We have a number of NW6.5 sp8 clusters, most of which are fine but we are having problems with logical space which drops over a period of time but seems to reach a critical point at which point the server goes to 100% utilisation on 1 cpu and our only way out is to reboot the server. The node in question is only running DS and the LDAP server but it does carry most of the master replicas for our entire tree. This has occured about three times now, at least one of those times was before sp8. (I am assuming that the three occcurences are linked) It seems that we get an awful lot of memory fragementation very quickly. I have unloaded netshld on this node because it is usually the named culprit when the server is unable to allocate short term cache. I have attaced a segstats.txt