Re: Centralized updates of debian-live systems

On Thu, Dec 6, 2012 at 7:40 AM, Davide Natalini <davide.natalini@gmail.com> wrote:

Il giorno 06/dic/12, alle ore 08:13, Daniel Baumann ha scritto:

On 12/05/2012 11:02 PM, Davide Natalini wrote:

Hello,

hi,

please do not hijack threads, start a new one by writing a new mail, not by replying to an existing one.

I'm sorry, I didn't realize I was doing that.

Any comments, ideas, suggestions or other about this is much appreciated

i'd probably use one read-only persistency over nfs for all clients (will need some patches to make it work) in such a scenario, and update the persistency on the nfs server whenever required.

I'm not willing to have a server for this, for a number of reasons:
* I would be forced to work only on the server, and probably only when all the clients are down
* a server would become crucial for the work of too many people
* a system with a single server does not scale up well: if the number of clients grows, the overall performance goes down, and/or its cost grows up
* the clients are in use in a school: at the beginning of a lesson all the clients will have simultaneous requests to the server
* every single client is powerful enough to run a standalone system with a good performance

I think a system like the one I'm proposing would be very interesting for all the applications with many identical systems, like schools.

Does anyone have had good/bad experiences in a similar context?

I have had excellent success with netboot images with redundant proximity boot servers. However, when netboot is not a choice for say notebooks, I do what I call a firmware install as plainroot and with persistence. And that as well has been successful, but what I am waiting for is what I call live-media install. The live-media install will install the squashfs to be booted from local hard drive . Here is a thread that mentions it http://lists.debian.org/debian-boot/2011/09/msg00197.html . I believe the live-media install type will allow for concise deployments of a given squashfs image to groups of machines as you describe.

Thank you,

Davide Natalini

--
To UNSUBSCRIBE, email to debian-live-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: [🔎] 7C8A0FE7-4355-4BA4-B7F7-41A87DB4FD6F@gmail.com" target="_blank">http://lists.debian.org/7C8A0FE7-4355-4BA4-B7F7-41A87DB4FD6F@gmail.com