IP network hearbeat does'nt fail too
you don't need to worry. (Only if ALL heartbeat networks fail that would be a real problem.)
To avoid this errormessage turning up again you could
- change non-IP network speed from normal to slow
- use more disks for I/O
- move the disk heartbeat device to another disk/LUN.

Last thought: In case you are unsure how to run an HACMP cluster don't touch it but ask a colleague who is aware of HA clusters to assist you.