RFC re: 798960 -- session timeout-induced bug

wes hayutin whayutin at redhat.com
Fri Mar 2 14:12:35 UTC 2012


On 03/01/2012 06:11 PM, Mo Morsi wrote:
> On 03/01/2012 05:48 PM, Hugh Brock wrote:
>> Hmmm... Problem is not timing out is a security hole. Is this anything more serious than a change to rails config?
>>
>> -h
>>
>> Matt Wagner<matt.wagner at redhat.com>  wrote:
>>
>> Hi folks,
>>
>> The other day, https://bugzilla.redhat.com/show_bug.cgi?id=794536 was
>> entered -- "conductor should have an inactivity timeout."
>>
>> I only noticed it after it was pushed, and I remarked on IRC that it
>> sounded more like an irritant than a good feature. Further, I suggested
>> that it was a new feature, not a bug, so it didn't belong in this
>> sprint. It sounded like there was some consensus on at least the second
>> point, but more in a "going forward" sense than in relation to this
>> specific bugfix.
>>
>> I just noticed https://bugzilla.redhat.com/show_bug.cgi?id=798960 as I
>> was packing up for the day, which is marked as an urgent bug that is
>> wreaking havoc on QE. While it's a legitimate bug, it seems like it's
>> directly caused by the patch in #794536.
>>
>> At the risk of sounding like a curmudgeon, I would propose that we
>> should revert the patch for 794536. The patch itself was good, but its
>> inclusion was out-of-scope for the sprint, and it's now causing a
>> high-priority bug.
>>
>> Thoughts?
>>
>> -- Matt
> Hrm why is the inactivity timeout causing issues w/ QE? Do any of the
> automated tests incorporate a delay>  15 minutes?
um.. running a build/push and waiting for it would take longer than 15 min.
I'm always of the opinion qe automation should work around any issues 
introduced by devel if there is a bug.

So.. I mentioned it in a bug, but I'm not expecting *any* fixes for qe 
automation's sake.
>
> Regardless, would tailoring the behaviour to the rails environment solve
> the issue? eg. when running the app in 'production' we incorporate this
> time out where as it is disabled in the 'test' environment.
>
>      -Mo
>
Changing the default behaviour for testing is always a bad idea IMHO
At any rate.. I just need some time to find the appropriate fix for the 
auto.




More information about the aeolus-devel mailing list