FWIW, my personal hgweb setup uses gunicorn instead of mod_wsgi partially for this reason.
On Aug 10, 2012, at 2:03 PM, Evan Kinney wrote: > mod_wsgi was configured to restart hgweb after 1000 requests, and since there > was only one process running, that made it kill your connection during huge > clones. > > The recommendation from Selenic is to leave that parameter unset, so I did > that. > > Cloning works fine now. > > /ek >
