We need some help trying to trace out a strange hang and system issue
in an SBS65 server with SP3 installed. Server is running GW654, Apache, and
BE91. We expect a high server utilization during backup, which we can see.
However, it appears that we are also seeing a utilization spike to 100% at
specific times, namely 7pm and 10pm, with regular frequency. I have sat on
the Profile/Debug during these periods but nothing ever shows. Added to
this mix is a random pegged utilization that of course knocks off any users
ability to do anything, and requires the server to be rebooted. A down
command issued at server shows "Java resources unloading" but usually hangs
right there, forcing a full poweroff and reboot.

The server is a P4 2.4G/865PERL/1G ram motherboard. Two SATA drives were
installed on the hopes we could use the onboard RAID, but NW did not
support the chipset so software mirroring was setup inside NW. This is a
small office, 3 regular users, and no real reason for these slowdowns.
Hyperthreading was turned off. I turned on e-mail notify for the CPU
utilization and failed logins, and received a notice just before full
lockup today that "Failed Logins Per Hour on server S_WALKER is in a
SUSPECT State, Please Check the server." If you try to use the link it
appears to hang, although I suspect that this is simply the server at 100%.
Local commands on the server are also sluggish but do work.

My question is first off what might be causing these peaks, and is there
ANY tool which can log exactly what is happening to a file rather than
screen only that I can activate to hopefully catch this culprit. My feeling
is it is something to do with either the mirroring or some scheduled
routine, but I can't find it.