> > Does this problem still persist?
yes
> > Is it always on the same clients?
It has happened on every client except (I think) the puppet server
itself. The clients are Solaris 10 boxes with a mix of sparc and
i386. It seems to be happening about 10x more on one particular
client, but I think that is explained by the fact that catalog runs
take much longer on that box due to a module that is only being
applied to it.
> > What webserver are you using (webbrick?)
Yes, though this will probably change soon due to what I have read in
the Puppet Scalability Notes. We currently have only deployed puppet
to 5 nodes, but plan on managing more than 50 nodes.
> > What OS is the puppet server running on?
Solaris 10