X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C730F1.1177B6B0@onstor-exch02.onstor.net>; Fri, 5 Jan 2007 09:43:54 -0800
MIME-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----_=_NextPart_001_01C730F1.1177B6B0"
Content-class: urn:content-classes:message
Subject: Definitive proof it's the bsd install program corrupting files in upgrade
Date: Fri, 5 Jan 2007 09:43:54 -0800
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E0A90E9@onstor-exch02.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Definitive proof it's the bsd install program corrupting files in upgrade
thread-index: Accw7FFgCkj69etjS42Sib+gZ6ZoWQ==
From: "Larry Scheer" <larry.scheer@onstor.com>
To: "Tim Gardner" <tim.gardner@onstor.com>,
	"Andy Sharp" <andy.sharp@onstor.com>
Cc: "Jay Michlin" <jay.michlin@onstor.com>

This is a multi-part message in MIME format.

------_=_NextPart_001_01C730F1.1177B6B0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

Tim, Andy, Jay,
   I confirmed my suspicions it is the OpenBSD install program that is =
corrupting files. Last night I ran a couple of tests on my filers that =
show:

1. Files do indeed get corrupted by the upgrade program, my test had a =
10% failure rate.

I wrote a script that does an upgrade of 2 different builds so a =
significant number of files are changing on the flash. The upgrade =
program I am using uses upgrade's internal system compare routine to =
verify the integrity of the installation immediately after the =
installation completes. Twenty installs were done two installs had a =
small number of files reported as being different.

2. Using tar or cp no files were corrupted. Using /usr/bin/install with =
the -S flag files are (presumably) corrupted.

I say presumably because their check sums are different from the source =
and don't match the check sum of the previously installed version of the =
file. My script doesn't compare hex dumps to confirm the corruption.

I wrote a script that simulates the upgrade program using =
/usr/bin/install, tar, and cp. The one difference in the script from the =
upgrade program is all files, except configuration files, are installed =
from the distribution tar file. The upgrade program skips files it =
deems, sometimes incrorrectly, are the same as the current version on =
the flash.=20

I only ran one pass of this test script, I would like to run more cycles =
of this test and also check if removing the -S flag to install changes =
anything.

If anyone wants to see my proof or check my math in the scripts let me =
know. I will gladly show you the results of my work.

Any questions, query for more details, or suggestions are welcome.

Larry
=20



------_=_NextPart_001_01C730F1.1177B6B0
Content-Type: text/html;
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Diso-8859-1">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
6.5.7650.28">
<TITLE>Definitive proof it's the bsd install program corrupting files in =
upgrade</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->

<P><FONT SIZE=3D2>Tim, Andy, Jay,<BR>
&nbsp;&nbsp; I confirmed my suspicions it is the OpenBSD install program =
that is corrupting files. Last night I ran a couple of tests on my =
filers that show:<BR>
<BR>
1. Files do indeed get corrupted by the upgrade program, my test had a =
10% failure rate.<BR>
<BR>
I wrote a script that does an upgrade of 2 different builds so a =
significant number of files are changing on the flash. The upgrade =
program I am using uses upgrade's internal system compare routine to =
verify the integrity of the installation immediately after the =
installation completes. Twenty installs were done two installs had a =
small number of files reported as being different.<BR>
<BR>
2. Using tar or cp no files were corrupted. Using /usr/bin/install with =
the -S flag files are (presumably) corrupted.<BR>
<BR>
I say presumably because their check sums are different from the source =
and don't match the check sum of the previously installed version of the =
file. My script doesn't compare hex dumps to confirm the corruption.<BR>
<BR>
I wrote a script that simulates the upgrade program using =
/usr/bin/install, tar, and cp. The one difference in the script from the =
upgrade program is all files, except configuration files, are installed =
from the distribution tar file. The upgrade program skips files it =
deems, sometimes incrorrectly, are the same as the current version on =
the flash.<BR>
<BR>
I only ran one pass of this test script, I would like to run more cycles =
of this test and also check if removing the -S flag to install changes =
anything.<BR>
<BR>
If anyone wants to see my proof or check my math in the scripts let me =
know. I will gladly show you the results of my work.<BR>
<BR>
Any questions, query for more details, or suggestions are welcome.<BR>
<BR>
Larry<BR>
<BR>
<BR>
<BR>
</FONT>
</P>

</BODY>
</HTML>
------_=_NextPart_001_01C730F1.1177B6B0--
