> It settled down ... we moved buildings ... everything > worked fine. > > then all hell broke loose our NIS+ decided not to > work properly and we had to reboot the 'temporary' > server (5.11 snv_64a) and now the locking has gone > totally to pot. [...]
Random thoughts: Heard someone say once that NIS+ (which I've never used myself) needs good time sync (like ntp) across all the systems. That can also affect some apps accessing files on an NFS server (like make), but barring an authentication scheme sensitive to such things, I don't know that NFS itself would care much. Naming services problems can can cause all sorts of interesting effects. Check that stuff out thoroughly. If systems had new IPs assigned, make sure they're being consistently used everywhere. And consider (but this may cause active client locks to be lost!) stopping the nfs server functionality, removing the contents of /var/statmon/sm, and restarting; especially if any of the IP -> name links in there no longer reflect the current state of affairs. Before any of that, snoop the traffic between some of the troublesome client/server pairs and make sure it looks like what you'd expect; until the network's good (including firewalls and switch port security configured correctly, if applicable), nothing else will help much. Probably a few other things too... This message posted from opensolaris.org
