From: Anton Starikov Subject: Re: NFS silent death Date: Sat, 19 Dec 2009 22:18:11 +0100 Message-ID: <60E7C084-8B89-4F6E-88D8-C3C41B1AC416@gmail.com> References: <143DD9C8-CD7D-4D18-8295-D9ADBD8A0972@gmail.com> <1240866287.8818.89.camel@heimdal.trondhjem.org> <38C9A462-2199-4CA7-9EEC-050400D71C56@gmail.com> Mime-Version: 1.0 (Apple Message framework v1077) Content-Type: text/plain; charset=us-ascii Cc: Trond Myklebust , "linux-nfs@vger.kernel.org" To: Anton Starikov Return-path: Received: from ey-out-2122.google.com ([74.125.78.24]:59174 "EHLO ey-out-2122.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753472AbZLSVSP (ORCPT ); Sat, 19 Dec 2009 16:18:15 -0500 Received: by ey-out-2122.google.com with SMTP id d26so1053020eyd.19 for ; Sat, 19 Dec 2009 13:18:14 -0800 (PST) In-Reply-To: Sender: linux-nfs-owner@vger.kernel.org List-ID: Hi! Quick question. Is this issue fixed by now On Apr 28, 2009, at 12:25 AM, Anton Starikov wrote: >>>> >>>> Hmm... My guess is that you are deadlocked because rpciod appears to be >>>> hanging on an idmapper upcall due to an NFSv4 open() call. At the same >>>> time, since /etc/passwd and /etc/group are on an NFSv3 partition, the >>>> idmapper will need to use rpciod in order to read them. >>>> >>>> We therefore apparently need to move the idmapper upcall out of the >>>> rpciod part of the open() call. I'll look into that... Thanks, Anton.