[sisuite-users] how to disable tmpfs staging?

2010-07-19 Thread Tristam MacDonald
Hello all,

I am setting up our university lab machines to image via system imager. The
image is around 7 GB, and on every attempt to install the image, I end up
running out of RAM (2 GB) on each client machine.

I am running system imager 4.1.6, from the apt repository on Ubuntu 10.04
amd64. I am using the default rsync mechanism, and using netboot to initiate
installs.

I have set option-144 to n in my dhcp server (which is separate to my
image server), and have passed TMPFS_STAGING=no as a kernel boot parameter,
either of which I believe should disable tmpfs staging. I also believe that
tmpfs staging is meant to be disabled by default, though the docs are
somewhat fuzzy on that particular issue.

Is there anything I am likely to have done wrong which is causing tmpfs to
be on in all cases? And anything I can do to remedy this?

I am happy to post configuration/logs/etc. if it will be of any help.

Thanks,

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/
--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] how to disable tmpfs staging?

2010-07-20 Thread Tristam MacDonald
On Tue, Jul 20, 2010 at 12:53 AM, Brian Elliott Finley
br...@thefinleys.com wrote:

 What you're doing should prevent tmpfs staging.  And yes -- it is off
 by default.

 Can you provide the output of df -k from the console of a failed
 autoinstall client?

 Thanks, -Brian

The output from df -k, which seems to indicate that it never mounted
the disk partitions:

Filesystem   1k-blocks  Used Available Use% Mounted on

rootfs 1028604   1015228 13376  99% /
/dev/root    15540 15540 0 100% /old_root
tmpfs  1028604   1015228 13376  99% /
devpts 1028604   164   1028440   0% /dev/pts
tmpfs  1028604   1015228 13376  99% /dev/.static/dev
udev   1028604   164   1028440   0% /dev

Also the output of fdisk -l, which indicates that it did create all of
the partitions successfully:

Disk /dev/sda: 160.0 GB, 1600 bytes
255 heads, 63 sectors/track, 19452 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

   Device Boot  Start End  Blocks  Id System
/dev/sda1   1 608 4881787  83 Linux
Partition 1 does not end on cylinder boundary
/dev/sda2   * 608    3040    19530273+ 83 Linux
/dev/sda3    3040    4255 9765625  83 Linux
/dev/sda4    4256   19452   122069902+ 83 Linux

My partition layout is somewhat unorthodox: 5GB swap, followed by 20GB
/, 10 GB /tmp and 115 GB /data. All of the above except for / are
empty, but could the swap-first layout be confusing things?

Thanks,

--
Tristam MacDonald
http://swiftcoder.wordpress.com/

--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] how to disable tmpfs staging?

2010-07-20 Thread Tristam MacDonald
On Tue, Jul 20, 2010 at 12:51 PM, Brian Elliott Finley
br...@thefinleys.com wrote:
 Tristam,

 Are you using ext4?  If so, I bet that's why. I'm in the process of
 adding support for ext4 and grub2, but its not there yet.

Right you are - I am imaging an ubuntu 10.04 box, so ext4 and grub2 by
default. I just tested switching to ext3, and the imaging appears to
be proceeding perfectly, so I will mess with downgrading grub in the
morning.

Thanks,

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/

--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] how to disable tmpfs staging?

2010-07-21 Thread Tristam MacDonald
On Wed, Jul 21, 2010 at 3:00 AM, Bas van der Vlies b...@sara.nl wrote:

 If you want ext4 and grub2 support. You can can grab the kernel/initrd from
 https://subtrac.sara.nl/oss/sali to replace the kernel/initrd that your are
 know using.

I saw the SALI kernel and gave it a spin, but it seems to take a
different set of parameters, and I haven't been able to figure out
which parameter I am feeding wrong.

With the default systemimager configuration, it boots fine, rsyncs the
script directory successfully, and then errors out with 'couldn't find
script'. Any idea how I pass it the image/script name?

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/

--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] how to disable tmpfs staging?

2010-07-21 Thread Tristam MacDonald
On Wed, Jul 21, 2010 at 7:40 AM, Bas van der Vlies b...@sara.nl wrote:
 On 21-07-10 13:34, Bas van der Vlies wrote:

 You have to supply the scriptname parameter ;-).

Excellent, one more hurdle overcome!

 There is also an example master script that use GRUB2. We expect a
 partition for grub2, see:
   * https://subtrac.sara.nl/oss/sali/browser/trunk/example/masterscript.new

 and search for grub2

I am a little confused how the master scripts work.

Do I manually edit (having made a backup copy) my image-name.master
script to include those features, or is there a way to modify
systemimager to generate a master script which works like that?

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/

--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] how to disable tmpfs staging?

2010-07-21 Thread Tristam MacDonald
On Wed, Jul 21, 2010 at 11:14 AM, Tristam MacDonald
swiftco...@gmail.com wrote:

 I am a little confused how the master scripts work.

 Do I manually edit (having made a backup copy) my image-name.master
 script to include those features, or is there a way to modify
 systemimager to generate a master script which works like that?

Having assumed that I manually create the master script at this point,
I have been able to create something that gets most of the way. Still
working on getting a valid grub config in place post imaging.

In case someone else is trying to use Ubuntu 10.04 with the SALI
kernel in the future, take note of the following:
- Ubuntu does not use a separate grub partition by default, so you
need to make one
- Ubuntu identifies partitions by UUIDs by default, so edit your
/etc/fstab to use partition names instead
- Ubuntu uses UUIDs in grub by default, so disable that in /etc/default/grub

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/

--
This SF.net email is sponsored by Sprint
What will you do first with EVO, the first 4G phone?
Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] How can I see the System Imager error message?

2010-08-03 Thread Tristam MacDonald
Did you try shift+pageup to use the kernel's built-in scrolling capabilities?

On Aug 3, 2010, at 15:45, Patrick Nolan patrick.no...@stanford.edu wrote:

 I've been banging my head against a problem installing OSCAR for a
 couple of weeks.  The network boot phase seems to be crashing because
 of a System Imager error, but I can't tell what type of error.

 The net boot begins OK and runs a downloaded kernel.  Then a lot of
 stuff scrolls by quickly.  It ends with a page which begins, Your
 autoinstall has failed.  You can use the shell prompt to try and see
 what happened.  This is followed by a page of advice, which causes
 all previous messages to scroll off the console.  Of course, there is
 no shell prompt, nor do the alt-f1 or other keys do anything.  It's
 just hung.

 I tried capturing the boot messages by echoing them out a serial port.
 (The AMI BIOS allows that.)  It works up to a point, but the output
 quits when the kernel boots.

 What can I do?


 --
 The Palm PDK Hot Apps Program offers developers who use the
 Plug-In Development Kit to bring their C/C++ apps to Palm for a share
 of $1 Million in cash or HP Products. Visit us here for more details:
 http://p.sf.net/sfu/dev2dev-palm
 ___
 sisuite-users mailing list
 sisuite-users@lists.sourceforge.net
 https://lists.sourceforge.net/lists/listinfo/sisuite-users

--
The Palm PDK Hot Apps Program offers developers who use the
Plug-In Development Kit to bring their C/C++ apps to Palm for a share
of $1 Million in cash or HP Products. Visit us here for more details:
http://p.sf.net/sfu/dev2dev-palm
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] How can I see the System Imager error message?

2010-08-04 Thread Tristam MacDonald
On Tue, Aug 3, 2010 at 3:56 PM, Patrick Nolan patrick.no...@stanford.eduwrote:

 Tristam MacDonald wrote:
  Did you try shift+pageup to use the kernel's built-in scrolling
 capabilities?

 Yes.  That's one of the things mentioned in the page of advice.  There
 doesn't seem to be a kernel running at that point. Nothing gets any sort of
 response except the Reset button on the front panel.


There certainly is a full kernel running by the time that failure message is
printed, so it sounds like you have an actual hard crash on your hands.


 I tried capturing the boot messages by echoing them out a serial port.
 (The AMI BIOS allows that.)  It works up to a point, but the output
 quits when the kernel boots.


Did you also tell your *kernel* to output over the console? If my memory
serves correctly, you need to add something like console=ttyS0,57600 to the
APPEND line in your pxelinux.cfg/whatever.

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/
--
The Palm PDK Hot Apps Program offers developers who use the
Plug-In Development Kit to bring their C/C++ apps to Palm for a share
of $1 Million in cash or HP Products. Visit us here for more details:
http://p.sf.net/sfu/dev2dev-palm___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] Booting from SALI, but install script doesn't run, fails when run manually

2010-08-16 Thread Tristam MacDonald
On Aug 16, 2010, at 5:07 PM, Chris Pepper pep...@reppep.com wrote:

With SALI our nodes connect to the imageserver and fetch scripts, but they
do not actually install. It's a bit confusing -- the VGA console shows the
correct hostname (c15), runs 99all.harmless_example_script, then says No
script defined, opening console, write_variables, and Starting
console


You need to pass the SCRIPTNAME=whatever option to the kernel. See my thread
about tmpfs a few weeks ago - about halfway through is a detailed
description.

--
Tristam MacDonald
http://swiftcoder.wordpress.com
Sent from my iPad
--
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev ___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] Booting from SALI, but install script doesn't run, fails when run manually

2010-08-16 Thread Tristam MacDonald
On Aug 16, 2010, at 7:24 PM, Chris Pepper pep...@reppep.com wrote:

 Tristam MacDonald wrote:
 On Aug 16, 2010, at 5:07 PM, Chris Pepper pep...@reppep.com
 mailto:pep...@reppep.com wrote:

 With SALI our nodes connect to the imageserver and fetch scripts, but
 they do not actually install. It's a bit confusing -- the VGA console
 shows the correct hostname (c15), runs 99all.harmless_example_script,
 then says No script defined, opening console, write_variables, and
 Starting console

 You need to pass the SCRIPTNAME=whatever option to the kernel. See my
 thread about tmpfs a few weeks ago - about halfway through is a detailed
 description.

 Tristam,

   I'm pretty sure our older installation of 4.0.2 detected this
 automatically from cluster.xml, but I will try tomorrow.

Systemimager does this fine, but it is listed as a current limitation
of SALI on the webpage (https://subtrac.sara.nl/oss/sali).


   Any ideas why parted is choking when I run the script manually?

Despite it being stated several places that SALI supports default
system imager scripts, I didn't find that to be the case on our Dell
cluster. I ended up writing a custom script using the newer
'partition' command, since I decided to support ext4 and grub2 as
well.

--
Tristam MacDonald
http://swiftcoder.wordpress.com
Sent from my iPad


--
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] Partitioning problem with SI 4.0.2 SALI 1.3.2

2010-09-28 Thread Tristam MacDonald
On Tue, Sep 28, 2010 at 3:50 PM, Chris Pepper pep...@reppep.com wrote:

 Tristam,

Last month, I made the minimal changes to the SI master script (as
 suggested on this list) to make it work with SALI. I have considered using
 the new SALI masterscript, but not tried it yet. Your suggestions look a bit
 more radical, and I haven't looked in the SALI documentation to find out
 what those verbs mean yet.

If possible, I'd like to patch the SI script, since I believe it's
 *almost* okay (it worked on PE1950s last month). If that won't fly, I'll
 test the SALI replacement.

Is there any documentation for your suggested modifications? Did you
 pull them from a public source that I can read?


The (semi-complete) official documentation is on the SALI wiki (
https://subtrac.sara.nl/oss/sali/wiki/SaliUsage/Functions). Most of it I
pulled from the various SALI-related posts on the mailing list, however.

I realise you would prefer to continue using the old-style script, but the
reason I recommend the newer method is that SALI-style functions result in a
much cleaner script that is far easier to read and maintain.

-- 
Tristam MacDonald
http://swiftcoder.wordpress.com/
--
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] Not seeing boel messages on serial console

2011-07-13 Thread Tristam MacDonald
On Wed, Jul 13, 2011 at 7:23 PM, Steven A. DuChene 
linux-clust...@mindspring.com wrote:


 The following messages about grub appear in the console log file:

 2011-7-13 15:7:36 [Boot::Grub :: Line 132] grub supports --no-floppy
 argument.
 2011-7-13 15:7:36 [Boot::Grub :: Line 107] Grub executable set to:
 /sbin/grub-install.
 2011-7-13 15:7:36 [Boot::Grub :: Line 151] bootloader = /sbin/grub-install
 2011-7-13 15:7:36 [Boot::Grub :: Line 338] mount: /boot
 2011-7-13 15:7:36 [Boot::Grub :: Line 338] mount: /
 2011-7-13 15:7:36 [Boot::Grub :: Line 368] calling /sbin/grub-install
 --no-floppy --recheck UUID=0b8f2f84-5a03-4a80-b853-a624d496e5e7

 Probing devices to guess BIOS drives. This may take a long time.
 Format of install_device not recognized.

 INSTALL_DEVICE can be a GRUB device name or a system device filename.


 Has anyone seen this before? Is there someway I can eliminate
 this --recheck option from being used with grub-install?
 Or is it that grub just does not like the
 UUID=0b8f2f84-5a03-4a80-b853-a624d496e5e7 device it is given?

 If I could figure out where the call was being made I would try
 to fix it. I tried inserting a hard-coded grub-install line into
 a couple of different places in the autoinstall master script
 but it seems no matter where I put ti the script complains that
 the path is not correct (it cannot find the executable).
 --
 Steven DuChene


My understanding is that the UUIDs are specific to a particular device. In
other words, unless you are imaging to the machine which served as the
golden client, the UUID will not match the device, and the grub install will
fail.

You can work around this by disabling the use of UUIDs by your OS. That
generally means editing your /etc/fstab to replace UUIDs with device paths,
and editing your /etc/default/grub to disable UUID support.

-- 
Tristam MacDonald
System Administrator, Suffolk University Math  CS Department
--
AppSumo Presents a FREE Video for the SourceForge Community by Eric 
Ries, the creator of the Lean Startup Methodology on Lean Startup 
Secrets Revealed. This video shows you how to validate your ideas, 
optimize your ideas and identify your business strategy.
http://p.sf.net/sfu/appsumosfdev2dev___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] si-suite site down?

2012-02-10 Thread Tristam MacDonald
On Fri, Feb 10, 2012 at 9:56 AM, Jan Groenewald j...@aims.ac.za wrote:


 Can you perhaps describe more extensively the relation between
 systemimager and SALI and your best guess as to the future developments
 over the next few years? And perhaps a specific note on the current status
 in SI and SALI of grub2, ext4, and systemconfigurator?

 Regards,
 Jan


Until the devs have time to weight in, I'll give you the perspective of
someone using them on a daily basis:

SALI is a drop-in replacement for the default systemimager boot image
(kernel and initrd). It has fairly good support for grub2 and ext4, and a
vastly simplified command set for writing system imager install scripts.
The rest of the systemimager suite you use as before.

We use systemimager + SALI to image a lab of Ubuntu 11.04 machines, which
works pretty much out of the box. The only caveat is that neither component
has support for Debian's new disk UUIDs, so monkeying with your /etc/fstab
and grub config may be required.

-- 
Tristam MacDonald
System Administrator, Suffolk University Math  CS Department
--
Virtualization  Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] si-suite site down?

2012-02-13 Thread Tristam MacDonald
On Mon, Feb 13, 2012 at 5:55 AM, Bas van der Vlies b...@sara.nl wrote:

 Tristam,

 Do you mean that the 'blkid` command is not supported in busybox or are
 you missing something else?

 For our installions we use LABEL functionality,  but we must also generate
 our /etc/fstab file. It is on out to do list to generate the fstab file.


To be honest, I never tried to generate the fstab, since switching the
fstab over to old-fashioned device references is trivial.

The grub2 config is a whole lot more troublesome, and it's very poorly
documented how to write a non-UUID grub config without manually overwriting
the output of update-grub.

-- 
Tristam MacDonald
System Administrator, Suffolk University Math  CS Department
http://swiftcoder.wordpress.com/
--
Try before you buy = See our experts in action!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-dev2___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


[sisuite-users] SALI + VMWare ESXi

2012-02-14 Thread Tristam MacDonald
Has anyone tried using SALI to boot a VM instance running under WMWare's
ESXi?

I'm getting the following output when the boot process fails, and I assume
it has something to do with SALI not recognising the particular virtual
ethernet adapter:

s*tart_network*
*IP Address not set with pre-boot settings.*
*
*
*Param DEVICE not set, guesing interface*
*
*
*DHCP configuration failed*
*
*
*Killing off running processes.*


And once at the recovery prompt, ifconfig won't show/manipulate any devices
except lo.

-- 
Tristam MacDonald
System Administrator, Suffolk University Math  CS Department
--
Keep Your Developer Skills Current with LearnDevNow!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-d2d___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users


Re: [sisuite-users] SALI + VMWare ESXi

2012-02-16 Thread Tristam MacDonald
On Wed, Feb 15, 2012 at 3:54 AM, Bas van der Vlies b...@sara.nl wrote:

 On 02/14/2012 08:57 PM, Tristam MacDonald wrote:

 Has anyone tried using SALI to boot a VM instance running under WMWare's
 ESXi?


  Which virtual network device did you choose? I just googled something and
 you can select the e1000 driver. The VMware VMXNET3 Ethernet driver is
 not currently enabled in our setup. I will enable it i build a new version
 of SALI


Ah, thank you. I must have glossed over network adapter selection when
creating the VM - the e1000 driver does indeed do the trick.

-- 
Tristam MacDonald
--
Virtualization  Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/___
sisuite-users mailing list
sisuite-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sisuite-users