Interesting problem today, had a Linux box that was providing slow NFS, intermittent SAMBA failures (of said NFS mounts), and the following KEY error messages:
Mar 12 10:59:05 somehost kernel: [X] portmap: server localhost not responding, timed out
Mar 12 10:59:05 somehost kernel: [X] lockd: cannot monitor XXX.XXX.XXX.XXX
Mar 12 10:59:05 somehost kernel: [X] lockd: failed to monitor XXX.XXX.XXX.XXX
This kind of looked like portmap or one of it’s RPC services weren’t running. On inspection, rpc.statd had failed and when restarting, it would time out because it couldn’t contact the portmapper. I decided to restart the system, because that is usually quicker than restarting portmapper and all RPC services (as required with a portmap restart) by hand and in this case the system users would be equally out of service from either one. After restarting, the NFS mounts didn’t mount and when we mounted them by hand we got the following KEY error messages.
Mar 12 14:46:42 somehost kernel: [X] portmap: server localhost not responding, timed out
Mar 12 14:46:42 somehost kernel: [X] RPC: failed to contact portmap (errno -5).
Mar 12 14:46:42 somehost kernel: [X] lockd_up: makesock failed, error=-5
Turns out that there was no route to localhost because ifconfig thought the loopback device was up, but route didn’t. Taking down the loopback device, bringing it back up and then adding the route to localhost manually (it’s
route add localhost dev lo, BTW) got us back to fully functional. I’m still investigating the root cause.
Tags: linux, portmap, samba, rpc