Here's a discussion that I had with Matt Ray on the subject, it explains some things that I didn't explain above:

[03:40pm] <@mrayzenoss> rmatte: take a look at this: docs/DOC-3219
[03:40pm] <@mrayzenoss> instead of Server.Linux just do all devics
[03:41pm] <@mrayzenoss> you want to loop over all devices right?
[03:41pm] <rmatte> correct
[03:41pm] <rmatte> you're sure that'll fix this particular issue?
[03:41pm] <@mrayzenoss> no, but I know that'll loop over all the devices
[03:42pm] <rmatte> well, I know how to loop over all devices, but when I try that with the code that I posted in the forum it errors out
[03:42pm] <rmatte> so I'm going to need to use some sort of check, I just don't know what
[03:42pm] <@mrayzenoss> oh, to see if the attribute is there or not?
[03:42pm] <rmatte> correct
[03:42pm] <rmatte> also, is removing the attribute going to hurt anything?
[03:42pm] <rmatte> I know it fixes the issue, but will it make the WMI pack stop working or something?
[03:43pm] <rmatte> like, I don't fully understand what caused it or what could happen by removing it?

I figured out what caused this.  I used Jane Curry's jc_batch_load script.  There are places to specify zWinUser and zWinPassword, but I always leave those blank.  The script doesn't check if there's a value specified and tries to set them at the device level with no value anyways.

The lines that cause the issue in the script are:

# Now for the bits from Scott's script - loadDevice doesn't handle Windows users / passwords...
      d=dmd.Devices.findDevice(deviceName)
      print "dmd device is ",d
      d.zWinUser=zWinUser
      d.zWinPassword=zWinPassword

I commented out the lines so that I don't run in to the same issue in the future.

Thanks to all those who helped me out with troubleshooting this.

Cheers.