Message-ID: <4B7A72CA.2030305@boozter.com>
Date: Tue, 16 Feb 2010 11:26:18 +0100
From: lionel duboeuf <lionel.duboeuf@boozter.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: pgsql-performance@postgresql.org
Subject: Re: Almost infinite query -> Different Query Plan when
	changing where clause value
References: <4B790387020000250002F2CF@gw.wicourts.gov>
In-Reply-To: <4B790387020000250002F2CF@gw.wicourts.gov>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit

Here is my log analysis:
Due to a database recovery task it appears that:

    I stopped postgresql
    I started postgresql (and as default autovacuum daemon)
    I restored the databases (need to restore 4 databases)
    It seems that after database 1 have been restored, autovacumm 
started on it and has been stopped while restoring database 2.
   
    @see log as attached


Here is backup/restore commands i use:

/usr/bin/pg_dump -i -h localhost -U postgres -F c -b -f  ... x 4 times

sudo -u postgres pg_restore -d 'database1' x 4 times


Does this kind of error can lead to the query problem  ?
If so, what would you suggest me to do to avoid the problem again ?.


Thanks
Lionel

   
Kevin Grittner a écrit :
> lionel duboeuf  wrote:
>   
>> Kevin Grittner a écrit :
>>     
>  
>   
>>> I just reread your original email, and I'm not sure I understand
>>> what you meant regarding VACUUM ANALYZE. If you run that right
>>> beforehand, do you still get the slow plan for user 10?
>>>       
>  
>   
>> I confirm by executing manual "VACUUM ANALYZE" that the problem is
>> solved.  But what i don't understand is that i would expect
>> autovacuum to do the job.
>>     
>  
> I think this is the crux of the issue.  Boosting the
> default_statistics_target or the statistics target for specific
> columns might help, reducing autovacuum_analyze_scale_factor might
> help, but I can't help wondering whether you inserted a large number
> of rows for user 10 and then ran the query to select user 10 before
> autovacuum had time to complete.  Does that seem possible?
>
> -Kevin
>
>