Re: HEADS UP: panic behaviour changed

[Arnaud Lacombe <lacombar%gmail.com@localhost>]

On Sun, Feb 1, 2009 at 2:31 PM, Robert Elz <kre%munnari.oz.au@localhost> wrote:
>    Date:        Sun, 1 Feb 2009 11:27:40 -0500
>    From:        Arnaud Lacombe <lacombar%gmail.com@localhost>
>    Message-ID:  
> <1a69a9d80902010827i492f5f25j2e795470d12e4547%mail.gmail.com@localhost>
>
> This (my comment) has nothing to do really with the previous topic of
> this threat, but is kind of related ...
>
>  | Nonetheless, I agree with the fact that release kernel must reboot on
>  | panic without user intervention,
>
> I have been wondering if perhaps ddb could simply cause a reboot if
> entered, and then left unattended for some period (like 10 minutes),
> with no commands at all - once any command was given it could assume a
> human is in control and just act like it does now.
>
> The period could perhaps be configurable - perhaps even change ddb.onpanic
> so it sets the number of minutes to wait, with 0 implying don't wait at all,
> (so don't bother entering ddb at all).   ('1' could be a special case meaning
> 10 or perhaps even infinite, so people who have ddb.onpanic=1 don't get too
> short a time to react -1 might be better for infinite, but keeping compat
> with what people have now is important).
>
> In any case, some of the bad feeling that the drop to ddb leads to might
> perhaps be alleviated if the system just waits a little, then reboots
> if no-one does anything with ddb for a reasonable (or configurable) period.
>
FWIW, Linux has this with /proc/sys/kernel/panic (or approaching).
Sounds a like nice feature to have.

 - Arnaud