Return-Path: linux-nfs-owner@vger.kernel.org Received: from fieldses.org ([174.143.236.118]:59274 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751735Ab3GHNfo (ORCPT ); Mon, 8 Jul 2013 09:35:44 -0400 Date: Mon, 8 Jul 2013 09:35:37 -0400 From: "J. Bruce Fields" To: Sven Geggus Cc: linux-nfs@vger.kernel.org Subject: Re: Kerberized NFS-Server Problem still present in 3.10.0-rc2 Message-ID: <20130708133537.GA29071@fieldses.org> References: <20130620144955.GB11728@fieldses.org> <20130620145209.GC11728@fieldses.org> <20130621083221.GA4103@geggus.net> <20130624215408.GD23596@fieldses.org> <20130625094612.GA2186@geggus.net> <20130701205234.GF19945@fieldses.org> <20130702080530.GA17953@geggus.net> <20130705181859.GA8288@fieldses.org> <20130705203435.GE8288@fieldses.org> <20130708080803.GB29990@geggus.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <20130708080803.GB29990@geggus.net> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, Jul 08, 2013 at 10:08:05AM +0200, Sven Geggus wrote: > Monday Monday, so good to me :) > > On Friday, 5. Jul 2013 22:34, J. Bruce Fields wrote: > > > Hm, might be interesting to see what mountd's doing too: > > > > strace -p $(pidof rpc.svcgssd) -s4096 -e trace=open,close,read,write Argh--I said "mountd", then I forgot to replace the rpc.svcgssd in the cut-n-pasted commandline; should have been: strace -p $(pidof rpc.mountd) -s4096 -e trace=open,close,read,write > > OK, output attached. While reading this output I recogniced, that I > overlooked the following error-messages from rpc.svcgssd and rpc.mountd: > > Jul 8 09:53:38 vnfsrv rpc.svcgssd[2914]: ERROR: GSS-API: error in gss_free_lucid_sec_context(): GSS_S_NO_CONTEXT (No context has been established) - Unknown error > Jul 8 09:53:38 vnfsrv rpc.svcgssd[2914]: WARN: failed to free lucid sec context > Jul 8 09:53:38 vnfsrv rpc.mountd[2927]: qword_eol: fprintf failed: errno 22 (Invalid argument) > > > Hm, maybe this gid upcall? Does the following (probably not completely > > right) help? > > Yes, we are getting there. While the strange messages from rpc.svcgssd > persist, the mount now seems to work again after your patch has been > applied. Anyway. OK, that more or less solves the mystery, though the strace output might still be interesting. --b.