"Panic" - that's my middle name. ;)
Had I known how to identify the database at fault, and that it would
have had no effect on the other databases, then I would have handled
this episode differently.
In the event, things seem to be OK. Our old slave db is now acting as
master and the old master rebuilt as the new slave ... courtesy of slon.
I'd like to move to 8.1 but I'm waiting for a quiet period when there's
less development/fire fighting so that I can test all the java
components of our webapp and then manage the upgrade properly.
Maybe suppressing other vacuums once a month, and running the "vacuumdb
-a" option instead wouldn't be a bad idea...
Many thanks for all your support and advice - you've been great help
(and comfort).
John
Tom Lane wrote:
> John Sidney-Woollett <johnsw@wardbrook.com> writes:
>
>>Just out of curiousity would the wraparound error (for mail_lxtreme)
>>actually have affected data in bp_live?
>>Could I just have deleted mail_lxtreme and then continued to use bp_live
>>as though nothing had happened?
>
>
> No, and yes, which is why panicking was not warranted ;-)
>
> Martijn's advice to be using "vacuumdb -a" every so often is well given,
> though.
>
> You could also consider switching over to autovacuum, particularly as of
> 8.1. (I'm not sure how much I trust the contrib version that exists in
> 8.0, and 7.4's is definitely pretty buggy, but I believe 8.1's can be
> relied on to prevent this sort of thing.)
>
> regards, tom lane