This is normally designed for legal reasons because the number assignment does not contain any gaps, and numbers are assigned chronologically. No buffering (This also applies when NUMBER_GET_NEXT is called with IGNORE_BUFFER, or when buffer size is set to 0):.In all cases, given number will be UNIQUE. Some others have legal restrictions, that need to be considered from application side, where buffering method cannot be used. Many number ranges have no legal constraints, meaning the numeration does not need to be neither ascending nor gapless, allowing for buffering techniques to be used in order to improve performance. This buffering method (as well as the amount of numbers in the buffer) is configured on the number range object attributes, on transaction SNRO (or SNUM). In order to improve performance, different buffering options can be used.
HOW TO ROLLBACK PARALLELS UPDATE UPDATE
This will be shown as locks on NRIV table, TIME_OUT dumps, and work processes hanging in the module SAPLSNR3, at the instruction: SELECT SINGLE FOR UPDATE * FROM NRIV This is why buffering concept is used. (See Note 678501 - System stoppage, locks on table NRIV). KNOWN ISSUE: It may happen that the exclusive lock with SELECT SINGLE FOR UPDATE takes long, until application does the COMMIT WORK and releases the interval to be used by a different application.This is solved using internal lock from database (SELECT SINGLE FOR UPDATE), so no other application can take that number. If many applications would take a number from same range at the same time, it may happen that same number could be given twice. Some applications request one single number and other applications request many numbers at the same time. If a dump is created on ST22, this can also be seen on the dump (NRIV-OBJECT). This should be displayed in the value of the field TNRO-OBJECT.
HOW TO ROLLBACK PARALLELS UPDATE HOW TO
0 kommentar(er)
