Here's the chain on the servers: volume->locks->io-threads Here's the chain on the client where cp doesn't work: volume io-threads (striped)->readahead->writebehind->quick-read->stat-prefetch cp works here: volume io-threads (striped)->readahead->writebehind or volume io-threads (striped)->readahead->writebehind->stat-prefetch On Tue, Mar 2, 2010 at 12:18 PM, Sabuj Pattanayek <sabujp at gmail.com> wrote: > I've tracked down the problem to the quick-read xlator. I tried the cp > -r while adding one xlator at a time. It worked until I added the > quick-read xlator where it crashed. > > On Tue, Mar 2, 2010 at 11:45 AM, Sabuj Pattanayek <sabujp at gmail.com> wrote: >> Also, I cannot duplicate the crash when I copy using cp off the >> distributed endpoint, i.e. cp works fine there. >> >