The lost interrupt error is growing to a crescendo as I put more work onto
this box (1.6.2_RC3, 7300+G3/500, 512MB RAM, Western Digital 40GB drive
connected to a Tempo Trio ATA/133 [Promise-type controller]).

pciide0:0:0: lost interrupt
        type: ata tc_bcount: 40960 tc_skip: 0
pciide0:0:0: bus-master DMA error: missing interrupt, status=0x21
wd0a: error reading fsbn 3337248 of 3337248-3337327 (wd0 bn 3337248; cn 3310 tn 12 sn 12), retrying
wd0: (uncorrectable data error)

The real problem now is that the error is now uncorrectable, as if the data
had been written wrong in the first place. I've tried setting flags to turn
off UltraDMA, and this doesn't help.

The fsbn is usually consistent in a run. When I back out the database and
go back to a new backup, a new fsbn suddenly starts going bad, but then reads
on that fsbn are consistently bad.

Suggestions of anything to do short of "replace the hardware" would be very
gratefully appreciated.

-- 
---------------------------------- personal: http://www.armory.com/~spectre/ --
 Cameron Kaiser, Floodgap Systems Ltd * So. Calif., USA * ckaiser@floodgap.com
-- You are not ready! ---------------------------------------------------------