John Scalia <jayknowsunix@gmail.com> writes:
> My wonderful colleagues in charge of our production database have screwed up royally. They managed to corrupt the
productiondatabase by having two instances of postgresql talking
> to the same database/same filesystem/same time. Now, they'd like me to fix it. Sigh...
Just outta curiosity, how'd they do that? There are interlocks that are
supposed to prevent it.
You can defeat the interlocks by manually deleting the postmaster lock
file, which is why you're *never* supposed to do that. I'm curious if
they did that, or found some new way.
regards, tom lane