Diagnosis
The statistical data is written to a cluster table. There is only a
certain amount of space for data in each row of the cluster table. If
a table row does not provide sufficient space, continuation rows
are written in the cluster table. All of these rows have a sequence
number, starting at 0. Each row also contains a field that contains
the amount of data stored in the row. All rows except the last row
contain the maximum value (2886) in this field; only the length of
the last row can deviate.
A bad entry is a data record for which continuation rows are definitely
missing. This can be the first row or a continuation row further down.
Rows are missing, if the highest continuation row is not 1 smaller than
the number of rows that exists.
Example: the rows with the continuation row counters 0,2,3, and 4 exist.
The row with the continuation row counter 1 is missing. The highest
continuation row counter is 4, and the number of rows is also 4.
Therefore, the highest continuation row counter is not 1 smaller than
the number => a row is missing, and the data record is definitely bad.
Bad data records can no longer be read, and the data in them is lost.
An attempt to read this data inevitably leads to a short dump. Data
records of this type are to be deleted.