From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:59387) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Yj19R-0004cf-ML for qemu-devel@nongnu.org; Fri, 17 Apr 2015 03:57:26 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Yj19O-00007I-5x for qemu-devel@nongnu.org; Fri, 17 Apr 2015 03:57:25 -0400 Received: from mx1.redhat.com ([209.132.183.28]:52315) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1Yj19O-00007B-0h for qemu-devel@nongnu.org; Fri, 17 Apr 2015 03:57:22 -0400 Date: Fri, 17 Apr 2015 09:57:17 +0200 From: Kevin Wolf Message-ID: <20150417075717.GA3716@noname.redhat.com> References: <55306E81.9000604@linux.vnet.ibm.com> <1679484709.357603914.1429245010449.JavaMail.zimbra@oxygem.tv> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <1679484709.357603914.1429245010449.JavaMail.zimbra@oxygem.tv> Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] qcow2 problem: while a qcow2 image run out of disk space the guest paused. List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Alexandre DERUMIER Cc: mimu@linux.vnet.ibm.com, qemu-devel , borntraeger@de.ibm.com, stefanha , Holger Wolf , tubo@cn.ibm.com, chenxg Am 17.04.2015 um 06:30 hat Alexandre DERUMIER geschrieben: > Hi, >=20 > Isn't it related to drive options ? >=20 >=20 > " > werror=3Daction,rerror=3Daction >=20 > Specify which action to take on write and read errors. Valid actions ar= e: =E2=80=9Cignore=E2=80=9D (ignore the error and try to continue), =E2=80= =9Cstop=E2=80=9D (pause QEMU), =E2=80=9Creport=E2=80=9D (report the error= to the guest), =E2=80=9Cenospc=E2=80=9D (pause QEMU only if the host dis= k is full; report the error to the guest otherwise). The default setting= is werror=3Denospc and rerror=3Dreport. > " Yes, it is. Generally you don't want to report such errors to the guest, because the guest sees that the disk is broken and stops using it. Even if you free some space, you wouldn't easily get the guest to use the disk again without rebooting. But if you really want, you can configure it so. It is not expected that the guest gets I/O errors (except possibly network timeouts if the VM is stopped for too long) or even qemu crashes. If it crashes again for you, please provide a gdb backtrace. Kevin > ----- Mail original ----- > De: "chenxg" > =C3=80: "Kevin Wolf" , "stefanha" , mimu@linux.vnet.ibm.com, tubo@cn.ibm.com, borntraeger@de.ibm.com, "Ho= lger Wolf" , "qemu-devel" > Envoy=C3=A9: Vendredi 17 Avril 2015 04:22:57 > Objet: [Qemu-devel] qcow2 problem: while a qcow2 image run out of disk = space the guest paused. >=20 > Hi Kevin,=20 >=20 > While installing a guest into a qcow2 image if the qcow2 image run out=20 > of disk space, the guest will stop work and change state to paused=20 > without messages.=20 >=20 > When we tried to free up disk space in the host and the virsh resume to= =20 > work. The guest then got I/O errors but later on QEMU also crashed.=20 >=20 > A guest in paused state is certainly not the best result.=20 >=20 > Probably good way to react are:=20 > - return I/O errors to the guest and the let the guest handle those=20 > (Linux file system code might just crash)=20 > - allow resume only when disk is free again.=20 >=20 > What's your opinion?=20 >=20