[Spacewalk-list] Postgres spacewalk
John Hodrien
J.H.Hodrien at leeds.ac.uk
Tue Nov 22 10:14:07 UTC 2011
On Mon, 21 Nov 2011, Parsons, Aron wrote:
> I did some digging into this query because it is one that is very obviously
> slow when working with real systems.
>
> It seems that the left join to rhnchecksum is the culprit of this slow query. Here's the effect on execution time by changing it to an inner join:
> LEFT JOIN: 2973.886 ms
> INNER JOIN: 1.444 ms
>
> Based on what this query is used for and the columns we're selecting, this
> seems safe to me; the inner join is just going to exclude rows that don't
> exist in both, but that should never happen under normal operation. I have
> not tested this on Oracle, but have verified that client operations are
> still operating normally. The real world effect is that this takes a ~50s
> 'rhncfg-client diff' on the client down to ~3s. I tested by updating the
> query in /usr/share/rhn/server/handlers/config/rhn_config_management.py.
Is that true of symlinks?
jh
More information about the Spacewalk-list
mailing list