Обсуждение: Postgres restart

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@postgresql.org
[mailto:pgsql-admin-owner@postgresql.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

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@gmail.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@builder6.centos.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@journyx.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@postgresql.org
    [mailto:pgsql-admin-owner@postgresql.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

On Mon, Sep 21, 2009 at 1:19 PM, S Arvind <arvindwill@gmail.com> wrote:

First, check performance after reboot versus no reboot and ensure it
really is making things faster.  It could just be that in the morning
the db is faster with less load and as load increases it gets slower.

> 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.

As the other Scott says, that's pretty normal.  I'd pay more attention
to the amount of swap used, what vmstat and iostat and possibly top
has to say.  How much IO wait, is it climbing, etc.

> 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.

Vacuum full hasn't been recommended for quite some time, and is even
being considered for removal from 8.5 (or renaming it to something
else, etc).  Vacuum FULL can bloat indexes.  But this wouldn't get
fixed by a reboot.  Vacuum full followed by reindex on certain tables
with certain access patterns may be required, but almost NEVER on the
whole db.

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

Is it keeping up?  I had a server where autovac couldn't keep up at
the default 20ms sleep and had to drop it down to 0.  But again a
reboot won't fix that.

> Is our problem is identifiable, from infrastructure side?

Maybe.  You need to get familiar with system monitoring tools so you
can watch what's happening to the system.  tail system logs in
/var/log, run vmstat, iostat, top etc.  Those tools can tell you a lot
about what's happening if you know how to read them.