Bug#740491: nfs-common: fails to upgrade (action "restart" failed)

[Petter Reinholdtsen <pere@hungry.com>]

I had a look at this issue, and the underlying problem seem to be that
rpcbind do not notice that rpc.statd was stopped, and believe the port
it request is occupied by another already running process when it try to
register its port when it start again.  The 'status' RPC service stay in
the list even after rpc.statd is stopped.

I was able to work around the problem by wiping the status of rpcbind
like this:

  service rpcbind stop ; rm /run/rpcbind/* ; service rpcbind stop 

But this should only be done if it is safe to restart all the RPC
services, as rpcbind will forget about all of them.  Check the output of
'rpcinfo -p' before and after to see which services need to be restarted
and registered again with rpcbind.

I have no idea why rpc.statd failed to tell rpcbind that its service was
no longer active, and no idea how the changes in libtirpc since 0.2.2-5
could cause this.

-- 
Happy hacking
Petter Reinholdtsen