Re: [GENERAL] Index pg_proc_prosrc_index: NUMBER OF INDEX' TUPLES (1071)ISNOT THE SAME AS HEAP' (1070)

> Bruce Momjian wrote:
> >
> > > Bruce Momjian wrote:
> > >
> > > > > Anyone seen this message or know what it means?
> > > > >
> > > > > NOTICE:  Index pg_proc_prosrc_index: NUMBER OF INDEX' TUPLES (1071) IS
> > > > > NOT THE SAME AS HEAP' (1070)
> > > >
> > > > Drop index and recreate.  Next release will be more specific in error
> > > > message.
> > >
> > > I have no idea *which* index to drop/recreate, and I have hundreds of them.
> > > Ouch.
> >
> > That will also be fixed.
>
> I thought that the index in question was, in fact,
> pg_proc_prosrc_index in the above example. If that's the
> case, then is it possible for Ed to rebuild a system index?
> The only absolutely surefire way is to dump/reload, isn't
> it? Maybe somewhere someone is doing a heap_insert(),
> heap_replace(), et al, and an event is happening which is
> causing the code to not get to the
> CatalogOpenIndices()/CatalogIndexInsert()/CatalogCloseIndices()...

Signe me up as a dope.  Yes, it is clearly that index.  I was thinking
of another place that has this problem, the famous "My bits moved off
the end of the world"  error message.  This one is clearly the
pg_proc_prosrc_index index.

The only way to fix that is to initdb, I think.  I would recommend
pg_upgrade, after removing the disable from the pg_upgrade script that
was added in 6.5.  That will fix it.  Not sure how it got that way,
though.


--
  Bruce Momjian                        |  http://www.op.net/~candle
  maillist@candle.pha.pa.us            |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026