I've had a system running at a clients since April of 1998. Data has been corrupted in two of the main tables, GROUP.DBF and PACSEA.DBF (they are related by GROUP, and Integer).

I normally use Stonefield to create a Reindex application, but the reindex fails, I suspect, due to the bad data.

There are groups of corrupted records in various locations in the file, i.e.,
- in one file the first 1327 records look good, the next 438 are bad, the rest of the 17,469 records are bad
- in the other file, the first 18,601 records are good, the next 366 are bad, the next 8,826 are good, the next 912 are bad, then 6,735 good records, 546 bad, and the last 32,000 are good.

On a browse, "Bad" data looks like:
logical fields: "|" or "A" or "+" or "u" or "z"
integer fields include: "||||" or "URAKO" or "F|||" along with characters in other alphabets

I think I can delete the offending records and limp along w/the missing data (and be able to run reindex successfully). Unfortunately, they have had back up problems for months now.

My question (finally) is how could this happen? what is most likely? disk problems/ operating system? any input is appreciated.

TIA, J