I am having endless trouble with a new news server.  All seemed to work
fine while I was testing it, but as soon as I unleashed some live load, it
started panicing every few hours.  Here are the most recent few panic
messages:

Jan 29 04:53:55 newnews savecore: reboot after panic: malloc: lost data
Jan 29 06:04:18 newnews savecore: reboot after panic: blkfree: freeing free frag
Jan 29 07:04:05 newnews savecore: reboot after panic: bad dir
Jan 29 11:40:01 newnews savecore: reboot after panic: ffs_alloccg: map corrupted

Should I suspect the hardware, or are these kernel bugs?

Kernel:       NetBSD-current 1996-01-19;
              maxusers=128 (which elicits a warning from "config");
Userland:     NetBSD-1.1
Architecture: i386
CPU:          Intel 80486DX4/100
Motherboard:  PCI with SiS chipset
Memory:       64MB main memory; 128KB external cache
    I plan to get more memory and more cache.
    I tried disabling the external cache, and that might have helped
    stability a little, but not enough.
Disks and SCSI controllers: two Adaptec 2940 controllers;
    one controller connected to three disks and one connected to two disks;
    using ccd to stripe the /var/spool/news partition across two of the 
    disks (one on each controller).

--apb (Alan Barrett)