On Thu, Jun 05, 2014 at 03:59:32PM -0000, James Hunt wrote:
> Public bug reported:
>
> The kerneloops daemon ignores WARNING-type oops's. This seems to be due
> to bug 346303 (2009).
Yes, it is due to that bug and users being told that a WARNING was a
"serious kernel problem". Additionally, there were was a tremendous
volume of these kernel oopses which were not being collated in Launchpad
and were clogging the kernel team's bug list.
> Is this behaviour still in line with our processes?
If we were to start including the WARNING type of oopses again I think
we need to do some consolidation of them (in addition to changing the
"serious kernel problem" message), such that they can be bucketed
together at errors.ubuntu.com. I seem to have worked on this at one
point in time.
> I'm guessing not since the kernel team don't seem to be aware of this
> silent filtering and it means that we would miss crash reports such as
> bug 1326300.
I've generally been handling kerneloops and the apport package hook for
the kernel for the kernel team (after discussing it with them), so that
may explain their lack of awareness. For example of such discussions
see:
On Thu, Jun 05, 2014 at 03:59:32PM -0000, James Hunt wrote:
> Public bug reported:
>
> The kerneloops daemon ignores WARNING-type oops's. This seems to be due
> to bug 346303 (2009).
Yes, it is due to that bug and users being told that a WARNING was a
"serious kernel problem". Additionally, there were was a tremendous
volume of these kernel oopses which were not being collated in Launchpad
and were clogging the kernel team's bug list.
> Is this behaviour still in line with our processes?
If we were to start including the WARNING type of oopses again I think
we need to do some consolidation of them (in addition to changing the
"serious kernel problem" message), such that they can be bucketed
together at errors.ubuntu.com. I seem to have worked on this at one
point in time.
https:/ /lists. ubuntu. com/archives/ kernel- team/2010- October/ 013051. html
> I'm guessing not since the kernel team don't seem to be aware of this
> silent filtering and it means that we would miss crash reports such as
> bug 1326300.
I've generally been handling kerneloops and the apport package hook for
the kernel for the kernel team (after discussing it with them), so that
may explain their lack of awareness. For example of such discussions
see:
http:// ubuntu. 5.x6.nabble. com/apport- kerneloops- reporting- td351124. html
> The fix is to remove the filter on "WARNING:" in pipe_oopses() - updated
> patch attached.
I don't believe that just re-enabling it is the right way to go.
--
Brian Murray
Ubuntu Bug Master