i use oracle logminer to analysis oracle redo log,and my program relied heavily on scn increase strictly，but sometimes i found that scn from result of logminer is not increase。
for example：
26942069
26942070
26942071
26942058
26942071
26942071
26942071
26942071
As we all know，scn in oracle is increased strictly,but why it break in logminer?
this is my sql：
BEGIN DBMS_LOGMNR.ADD_LOGFILE(logfilename => '/home/oracle/oracle/product/12.2.0/dbhome_2/dbs/arch1_1917_1005421687.dbf' ,options => DBMS_LOGMNR.NEW); END;
BEGIN DBMS_LOGMNR.START_LOGMNR(STARTSCN=>26942032,OPTIONS => DBMS_LOGMNR.NEW); END;
my program relied heavily on scn increase strictly
Well.. your program should not :-)
The SCN is an increasing mechanism, but that is not the same as saying it *must* increase for every single commit. The SCN is all about *our* (the database's) ability to ensure changes are applied in order, including during a recovery.
If you'd like to really dig into the nitty gritty, Frits Hoogland did some very detailed analysis on this
https://fritshoogland.wordpress.com/2016/10/22/transactions-and-scns/