NFS umount takes ages when no DNS available

[Robert Schulze]

Hi,

first of all: thanks to your answers.

Am 22.09.2011 01:50, schrieb Rick Macklem:
>>
> Well, here is the code snippet. (I'm not sure why the author felt that
> the getaddinfo() needed to be done before the check for the need to do
> an rpc?):
 > [...]
>
> These protocols were done in the days when servers would have been in
> /etc/host files, so the names always resolved. (mid to late 1980s)

this all would make sense (at least to me) when using hostnames. But I 
don't use them anywhere on the server or client regarding NFS. 
Furthermore a getaddrinfo() with numeric ip address should return 
instantly, or is it supposed to do a reverse lookup?

The addresses we use for NFS are in the 10.x.y.z/24 range which are not 
declared in our nameservers, so the client will get NXDOMAIN for the 
address.

Regarding the hostname which statd complains with: I've looked into the 
sources of /usr/sbin/rpc.statd and found out one place, where a 
gethostname() is called, but I can't figure out whether this value is 
also handed out over the wire or just for logging purposes. When looking 
at a multihomed setup like we use it (external public ip address for 
non-nfs and internal local address for nfs-only) the hostname of the 
machine is true for the external interface only.

with kind regards,
Robert Schulze