bootloader hangs on IBM x306 w/2 SATA disks

I have an IBM eServer x306 (1U) that I've been using as a build host
for i386-8.0_BETA and -current packages.  It has until recently only
had a single 80GB SATA hard disk installed in the bays accessible
behind the front bezel of the machine.

It has i386-8.0_BETA w/native Xorg on the 80GB disk, it runs -current
via netboot/NFS semi-disklessly (swap "/tmp", and auxiliary file systems
on the local disk).

I installed a second hard disk (40GB) in order to have a separate
local-disk installation of 8.0_BETA using modular/pkgsrc Xorg.

With both disks installed, the bootloader hangs.  This is true whether
booting from the disk or netboot/PXE.

The disks have the stock "/usr/mdec/mbr" installed and the NetBSD
partition is marked active on the original 80GB disk.  The machine's
BIOS is set to try that disk first.  The usual "NetBSD MBR Boot" message
never appears.  I don't know if it got as far as loading the boot program
from the active partition.

The PXE bootloader loads the kernel, but hangs right after that (I have
mine set to enable "boot.cfg" and module loading).  It prints the
kernel segment sizes, but no module-loading messages appear after that
and it is stuck indefinitely.

In either case, Ctrl-Alt-Delete will reboot the machine, so perhaps
that tells a little more about where the CPU is getting hung up.

With either disk installed alone, the machine boots fine.  (The 40GB
disk, not having an active partition, fails over to netboot/PXE.)

In case it matters, the 80GB disk is a Maxtor and the 40GB disk is
Western Digital.

Anyone else ever run across something like this? 

I'll see if a bootable CD using some other bootloader has any trouble.

-- 
|/"\ John D. Baker, KN5UKS               NetBSD     Darwin/MacOS X
|\ / jdbaker[snail]mylinuxisp[flyspeck]com    OpenBSD            FreeBSD
| X  No HTML/proprietary data in email.   BSD just sits there and works!
|/ \ GPGkeyID:  D703 4A7E 479F 63F8 D3F4  BD99 9572 8F23 E4AD 1645