On Mon, Mar 4, 2013 at 5:20 PM, Chris Evich <[email protected]> wrote:
>
>> And another problem is the crash handler set by autotest is gone if
>> I use the standalone runner, I guess we should bring it back again.
>
>
> I vaguely remember lmr explaining the reasoning behind this as:  If you're
> using the runner, you'll likely notice a crash and don't need the code to do
> the work of your eyeballs.  I could be mistaken though.

You're mistaken indeed. The gist of the crash handler not being
present is that in any modern-ish distro that people use for
development there is a crash handling system [1] and therefore we
don't need the custom one that autotest sets up. With those crash
handling systems you can get a nice notification that there was a
crash, and a wizard kind of thing that helps you open a bug on the
distro's bug tracker.

Now, if the standalone runner is somehow missing crashes during
postprocess, this is a bug and we should fix it (even if the
infrastructure is not analyzing or processing the crashes as
mentioned). Need to see what is happening and fix it.

If I had to guess right now without looking at the code, I'd say we
are lacking some functions to verify kernel and QEMU crashes during
postprocessing.

[1] There's apport for Ubuntu (and OpenSUSE), ABRT for Fedora. All of
them do what the autotest crash handler does, in a more sophisticated
way. Then you'd ask: Why autotest has its own? Some reasons:
 1) Back then we didn't have ABRT
 2) We still test distros without a crash handling system
 3) We still need something that outputs the core dumps to autotest
directories, and figuring out how to set up each individual system to
output to autotest dirs is not worth the trouble

-- 
Lucas

_______________________________________________
Virt-test-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/virt-test-devel

Reply via email to