Dear Debian Project Team,
First of all, thanks you very much for the final product that you always give to us. We are from Montevideo City capital of Uruguay and we are working at TI area, specifically in Infrastructure and Servers from several years ago (we are using Debian since Woody).
At present we are working for one of the most outstanding technological project at the country, called Ceibal Project (http://www.ceibal.org.uy) (http://en.wikipedia.org/wiki/Ceibal_project). In your country this is known like OLPC (http://one.laptop.org/), please view: Next Story/Photo “Uruguay Students in Artigas”. For your information in more than 2000 servers that currently exist to support the Plan in all country we are using Debian.
One of our tasks in the Plan is to create images servers, which provide the support of each School http://www.ceibal.org.uy/index.php?option=com_content&view=article&id=177:servidores&catid=40:tecnologia&Itemid=61). To develop it we were using “Etch”, “Lenny” and now we are using “Squeeze” every day.
We have known about the improvements that this new version has like update the kernel, faster booting, ext4 support inclusion, and control units through sysv-rc-conf in runlevels and identification devices between others points. All of this had a pleasant result we started to use a new method of work more strict and orderly.
On the level control device UUID we focus on doing the initrd and the current grub2 (grub-pc.) It is on this last thing we would like to direct a query to the list Grub-Devel <firstname.lastname@example.org> (QA page) if you enable this medium.
We have a solution Raid + LVM where generated images to scale on school servers, both as Lenny as ETCH this methodology applied to generate the image and then dump the new server where we generated by scripting the partition / file system and then we added the solution.
Currently in Squeeze this new device-level control by UUID, forced us to remove the native UUID server and add to that generated by the image. Ex: (q-mdadm-create / dev/md0-run-level = 1-raid-devices = 2-uuid = xxxxx: xxxx: xxxx: xxxxx / dev/sda1 / dev/sdb1) (idem with assemble) this to initrd level is supported to move from a master image to a server. The problem is generated by GRUB2 which this procedure is letting us do not dump the image.
Therefore we disabled the control in Grub2 UUID (GRUB_DISABLE_LINUX_UUID = bring), then update-grub, we generate a new grub.cfg, core.img, etc etc to not corroborate anything regarding UUID at boot. But to dump the master image to a new server, this does not boot, limiting the incident in the new boot loader, for that reason, we use the old package grub (1) holding in Squeeze distribution.
If you have any comments (we did not find anything similar in the mailing list) or a special treatment for this particular point, we appreciate the information, however we will continuous very attached to DEBIAN.
Gonzalo and Eduardo
Reference for the Ceibal Plan:
“Uruguay (380,000+ given to 1-6 students and teachers, completing the initial objective. 90,000 ordered in 2010 for high-school students).”
To consult: http://www.debian.org/devel/people
Debian QA Group <email@example.com> (QA page)
Debian Policy List <firstname.lastname@example.org> (QA page)
Grub-Devel <email@example.com> (QA page)
main: grub-splashimages, grub2-splashimages