"Edoardo Ceccarelli" <eddy@axa.it> writes:
> I bet a crashing process could cause that data corruption, right?
No, Postgres is remarkably resistant to causing corruption during a
crash; it's deliberately engineered that way. I don't say this
*couldn't* be the explanation but it shouldn't be your first guess.
Frankly I think hardware flakiness is more likely.
There are currently two known crashing bugs in 8.2.0, and both of them
will result in a simple crash, no data corruption.
regards, tom lane