On 03/23/12 14:37, Marc Espie wrote:
Since we are talking about a cluster here, it's likely you're mixing&matching
stuff and confusing what happens on the local box and on the distant box.
I'm pretty sure that, if you take the time to REALLY look at what's going
on on your machine(s), you'll figure it out.
I'm convinced that there's no problem in the test itself. The problem lies
within your building process, until you prove otherwise.
You are right. My problem was in the other hosts in the cluster where I forgot
to add the PORTSDIR=nfs_mounted_directory to their respective /etc/mk.conf. As
Jan Stary pointed out, my patch allowed the symlink to be followed on those
other hosts which masked my problem.
I have successfully run a partial dpb build on both i386 and amd64 with the
updated /etc/mk.conf.
Thanks for the feedback.
