Re: My server is oddly very slow

From: adrien ducos <aducos(at)hbs-research(dot)com>
To: Kevin Grittner <Kevin(dot)Grittner(at)wicourts(dot)gov>
Cc: "pgsql-admin(at)postgresql(dot)org" <pgsql-admin(at)postgresql(dot)org>
Subject: Re: My server is oddly very slow
Date: 2011-05-05 10:02:16
Message-ID: 4DC275A8.8000400@hbs-research.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

Kevin Grittner a écrit :
> adrien ducos <aducos(at)hbs-research(dot)com> wrote:
>
> [rearranged somewhat]
>
>
>> The version of both databases is postgres 8.4.1
>>
>
> [sigh] You really should upgrade.
>
> http://www.postgresql.org/support/versioning
>
> http://www.postgresql.org/docs/8.4/static/release.html
>
I know I should but only my system administrator is allowed to do so, we
will soon change to postgres 9.1 anyway.

>
>
>> So I checked the memory on prod during my query execution:
>> vmstat
>>
>
>
>> procs -----------memory----------
>> r b swpd free buff cache
>> 0 1 1280 23320 6356 2975956
>>
>
>
>> The swap is not used but something is a bit odd: the cache is
>> using 98% of the RAM
>>
>
> The first line in vmstat (which is all you get if you run it without
> a number) is averages since the OS was booted. That's not usually
> very helpful. What does this show?:
>
> free -m
>
> or run the query while watching the output from:
>
> vmstat 1
>
>
So I did that also, and have the same results:

free -m
total used free shared buffers cached
Mem: 3042 3000 41 0 4 2896
-/+ buffers/cache: 99 2942
Swap: 1983 1 1982

same for vmstat 1 while runing the query

vmstat 1
procs -----------memory---------- ---swap-- -----io---- --system--
----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
0 0 1280 45752 4496 2963460 0 0 29 45 0 1 2 0 95 3

Almost everything is "cached", I read on some documents that cache means
linux uses the free memory to cache disk data (and avoid waisting the
free memory).

On the other hand, when I do ps -aux while the query is running I see
the query is using 17% of the RAM, so I guess it is using the RAM ok.

The question could be, why is linux marking that 95% of the memory is
cache while in the same time ps -aux shows my query uses 17% of the RAM.

>> the copy of files on the SAN is about 3 times faster than on the
>> development hard drive.
>>
>
> Copy time doesn't always correlate real well with database
> performance. Have you tried running bonnie++ or similar?
>
Ok, we didn't check on that but we will. I'll see with my system admin
next week (he is on hollydays now). On the paper thought it is a SAN
composed of several velociraptors. It is supposed to be faster than a
normal and rather old hard drive (speed and access time).


Thanks,

Adrien

--

Logo_HBS_mail.jpg
Adrien DUCOS
Analyste développeur
aducos(at)hbs-research(dot)com <mailto:aducos(at)hbs-research(dot)com>
www.hbs-research.com <http://www.hbs-research.com/>
+33 (0)9 70 44 64 10
24 rue de l'Est
75020 Paris

In response to

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Ross J. Reedstrom 2011-05-05 16:32:49 Re: Unable to Insert Row
Previous Message ktm 2011-05-05 00:04:51 Re: Re: best practice for moving millions of rows to child table when setting up partitioning?