I'm using BerkeleyDB 4.2.52.

One of our applications using BDB crashed at the same time a disk
defragmenter (Raxco Perfect Disk) started working on the disk. I don't
have much information (like a dump) so I'm trying to figure out what
might have happened.

The following errors were written to the BDB error file around the time
of the crash:
write: 0xa3cfeac, 4133: Input/output error
txn_checkpoint: log failed at LSN [4750 8846124] Input/output error

Any thoughts or help would be appreciated? Would a disk defragmenter be
known to cause problems with some BDB files?

Hi Craig,

The EIO error that Berkeley DB is reporting here is how we treat all
Windows errors that we don't handle specially. So it's difficult to
say what the real cause of this error is -- some interaction with the
defragmenter may be generating an error that we haven't encountered
before.

If you could identify the Windows error code (say, by logging it from
the __os_win32_errno routine in os_win32/os_errno.c), then the solution
may be to map this error to EBUSY or EINTR so that Berkeley DB will
retry the operation.

Regards,
Michael.

You're running a defragmenter on a live system ?? You're brave IMHO.

Anyway, is it possible that the disk filled up temporarily ?

The defragmenter was installed and run by someone other than me.

The disk has plenty of space. Somethine like 100 GB.

Thanks for the replies. I will look into identifying the Windows error
code.