View Single Post
  #8   (View Single Post)  
Old 26th August 2010
jggimi's Avatar
jggimi jggimi is offline
More noise than signal
 
Join Date: May 2008
Location: USA
Posts: 7,977
Default

I have OBSD systems with NFS mounts working fine. I have also had trouble in the deep dark past, which I will reiterate in short form, in the hope it may help:

The situation:

Mounts would hang using the UDP default protocol. Connections via TCP would work fine.

The root cause:

My server had multiple IP addresses sharing the same NIC. My client was using one of the alias address in the mount request. Traces with tcpdump(8) showed the UDP the mount request arriving at the alias address, but server responses were coming from the base IP address. Because UDP is stateless, the client was ignoring the responses from this address that it did not associate with the server. (TCP, being stateful, did not have this issue.)

The moral of my story:

You may not be using an alias address on your server, as I was. But you may want to still test TCP protocol, to see if there is a different result. As I did, you may discover that learning to use tcpdump(8) to analyze network problems will be more helpful for more things than you'd ever imagined.
Reply With Quote