nfsv4 not working on new server

depontius · Advocate Joined: 05 May 2004 Posts: 3509

I'm in the process of setting up a new server, bringing functions up one-by-one. I'm stuck with nfsv4 at the moment.

Both old and new servers, as well as the client I'm using for debug are running nfs-utils-1.2.9 with identical USE="caps libmount nfsidmap nfsv4 tcpd uuid -ipv6 -kerberos -nfsdcld -nfsv41 (-selinux)"

When I try to mount a filesystem from the new server in /var/log/messages I get:
Dec 31 19:52:57 anastasia kernel: RPC: AUTH_GSS upcall timed out.
Dec 31 19:52:57 anastasia kernel: Please check user daemon is running.
But as can be seen above, none of these installations use Kerberos, yet GSS makes me think that this is a Kerberos-related sugestion, and it isn't necessary to connect to the old server, anyway.

Any suggestions?
_________________
.sigs waste space and bandwidth

krinn · Watchman Joined: 02 May 2003 Posts: 7470

try mount it with -o sec=sys on the client

Kompi · Posted: Sun Jan 05, 2014 8:57 am Post subject:

Have you tried to explicitly mount it with sec=sys to disable any kerberos auth? Like this:

depontius · Advocate Joined: 05 May 2004 Posts: 3509

I seem to be working. I'm not sure that the "sec=sys" option did it, there appear to be a number of little problems here, and they all need to be flattened in order to get a clean mount.

First, half of my RAID-1 is an eSATA drive that doesn't reliably mount at power up. So I've de-auto'ed all of that stuff. After power up, I need to:
1 - See if the eSATA drive is there, and replug it if it isn't.
2 - Start the RAID
3 - Mount the RAID
4 - Bind-mount the /exports/home

Next, it appears that nfs doesn't do a clean start, first time. I get this error message:

BitPit · n00b Joined: 10 Jun 2003 Posts: 11 Location: Austin, TX

Caution: changing krb5 to sys allows connection without kerberos authentication. It no longer uses kerberos security.

There is still something wrong with NFS4 using kerberos. I too have this problem, but have not figured out why it came about or how to fix it.

It worked until the last month on a gentoo mostly stable system. It may have broken in linux 3.10.17, but I didn't see really bad things until linux 3.12.13 where is stopped working. I noticed it in linux 3.12.13 because I share /usr/portage with several machines and a remote update suddenly wanted to revert a lot of stuff to a much older version. /usr/portage was not NFS mounted over the original portage files.

ssh with kerberos still works, but it is real slow making a connection.

Kompi · Posted: Mon Mar 17, 2014 3:49 pm Post subject: