Re: Slow count(*) again...

On Oct 12, 2010, at 8:39 AM, Dan Harris wrote:

>  On 10/11/10 8:02 PM, Scott Carey wrote:
>> would give you a 1MB read-ahead.  Also, consider XFS and its built-in defragmentation.  I have found that a longer lived postgres DB will get extreme
>> file fragmentation over time and sequential scans end up mostly random.  On-line file defrag helps tremendously.
>> 
> We just had a corrupt table caused by an XFS online defrag.  I'm scared 
> to use this again while the db is live.  Has anyone else found this to 
> be safe?  But, I can vouch for the fragmentation issue, it happens very 
> quickly in our system.
> 

What version?  I'm using the latest CentoOS extras build.

We've been doing online defrag for a while now on a very busy database with > 8TB of data.  Not that that means there are no bugs... 

It is a relatively simple thing in xfs -- it writes a new file to temp in a way that allocates contiguous space if available, then if the file has not been modified since it was re-written it is essentially moved on top of the other one.  This should be safe provided the journaling and storage is safe, etc.


> -Dan
> 
> -- 
> Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-performance