[LRUG] To debug the impossible bug

Jonathan del Strother maillist at steelskies.com
Wed Aug 31 04:06:36 PDT 2011


On 4 August 2011 18:18, Jonathan del Strother <maillist at steelskies.com> wrote:
> On 3 August 2011 14:15, Simon Coffey <simon at tribesports.com> wrote:
>> Hi Tatsuya,
>>
>> Thanks very much for the response. Instrumenting this particular
>> method is definitely next on the list - I was a bit loath to stick
>> debugging code on production, but it looks like the only option.
>>
>> On 3 August 2011 12:59, Tatsuya Ono <ononoma at gmail.com> wrote:
>>> 2) Error happened different place.
>>> e.g.
>>> class Question
>>>   def background
>>>     return @something.background # <= NoMethod error @something is nil here
>>>   end
>>> end
>>> I know that unless you are bit careless, you wouldn't miss the backtrace.
>>
>> Yep, sorry for not including the backtrace - unfortunately Hoptoad
>> (now AirBrake), the service we use to trap exceptions, has scrubbed
>> the backtrace so I can't copy it here. However as I said in my reply
>> to Murray, Question#background isn't a particularly interesting
>> method, it's just an ActiveRecord attribute accessor, and not one
>> we've overridden or delegated in any way.
>>
>> Thanks again - I'll definitely be posting back results as and when we
>> get to the bottom of this. :-)
>
> I'm not sure how helpful this is going to be - it's a bit handwavy -
> but we've started migrating to 1.9.2 recently and coincidentally have
> also started seeing a couple of really strange errors that seem
> somewhat similar to what you're seeing.  They're pretty rare,
> definitely not reproducible at will, but it does almost seem like an
> object is getting swapped out and replaced with something totally
> unrelated.  I haven't yet had chance to do a proper investigation of
> it (and a bug like that in the ruby interpreter does seem pretty
> unlikely), but I thought I'd mention it...
>
> -Jonathan
>



Heya - did you ever get anywhere with this, or find any
officially-filed bug reports about it?  We're continuing to see
occasional errors crop up that sound rather similar to this on our
Solaris10 / Ruby 1.9.2 machines.



More information about the Chat mailing list