Philippe MILLET, PhD

Chunk delimiter was invalid

2011-06-18T07:37:00.000-07:00

You may experience this error with subversion (SVN).
- this is a server error
- this happens when SVN is running over http (Apache)

If your SVN/Apache servers are working on Linux, you may try the following:
1) change the MTU of you ethernet board from 1500 to 1492
- in the file /etc/sysconfig/network-scripts/ifcfg-
e.g. -> vi /etc/sysconfig/network-scripts/ifcfg-eth0
- add the line:
MTU=1500

2) change the Apache Timeout from 120 to 1200
- edit the file : /etc/httpd/conf/httpd.conf
- change the value of the Timeout

It should solve you chunk problems!

Create a bootable USB key with Windows using Barte PE / PE Builder

2011-02-02T10:29:00.000-08:00

How to boot ISO image from USB device NTFS formatted.

# The situation:
- You have an USB drive (HDD or Stick) containing only one NTFS partition.
- You already have data on this USB drive and you don't want to format it or repartition it.
- You have a computer or several computers that can boot on USB devices
- You want to dump the partitions of these computers into files on the USB device and use that USB device as a recovery medium in case of computer crash or hard drive failure.

# What you need:
- computer with a BIOS allowing to boot on USB devices
- a USB device
- a windows XP SP2 installation CD or DVD
this is the system we want to boot
- when XP SP2 installation CD comes from DELL we need subinacl tool from Microsoft (download here: http://www.microsoft.com/downloads/details.aspx?FamilyID=E8BA3E56-D8FE-4A91-93CF-ED6985E3927B&displaylang=en)
- the windows XP server 2003 SP1 (here to download from microsoft: http://download.microsoft.com/download/1/2/7/127c5938-d36a-4405-9df1-f00d57495652/WindowsServer2003-KB889101-SP1-x86-ENU.exe)
Some of the files from this release are needed to update the XP SP2 boot files.
- BartPE (here to download: http://www.nu2.nu/pebuilder/)
BartPE is the free software that you will be using to create the bootable "live" Windows rescue disk
- Sherpya's XPE (here to download: http://sourceforge.net/project/showfiles.php?group_id=126922&package_id=140740)
XPE is an addition to BartPE (that is, a "BartPE plugin") that allows you to have a Windows-like desktop, Windows Explorer and other Windows tools (like the disk defragmenter) when you boot to the rescue environment

# Important notices
In the following, we assume that USB_DRIVE: is the drive letter of you USB device
Do not use any space in directory and file names as it may not always be supported by the tools we are using here.

# Make the USB drive bootable
1 -> Install windows XP on the USB drive from the installation CP
The easiest way to do so, is to install XP on the USB drive from the installation CD. For this purpose you may use any XP installation CD.
To do so, I changed the boot devices in order to boot from CD.
When you are asked about disk preparation, you may choose to install directly on the drive without formatting it.
This will modify the MBR of the disk so that it can boot windows XP.
When the files are copied you will be asked to reboot the PC.
You must do it, but don't try to reboot on the USB hard drive, windows XP would not boot and would show a blue screen!
This is because windows XP resets the USB ports and looses the USB device during boot...
You must then boot the PC on the internal hard drive of your PC.

2 -> Remove installed files form the USB drive.
Since we just wanted windows to make it bootable and need to used the server 2003 files, we must remove the XP SP2 installed files from the USB drive.
Remove the windows directory, as well as the NTDETECT.COM, ntldr, and any other file that was created by the windows installation. (you must check that you removed all hiden files like AUTOEXEC.BAT, boot.ini, CONFIG.SYS, IO.SYS, MSDOS.SYS, ...)

3 -> Copy USB boot files from XP server 2003 SP1
To solve this boot problem from XP, we use WP server 2003 SP1.
Extract the downloaded WindowsServer2003-KB889101-SP1-x86-ENU.exe file into a temporary directory.

> WindowsServer2003-KB889101-SP1-x86-ENU.exe -x

You will get a I386 folder.
From this folder:
copy setupldr.bin to USB_DRIVE:\ntldr
copy ramdisk.sy_ to USB_DRIVE:\ramdisk.sy_
copy NTDETECT.COM to USB_DRIVE:\
USB_DRIVE:
cd \
expand -r ramdisk.sy_

create a text file named winnt.sif with the following lines:
[SetupData]
BootDevice = "ramdisk(0)"
BootPath = "\I386\SYSTEM32\"
OsLoadOptions = "/noguiboot /fastdetect /minint /rdexportascd /rdpath=boot.iso"

This file will tell windows to copy the boot.iso image into a ramdisk and boot from it.
The RAMDISK is limited to 512MB, the ISO image has to fit this maximum size.

note: be careful, these files must not be compressed on the NTFS of your USB drive.

# create a ISO bootable windows XP Preinstalled Environment (PE) with BartPE

0 - Install Tools
a) install bartPE
let say you have installed it in c:\pebuilder3110a
b) install Macrium Reflect
let say you have installed it in c:\reflect

1 - Prepare the I386 folder
In order to create such an environment, BartPE requires you to have a windows installation CD.

a) copy the I386 folder from the CD to the harddrive
Some of the files must be modified with the server 2003 files, so the I386 folder must be copied onto an harddrive and updated.
Let say we copy the I386 folder into c:\pebuilder3110a\i386

b) update the I386 folder
create a srsp1 folder in c:\pebuilder3110a
From the previously extracted I386 XP server 2003 folder
copy setupldr.bin to c:\pebuilder3110a\i386\ntldr
copy ramdisk.sy_ to c:\pebuilder3110a\i386\ramdisk.sy_
copy NTDETECT.COM to c:\pebuilder3110a\i386\NTDETECT.COM

c) When the XP SP2 CD comes from DELL, the following steps must be done:
Dell delivery CD bring some issues to BartPE with file permission.
Using the DELL installation CD will lead BartPE to be unable to delete some file during the environment creation, so that it will not create the ISO file.
I found the solution on 911cd's forum (here: http://www.911cd.net/forums//index.php?showtopic=15138&st=0) and copy it here for convenience.

c.1) In c:\pebuilder3110a\i386, create a text file named fixdell.cmd with the following lines:
REM Version 1.4
reg query HKU | find /i "pebuilder" > fixdell.txt
for /f %%a in (fixdell.txt) do reg unload %%a
reg load HKLM\DELL setupreg.hiv
subinacl /subkeyreg hkey_local_machine\dell\controlset001\services\*\ /objectcopysecurity=hkey_local_machine\dell\controlset001\services
reg unload HKLM\DELL
del fixdell.txt
del /ah setupreg.hiv.log
echo Check output to see if there is any error.
pause

c.2) copy the previously downloaded subinacl.exe into c:\pebuilder3110a\i386.

c.3) run the fixdell.cmd script.

c.4) Create a folder fixdell in c:\pebuilder3110a\plugin
In c:\pebuilder3110a\plugin\fixdell create a text file named fixdell.inf with the following lines:
; Original file from Bart
; Updated 2009-12-01

[Version]
Signature= "$Windows NT$"

[PEBuilder]
Name="Fix Dell Windows XP OEM boot problems"
Enable=1

[SourceDisksFiles]
a320raid.sys=4,,4
aac.sys=4,,4
aarich.sys=4,,4
afamgt.sys=4,,4
ahcix86.sys=4,,4;required at least by Latitude D620 and D630
cercsr6.sys=4,,4
iastor.sys=4,,4
megasas.sys=4,,4
NvAtaBus.sys=4,,4
nvgts.sys=4,,4;required at least by Latitude D620 and D630
nvraid.sys=4,,4
nvrd32.sys=4,,4;required at least by Latitude D620 and D630
sisraid4.sys=4,,4;required at least by Latitude D620 and D630
symmpi.sys=4,,4

d) copy Macrim plugin
copy the Macrium folder from c:\reflect\PEBuilder to c:\pebuilder3110a\plugin

e) launch BartPE
Click on Plugin and use the button Activer to activate and unactivate a given plugin.
As it will reduce the ISO we will only activate the following:
[Sherpya] Windows XPE
[Sherpya] Windows XPE: Crypto
[Sherpya] Windows XPE: Defaults
[Sherpya] Windows XPE: DirectX
[Sherpya] Windows XPE: Help System
[Sherpya] Windows XPE: MMC
[Sherpya] Windows XPE: PnP & Multimedia
[Sherpya] Windows XPE: WMI
Check Disk (chkdsk.exe)
Customize
Fix Dell Windows XP OEM boot problems
Keyboard Layout (read PluginHelp)
Macrium Reflext
Nu2Menu
RAMDisk [Nu2 Productions]
Serial Mouse

Unactivate the other ones.

Then configure :
- the Source directory to point to your c:\pebuilder3110a folder
- the output directory where the environment will be generated (this is a temporary directory)
- select "create an ISO image" (or créer une Image ISO)

TS101: gcc compiler

2010-09-11T02:18:00.000-07:00

TS101 is a powerpc based architecture.
There is a mini-native-powerpc compiler that one can find at uClibc that will work for the TS101.
I tested the 0.9.30 (click here to get it), at the time of this post there is also a 0.9.30.1.
Note that this post uses information and directory structure from previous posts.
When you have the tar ball on your NAS, log as administrator and do the following commands:
mkdir /mnt/HDA_ROOT/mini-native-powerpc
cd /mnt/HDA_ROOT/mini-native-powerpc
tar -xjf mini-native-powerpc.tar.bz2

edit /share/HDA_DATA/adminfiles/addprofile.sh2 and add the following lines:
/bin/echo "export PATH=$PATH:/mnt/HDA_ROOT/mini-native-powerpc/bin" >> /etc/profile

This sh2 file will be automatically called at next startup of the TS (see previous post).
So restarting the TS will give you a gcc compiler.

TS101: improve ethernet performances

2010-09-10T14:00:00.000-07:00

This post is using information from the SpeedGuide.net that should improve the performance of the ethernet link of the TS.
This configuration was tested and is working but I have not done any bench to check if I really get improvement. Anyway, it should not be worth. ;-)
Assuming you are using the previous post method for having .sh2 files called at startup,
create a file named /share/HDA_DATA/adminfiles/config_ethernet.sh2
add the following lines in it:
ifconfig eth0 txqueuelen 50000
echo 1 > /proc/sys/net/ipv4/tcp_rfc1337
echo 2 > /proc/sys/net/ipv4/tcp_frto
echo 1 > /proc/sys/net/ipv4/tcp_window_scaling
echo 1 > /proc/sys/net/ipv4/tcp_tw_reuse
echo 0 > /proc/sys/net/ipv4/tcp_tw_recycle
echo 1 > /proc/sys/net/ipv4/tcp_low_latency
echo 1 > /proc/sys/net/ipv4/tcp_ecn
echo 0 > /proc/sys/net/ipv4/tcp_timestamps
echo 1 > /proc/sys/net/ipv4/tcp_sack
echo 1 > /proc/sys/net/ipv4/tcp_window_scaling

TS101:setup an autorun.sh on your harddrive

2010-09-10T08:40:00.000-07:00

The autorun.sh is a script used by the TS to initialise and launch application at startup.
This file is unfortunatly located in an ext2 filesystem in the flash memory of the TS.
This filesystem is unmounted after bootup. Modifying this file can then endanger the system if you forget to unmount it after modification.
To overcome this issue, I propose to setup a file on your harddrive that will be called by autorun.sh at startup.

1. create the file on your harddrive

mkdir /share/HDA_DATA/adminfiles
cd /share/HDA_DATA/adminfiles
touch myautorun.sh
chmod +x myautorun.sh

2. call myautorun.sh from autorun.sh
(see previous post for editing autorun.sh)
add the following lines:
#call user script located on harddrive
[ -f /share/HDA_DATA/adminfiles/myautorun.sh ] && /share/HDA_DATA/adminfiles/myautorun.sh

Don't forget to unmount /tmp/config after modifications

3. I propose that you give your myautorun.sh the following lines:
#!/bin/sh
echo "Starting myautorun script:"
for file in /share/HDA_DATA/adminfiles/*.sh2;
do
echo $file;
[ -f $file ] && $file;
done

Then any file you may add in this directory with extension .sh2 will be executed at startup

Install SVN client to TS 101

2010-09-09T14:12:00.000-07:00

1. Update the list of available packages
/opt/bin/ipkg update

2. Install svn
/opt/bin/ipkg install svn

This actually also install some svn server stuff, but there is some more configuration to be done to have an svn server up and running. Have to check if this is possible over apache...

optware Installation on TS-101

2010-09-09T13:31:00.000-07:00

With optware installed on the turbostation you can manage additional packages easily.
This post follows the post from glomde but add some corrections while doing it at the same time.

1. Change /opt location
a. create a /mnt/HDA_ROOT/opt directory
mkdir /mnt/HDA_ROOT/opt

b. edit your autorun.sh (see previous post for that) and add the following lines
if [ -d /opt ]; then rmdir /opt; fi
if [ -e /opt ]; then true; else ln -sf /mnt/HDA_ROOT/opt /opt

c. reboot your TS

2. Get and install latest ipkg. Currently the below is latest
a. find the name of the lattest version
cd /tmp
wget http://ipkg.nslu2-linux.org/feeds/optware/ts101/cross/unstable
grep "ipkg-opt" index.html | grep "powerpc\.ipk"

at the time of this post, the result was ipkg-opt_0.99.163-10_powerpc.ipk

b. download it
wget http://ipkg.nslu2-linux.org/feeds/optware/ts101/cross/unstable/ipkg-opt_0.99.163-10_powerpc.ipk
tar -zxOf ipkg-opt_0.99.163-10_powerpc.ipk ./data.tar.gz | tar -zx -C /

c. Configure it
echo src ts101a http://sources.nslu2-linux.org/feeds/optware/ts101/cross/stable/ >> /opt/etc/ipkg.conf
echo src ts101b http://ipkg.nslu2-linux.org/feeds/optware/ts101/cross/unstable >> /opt/etc/ipkg.conf

3. Backitup
cd /share/Public
tar -cf backup.tar /opt

MySQL on TS101

2009-07-22T02:48:00.000-07:00

There is a pre-compiled MySQL that can be downloaded from site1 or from site2.
I Guess both are the same, I personally tested the one from site2.

The installation procedure is well described in a README included in the delivery, but when you reboot your TS station, MySQL will not be launched.

This is due to some symbolic links not kept at restart, because written in a filesystem placed in the RAM.

Edit the autostart.sh file and add the following lines before the line added by MySQL installation script :

cd /share
ln -s /share/HDA_DATA/MySQL
export PATH=$PATH:/share/MySQL/bin

This will allow MySQL to work correctly after reboot, and thus to be launched automatically after reboot.

Blog list and links about TS101

2009-07-22T02:42:00.000-07:00

This is a list of blogs websites and Wikis about TS101

Launch application at TS-101 startup : autostart.sh

2009-07-22T02:25:00.000-07:00

On Turbo Station TS101, there is a hidden partition that one can mount to access a file (autorun.sh) that is launched and executed at each startup.

To mount this hidden partition :
1- Log to you TS station with administrator login
Note : To be able to do that you must have a updated firmware from QNAP.
The one you have when you receive the TS does not allow telnet services.
When the firmware is flashed, then telnet service is connected to port 13131.
When it is installed you access your QNAP with the following command (on windows console):
telnet ip_of_your_qnap 13131

2- execute the following to mount the hidden partition
mount -t ext2 /dev/mtdblock4 /tmp/config

3- Then edit the autorun.sh file :
vi /tmp/config/autorun.sh

4- If you are as paranoid as me, then sync and unmount your file system after modification:
sync
umount /dev/mtdblock4

Be careful, this file system is in the flash, so :
- do not put big files in it
- do not erase files in it
- use as much care as possible to not destroying it

Information on other TS can be found here

Internet Time with NTP protocol on windows XP

2008-10-26T03:43:00.000-07:00

To be able to use NTP on windows few actions has to be made :
1 - open a command prompt and type :
regsvr32 w32time.dll

2 - activate the service :
windows time

3 - open the Adjust date and time properties and select the tab "Internet Time"
There, check the box "Automatically synchronize with an Internet time server"
In the server list you may use the Windows server :
time.windows.com

4 - press the "Update Now" button to check that the configuration is working.

CVS : How to checkout a branch

2008-08-21T05:03:00.000-07:00

Checkout a branch version :
cvs checkout -r branch_name

Then an update can be done as usual :
cvs update

An alternative way is to update an existing checkout to switch to a branch :
cvs update -r branch_name

CVS : How to create a branch

2008-08-21T05:01:00.000-07:00

Checkout the root version.

Put a Tag at that root from where to create the branch :
cvs tag -F -R root_branch_name

Create the branch from that tag :
cvs tag -b -r root_branch_name branch_name

How to change the way how LaTeX gives numbers to chapters, sections...

2008-08-19T03:21:00.001-07:00

The best way is to use a renewcommand.
For instance :
\renewcommand\thesection{\arabic{part}.\arabic{chapter}.\arabic{section}}

gives 1.1.1 numbers to sections while :
\renewcommand\thesection{\arabic{section}}

will only display the section number of the given section without chapter or part information.

this can be used to redefine :
thechapter, thesection, thesubsection, and so on...

Clipboard between Linux and Windows using GDM

2008-08-13T03:12:00.001-07:00

When using GDM on linux, one may receive the following find of message :

winProcSetSelectionOwner - OpenClipboard () failed: 00000578

This can be solved with setting in the [daemon] section :
KillInitClients=false

How to install SAMBA on a red Hat entreprise linux 5

2008-08-11T09:56:00.001-07:00

Log as root on the PC.

if samba is not yet installed use the console to install the packages, for instance with yum :
> yum install samba
> yum install samba-common
> yum install samba-client
> yum install samba-swat
> yum install system-config-samba

How to configure samba :

Edit the file [/etc/samba/smb.conf]
In the [global] section add those lines :

workgroup = <your workgroup name>
guest account = nobody

then, in the same file, after section [homes], add your own shared directories for instance :

[shared]
comment = my shared directory
path = /the/path/to/the/directory
read only = no
public = yes
writable = yes
share modes = yes

write and close the smb.conf file.
Edit the file /etc/samba/smbusers and add the users you want to have access through samba,
with the format : linux_user_name = samba_user_name
I suggest that you give the same name for the samba user_name and for the linux user_name.
For instance :

user1 = user1
user2 = user2

write and close the smbusers file.
Usnig the shell prompt, execute the following command :

cat /etc/passwd | mksmbpasswd.sh > /etc/samba/smbpasswd

This should create a password file for samba users.
When this does not gives you access to your samba account, then use the following for each user :

smbpasswd -a <username to add>

you will be prompted for a password for the specified user.

NOTE 1 : After modification of a configuration file, restard the smb daemon.
NOTE 2 : The shared directory is a place where one may setup symbolic links so that only logging to this directory gives you access to any desired directory
NOTE 3 : SElinux is a kind of firewall, and your should configure it prior to use samba because it can prevent samba from running proprely, best to check your samba configuration is to disable SElinux.
NOTE 4 : The installed packages also include a graphical user interface to configure samba.

To mount a samba drive in windows :
- open explorer
- [Tools]->[Connect to a network harddrive]
- select a letter
- give the folder name : \\computer_name\shared_directory (e.g. \\mypc\shared)
- check the box "Reconnect at loggon"
- Click on "Connect with a different user name"
- Give a login and password for the shared directory you just defined with samba
- click OK twice.

To mount a samba drive in linux :
- smbclient //computer_name/shared_directory -U samba_user_name
- you will be prompted for a password.

Using CentOS 5 packages with yum

2008-08-03T14:52:00.000-07:00

Create the following new file : /etc/yum.repos.d/centos.repos
and put the following lines in it :

[CentOS5 base]
name=CentOS-5-Base
mirrorlist=http://mirrorlist.centos.org/?release=5&arch=$basearch&repo=os
gpgcheck=0
enabled=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-5

[CentOS5 updates]
name=CentOS-5-Updates
mirrorlist=http://mirrorlist.centos.org/?release=5&arch=$basearch&repo=updates
gpgcheck=0
enabled=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-5

[CentOS5plus]
name=CentOS-5-Plus
mirrorlist=http://mirrorlist.centos.org/?release=5&arch=$basearch&repo=centosplus
gpgcheck=0
enabled=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-5

Automatic SSH login (key based authorization)

2008-07-16T05:44:00.001-07:00

The purpose is to be able to connect on an SSH based server without having to enter a password at each stage. For instance a CVS server using SSH will ask you to enter your password at each command you enter.

Steps for having such an automatic connection :
1. Log on the client computer with the account used to access the server. (e.g. local_user)
Note, we consider here that you have a local_user account on your client, and a remote_user account on the server. Using local_user on client, you should currently connect the server using : ssh remote_user@remoteserver Which should prompt you for a password.

2. On the client, create a public key :
ssh-keygen -t dsa

3. Append the generated public key to the remote server
cat ~/.ssh/id_dsa.pub | ssh -l remote_user remoteserver 'cat >> ~/.ssh/authorized_keys'

4. Access from client to server
ssh remote_user@remoteserver

5. You should not be prompted

Philippe MILLET

2006-05-01T02:42:00.000-07:00

Philippe MILLET
This is my first Edito.
I am currently in charge of parallel computer architecture modification in my company.
This company has been working up to now with private technology.
If looking twenty years ago, we were doing everything :
software, boards, but chips also (mainly ASGa).
My company busines is not computer but big systems,
anyway, there were no choice in this time because COTS was not at a level high enough for the realtime constraint required.
Time going on, chips from the market became more sofisticated and it was not possible to continue to develop the activity at a correct level, so the company started to buy standard chips to build up there own parallel computer architecture based on DSPs and FPGAs.
From this time, the middleware is completly "home made" from C language, with "home made" message formats, "home made" interrupt routines... Everything "home made".
Off course, it correspond well to the architecture, may be too well, but it is now old enough (16 years) for the company to ask it-self the question :
Is there enough interrest for the company to continue developping such a middleware or is there any at the market which could be used for our business.
I am thus investigating technology which could be used to serve us.
While I will learn and train, I will maintain this blog to keep ideas of my researches.