Ok, I've been fighting with this one for a while. Google is just about sick of me asking the same question in a number of different ways.

Environment:

I have a FreeBSD 5.3 RELEASE fileserver - from here I share the /usr/ports, /usr/obj /usr/src to my four other FreeBSD boxes using nfs - works fine.

I also share a mount for my OpenBSD router's ports collection - works fine.

My problem lies within the mount for my primary data. This data is located at '/mnt/200Gig' and is accessed by both Windows machines (Samba 3.x) and two Gentoo 2005.0 boxes (using nfs).

Samba is working just fine. I have all my users able to log in and users can only access what they should be able to access. Fantastic.

NFS is a totally different story, and so far it doesn't have a happy ending.

Firstly, the relevant line of my fstab file on one of the Gentoo boxes:
# Network Mounts
fileserver:/mnt/200Gig /mnt/network/fileserver nfs noauto,noatime,user 0 0

Now if I'm to try and do '#mount -t nfs fileserver:/mnt/200Gig /mnt/network/fileserver' as either my everyday user OR as root - the mount can take literally close to 5 minutes. Sure enough, the mount will be successful, but hardly practical in timing. It seems (by google) this is a common problem, but no one else's resolutions seem to be able to help me.

At one stage the Gentoo box would complain something like 'NFS version X failed, trying version Y' and the version Y would work. So, the best option I could find was to recompile my Gentoo kernel with no version X support, theoretically it would no longer try this. Well, the error stopped, but the mount still takes AGES. Im stuck for ideas after this. At this stage neither box is running a firewall until I can get this sorted.

There is another (smaller) problem following this which is related to getting users accessing their respective directories with an nfs mount. I'll hassle with that after this first problem is resolved.

I hope SOMEONE can suggest a resolution for this because I really need nfs mounting to work.

Thanks for your time,

Gsee

My exports file from the server:

/OBSD -alldirs -maproot=root router
/mnt/200Gig -mapall=root 192.168.1.103 laptoo 192.168.1.111 beastie
/usr/ports /usr/src /usr/obj -maproot=root mailserver beastie dnsserver server
/ftp/SeeResults/online -maproot=user42 mailserver

N.B. There is only one space between each hostname in the actual file, I've only increased this for readability in a web browser.

Thanks again.

The problem may be on the Gentoo side. Check client syslog messages for RPC errors. This may indicate that the portmap service is either not installed or not running.

NFS uses RPC rather frequently, and the fallback alternatives take a long time to negotiate. This should also alleviate long delays for writes and locking. I am not a Gentoo user, but this fixed my problems on Debian/Ubuntu connecting to NFS shares on both FreeBSD as well as NetworkAppliance Filers (nfs2/UDP).

Good luck!

I had exactly the same problem: FreeBSD nfs server, Gentoo client, mounting takes a lot of time.
Starting portmap solved it, the nfs share is mounted immediately.

Thank you!

Had the same problem between ubuntu (client) and fedora core 5 (server). It seems that portmap is not part of the depenancies for nfs client packages. bit silly cos although not required it does seem to be pointless without.
I assume there are security issues associated with portmap... dunno.
Also to Gsee, you may well run in some trouble with firewalling nfs as it uses dynamic port allocations and (depending on the firewall software/hardware) you need to set the server to use static ports for the 4/5 different services that nfs relies on.