Database crashes with block checksum error 14410 but database scan returns no errors.

Database crashes with block checksum error 14410 but database scan returns no error
Attempts to test the same block after the crash came up clean so it is possible the error was caused by a memory problem


 

The exact cause is not known at this time.
Suspect a memory / hardware problem in system caused an in correct checksum validation.

See Steps in note section below for methods to check if the error may be related to memory / hardware problems.

In at least one case, data was moved to a new location on disk (separate area was created and the data was moved to new area). This corrected the problem.

Another case, 3rd Party software was accessing the runtime database without a confirmed PROQUIET raised. While database block checks were clean, data / index integrity checks raised specific blocks with logical errors. Since the database blocksize was larger than the filesystem blocksize, a torn page resulted. 

The following Steps are intended to validate if the problem may be a memory / hardware issue.

0)  ​With the database shutdown (so as to avoid crashing the database if the corruption were real) perform either of the following tests:

1)  Try to read the record reported in the 14410 error

find first <tablename> where recid(<tablename>) = <dbkey reported in 14410 error>.
display <tablename>.

2)  Determine if there are block checksum errors before and after rebooting the server
Note: the DBRPR utilitiy disables after imaging.

proutil <dbname> -C truncate bi
proutil <dbname> -C dbrpr
Choose Option 9 to change the scanned area to the area containing the table and the reported dbkey in the 14410 error.
Choose Option 16 to scan / fix any block checksum errors within area.

Choose option 1 just report bad block checksum errors.
or
Choose option 2 to scan and fix bad block checksum errors.

Choose option G
Choose Y
Choose Y
Choose Q

Review the database log for the output:

• If no bad block checksums are reported then the database area is clean.
• If bad block checksums are reported then reboot the machine to clean memory and repeat the steps above.
• If bad block checksum errors are reported after the above steps contact technical support to review available Options.