From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from cuda.sgi.com (cuda2.sgi.com [192.48.176.25]) by oss.sgi.com (8.14.3/8.14.3/SuSE Linux 0.8) with ESMTP id q3H9K2e5036927 for ; Tue, 17 Apr 2012 04:20:03 -0500 Received: from mail.profihost.ag (mail.profihost.ag [85.158.179.208]) by cuda.sgi.com with ESMTP id HdFEuZilLdiNrk0G (version=TLSv1 cipher=AES256-SHA bits=256 verify=NO) for ; Tue, 17 Apr 2012 02:20:00 -0700 (PDT) Message-ID: <4F8D35B8.4090507@profihost.ag> Date: Tue, 17 Apr 2012 11:19:52 +0200 From: Stefan Priebe - Profihost AG MIME-Version: 1.0 Subject: Re: XFS blocked task in xlog_cil_force_lsn References: <4F8BC0B9.90604@profihost.ag> <20120416235647.GA6734@dastard> <4F8D27A1.2090809@profihost.ag> <20120417083210.GI6734@dastard> In-Reply-To: <20120417083210.GI6734@dastard> List-Id: XFS Filesystem from SGI List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: xfs-bounces@oss.sgi.com Errors-To: xfs-bounces@oss.sgi.com To: Dave Chinner Cc: Christoph Hellwig , Greg KH , Philipp Herz - Profihost AG , stable@vger.kernel.org, "xfs@oss.sgi.com" Am 17.04.2012 10:32, schrieb Dave Chinner: >>> What's your storage hardware, application, workload, etc? >> It a Raid 10 of 1TB SATA Drives in front of a 3Ware Controller doing >> heavy rsync backups. > > Hmmm. It sounds like it is just taking a long time for the IO to > complete, not that there is any other problem. If the IO is realy, > really slow, the a log force is definitely a place it coul dstall > for some time. OK as i've seen this just once it just might be that the controller / disks were too slow in this case. > Does the RAID controller have a BBWC? No but doesn't matter ;-) If it crashes i can just repair the FS and restart the backup again. > What IO scheduler are you using? [root@backup1 ~]# cat /sys/block/sdc/queue/scheduler [noop] deadline cfq > Does mounting with nobarriers reduce the stall time? It is already mounted with nobarrier: /dev/sdc3 on /sdc3 type xfs (rw,noatime,nodiratime,nobarrier,prjquota) > Can you reduce the IO load so that it doesn't end up in this state? Can try that by just doing less backups per time period. Stefan _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs