"Amaya Gamarra" <amaya(dot)gamarra(at)gtd(dot)es> writes:
> We've got a Slony-I cluster over 2 postgres 8.1.11 servers. Every 4 or 5
> days the DB access gets locked and an the next error appears on the log:
> "redo starts at 2/XXXXXX"
> "incorrect resource manager data checksum in record at 2/YYYYYYYY"
> "redo done at 2/ZZZZZZZZ"
The "incorrect data checksum" message does not necessarily show that
anything is wrong --- it's entirely possible to get that message during
normal reading of the WAL after a crash. What seems more useful to
investigate is why you're crashing in the first place. But you did not
provide any information that could reveal that.
regards, tom lane