[SoaS] An Introduction

[Xander Pirdy]

Hello Everyone,
My name is Alexander Pirdy. I am a new follower of this mailing list (as of
Sunday) and just wanted to take a minute to introduce myself, and what I am
doing

I was assigned to work on this project by Caroline Meeks, in the hopes of
being able to help figure out a solution, or at least explore options to the
problems that some sticks are having with the overlay becoming corrupted and
preventing booting.

I have been trying to follow these posts but I am getting lost in all of the
nested responses; the first thing that I am going to do is to create a wiki
page so that this can be a bit more organized and we can hopefully take a
more systematic approach. I am still just trying to understand this problem.

To me it seems likely that this is just linked to running out of space on
the squashfs file, if not then it seems like we have two distinct problems.

I will create a brief wiki about it tonight and will expand it over the next
couple of days.

It would be really helpful for me if everyone that has seen this problem
could detail how they set their stick up, what size they made the overlay
file, what happened when they ran into trouble, what they were doing when
trouble was noticed and their ideas on potential solutions, and the
shortcomings of each solution. For example moving to a full installation on
the stick may incur a penalty for lacking as diverse of hardware support,
and reducing drive life (www.wikipedia.org/wiki/Live_usb) though this may be
insignificant. It also may make creation of these sticks more difficult, and
may reduce the robustness of the system (replacing the overlay fixes the
problem right?).  From what I have heard from David Farning, full installs
last an order of magnitude longer than the live-type of installs with his
setups and tests, though I am unsure if they crash in different ways, or if
there is just more free space on a full install (I doubt by a factor of 10
though).

It would also be helpful for me if anyone has tests or suites of tests that
they would like to see run successfully on the sticks. (I think I saw
something about dd'ing a bunch of info). Also if anyone has already worked
out a roadmap towards a solution and want to share it that could save me
some time.

At the very least it looks like this system needs to be made to crash more
gracefully (if running out of space can't be avoided), even with just a
warning.

One last note: If the plan is to move to SoaS being based on Ubuntu I doubt
that this issue will persist in the same form. Can anyone replicate the
issue with Ubuntu?  What are the thoughts on the architecture of running it
on Ubuntu, should it switch to a casper-rw/home-rw partition type
persistence? Can that setup be accomplished with a squashfs file (so that
the sticks don't have to be formated or partitioned)?

I look forward to working with all of you, and hopefully working this (and
future) issue(s) out,
Thank you,
Alexander
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sugarlabs.org/archive/soas/attachments/20100127/9b3a62bf/attachment.htm