Another bug that would have been caught if packages went through a boot/acceptance test

[James Laska]

On Wed, 2010-09-29 at 09:55 +0100, Richard W.M. Jones wrote:
> If you've been following updates to F14 alpha/beta, you'll undoubtedly
> have been hit by:
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=638091
> 
> which makes everything segfault and is unrecoverable without a live CD
> and a lot of manual intervention.
> 
> A simple "does it boot" test in fact wouldn't have caught this, but a
> "does it boot, then run prelink" (completely automatic as part of the
> Koji build or AutoQA acceptance) would have.

In fairness, reboot was tested by a proventester for this update.  The
bug doesn't surface on reboot alone.  The problem happens after prelink
has run ... then you're hosed.  

I've added this issue to the f14 qa retrospective [1].  In addition to
the current automation milestones, we're also thinking about documenting
basic test procedures for several critpath components (including glibc).
I'd like to see test procedures documented before think about how best
to automate.

Thanks,
James

[1] https://fedoraproject.org/wiki/Fedora_14_QA_Retrospective
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
Url : http://lists.fedoraproject.org/pipermail/devel/attachments/20100929/0b0bb037/attachment.bin