Re: buildds screwed?
On Mon, May 13, 2013 at 01:05:39PM +0200, Geert Uytterhoeven wrote:
> On Mon, May 13, 2013 at 12:40 PM, Christian T. Steigies <cts@debian.org> wrote:
> >> > Well, I had this example last week where during two builds different test
> >> > cases failed, and the third build succeeded. If this is the hardware, what
> >> > should I do, reseat the CPU? I think Michael gave me a spare 060 some time
> >> > ago... but Ingo has similar problems, but not with old kernels?
> >>
> >> Could be a kernel issue with '060s.
> >
> > Thats what I want to hear! And now that our kernel hacker is listening, what
> > can we do to track this down and fix it?
>
> If there are no problems with older kernels, you can try to bisect it.
Ingo, what was the last working one, something from etch, 2.6.something I
guess?
> Is it easy to trigger?
No, unfortunately(?) not. I have the impression it happens more often when
the load is high (or maybe when swap is used), but I don't have any proof
for that. Sometimes every buildd-[uploader|watcher] fails a few times in a
row, and then again everything runs smoothly for days.
Christian
Reply to: