From: "Kevin Coffman" Subject: Re: stuck/hung nfsv4 mounts Date: Mon, 3 Nov 2008 15:28:11 -0500 Message-ID: <4d569c330811031228r5bb9aefs7a970303910810e2@mail.gmail.com> References: <1225724721.2247.29.camel@brian-laptop> <1225731544.6958.6.camel@heimdal.trondhjem.org> <1225734631.2247.76.camel@brian-laptop> <4d569c330811031158r26963e0w5bcf8331e0fb14b7@mail.gmail.com> <1225742958.2247.128.camel@brian-laptop> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Cc: linux-nfs@vger.kernel.org To: "Brian J. Murrell" Return-path: Received: from wf-out-1314.google.com ([209.85.200.171]:4760 "EHLO wf-out-1314.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753958AbYKCU2M (ORCPT ); Mon, 3 Nov 2008 15:28:12 -0500 Received: by wf-out-1314.google.com with SMTP id 27so2728998wfd.4 for ; Mon, 03 Nov 2008 12:28:12 -0800 (PST) In-Reply-To: <1225742958.2247.128.camel@brian-laptop> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, Nov 3, 2008 at 3:09 PM, Brian J. Murrell wrote: > On Mon, 2008-11-03 at 14:58 -0500, Kevin Coffman wrote: >> >> These are all from librpcsecgss (from the -rrr). I don't see an error here. > > Good. > >> You should have gotten output from gssd itself as well. You might >> also look for any errors on the server from rpc.svcgssd. > > Nothing. > >> This looks fine. (Assuming linux.interlinx.bc.ca is the NFS server >> and pc.interlinx.bc.ca is the client.) > > Indeed, they both are as you assume. > > So I've restarted both the rpc.svcgssd and rpc.gssd with -rrr -vvv on > the server and client (respectively) to see what more I can see but I > can't get the upcall to trigger. I would have thought simply issuing a > mount would do that, but it seems not. What exactly causes the upcall > to be made and what might prevent it from being made? There should definitely be corresponding debug output from rpc.gssd (-vvv) if you got those librpcsecgss (-rrr) messages. I would suggest that you leave off the -rrr for now as it appears to be extra clutter at this point. It might help if you run rpc.gssd in the foreground (-f) while debugging. Also, make sure this is the only rpc.gssd running. Re: the upcall. Any new mount using Kerberos should cause the upcall to create a new context with the server. Make sure the filesystem is not already mounted. After the mount, any new user accessing that mounted filesystem should cause an upcall to create a context for that user. K.C.