From pkerrwall at psu.edu  Thu Feb  6 12:39:25 2003
From: pkerrwall at psu.edu (Kerr Wall)
Date: Thu, 06 Feb 2003 12:39:25 -0500
Subject: [Bio-linux] Clone Questions
Message-ID: <BA6807FD.7C21%pkerrwall@psu.edu>

Hi Milo et all,

First off, I'd like to again thank you guys for such a wonderful machine.  I
couldn't be happier with your clone.

We just hired a new computer science student that will be helping with
system administration and programming.  His name is Siddharth Ray and his
email is rays at psu.edu.  I would like to have his name added to the list and
let you guys know that he might be shooting you emails from time to time as
he learns the administration on the clone.

The clone that I have running here has 2 80 GB hard drives.  I am wondering
what commands that I need to use to format the second hard drive (It came
with Windows XP so I think it would be formatted in FAT32).  We would like
to use reiserfs to allow for large files and would like to have it mounted
on '/usr/db' and set up its configuration to automatically mount when the
machine boots.

We have two other Dell machines that are nearly identical to the ones you
guys have in Oxford.  I want to clone both of these machines.  The first one
that I want to clone has a bad floppy drive.  The machine is running Windows
2000 right now and I have a zip drive attached through USB.  Can we still
clone without a floppy drive?  If so, I would like to do this next week.  I
still have the CD that I made back in October/November.  Once I clone this
machine, I will then clone the second machine that has a working floppy
drive.  We are cloning all of our linux machines in the lab to hopefully
make administration easier.

Two other issues that I'd like to get your thoughts on.  We are wanting to
set up one of the machines as a print server.  Do you guys have any thoughts
on this.  We need to set it up to allow for windows, mac os x, and other
linux (clones) to print to a laser jet printer and color ink jet printer
that I would assume would be attached to one of the clones.  Our previous
computer science dude set up cups, samba, and appletalk on a suse linux
computer that worked fine for about a month but I have been unable to
resolve the current issues.  He didn't leave any documentation on how he set
it up and I don't want to use Suse anymore.  Therefore, we will be starting
from scratch again.

We are in the process of resolving our backup problems and have gone to a
group here on campus that will automatically login every night to one of the
machines to backup.  We would like to set up all of the clones and me and
Jim's os x laptops in a  Network File System to allow for us to easily share
resources and get important files to the backup machine.  Do you guys have
any thoughts on NFS?  I am really tired of having to ssh to each of the
machines or scp things from my laptop to other computers and thought that
NFS would help remedy the situation.  But if this is too complicated, I
understand.

Thanks again,

Kerr
 



From dfield at ceh.ac.uk  Thu Feb  6 12:45:51 2003
From: dfield at ceh.ac.uk (Dawn Field)
Date: Thu, 06 Feb 2003 17:45:51 +0000
Subject: [Bio-linux] Clone Questions
Message-ID: <se429f69.072@wpo.nerc.ac.uk>

Hi Kerr,

Glad to hear you are using the Bio-Linux machine and have some help on the sys admin side! 

Dan, our new EGTDC Linux sysadmin is now here (to introduce you - you'll meet in April when you visit).  Very importantly he is now officially maintaining the Bio-Linux FAQ.

Dan - are you already on the Bio-Linux list Milo started?  Kerr's questions, and any answers you provide would be great material for an "Advanced Bio-Linux Technical FAQ" that we could start writing.  If we start from today framing everything in the form of Q&A we'll very quickly hit a lot of the key issues.

cheers,
Dawn

>>> Kerr Wall <pkerrwall at psu.edu> 2/6/03 17:39:25 >>>
Hi Milo et all,

First off, I'd like to again thank you guys for such a wonderful machine.  I
couldn't be happier with your clone.

We just hired a new computer science student that will be helping with
system administration and programming.  His name is Siddharth Ray and his
email is rays at psu.edu.  I would like to have his name added to the list and
let you guys know that he might be shooting you emails from time to time as
he learns the administration on the clone.

The clone that I have running here has 2 80 GB hard drives.  I am wondering
what commands that I need to use to format the second hard drive (It came
with Windows XP so I think it would be formatted in FAT32).  We would like
to use reiserfs to allow for large files and would like to have it mounted
on '/usr/db' and set up its configuration to automatically mount when the
machine boots.

We have two other Dell machines that are nearly identical to the ones you
guys have in Oxford.  I want to clone both of these machines.  The first one
that I want to clone has a bad floppy drive.  The machine is running Windows
2000 right now and I have a zip drive attached through USB.  Can we still
clone without a floppy drive?  If so, I would like to do this next week.  I
still have the CD that I made back in October/November.  Once I clone this
machine, I will then clone the second machine that has a working floppy
drive.  We are cloning all of our linux machines in the lab to hopefully
make administration easier.

Two other issues that I'd like to get your thoughts on.  We are wanting to
set up one of the machines as a print server.  Do you guys have any thoughts
on this.  We need to set it up to allow for windows, mac os x, and other
linux (clones) to print to a laser jet printer and color ink jet printer
that I would assume would be attached to one of the clones.  Our previous
computer science dude set up cups, samba, and appletalk on a suse linux
computer that worked fine for about a month but I have been unable to
resolve the current issues.  He didn't leave any documentation on how he set
it up and I don't want to use Suse anymore.  Therefore, we will be starting
from scratch again.

We are in the process of resolving our backup problems and have gone to a
group here on campus that will automatically login every night to one of the
machines to backup.  We would like to set up all of the clones and me and
Jim's os x laptops in a  Network File System to allow for us to easily share
resources and get important files to the backup machine.  Do you guys have
any thoughts on NFS?  I am really tired of having to ssh to each of the
machines or scp things from my laptop to other computers and thought that
NFS would help remedy the situation.  But if this is too complicated, I
understand.

Thanks again,

Kerr
 

_______________________________________________
Bio-linux mailing list
Bio-linux at ivsun01.nerc-oxford.ac.uk 
http://www.bioinf.ceh.ac.uk/mailman/listinfo/bio-linux 









From dswan at ceh.ac.uk  Mon Feb 10 06:36:36 2003
From: dswan at ceh.ac.uk (Dr Dan Swan)
Date: 10 Feb 2003 11:36:36 +0000
Subject: [Bio-Linux] Re: Clone Questions (Kerr Wall)
In-Reply-To: <200302071201.h17C11a20355@ivsun01.nerc-oxford.ac.uk>
References: <200302071201.h17C11a20355@ivsun01.nerc-oxford.ac.uk>
Message-ID: <1044876995.19086.370.camel@bioinf4.nox.ac.uk>

On Fri, 2003-02-07 at 12:01, bio-linux-request at bioinf.ceh.ac.uk wrote:

> Hi Milo et all,

Hi Kerr, and everyone else on the list.  I would just like to introduce
myself.  My name is Dan and I have taken over the Bio-Linux project from
Milo as of last week.  This includes the development of Bio-Linux v2.0,
maintenance of all Bio-Linux documentation and mailing list
administration and support.

> We just hired a new computer science student that will be helping with
> system administration and programming.  His name is Siddharth Ray and his
> email is rays at psu.edu.  I would like to have his name added to the list and
> let you guys know that he might be shooting you emails from time to time as
> he learns the administration on the clone.

I have manually subscribed Siddharth to the list.  If he would like to
change to a digest form of the Bio-Linux list then he can contact me
privately at dswan at ceh.ac.uk and I will make the necessary changes.
For future people subscribing there is a mailman interface for the
Bio-Linux list at http://www.bioinf.ceh.ac.uk/mailman/listinfo/bio-linux

> The clone that I have running here has 2 80 GB hard drives.  I am wondering
> what commands that I need to use to format the second hard drive (It came
> with Windows XP so I think it would be formatted in FAT32).  We would like
> to use reiserfs to allow for large files and would like to have it mounted
> on '/usr/db' and set up its configuration to automatically mount when the
> machine boots.

There is a comprehensive article on managing the addition of further
hard drives at http://www.linuxplanet.com/linuxplanet/tutorials/4232/1/
The process is relatively simple (fdisk, mkfs, and additions to
/etc/fstab).  The document includes information on how to set up the
ReiserFS.  If you have any problems, please don't hesitate to get in
touch.

> We have two other Dell machines that are nearly identical to the ones you
> guys have in Oxford.  I want to clone both of these machines.  The first one
> that I want to clone has a bad floppy drive.  The machine is running Windows
> 2000 right now and I have a zip drive attached through USB.  Can we still
> clone without a floppy drive?  If so, I would like to do this next week.  I
> still have the CD that I made back in October/November.  Once I clone this
> machine, I will then clone the second machine that has a working floppy
> drive.  We are cloning all of our linux machines in the lab to hopefully
> make administration easier.

No, you can't clone without a floppy drive - can I suggest that you take
a working floppy drive from one of the machines in which it is known to
work, and simply use it during the installation process?  You can swap
it back out after the installation if theres one machine you absolutely
must have a working floppy drive in.

Please note that for the installation I will still require all the
information passed to us for the installation of the initial clone
(network, computer name, IP, DNS servers etc etc.) as I will need this
a) for our records and b) to prepare a configuration floppy for you (I
can send you this via email, all you will need to do is write the config
files to a floppy disk).

> Two other issues that I'd like to get your thoughts on.  We are wanting to
> set up one of the machines as a print server.  Do you guys have any thoughts
> on this.  We need to set it up to allow for windows, mac os x, and other
> linux (clones) to print to a laser jet printer and color ink jet printer
> that I would assume would be attached to one of the clones.  Our previous
> computer science dude set up cups, samba, and appletalk on a suse linux
> computer that worked fine for about a month but I have been unable to
> resolve the current issues.  He didn't leave any documentation on how he set
> it up and I don't want to use Suse anymore.  Therefore, we will be starting
> from scratch again.

I am not aware of cups (http://www.cups.org/) I'm afraid. Samba
(http://us2.samba.org/samba/samba.html) is relatively simple to set up,
netatalk (http://rsug.itd.umich.edu/software/netatalk.html) (the
appletalk implementation) is a little more tricky (or it was when I last
did it a couple of years ago).

Does the laserjet have an ethernet card in it?  If so it might be easier
to set it up plugged into the network, than trying to set up a print
server on a Linux clone with the printer attached.  Drivers for network
printing with Mac/Win platforms would be trivial to install, and I have
previously used rlpr (http://freshmeat.net/projects/rlpr/?topic_id=154)
for network printing under Linux with a variety of laser printers.

> We are in the process of resolving our backup problems and have gone to a
> group here on campus that will automatically login every night to one of the
> machines to backup.  We would like to set up all of the clones and me and
> Jim's os x laptops in a  Network File System to allow for us to easily share
> resources and get important files to the backup machine.  Do you guys have
> any thoughts on NFS?  I am really tired of having to ssh to each of the
> machines or scp things from my laptop to other computers and thought that
> NFS would help remedy the situation.  But if this is too complicated, I
> understand.

NFS is again, relatively simple to set up.  However it is a security
concern if set up incorrectly.  I am unsure as to why you would want all
the clones to export NFS file systems (or maybe thats not what you're
suggesting!), what you need to do is set up the backup server as an NFS
file server, and then the clones would act as NFS clients, so any data
you wanted shared would be placed in an NFS mounted directory.  The
NFS-HOWTO (http://nfs.sourceforge.net/nfs-howto/)is the definitive
guide, and I have set up NFS so many times that if you need any help
just ask.  We will need to secure access to any NFS shares on a per-IP
basis.  Exporting shares to the world will make a 14 year old computer
hackers day...  

I hope this is of some use,

regards,

Dan

-- 
Dr Dan Swan - Linux Developer/Bioinformatician
EGTDC, CEH, Mansfield Road, Oxford, OX1 3SR
Tel: 01865 281 976 Fax: 01865 281 696
http://envgen.nox.ac.uk/ | dswan at ceh.ac.uk



From mith at ceh.ac.uk  Mon Feb 10 06:01:16 2003
From: mith at ceh.ac.uk (Milo Thurston)
Date: Mon, 10 Feb 2003 11:01:16 +0000
Subject: [Bio-Linux] Re: Clone Questions (Kerr Wall) 
In-Reply-To: Your message of "10 Feb 2003 11:36:36 GMT."
             <1044876995.19086.370.camel@bioinf4.nox.ac.uk> 
Message-ID: <200302101101.h1AB1GP3007285@bioinf1.nox.ac.uk>

In message <1044876995.19086.370.camel at bioinf4.nox.ac.uk> you wrote:
>I am not aware of cups (http://www.cups.org/) I'm afraid. Samba

I have been using CUPS on a couple of machines. It is relatively simple,
and has an administrative interface accessed through a web browser
(at http://localhost:631/) that also has full user documentation.
My suggestion would be to use Red Hat's lprng implimentation to set
up a print server (/sbin/printconf-gui is good), and only use CUPS
if you are printing from non-Red Hat machines.
Milo.

--
Milo Thurston, CEH Oxford, Mansfield Road, Oxford, OX1 3SR.
'phone 01865 281658,  fax 01865 281696.
http://www.bioinf.ceh.ac.uk/lab/


From sur123 at psu.edu  Tue Feb 18 13:55:37 2003
From: sur123 at psu.edu (Siddharth Ray)
Date: Tue, 18 Feb 2003 13:55:37 -0500 (EST)
Subject: [Bio-Linux] Cylinder size
Message-ID: <200302181855.h1IItb127347@webmail2.cac.psu.edu>

This is a totally technical question. The cylinder size has been set to 9726
while it should be 1024 (ideally). Does it make much of a difference??? I hope
there won't be any portability problem etc...
-Siddharth
"Keep me away from the wisdom which does not cry, the philosophy which does not laugh and the greatness which does not bow before children."
- Kahlil Gibran

MS, Department of Computer Sc. & Engg,
Pond Lab, State College,
PA - 16802, USA.


From sur123 at psu.edu  Fri Feb 21 11:16:48 2003
From: sur123 at psu.edu (Siddharth Ray)
Date: Fri, 21 Feb 2003 11:16:48 -0500 (EST)
Subject: [Bio-Linux] (no subject)
Message-ID: <200302211616.LAA14073@webmail3.cac.psu.edu>

Siddharth Ray wrote:

I am facing the following problems :

1.)  I used fdisk on /dev/hda and got the following:
     (This machine happens to be a Dell cloned with biolinux)

The number of cylinders for this disk is set to 9726.
There is nothing wrong with that, but this is larger than 1024,
and could in certain setups cause problems with:
1) software that runs at boot time (e.g., old versions of LILO)
2) booting and partitioning software from other OSs
(e.g., DOS FDISK, OS/2 FDISK)

For /dev/hdb it is showing 16127 unallocated sectors (using v of fdisk)
p shows /dev/hdb1  start->1 End->9725 Blocks->78116031 Id->7
System->HPFS/NTFS

Another error message which was generated was that:
The cylinders are overlapping...

The cylinder size is set at the time of loading linux. However, is there any
specific reason for setting the cylinder size to 9726. I think the default is
1024 and at least it can be a multiple of that. I know a couple of systems
having cylinder size other than 1024 but the main thing that worries me is the
cylinder overlapping and stuff. Another thing which I wanted to point to is
that the machine which was cloned had 2 hard disks, but the 2nd hard disk was
never mounted. Hence, if the linux system is being loaded via CD etc then it
should acknowledge the 2nd hard disk.

I am not exactly worried as of now of the 9726 thing but the error message which
is generated.


2.) Is it that biolinux can only be installed on a Dell machine. Is it company
specific because to my knowledge I have never heard of any OS to be company
specific unless their architecture is drastically different. I am new to this
lab and the notion that people are having is that only Dell machines can be
cloned. 



Any suggestions...
-Siddharth
"Keep me away from the wisdom which does not cry, the philosophy which does not laugh and the greatness which does not bow before children."
- Kahlil Gibran

MS, Department of Computer Sc. & Engg,
Pond Lab, State College,
PA - 16802, USA.


From dswan at ceh.ac.uk  Mon Feb 24 05:42:28 2003
From: dswan at ceh.ac.uk (Dr Dan Swan)
Date: 24 Feb 2003 10:42:28 +0000
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray)
In-Reply-To: <200302221201.h1MC10a13989@ivsun01.nerc-oxford.ac.uk>
References: <200302221201.h1MC10a13989@ivsun01.nerc-oxford.ac.uk>
Message-ID: <1046083348.1566.91.camel@bioinf4.nox.ac.uk>

On Sat, 2003-02-22 at 12:01, bio-linux-request at bioinf.ceh.ac.uk wrote:

Hi Siddharth,

> I am facing the following problems :
> 
> 1.)  I used fdisk on /dev/hda and got the following:
>      (This machine happens to be a Dell cloned with biolinux)
> The number of cylinders for this disk is set to 9726.
> There is nothing wrong with that, but this is larger than 1024,
> and could in certain setups cause problems with:

I have to say I'm rather confused about what is going on here.  The 1024
cylinder limitation is a rather historic point in Linux these days. Ever
since LILO supported Logical Block Addressing, this has been somewhat of
a non-issue.  Some people still like to keep the boot partition under
1024 cylinders but there is no need to do this.

> 1) software that runs at boot time (e.g., old versions of LILO)

If this is a Bio-Linux clone - why is it running LILO?  We have chosen
the far more flexible GRUB bootloader for Bio-Linux.  Have you changed
this on your clone?

> 2) booting and partitioning software from other OSs
> (e.g., DOS FDISK, OS/2 FDISK)

Bio-Linux was not developed with the intention of having dual boot
systems.  It is meant to be a dedicated Linux workstation and we
recommend that if people wish to run additional OS's should use VMWARE
within Bio-Linux.  

> For /dev/hdb it is showing 16127 unallocated sectors (using v of fdisk)
> p shows /dev/hdb1  start->1 End->9725 Blocks->78116031 Id->7
> System->HPFS/NTFS

None of the Bio-Linux clones have a /dev/hdb as far as I know.  Have you
attempted to add and configure a second hard drive on your Bio-Linux
clone?  I assume you have - are you working with OS/2 (from the HPFS
clue)

> Another error message which was generated was that:
> The cylinders are overlapping...

Is this on /dev/hda? or /dev/hdb?  To be honest Linux doesn't actually
actually have an alignment requirement for cylinders, and its more
likely an error from over fussy partioning software. NT can handle
non-aligned partions as well.  The only OS I can think of that really
doesn't like it is DOS - hence DOS based partition tools can be very
particular.

> The cylinder size is set at the time of loading linux. However, is there any
> specific reason for setting the cylinder size to 9726. I think the default is
> 1024 and at least it can be a multiple of that. I know a couple of systems
> having cylinder size other than 1024 but the main thing that worries me is the
> cylinder overlapping and stuff. Another thing which I wanted to point to is
> that the machine which was cloned had 2 hard disks, but the 2nd hard disk was
> never mounted. Hence, if the linux system is being loaded via CD etc then it
> should acknowledge the 2nd hard disk.

I see.  Have you added a second hard drive to a Linux machine before? 
Until you tell Linux it its there, it may autodetect its presence but it
will never be utilised.  The Bio-Linux v1.0 clones have a single hard
drive.

> 2.) Is it that biolinux can only be installed on a Dell machine. Is it company
> specific because to my knowledge I have never heard of any OS to be company
> specific unless their architecture is drastically different. I am new to this
> lab and the notion that people are having is that only Dell machines can be
> cloned. 

No that's not strictly true, but it certainly helps.  Do you know how
Bio-Linux is made and installed?  It's not a standard Linux
distribution, its a system image - therefore its preconfigured to
certain aspects of the Dell hardware its developed on - especially the
video cards/NICs.  If you attempt to install it on other hardware we
cannot make any guarantees that it will work.  We have managed to put it
onto Dell Laptops which only required X to be configured.

Hope this helps,

Dan

-- 
Dr Dan Swan - Bio-Linux Developer
EGTDC, CEH, Mansfield Road, Oxford, OX1 3SR
Tel: 01865 281 630 Fax: 01865 281 696
http://envgen.nox.ac.uk/ | dswan at ceh.ac.uk



From sur123 at psu.edu  Mon Feb 24 17:51:05 2003
From: sur123 at psu.edu (Siddharth Ray)
Date: Mon, 24 Feb 2003 17:51:05 -0500 (EST)
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray)
Message-ID: <200302242251.RAA04905@webmail9.cac.psu.edu>

Hello Dr. Swan,
I now understand the need for Dell machines. The main part is the X-windows
configuration which needs very specific system information. Anyways, I feel
that two changes can be incorporated to make biolinux highly portable:
1. Machine independence - I think the X-windows can be configured on the fly.
2. Hard drives -  The main problem I think is; when cloning was done the
computer already had 2 hard drives and earlier Windows XP was installed, I
think may be for that reason its showing the cylinder overlapping problem. Some
changes should be made so that the kernel properly mounts all the hard drives.
I know giving ideas is easy and implementing it is twice as tough but I think
its doable. Anyways I also plan to work on kernel level programming and might
be able to help in some aspects.

Thanks,
Siddharth Ray















On Mon, 24 Feb 2003 10:42:28 +0000, Dr Dan Swan wrote:

> On Sat, 2003-02-22 at 12:01, bio-linux-request at bioinf.ceh.ac.uk wrote:
> 
> Hi Siddharth,
> 
> > I am facing the following problems :
> > 
> > 1.)  I used fdisk on /dev/hda and got the following:
> >      (This machine happens to be a Dell cloned with biolinux)
> > The number of cylinders for this disk is set to 9726.
> > There is nothing wrong with that, but this is larger than 1024,
> > and could in certain setups cause problems with:
> 
> I have to say I'm rather confused about what is going on here.  The 1024
> cylinder limitation is a rather historic point in Linux these days. Ever
> since LILO supported Logical Block Addressing, this has been somewhat of
> a non-issue.  Some people still like to keep the boot partition under
> 1024 cylinders but there is no need to do this.
> 
> > 1) software that runs at boot time (e.g., old versions of LILO)
> 
> If this is a Bio-Linux clone - why is it running LILO?  We have chosen
> the far more flexible GRUB bootloader for Bio-Linux.  Have you changed
> this on your clone?
> 
> > 2) booting and partitioning software from other OSs
> > (e.g., DOS FDISK, OS/2 FDISK)
> 
> Bio-Linux was not developed with the intention of having dual boot
> systems.  It is meant to be a dedicated Linux workstation and we
> recommend that if people wish to run additional OS's should use VMWARE
> within Bio-Linux.  
> 
> > For /dev/hdb it is showing 16127 unallocated sectors (using v of fdisk)
> > p shows /dev/hdb1  start->1 End->9725 Blocks->78116031 Id->7
> > System->HPFS/NTFS
> 
> None of the Bio-Linux clones have a /dev/hdb as far as I know.  Have you
> attempted to add and configure a second hard drive on your Bio-Linux
> clone?  I assume you have - are you working with OS/2 (from the HPFS
> clue)
> 
> > Another error message which was generated was that:
> > The cylinders are overlapping...
> 
> Is this on /dev/hda? or /dev/hdb?  To be honest Linux doesn't actually
> actually have an alignment requirement for cylinders, and its more
> likely an error from over fussy partioning software. NT can handle
> non-aligned partions as well.  The only OS I can think of that really
> doesn't like it is DOS - hence DOS based partition tools can be very
> particular.
> 
> > The cylinder size is set at the time of loading linux. However, is there any
> > specific reason for setting the cylinder size to 9726. I think the default
is
> > 1024 and at least it can be a multiple of that. I know a couple of systems
> > having cylinder size other than 1024 but the main thing that worries me is
the
> > cylinder overlapping and stuff. Another thing which I wanted to point to is
> > that the machine which was cloned had 2 hard disks, but the 2nd hard disk
was
> > never mounted. Hence, if the linux system is being loaded via CD etc then it
> > should acknowledge the 2nd hard disk.
> 
> I see.  Have you added a second hard drive to a Linux machine before? 
> Until you tell Linux it its there, it may autodetect its presence but it
> will never be utilised.  The Bio-Linux v1.0 clones have a single hard
> drive.
> 
> > 2.) Is it that biolinux can only be installed on a Dell machine. Is it
company
> > specific because to my knowledge I have never heard of any OS to be company
> > specific unless their architecture is drastically different. I am new to
this
> > lab and the notion that people are having is that only Dell machines can be
> > cloned. 
> 
> No that's not strictly true, but it certainly helps.  Do you know how
> Bio-Linux is made and installed?  It's not a standard Linux
> distribution, its a system image - therefore its preconfigured to
> certain aspects of the Dell hardware its developed on - especially the
> video cards/NICs.  If you attempt to install it on other hardware we
> cannot make any guarantees that it will work.  We have managed to put it
> onto Dell Laptops which only required X to be configured.
> 
> Hope this helps,
> 
> Dan
> 
> -- 
> Dr Dan Swan - Bio-Linux Developer
> EGTDC, CEH, Mansfield Road, Oxford, OX1 3SR
> Tel: 01865 281 630 Fax: 01865 281 696
> http://envgen.nox.ac.uk/ | dswan at ceh.ac.uk
> 
> _______________________________________________
> Bio-Linux mailing list
> Bio-Linux at ivsun01.nerc-oxford.ac.uk
> http://www.bioinf.ceh.ac.uk/mailman/listinfo/bio-linux
> 
> 

"Keep me away from the wisdom which does not cry, the philosophy which does not laugh and the greatness which does not bow before children."
- Kahlil Gibran

MS, Department of Computer Sc. & Engg,
Pond Lab, State College,
PA - 16802, USA.


From mith at ceh.ac.uk  Tue Feb 25 05:06:00 2003
From: mith at ceh.ac.uk (Milo Thurston)
Date: Tue, 25 Feb 2003 10:06:00 +0000
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray) 
In-Reply-To: Your message of "Mon, 24 Feb 2003 17:51:05 EST."
             <200302242251.RAA04905@webmail9.cac.psu.edu> 
Message-ID: <200302251006.h1PA60P3017081@bioinf1.nox.ac.uk>

Dear Siddharth,
I hope that you don't mind me responding to your comments.

>The main part is the X-window system
>configuration which needs very specific system information. 

Indeed it does.

>1. Machine independence - I think the X-windows can be configured on the fly.

In order to configure X for a new system, there are a couple of methods. One
could run Xconfigurator manually (but it is nasty), or run "xfree86 --configure".
This second method is probably what you are after, but it is not foolproof and often
requires editing of /etc/X11/XF86Config after it has been run.
I think that a far simpler solution to the problem of X not working on different
hardware is to buy the appropriate hardware in the first place, or, run the
above commands yourself.

>Some
>changes should be made so that the kernel properly mounts all the hard drives.

The kernel's involvement in the mounting of drives is that it must have appropriate
drivers for the filesystem on the drive. If the drive has no filesystem on it, then
the kernel will not recognise it and the mount command will not succeed. In order
to use a second drive on the machine when the image does not account for it,
one must follow these steps:

1. Partition the drive. Fdisk is the necessary tool.
2. Format the partitions. "mke2fs -j /dev/hdb[insert partition number here]" is the
command you need.
3. Edit /etc/fstab to contain appropriate entries for the new partitions.

It's not possible to automate this process unless we know:

1. How big your disk is.
2. How many partitions you want on it, and of what filesystem.
3. Where you want the partitions mounted.

...and then modify the install script used and the image.
Again, rather than write a custom script for each unusual machine, it is
easier for the user to run the commands above (or use the correct hardware).
Of course, any of your programming assistance would be gratefully received.

I hope that this helps.
'bye,
Milo.

--
Milo Thurston, CEH Oxford, Mansfield Road, Oxford, OX1 3SR.
'phone 01865 281658,  fax 01865 281696.
http://www.bioinf.ceh.ac.uk/lab/


From john.parkinson at ed.ac.uk  Tue Feb 25 06:35:42 2003
From: john.parkinson at ed.ac.uk (John Parkinson)
Date: 25 Feb 2003 11:35:42 +0000
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray)
In-Reply-To: <200302251006.h1PA60P3017081@bioinf1.nox.ac.uk>
References: <200302251006.h1PA60P3017081@bioinf1.nox.ac.uk>
Message-ID: <1046172940.4480.6393.camel@nema>

Just to add my 2p's worth

Would it be possible to have a script run at inititialisation which asks
the user (or queries the system) for information on the hardware (Video,
Extra Disks etc) which if non-standard drops the user into single user
non-graphics mode so that the appropriate configuration executables can
be run. As it stands at the moment, it is very difficult to get to
single-user mode, which means you get stuck with an unconfigured blank
screen.

Before this happens the installation can tell the user what they need to
do and if this fails then point the user to the appropriate Linux forum.

John

On Tue, 2003-02-25 at 10:06, Milo Thurston wrote:
> Dear Siddharth,
> I hope that you don't mind me responding to your comments.
> 
> >The main part is the X-window system
> >configuration which needs very specific system information. 
> 
> Indeed it does.
> 
> >1. Machine independence - I think the X-windows can be configured on the fly.
> 
> In order to configure X for a new system, there are a couple of methods. One
> could run Xconfigurator manually (but it is nasty), or run "xfree86 --configure".
> This second method is probably what you are after, but it is not foolproof and often
> requires editing of /etc/X11/XF86Config after it has been run.
> I think that a far simpler solution to the problem of X not working on different
> hardware is to buy the appropriate hardware in the first place, or, run the
> above commands yourself.
> 
> >Some
> >changes should be made so that the kernel properly mounts all the hard drives.
> 
> The kernel's involvement in the mounting of drives is that it must have appropriate
> drivers for the filesystem on the drive. If the drive has no filesystem on it, then
> the kernel will not recognise it and the mount command will not succeed. In order
> to use a second drive on the machine when the image does not account for it,
> one must follow these steps:
> 
> 1. Partition the drive. Fdisk is the necessary tool.
> 2. Format the partitions. "mke2fs -j /dev/hdb[insert partition number here]" is the
> command you need.
> 3. Edit /etc/fstab to contain appropriate entries for the new partitions.
> 
> It's not possible to automate this process unless we know:
> 
> 1. How big your disk is.
> 2. How many partitions you want on it, and of what filesystem.
> 3. Where you want the partitions mounted.
> 
> ...and then modify the install script used and the image.
> Again, rather than write a custom script for each unusual machine, it is
> easier for the user to run the commands above (or use the correct hardware).
> Of course, any of your programming assistance would be gratefully received.
> 
> I hope that this helps.
> 'bye,
> Milo.
> 
> --
> Milo Thurston, CEH Oxford, Mansfield Road, Oxford, OX1 3SR.
> 'phone 01865 281658,  fax 01865 281696.
> http://www.bioinf.ceh.ac.uk/lab/
> _______________________________________________
> Bio-Linux mailing list
> Bio-Linux at ivsun01.nerc-oxford.ac.uk
> http://www.bioinf.ceh.ac.uk/mailman/listinfo/bio-linux



From mith at ceh.ac.uk  Tue Feb 25 07:57:34 2003
From: mith at ceh.ac.uk (Milo Thurston)
Date: Tue, 25 Feb 2003 12:57:34 +0000
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray) 
In-Reply-To: Your message of "25 Feb 2003 11:35:42 GMT."
             <1046172940.4480.6393.camel@nema> 
Message-ID: <200302251257.h1PCvYP3017482@bioinf1.nox.ac.uk>

In message <1046172940.4480.6393.camel at nema> you wrote:
>Would it be possible to have a script run at inititialisation which asks
>the user (or queries the system) for information on the hardware (Video,
>Extra Disks etc) 

At present, anyone installing on non-standard hardware will encounter
kudzu - Red Hat's hardware detection system. Assuming that one answers
yes to each question (I didn't find this, shall I remove the config.?, and
I found this, shall I configure it?) then all the changes in hardware
will be dealt with efficiently.
Kudzu won't: 
1. Write a new XF86Config file.
2. Run fdisk/mkfs/vi fstab for you.

>which if non-standard drops the user into single user
>non-graphics mode so that the appropriate configuration executables can
>be run. 

By default, the machine boots to run level 3, so X will not run. One
can then log in as manager and run xfree86 --configure as well as
the usual setup script, fdisk etc.

>As it stands at the moment, it is very difficult to get to
>single-user mode, 

By design, because it is not necessary to do so in order to set the machine
up.

>which means you get stuck with an unconfigured blank
>screen.

In some of the older releases this may have been the case, but before
handing it all over to Dan I changed things so that run level 3 is the
default. Thus, the machine will always be usable no matter how odd
the hardware you install it on.

>Before this happens the installation can tell the user what they need to
>do and if this fails then point the user to the appropriate Linux forum.

What the user needs to do is install on appropriate hardware ;-)
'bye,
Milo.

--
Milo Thurston, CEH Oxford, Mansfield Road, Oxford, OX1 3SR.
'phone 01865 281658,  fax 01865 281696.
http://www.bioinf.ceh.ac.uk/lab/


From john.parkinson at ed.ac.uk  Tue Feb 25 08:40:00 2003
From: john.parkinson at ed.ac.uk (John Parkinson)
Date: 25 Feb 2003 13:40:00 +0000
Subject: [Bio-Linux] Re: (no subject) (Siddharth Ray)
In-Reply-To: <200302251257.h1PCvYP3017482@bioinf1.nox.ac.uk>
References: <200302251257.h1PCvYP3017482@bioinf1.nox.ac.uk>
Message-ID: <1046180400.18188.37.camel@nema>

> At present, anyone installing on non-standard hardware will encounter
> kudzu - Red Hat's hardware detection system. Assuming that one answers
> yes to each question (I didn't find this, shall I remove the config.?, and
> I found this, shall I configure it?) then all the changes in hardware
> will be dealt with efficiently.
> Kudzu won't: 
> 1. Write a new XF86Config file.
> 2. Run fdisk/mkfs/vi fstab for you.

Sure, but this is what I was wondering if scripts could be created for.

> >which if non-standard drops the user into single user
> >non-graphics mode so that the appropriate configuration executables can
> >be run. 
> 
> By default, the machine boots to run level 3, so X will not run. One
> can then log in as manager and run xfree86 --configure as well as
> the usual setup script, fdisk etc.

Our recent experience saw the machine try to load the graphics login
page unsuccessfully and then gave a text login screen. Login to the
manager account was then not possible due to some error which appeared
and disappeared so quickly we never found out what it was - I assumed it
was something to do with a startup login script which conflicted with
the hardware (X-server ??)........

> >As it stands at the moment, it is very difficult to get to
> >single-user mode, 
> 
> By design, because it is not necessary to do so in order to set the machine
> up.

.....hence the need to drop into single user mode in none standard
machines.

> >which means you get stuck with an unconfigured blank
> >screen.
> 
> In some of the older releases this may have been the case, but before
> handing it all over to Dan I changed things so that run level 3 is the
> default. Thus, the machine will always be usable no matter how odd
> the hardware you install it on.

We tried our last install three weeks ago, was this one of the older
releases ?

> >Before this happens the installation can tell the user what they need to
> >do and if this fails then point the user to the appropriate Linux forum.
> 
> What the user needs to do is install on appropriate hardware ;-)

It certainly makes things easier :-)

John



From pkerrwall at psu.edu  Tue Feb 25 10:46:12 2003
From: pkerrwall at psu.edu (Kerr Wall)
Date: Tue, 25 Feb 2003 10:46:12 -0500
Subject: [Bio-Linux] SSH Question
Message-ID: <BA80F9F4.7ED4%pkerrwall@psu.edu>

Hi Milo and Dan,

I have a question concerning ssh.  I have received a request from the
programmer at Cornell who is responsible for getting FGP tracefiles from our
server in an automated fashion to switch from the openSSH server that came
with the clone to the comercial version at ssh.com.  He has stated that the
script he has written to automatically login to our server doesn't work with
openSSH and that it only works with the version from ssh.com.

Would you guys recommend this (ie, would this cause any dependency problems
or are there any security concerns that I don't know about)?  If so, what is
the best way to uninstall openSSH and then install SSH (I'm guessing that
using the rpm command will accomplish this)?  If not, do you have any
suggestions for him to get a working perl script to automatically login with
openSSH?

Thanks,

Kerr



From dswan at ceh.ac.uk  Tue Feb 25 11:23:22 2003
From: dswan at ceh.ac.uk (Dr Dan Swan)
Date: 25 Feb 2003 16:23:22 +0000
Subject: [Bio-Linux] SSH Question
In-Reply-To: <BA80F9F4.7ED4%pkerrwall@psu.edu>
References: <BA80F9F4.7ED4%pkerrwall@psu.edu>
Message-ID: <1046190202.25384.72.camel@bioinf4.nox.ac.uk>

On Tue, 2003-02-25 at 15:46, Kerr Wall wrote:

Hi Kerr,

> I have a question concerning ssh.  I have received a request from the
> programmer at Cornell who is responsible for getting FGP tracefiles from our
> server in an automated fashion to switch from the openSSH server that came
> with the clone to the comercial version at ssh.com.  He has stated that the
> script he has written to automatically login to our server doesn't work with
> openSSH and that it only works with the version from ssh.com.
> 
> Would you guys recommend this (ie, would this cause any dependency problems
> or are there any security concerns that I don't know about)?  If so, what is
> the best way to uninstall openSSH and then install SSH (I'm guessing that
> using the rpm command will accomplish this)?  If not, do you have any
> suggestions for him to get a working perl script to automatically login with
> openSSH?

So he wants you to change the OpenSSH daemon on your clone because he's
using a commercial version of the ssh client?  That's a bit unfair ;)

Its perfectly possible to install the commercial version of the ssh
daemon on the clone, although you will not be able to do it using an rpm
as it supplied as a source tarball.  There's no licencing issue as its
free for academics.  I would suggest however that your colleague
downloads the client/base rpms from www.openssh.org and installs those
on his machine.  I am suprised though, I have not had any
interoperability issues with open/commercial ssh : 

[dswan at bioinf4 downloads]$ ssh -V
OpenSSH_3.4p1, SSH protocols 1.5/2.0, OpenSSL 0x0090602f
[dswan at bioinf4 downloads]$ ssh XXX.XXX.XXX.XXX
dswan at 138.37.232.110's password: 

Last login: Wed Feb 19 2003 14:17:54 from XXX.XXX.XXX.XXX

Welcome to toybox, take a vigil in the wilderness of mirrors.
--
28/5/2002 - outages due to Lonon Electricity power failure
No mail.

[dswan at toybox dswan]$ /usr/local/sbin/sshd2 -V
sshd2: SSH Secure Shell 3.2.0 (non-commercial version) on
i686-pc-linux-gnu

Are you sure you're both using ssh protocol version 2?  They should play
fair together unless there's a protocol mismatch.

If he really isn't prepared to do make the switch then I can send you a
blow by blow account on installing the ssh.com version of sshd and
replacing OpenSSH.  But I'm not going to recommend it as the best
solution!

If he has written the script to log into your clone.. how edoes he know
it works with the commercial ssh daemon anyway?  Just curious!  Any
chance we can get a peek at the script (with usernames and passwords
removed of course!).

regards,

Dan

-- 
Dr Dan Swan - Bio-Linux Developer
EGTDC, CEH, Mansfield Road, Oxford, OX1 3SR
Tel: 01865 281 630 Fax: 01865 281 696
http://envgen.nox.ac.uk/ | dswan at ceh.ac.uk



From pkerrwall at psu.edu  Tue Feb 25 12:02:41 2003
From: pkerrwall at psu.edu (Kerr Wall)
Date: Tue, 25 Feb 2003 12:02:41 -0500
Subject: [Bio-Linux] SSH Question
In-Reply-To: <1046190202.25384.72.camel@bioinf4.nox.ac.uk>
Message-ID: <BA810BE1.7EEB%pkerrwall@psu.edu>

On 2/25/03 11:23 AM, "Dr Dan Swan" <dswan at ceh.ac.uk> wrote:

> On Tue, 2003-02-25 at 15:46, Kerr Wall wrote:
> 
> Hi Kerr,
> 
>> I have a question concerning ssh.  I have received a request from the
>> programmer at Cornell who is responsible for getting FGP tracefiles from our
>> server in an automated fashion to switch from the openSSH server that came
>> with the clone to the comercial version at ssh.com.  He has stated that the
>> script he has written to automatically login to our server doesn't work with
>> openSSH and that it only works with the version from ssh.com.
>> 
>> Would you guys recommend this (ie, would this cause any dependency problems
>> or are there any security concerns that I don't know about)?  If so, what is
>> the best way to uninstall openSSH and then install SSH (I'm guessing that
>> using the rpm command will accomplish this)?  If not, do you have any
>> suggestions for him to get a working perl script to automatically login with
>> openSSH?
> 
> So he wants you to change the OpenSSH daemon on your clone because he's
> using a commercial version of the ssh client?  That's a bit unfair ;)
> 
> Its perfectly possible to install the commercial version of the ssh
> daemon on the clone, although you will not be able to do it using an rpm
> as it supplied as a source tarball.  There's no licencing issue as its
> free for academics.  I would suggest however that your colleague
> downloads the client/base rpms from www.openssh.org and installs those
> on his machine.  I am suprised though, I have not had any
> interoperability issues with open/commercial ssh :
Thanks,  I've forwarded all replies and he has agreed to install openSSH.

> Are you sure you're both using ssh protocol version 2?  They should play
> fair together unless there's a protocol mismatch.
We are both using version 2.

> If he really isn't prepared to do make the switch then I can send you a
> blow by blow account on installing the ssh.com version of sshd and
> replacing OpenSSH.  But I'm not going to recommend it as the best
> solution!
> 
> If he has written the script to log into your clone.. how edoes he know
> it works with the commercial ssh daemon anyway?  Just curious!  Any
> chance we can get a peek at the script (with usernames and passwords
> removed of course!).
Here is the response from Dan (at Cornell):

************************************************************************
I just use key based authentication with a blank passphrase.  The syntax
for login then looks like:

ssh -l username -i path_to_identification_file ip_address

If you're logging in from a user login on your local machine that is the
same as the username on the remote machine, you can oming the -l and -i
parameters since by default it uses your current username and it look
for the ~/.ssh2/identification file.

For further info on setting up key based authentication take a look at
the ssh man file.

In scripts I use it to do lots of things such as running commands
remotely.  For example, to open a file handle in perl that feeds in a
list of all the zip files in the directory (so I get a listing of all
the plates) I was doing this:

open PS_FILES, "ssh -l transfer -i /home/dci1/.ssh2/identification
128.118.180.140 'ls *.zip' |";

I keep the identification file with the private key in my home dir, but
it could be anywhere as long as you tell it how to get to it.
************************************************************************

Thanks,

Kerr