-
Bug
-
Resolution: Done
-
Major
-
3.0.7.Final, 3.1.0.Alpha2
-
None
-
False
-
None
-
False
-
-
When an Oracle RAC node is brought offline and brought back online after maintenance, the LogFileCollector will treat that specific node's redo thread as inconsistent indefinitely. This is because during the downtime window, while there is no gap in log sequences, the expectation is that a log has the specified SCN, but that's not always the case.
Given redo thread 1:
| File | Sequence | First SCN | Next SCN |
|---|---|---|---|
| thread_1_seq_16546.2850.119354407 | 16546 | 6642813012960 | 6642813833063 |
| thread_1_seq_16547.24874.1193544929 | 16547 | 6642813833063 | 6642814020839 |
| thread_1_seq.16548.21380.1193544939 | 16548 | 6642814020839 | 6642814020841 |
| thread_1_seq_16549.40606.1193551121 | 16549 | 6642815622343 | 6642815622349 |
| group_1.6537.1178059925 | 16550 | 6642815622349 | MAX |
If the connector wishes to mine from 6642815621860, the connector will report that no log has that SCN as part of redo thread 1, which makes sense, given that the node was stopped at sequence 16548 and brought back online with sequence 16549.