Hi! I've fixed that lockup by now, but unfortunately haven't spot the cause - just deleted some problem files... The "problem files" was found by running glusterfs client in non-daemon mode. Unfortunately, there is no log left - the messages were going to /dev/stdout. There were errors like EBADFD and EINVAL concerning files /danilov/__DCOM_server*. So I removed them from other client (all clients were working fine, except the head-node one), and head-node client returned to life. :) Will report if the issue raise again. WBR, Andrey
