This error can present due to one of several issues, the following outlines the areas to investigate:
0. Application code issueThere have been several ABL code bugs in the past related to this error, none in the current 10.2B stream. Knowing under what conditions this error is raised, examining the resutling
protrace.<pid> files created thereby identifying the code related in order to provide a reproducible to Technical Support would first need to be done.
1. Physical record or index corruption This corruption would still be present if the records still exist the next time they're accessed, they may have been deleted in the meantime which is why while running database integrity checks the evidence will not be conclusive but never-the-less needs to be confirmed and corrective actions taken as long as this error manifests.
$ dbtool dbname, Option 3. Record Validation
$ proutil dbname -C idxcheck
2. Memory corruptionCould hit multiple users simultaneously touching a buffer or multiple buffers in memory. This type of corruption would not be detectable after memory was scrubbed (overwritten by something else).
3. Schema corruptionWhile possible, it would then affect all users sessions. Example:
Creating a record errors with bffld: nxtfld: scan past last field 4. Network problemsWhile possible, it would then affect all users sessions connecting client/server at the time.