We have several sparcs (solaris 2.6) sharing user's maildirs via NFS on a Netapp 760 filer. We have noticed that when we first boot the machines, we get great performance. After the first 24 hours though, the same # of users cause the load on these machines to slow down substantially. By the 2nd day, these machines feel like molasses, and the only way to fix this is a reboot. The bottleneck appears to be NFS related. We run NFS vers 2/UDP between the SPARCS and the NetApp. Has anyone seen this kind of behavior before while running qmail? [It doesn't seem to be related to qmail at all as far as we can tell] Thanks, --curtis
