The torn page problem refers to a problem that may exist when the data block is bigger than the filesystem page size and the data block is not completely written into disk due to a hardware problem or energy failure.
Torn pages can happen to both Type I or Type II areas when a data block write is interrupted.
- With Type I Storage Areas there is no way to know that a torn page happened.
- The difference with Type II areas is that they store a check sum stored in all data block headers, enabling the detection of torn pages with block checksum error 14410:
Checksum validation of dbkey <Arg1> block type <Arg2> in area <Arg3> does not match data. Expected: <Arg4> received <Arg5>. (14410)
For further information on Torn Pages, refer to :The attached Exchange 2006 presentation by GUS BJORKLUND
"OpenEdge R RDBMS Performance Tuning Made Simple", 5.4 Choosing Data Block Size
The "Using a larger database blocksize than the filesystem blocksize" Section in Article:
The consequence that may arise when using non-Progress backups in Article:
