[isf-wifidog] Server Capacity Question

[Benoit Grégoire]

On 25 January 2008, Andrew wrote:
> Thank You.
>
> We found that a VACUUM FULL running once per day has greatly improved the
> performance (8x!).  We were only running a VACUUM before once per hour.

Be carefull, a regular VACUUM FULL is normally not necessary, and is a 
blocking operation that will have a serious (and growing over time) impact on 
your server while it runs.

Either one of the following is probably true:
-You initially didn't run VACUUM regularly enough, and you carried forward an 
important burden that could only be fixed by a VACUUM FULL.
-You need to run VACUUM even more often considering your workload and the 
amout of ram available to postgres.



-- 
Benoit Grégoire
Technologies Coeus inc.