On 10/27/06, Joshua D. Drake <jd@commandprompt.com> wrote:
> Merlin Moncure wrote:
> > My previous employer contacted me today. Apparently they are having
> > an increasing frequency of occurances where they disocover violations
> > of the primary key constraint not being caught by the database. This
> > system is an ISAM emulation system, and it relies heavily on the
> > database throwing unique constraint violations to approximately
> > provide a cobol rewrite statement.
> >
> > These errors are never caught when they occur. One way they are
> > caught is during a dump/reload, the record fails to reinsert.
> >
> > I brought this up around a year ago, and the issue was never
> > successfully resolved. These are windows systems running PostgreSQL
> > 8.0.x.
> >
> > any thoughts?
>
> Do they vacuum enough? I have seen problems with PostgreSQL (albeit not
> since 7.3) where a unique constraint would not enforce because of index
> bloat.
iirc vacuum was set up to run weekly on cron. based on turnover which
was high but not super high that seemed appropriate.
merlin