Re: Are WALs affected by 'duplicate key' errors?

"Jeff Boes" <jboes(at)nexcerpt(dot)com> writes:
> We have an application that attempts many insertions into a table,
> detecting the prior existance of the row by whether we get a 'duplicate
> key in unique index' error. When this application is running as multiple
> copies and really cranking, we get 130K or more of these in a day. Are
> these filling up (and possibly messing up) our WALs?

Shouldn't be a problem.

> We have experienced a few shared-memory corruption errors, and a
> full-blown corruption (although of a different table).

The last couple of reports we got like that turned out to be hardware
problems (bad RAM, flaky disk controller, that sort of thing). It
wouldn't be a waste of time to run some hardware diagnostics.

I have been thinking about your report of trouble with the 7.1.2 to
7.1.3 upgrade, and I have to confess bafflement. There shouldn't be
any compatibility issue there AFAICS. Are you sure there wasn't
anything else that changed at the same time?

regards, tom lane