> It can be when building any package, and I haven't seen a pattern
> forming yet, but it seems that there are a couple of queues left over
> after some builds, and then a failed build increases the number even
> more. Then it reaches the point where there are no spare ones, and I
> have to manually clean them. This then allows me to continue a bit.
>
> Attached are a couple of my build logs, I have just modified the script
> to include ipcs output before and afterwards. Hopefully this will help
> track the problem down.
It would be nice to know exactly when this happens.
There are obviously two cases:
1. fakeroot is not doing any clean-up after run due to some weird
interaction with threads.
This is a problem with fakeroot, and needs to be fixed with
fakeroot side. So if you can reproduce the problem without using
cowdancer, you will know it's fakeroot.
2. fakeroot is failing to clean-up after run due to cowdancer.
This will be evident if you run 'strace fakeroot' instead of
'fakeroot' inside pbuilder.
I'm not quite sure if
BUILDSOURCEROOTCMD="strace fakeroot"
or
BUILDSOURCEROOTCMD="ltrace fakeroot"
would work,
It should manifest in the form of ENOMEM (out of memory), or
something similar.
> > Something I would suspect would be the environment. checking 'env'
> > output
>
> attached, I can't see anything odd.
>
> > and looking at ldd output of packages and checking their
> > versions might be worthwhile.
>
> ldd of the packages I am building, or the ones I am using?
Erm sorry, I was being vague. I was suspecting that some env vars
resulting in weird linkage; so I was thinking of finding out if
anything is wrong inside/outside the chroot.
> > At least, it doesn't seem to be a problem reproduced on many systems,
> > which is just weird.
> >
>
> Yes, I'm not sure what the crucial difference is.
I don't really have an idea right now.
regards,
junichi
--
[EMAIL PROTECTED],netfort.gr.jp} Debian Project
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
