Re: [HACKERS] Index corruption

> No, that's a new one AFAIK. I don't suppose you saved the state of your
> DB before rebuilding it? I'd like to try to reproduce the problem...

No, sorry. I got increasing desperate as this was a production system and I
was under a bit of pressure to get it back up. A day earlier I had had a
complaint about the number of tuples in the index being incorrect. At the
third attempt I managed to run vacuum over it without the backend crashing
and the it seemed to behave well. Next morning I ran vacuum again and then I
ended up with the endless file-creation loop. Oh yes, to get it to vacuum
I had to delete all my functions (pg_proc) and then reload them. I know that
all my procedures are small enough not to break the 8K limit, as I used to
have trouble with that. I tried the same trick, i.e. dropping and reloading
my functions, but no luck. As most of what they do is to enforce referential
integrity, Jan's foreign key stuff may solve a large part of the problem!

I had the system logging with debug level 3 and there was nothing in the
logs. Did anything get fixed in this area between 6.5.2 and 6.5.3? I.e.
should I upgrade? I'd rather not just at the moment.

Merry Christmas!

Adriaan