Skip site navigation (1) Skip section navigation (2)

Re: Postgres restart

From: "Scott Whitney" <swhitney(at)journyx(dot)com>
To: "'S Arvind'" <arvindwill(at)gmail(dot)com>,"'pgsql-admin'" <pgsql-admin(at)postgresql(dot)org>
Subject: Re: Postgres restart
Date: 2009-09-21 19:23:50
Message-ID: 20090921192413.D5585C4001@mail.int.journyx.com (view raw or flat)
Thread:
Lists: pgsql-admin
The 2.5GB -> 50MB "problem" could mean a number of things. It could also
mean nothing at all. It's much more about the amount of swap (paging) space
that's being used at the time of the slowdown.

The fact that it runs fine for a few hour then degrades significantly would
point me in the direction of watching paging space consumption. You may have
added just one too many rows, making one query just big enough to make your
server start paging more than it ever used to do.

Check your vmstat and iostat information at the time of the slowdown, and
see if you can determine whether you're paging or waiting on storage. I
suspect you'll find that throwing hardware (memory, as a guess) at this
problem will solve it. 

-----Original Message-----
From: S Arvind [mailto:arvindwill(at)gmail(dot)com] 
Sent: Monday, September 21, 2009 2:19 PM
To: Scott Whitney; pgsql-admin
Subject: Re: [ADMIN] Postgres restart

Thanks Scott.
1. Well, is this a dedicated database server?
YES, it runs only Postgres with some Back-up script for that DBs alone
daily.

2. What O/S? 
CentOS (Linux version 2.6.18-8.1.4.el5 (mockbuild(at)builder6(dot)centos(dot)org) (gcc
version 4.1.1 20070105 (Red Hat 4.1.1-52)) )

3. What version of PG?
 8.3.7

4.  Have you check out memory usage? 
 When rebooted it has more then 2.5 GB free space but after few hours it
will reach 50MB. This is usual in our DB server, since this decrease never
affected our performance for past years. And also for 5 months we never
rebooted our system and also we had restart the postgres likely once in a
month, before this problem.

5. Also, when was the last time you vacuumed the database(s)?
As per advise from postgres team we are running full vaccum for every week
and frequently-used table(30) vacum daily. We have nearly 640 tables in each
DB.

6. Is auto-vac on?
Yes ( postgres: autovacuum launcher process running)

Is our problem is identifiable, from infrastructure side?

-Arvind S



"Many of lifes failure are people who did not realize how close they were to
success when they gave up."
-Thomas Edison



On Mon, Sep 21, 2009 at 11:33 PM, Scott Whitney <swhitney(at)journyx(dot)com>
wrote:


	There's an awful lot of information left out that would be very
useful to
	help advise you.
	
	Restarting the postgres services on a daily basis is certainly
nothing
	that's going to corrupt your data or hurt your system, PROVIDED that
it is
	done correctly (ie: not killing the backend postmaster when
something is
	happening, not hard-booting the system while a RAID card is trying
to write
	its cache, etc.)
	
	However, I think you'd be postponing the problem. The better answer
might be
	to define and resolve the issue.
	
	"Its performance is good but gradually going down?" Well, is this a
	dedicated database server? What O/S? What version of PG? Have you
check out
	memory usage? 4GB seems a bit low for the amount of data you're
using. In a
	similar environment, I've got 12GB, and from time to time I'm
paging.
	
	If "nothing changed" (TRULY, that is), you're most likely finding
that
	you're either CPU, memory, or I/O bound, and the most likely
culprits are
	the last 2 unless you've suddenly started some massive queries that
didn't
	happen a few weeks ago.
	
	Also, when was the last time you vacuumed the database(s)? Is
auto-vac on? I
	know, I know, I'm not _supposed_ to have to perform a full vacuum
and
	analyze on my databases with auto-vac on, but if I don't, I run into
	performance problems, so I do that once per week, myself.
	

	-----Original Message-----
	From: pgsql-admin-owner(at)postgresql(dot)org
	[mailto:pgsql-admin-owner(at)postgresql(dot)org] On Behalf Of S Arvind
	Sent: Monday, September 21, 2009 12:46 PM
	To: pgsql-admin
	Subject: [ADMIN] Postgres restart
	
	Recently due to some problem(not yet diagnosed) , our DB server
Postgres is
	getting very slow after few hours. We didnt changed any settings for
6
	months , so we dont know y its happening suddenly in this week. Our
data
	folder is 118GB with 160 DBs. System is 2 Quad core with RAM 4GB. In
last
	two days when it was restarted its performance is good but gradually
going
	down. So few planned to restart the posgres process daily. Is it
advisable
	to restart server daily ? since daily we can have 30 mins downtime.
Please
	advise is it advisable or not?
	
	-Arvind S
	
	"Many of lifes failure are people who did not realize how close they
were to
	success when they gave up."
	-Thomas Edison
	
	
	




In response to

Responses

pgsql-admin by date

Next:From: Scott MarloweDate: 2009-09-21 19:36:35
Subject: Re: Postgres restart
Previous:From: S ArvindDate: 2009-09-21 19:19:05
Subject: Re: Postgres restart

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group