On Sun, Sep 01, 2013 at 03:57:56PM +0200, Matthieu Herrb wrote:
> Hi,
>
> rencently, while rebuilding ports on 2 machines (one amd64 and one
> i386) which share /usr/ports over NFS, they entered a loop fetching
> the same distfile in parallel (and failing because the resulting file
> gets corrupted) over and over, until I stopped one of those.
>
> Shouldn't the lock for fetch operations be in a mi location (instead
> f ${PORTSDIR}/logs/$[MACHINE}/locks) ?
It should be settable independently, indeed. However, changing the default is
going to be breaking some setups.
The loop is weird. They should both fail, eventually, when they run out of
distsites. Are you sure it was not a distsite with lots of mirrors and that
you didn't lose patience before the en ?
