7468 Too many rows %VSTR.
The node-level buffered row redistribution service has detected an overflow in the count of rows sent to the indicated AMP in a hashed redistribution.
AMP row redistribution service (sutrrd).
The row count at issue is a 32-bit unsigned value and will therefore support redistributions that send up to 4G rows to any given AMP. Since this number of rows would require at least 48GB of disk storage on one AMP, it is expected that an out-of-space error would usually occur long before the 4G row count overflows. This error is therefore not expected to occur unless (1) disk storage limits are increased significantly, or (2) a new DBS service processes the redistributed rows without writing all, or any, of the row data to disk. If either of these conditions changes and queries begin failing with the 7468 error, then the 32-bit row count limit may have to be extended to prevent the overflow.
Examine the failing query carefully to make sure there are no unintended Cartesian product joins or other errors that could cause runaway spool generation.