Hello,
Trying to understand where I could be wrong.
Oracle 12.1.0.2 patchset 20210420
The sequence of events:
-
at some noon a data corruption happened (app tables were updated in the wrong way)
-
at 6 pm a cold RMAN backup was created (took a few hours)
-
after the backup finished, the DB was open in read-only mode and we managed to read successfully the data “as of” before the corruption
-
the backup taken at step 2 was restored at a separate host and opened in read-only (Oracle version and patching level are the same on both source and destination hosts)
-
the same “as of” SQLs from the step 3 failed with ORA-01555 at the restored DB
-
also tried to “shift” time back at the host where the restoration happened, to match roughly the time when the backup was made, same result - ORA-01555.
undo_retention = 12h
Failures with ORA-01555 happened when the SQL tried to read the table itself, if to run something like "select count(*) from .. as of" it, was able to provide the result reading just PK index.
Is it even possible to run retrospective SQLs in that way?