From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: out of memory error leads to unbootable VM Date: Tue, 02 Jun 2009 19:27:51 +0300 Message-ID: <4A255307.2070005@redhat.com> References: <1243959325.14381.27.camel@corn.betterworld.us> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm@vger.kernel.org To: Ross Boylan Return-path: Received: from mx2.redhat.com ([66.187.237.31]:50715 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754089AbZFBQ2r (ORCPT ); Tue, 2 Jun 2009 12:28:47 -0400 In-Reply-To: <1243959325.14381.27.camel@corn.betterworld.us> Sender: kvm-owner@vger.kernel.org List-ID: Ross Boylan wrote: > I had a VM running Linux. While running that SAS installer the kernel > ran out of memory and killed the installer. After that, any attempt to > run a binary inside the VM produced "cannot execute binary file". The > log (inside the VM) also showed messages like > "Buffer I/O error on device hda1, logical block 12243549368". > hda1 is 86\% full. > > I killed the VM and attempted to restart. However, the virtual BIOS > says the disk is unrecognizeable. > > lvscan shows, among other things, > inactive Original '/dev/turtle/Lenny00' [5.00 GB] inherit > ACTIVE '/dev/turtle/SAS92' [33.00 GB] inherit > inactive Snapshot '/dev/turtle/Lenny01SAS' [3.00 GB] inherit > The active volume has the installation disk; Lenny01SAS was the "disk" > the VM was using; it is a snapshot of Lenny00. I have not deliberately > deactivated either. > > lvdisplay does not give any indication of COW overflow, though that may > be because the volumes are inactive. > > Only about 11G of SAS92 are in use. The SAS installation docs seem to > indicate 1 to 1.5G for a full installation. Although it looks as if I > had enough room, I suppose a snapshot overflow could account for the > inactive snapshot (and inactive original?). > > It probably did run out of virtual RAM, since kvm started > (inadvertently) with the default RAM. > > Any ideas what's going on? > > Sounds like a guest bug that caused guest disk corruption. Try with more memory (on new volumes to be sure). -- error compiling committee.c: too many arguments to function