[Dev] [RFC] Tizen system rollback

Stoppa, Igor igor.stoppa at intel.com
Mon Dec 2 10:04:10 GMT 2013


On 2 December 2013 11:45, Łukasz Stelmach <l.stelmach at samsung.com> wrote:

> Please consider upgrading tens (hundreds) of thousands units. If you
> break them OTA you have to fix them OTA. You won't make N (where N ~
> 1E5, you may be lucky enough to make 10% of you customers download
> flashing software for PC that still gives 90k) customers visit repair
> shops. You don't have that many repair shops to fix such failure (in
> reasonable time).

​I agree that, presented from this perspective there is a point to it.
Questions: how likely is it?
Of course even if it's unlikely one doesn't want to be the responsible for
the failure.
AFAIK nobody has yet managed to cause such massive failure on the field.

But let's assume that, since it's better to be safe than sorry, some
mechanism has to be deployed,
I'm still not sure that btrfs is the right answer. Not for btrfs per se,
but rather because the rollback
would happen at a very low level which has no knowledge of what is really
going on. ​

There should be also some mechanism to re-align with the user data stored
in the cloud,
which ​might have diverged significantly, in case the rollback happens
significantly later on, after the upgrade.

cheers, igor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.tizen.org/pipermail/dev/attachments/20131202/e75fde48/attachment.html>

More information about the Dev mailing list