Re: Alter Fragment on a table in syscdr
If the table is empty, then the cluster index will be faster than dropping
the table. The cost of a cluster index is that we have to scan and sort the
table. If there is nothing in the table, then all we have to do is to
create a new (and empty) partition.
"Heinz Weitkamp" <heinz.weitkamp@westfleisch.de> wrote in message
news:c8cghv$tpa$1@terabinaries.xmission.com...
>
> Thank you, Madison Pruet.
> Another question:
>
> Can I do: (because it's faster)
>
> cdr stop
> dbschema -d syscdr -t trg_send_sbuf -ss > file> Drop the table and than create the table new
> cdr start
> without any trouble with ER to reclaim the space.
>
> (IDS 7.31UD4)
> Thank you.
>
> Heinz Weitkamp
>
> -----Urspr'ngliche Nachricht-----
> Von: owner-informix-list@iiug.org
> [mailto:owner-informix-list@iiug.org]Im Auftrag von Madison Pruet
> Gesendet: Montag, 17. Mai 2004 18:04
> An: informix-list@iiug.org
> Betreff: Re: Alter Fragment on a table in syscdr
>
>
> In 7.31 and 9.21, the space allocated in the send queue is not reclaimed
> because the data is stored as a partition blob. With 9.3, we have moved
the
> disk queue to a smart blob, which enables us to reclaim the space.
>
> I would not attempt to use fragmentation to reclaim the space. Instead I
> would attempt to use an index clustering to do the same thing.
>
> You will need to bring down ER, using your favorite technique (i.e. cdr
> stop, blockout, etc) and then attempt to convert one of the indexes on
the
> send queue to a clustered index. Then restart ER.
>
> "Heinz Weitkamp" <heinz.weitkamp@westfleisch.de> wrote in message
> news:c8aknh$8qo$1@terabinaries.xmission.com...
> >
> >
> > Hallo,
> >
> > I have a large table "trg_send_sbuf" in dbspace "cdrquedbs" belonging to
> the
> > database syscdr (Enterprise Replication).
> > The table has 1 GB with no rows.
> >
> > Can I do an
> > ALTER FRAGMENT ON trg_send_sbuf INIT IN <cdrquedbs>;> > to reclaim the space without any trouble with ER.
> >
> > Must I stop the Replication?
> >
> > I appreciate any ideas anyone can give me.
> > Thank you.
> > (Excuse my poor english)
> >
> > Heinz Weitkamp
> >
> > sending to informix-list
>
>
>
> sending to informix-list