Return-Path: Received: from linuxhacker.ru ([217.76.32.60]:58272 "EHLO fiona.linuxhacker.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932500AbcIHO4A (ORCPT ); Thu, 8 Sep 2016 10:56:00 -0400 Subject: Re: NFS client stuck in a tight loop Mime-Version: 1.0 (Apple Message framework v1283) Content-Type: text/plain; charset=windows-1252 From: Oleg Drokin In-Reply-To: <59CD8653-D109-408E-A5BA-124220051094@primarydata.com> Date: Thu, 8 Sep 2016 10:55:50 -0400 Cc: List Linux NFS Mailing Message-Id: <730673D2-B564-4B0A-B69F-170192BB5302@linuxhacker.ru> References: <0ABF496B-2E1F-46B6-A3EA-55C447F71073@linuxhacker.ru> <59CD8653-D109-408E-A5BA-124220051094@primarydata.com> To: Trond Myklebust Sender: linux-nfs-owner@vger.kernel.org List-ID: On Sep 7, 2016, at 10:35 AM, Trond Myklebust wrote: >> You can do umount -f and then unmount does work on the second try, >> this also results in the following dmesg message: >> [155381.995368] NFS: nfs4_reclaim_open_state: unhandled error -5 >> >> I don't know this code well enough to take my own stab at a fix, but I can >> help testing if needed. > > I did send out a set of patches a week ago that addresses an issue in this area, and I know of a couple more holes that need to be plugged. I tried the tip of your testing branch (hash 05c974669ecec510a85d8534099bb75404e82c41) and the problem is still there, though possibly a bit harder to trigger. Bye, Oleg