NFS problems - internal IPs cause it to hang?
 

Hello,

I have a fileserver that has been serving files via NFS for 3 years that suddenly started having problems. I had tried everything I could think of at the time to trouble shoot, but nothing worked so I formatted the drives and installed a fresh copy of CentOS4. After getting basically everything configured again, I copied over my old trusty /etc/exports file from backup, but the problem still persisted.

Any time I run "/usr/sbin/exportfs -a" it just hangs. (and the /etc/init.d/nfs script just runs "/usr/sbin/exportfs", hence the same stalling occurs.)

I have now figured out that if I use only one line like this in my /etc/exports file:
/share/server1 192.168.1.5(rw,sync,no_root_squash)

exportfs -a will hang and the client will not be able to connect.

if I instead have a line like:
/share/server1 144.96.55.155(rw,sync,no_root_squash)

where I specify the client's external IP address, exportfs -a works great, and the client can connect. The weird thing is if I specify a host name like:
/share/server1 server1.domain.com(rw,sync,no_root_squash)

and put an entry in my /etc/hosts file like:
192.168.1.5 server1.domain.com server1

then running exportfs -a works, and the client can connect fine. So for some reason, my /etc/exports file just stopped liking internal addresses - even though a host name mapped to an internal address works.

The most confusing thing about all of this is the fact that it just stopped working one day. My /etc/exports file has been full of lines containing internal IPs for 3 years.

Any ideas?

Thanks,
Devin

What else changed?
 

I'm guessing something changed in your network. If there are other people changing things, I would get with the firewall admin, or start looking into any network settings on that box / clients that might have changed.

Re: What else changed?
 

Agreed - but as I am the one doing all of the sysadmin work, I can't blame anyone else.
I have installed several new servers this week, (connected to both our internal network
and external network) but I can't figure out why the NFS server all of a sudden won't
accept internal IPs in its /etc/exports file. From everything I can gather, the internal
network and switch is operating normally. But then again, I'm not exactly sure how
to test for small glitches in the internal network. I can say that all of the machines on
the internal network can ping and connect to each other.

Regards,
Devin

I'm having the same problem
 

Did you ever figure out what the problem was?

Kirk

A very late reply, but it may help other googlers like me. Solution is to zero out the file:

Hope this helps.