From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zenon Panoussis Subject: Re: kernel errors, timeouts and qemu-img usage Date: Tue, 03 May 2011 18:14:28 +0200 Message-ID: <4DC029E4.3@provocation.net> References: <4DBFEC14.1040102@science-computing.de> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: Received: from host-88-80-6-138.cust.prq.se ([88.80.6.138]:51079 "EHLO mail.nettheatre.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751680Ab1ECQOf (ORCPT ); Tue, 3 May 2011 12:14:35 -0400 In-Reply-To: <4DBFEC14.1040102@science-computing.de> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: ceph-devel@vger.kernel.org On 05/03/2011 01:50 PM, Christoph Raible wrote: > First I alwas get on ceph -w following "error": > "[WRN] message from mon2 was stamped 12.271440s in the future clocks not synchronized" > But i have synchronized my clocks 1 min befor with the same ntp-server.. But are they really in sync? > The second error is, that I can't create / start an qemu-image on the ceph-filesystem. > I want to start a kvm virtual machine with the virt-manager. > I create an image with > "qemu-img create -f qcow2 Platte-qcow2.img 10G" > The virtual machine never starts. It hangs on look for the "harddisk" The image file has to appear as a block device; I suspect you need http://ceph.newdream.net/wiki/QEMU-RBD . I have no clue about the third error; what were you doing when you got it? The fourth error is a timeout too. The file you are trying to write is 2.4 GiB in size (dd defaults to 512 bytes, so 512*5000000 bytes). Your journal is only 500 MB and you are replicating the data and you are logging 20 on the same physical hard drive, so I suspect very serious I/O saturation. Try bs= together with count= for better dd control and write a much smaller file. Try disabling hung_task_timeout as the error message suggested. Try also nice and ionice on dd. The problem might be something unrelated, but this is a quick way to start figuring what's happening. Z