Redo sequences in log files not always ascending

In the logfiles of Attunity Replicate 5.0.1 32 from Oracle to SQLServer I noticed that the sometimes the first redo sequence is lower than the previous last redo sequence number. Is this normal behavior or is this something that needs investigation.
During 1 occurrence we have had an issue in the target system with duplicate entries when the above happened.
In a logfile the last entry is:
00007428: 2016-09-24T21:10:13 [SOURCE_CAPTURE ]I: New Log Miner boundaries in thread '1' : First REDO Sequence is '2038285', Last REDO Sequence is '2038285'
In the following logfile the first actual source capture line after restart of replication is:
00010544: 2016-09-24T21:10:32 [TASK_MANAGER ]I: Starting replication now (replicationtask.c:1912)
00011168: 2016-09-24T21:10:33 [SOURCE_CAPTURE ]I: New Log Miner boundaries in thread '1' : First REDO Sequence is '2038283', , Last REDO Sequence is '2038287' (oracdc_reader.c:595)

But I can also find the following within a log file:
00008608: 2016-09-18T04:16:53 [SOURCE_CAPTURE ]I: New Log Miner boundaries in thread '1' : First REDO Sequence is '1983972', Last REDO Sequence is '1983974' (oracdc_reader.c:595)
00008608: 2016-09-18T04:17:07 [SOURCE_CAPTURE ]I: New Log Miner boundaries in thread '1' : First REDO Sequence is '1983973', Last REDO Sequence is '1983974' (oracdc_reader.c:595)
Were the last line show number 73 while the previous line showed 74 as last number.