Date: Thu, 10 Mar 2011 07:01:10 -0600 (CST)
From: Scott Whitney <scott@journyx.com>
Reply-To: Scott Whitney <swhitney@journyx.com>
To: pgsql-admin@postgresql.org
Message-ID: 
 <11997393.14072.1299762070874.JavaMail.root@zimbra.int.journyx.com>
In-Reply-To: <358079.14070.1299761935620.JavaMail.root@zimbra.int.journyx.com>
Subject: Re: pg_clogs hanging around
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit

Ooops...I accidentally took this off list, as Kevin was nice enough to point out.


>> What am I looking for? 

>Outliers. 

> Yeah. It's just those 2. I'd assume that the db I created
> yesterday would be an outlier, but template0 has been there all along
> (of course) and is still listed as 648, a significantly smaller number. 


>> The output shows me 345 rows, most of which are 132xxxxx numbers. 
>> Two of them (template0 and a database created yesterday) say 648. 

>The template0 database is what's keeping the clog files from being 
>cleaned up, but I guess the big question is why you care. They will 
>go away eventually, and shouldn't affect performance. Are they 
>taking enough space to merit extraordinary effort to clean them up? 
> -Kevin 


My concern is that when we had a failure a few years ago, and one of the clog files went bad. I had to manually recreate some customer data after bringing up the previous backup. So, I'd rather have them not there, because, well, if there are 200 of them in the dir, there's a higher chance in a case of a crash that one goes bad than if I have 15. 

Would adding -f (full) clean these up? I seem to recall it did in earlier versions. I've added the -F to it already, and that didn't seem to help.