[Info-vax] DNFS1ACP using 100% of CPU

[Sum1]

Hi Paul

OSX 10.6.8 patched, last rebooted 106 days ago.  No recent updates 
requiring a reboot :)

At time of VMS 100% CPU, OSX activity monitor remains the same except 
for disk I/O which drops (no longer sending NFS data).  Monitoring nfsd 
shows no change in process or memory usage.

OSX has 4Gb memory, 256Mb free which doesn't change when VMS has a 
problem.  No NFS or other messages in the console log for 5 minutes 
each side of the events.

On the VMS side, when DNFS1ACP hits 100% CPU:

- MONITOR MODE shows Kernel=20, User=80; normally it is Idle=93, 
User=5, Kernel=1, Exec=1
- Buffered I/O rate shows 1200, usually around 200-300
- Modified List fault rate exceeds 48000, all soft faults
- MONITOR PAGE shows Global Valid fault rate is 12, usually 0; Demand 
Zero fault rate is 15, usually 0;  all other fault rates remain the same
- MONITOR DISK shows NFSdisk I/O rate 0, usually 28; SystemDisk rate 
doesn't change
- MONITOR LOCK shows locks drop to 0

SHOW PROC/CONT/ID=xx (the DNFS1ACP process) shows the process operating 
normally, priority 8, and then get into a high page fault rate when the 
CPU hits 100% CPU.  SP stays in roughly the same address.

Running AUTOGEN before, during or after the events (which can last up 
to 2 minutes and happen every 15 minutes or so) provides nothing useful 
as it doesn't recommend any changes to parameters.

So, something in the process has "spun out", causing massive soft 
faults, and the NFS server on OSX isn't showing any changes when the 
VMS system has its problems.  No messages in OPERATOR.LOG and I can't 
see that there is any other TCP/IP or NFS-client logging available.

Cheers