Automated kernel crash reporting system

[Robert Watson]

On Thu, 4 Mar 2010, sean connolly wrote:

> Automatic reporting would end up being a mess given that panics can be 
> caused by hardware problems. Having an autoreport check if memtest was run 
> before it reports, or having it only run with -CURRENTmight be useful.

Hi Sean, Dan, et al:

I'm not sure I agree with this view.  For releases, it's true that many 
reported panics are a result of bad hardware.  However, on active development 
branches, especially -CURRENT, that's not the case.  An automated scheme to 
track bug reports and find common themes could be incredibly valuable in the 
development environment.

And, to be honest, even if a fair number of reports are due to hardware 
failures, these often have common themes themselves, so it would be quite 
educational to be able to reason about panics on a large scale.  Not to 
mention using it to identify potentially flakey hardware that users could then 
be warned about :-).


Collecting crash reports is widespread in industry for both operating systems 
and applications for these reasons.  Certainly, the crashinfo summary gathered 
on recent FreeBSD versions is an excellent starting point for building such a 
system.  If we were to move ahead with it, we'd need to pay very close 
attention to scrubbing potentially sensitive information from panic reports, 
however.

Robert


>
> Sean
>
>
>
>
> ________________________________
> From: jhell <jhell at DataIX.net>
> To: Dan Naumov <dan.naumov at gmail.com>
> Cc: FreeBSD Hackers <freebsd-hackers at freebsd.org>; freebsd-questions at freebsd.org
> Sent: Thu, March 4, 2010 8:06:50 AM
> Subject: Re: Automated kernel crash reporting system
>
>
> On Thu, 4 Mar 2010 07:09, dan.naumov@ wrote:
>> Hello
>>
>> I noticed the following on the FreeBSD website:
>> http://www.freebsd.org/projects/ideas/ideas.html#p-autoreport Has
>> there been any progress/work done on the automated kernel crash
>> reporting system? The current ways of enabling and gathering the
>> information required by developers for investigating panics and
>> similar issues are unintuitive and user-hostile to say the least and
>> anything to automate the process would be a very welcome addition.
>>
>>
>> - Sincerely,
>> Dan Naumov
>>
>
> Hi Dan,
>
> I am assuming that the output of crashinfo_enable="YES" is not what you
> are talking about is it ? are you aware of it ?
>
> The info contained in the crashinfo.txt.N is pretty informative for
> developers, maybe your talking about another way of submitting it ?
>
> Regards,
>
> -- 
>
>  jhell
>
> _______________________________________________
> freebsd-questions at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-questions
> To unsubscribe, send any mail to "freebsd-questions-unsubscribe at freebsd.org"
>
>
>
>
> _______________________________________________
> freebsd-hackers at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
> To unsubscribe, send any mail to "freebsd-hackers-unsubscribe at freebsd.org"
>