#161 Issue closed: Unattended recovery¶
Labels: enhancement
kkaiser opened issue at 2012-09-20 09:49:¶
I have a question concerning the recovery process. Is it possible to fully automate the backup recovery? So that there are no questions during the recovery no login etc. just an automated process with default values as answer.
I know that is not the main design goal but it would be a nice way to migrate or install many machines by just deploying the "golden client" image.
If you need any further information don't hesitate to ask me.
dagwieers commented at 2012-09-20 10:09:¶
It is certainly something we have thought about before and I am not against this at all.
The implementation for this would work through a kernel command line
option or a configuration variable that the init-scripts would intercept
to execute rear recover as the final step. The reason why this is not
a priority (for me at least) at this time is because we have a lot of
more important stuff to cover before this is nicely integrated.
Side-note: The advantage of a kernel command line option is that one can decide to add it during boot, even if the rescue image was not prepared for it.
There are various reasons why an (unattended) restore could stop at the moment, and without having a prompt available or even the output on screen (because it may be replaced by a login, or the output may be on a different terminal) it will be more confusing to the person in charge of doing a disaster recovery. So an unattended process makes more sense if we offer menus and/or options to the user in (almost) all cases where the process can stop unexpectedly.
We already added some (line-based) menus for the layout-code that can help the person in charge during recovery, but there are currently too many corner-case to make this work (for everyone). That said, there's definitely room for an implementation that works in a controlled environment like yours and we would accept a pull-request adding this kind of functionality.
gdha commented at 2012-11-23 13:10:¶
During a recovery we do device migration (sometimes) that requires
manual input, or during cloning it may be need to manual edit the
restore script. So, for various reasons automating is not easy and
perhaps not desirable at this moment.
For an instant recovery on the same hardware it could be feasible, but
even then when an error occurs it should offer a prompt. However, in an
background recovery process how would you know?
gdha commented at 2013-07-03 11:48:¶
is in fact already implemented by issue #256 and will be part of rear-1.15 (yet to be released)
dagwieers commented at 2013-07-11 23:47:¶
@kkaiser Could you please test this new functionality again ? Feel free to reopen this issue in case there is a problem, or it doesn't do exactly what you were expecting. Thanks for helping us out !
[Export of Github issue for rear/rear.]