Message-ID: <50C7E598.2070902@2ndQuadrant.com>
Date: Wed, 12 Dec 2012 10:02:00 +0800
From: Craig Ringer <craig@2ndQuadrant.com>
Organization: 2nd Quadrant
User-Agent: Mozilla/5.0 (X11; Linux x86_64;
	rv:17.0) Gecko/17.0 Thunderbird/17.0
MIME-Version: 1.0
To: Evgeny Shishkin <itparanoia@gmail.com>
CC: =?windows-1252?Q?Niels_Kristian_Schj=F8dt?= <nielskristian@autouncle.com>,
	Jeff Janes <jeff.janes@gmail.com>,
	"pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Subject: Re: Do I have a hardware or a software problem?
References: <AE27BD07-7A3A-449A-BD3F-A0DF4EF76C04@autouncle.com>
	<CAMkU=1zcXD5YWMD=QOAdoNG0BKoB-DAXKbzBKybE-3w-YzhSgw@mail.gmail.com>
	<8F7661E7-D6D4-4334-8C6A-A8B6C56A2C3F@autouncle.com>
	<50C7351C.30407@2ndQuadrant.com>
	<4507A68A-3CD5-4283-AFA5-1C0593427E14@autouncle.com>
	<67FBF3AF-5C6A-45D4-9206-7257C8C01C5E@gmail.com>
	<CADZnK0nmm=2WnVVsveu3j0sJ95Ct13SqKbjQ_9DeVCGAmrcggw@mail.gmail.com>
	<E2A43856-FE0A-44BC-9764-C3CA533D94D9@gmail.com>
	<CADZnK0=0q9Ti9jVpCX3VX=3X_Rs=AxONhxC9cejnbOd5GKjb4A@mail.gmail.com>
	<ED7DEE06-B822-43E0-AF49-D89BB40C4C83@gmail.com>
	<50C7D7D2.4070605@2ndQuadrant.com>
	<2BCB6843-1396-42C9-A5BF-21C7C978FF4B@gmail.com>
	<50C7DDDE.1000000@2ndQuadrant.com>
	<3ADBD9A3-966F-40E8-AF78-0F2093D79BEE@gmail.com>
In-Reply-To: <3ADBD9A3-966F-40E8-AF78-0F2093D79BEE@gmail.com>
Content-Type: multipart/alternative;
	boundary="------------070908010903070401020607"
Precedence: bulk
Sender: pgsql-performance-owner@postgresql.org

This is a multi-part message in MIME format.
--------------070908010903070401020607
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit

On 12/12/2012 09:44 AM, Evgeny Shishkin wrote:
> So far, more than a year already, i bought consumer ssds with 300-400$
> hw raid. Cost effective and fast, may be not very safe, but so far so
> good. All data protection measures from postgresql are on, of course.

You're aware that many low end SSDs lie to the RAID controller about
having written data, right? Even if the RAID controller sends a flush
command, the SSD might cache the write in non-durable cache. If you're
using such SSDs and you lose power, data corruption is extremely likely,
because your SSDs are essentially ignoring fsync.

Your RAID controller's BBU won't save you, because once the disks tell
the RAID controller the data has hit durable storage, the RAID
controller feels free to flush it from its battery backed cache. If the
disks are lying...

The only solid way to find out if this is an issue with your SSDs is to
do plug-pull testing and find out.

-- 
 Craig Ringer                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services


--------------070908010903070401020607
Content-Type: text/html; charset=windows-1252
Content-Transfer-Encoding: 7bit

<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">On 12/12/2012 09:44 AM, Evgeny Shishkin
      wrote:<br>
    </div>
    <blockquote
      cite="mid:3ADBD9A3-966F-40E8-AF78-0F2093D79BEE@gmail.com"
      type="cite">
      <meta http-equiv="Content-Type" content="text/html;
        charset=windows-1252">
      So far, more than a year already, i bought consumer ssds with
      300-400$ hw raid. Cost effective and fast, may be not very safe,
      but so far so good. All data protection measures from postgresql
      are on, of course.<br>
    </blockquote>
    <br>
    You're aware that many low end SSDs lie to the RAID controller about
    having written data, right? Even if the RAID controller sends a
    flush command, the SSD might cache the write in non-durable cache.
    If you're using such SSDs and you lose power, data corruption is
    extremely likely, because your SSDs are essentially ignoring fsync.<br>
    <br>
    Your RAID controller's BBU won't save you, because once the disks
    tell the RAID controller the data has hit durable storage, the RAID
    controller feels free to flush it from its battery backed cache. If
    the disks are lying...<br>
    <br>
    The only solid way to find out if this is an issue with your SSDs is
    to do plug-pull testing and find out.<br>
    <br>
    <pre class="moz-signature" cols="72">-- 
 Craig Ringer                   <a class="moz-txt-link-freetext" href="http://www.2ndQuadrant.com/">http://www.2ndQuadrant.com/</a>
 PostgreSQL Development, 24x7 Support, Training &amp; Services</pre>
  </body>
</html>

--------------070908010903070401020607--