Dear List, I have a simple setup where VMs are created using the following command line xm create /dev/null ramdisk=/boot/initrd.img-2.6.32-5-xen-amd64 kernel=/boot/vmlinuz-2.6.32-5-xen-amd64 name=vm0 vif="bridge=br0,mac=52:54:00:13:37:00" vcpus=8 memory=512 disk="file:/home/thomas/euca-debian-5.0-x86_64/vm1.img,xvda1,w" root=/dev/xvda1 file:/home/thomas/euca-debian-5.0-x86_64/vm1.img is an NFS mount For some reason, the virtual disk seems to get corrupted very frequently. Running e2fsck on the disk file (after powering down the VM with "xm shutdown vm0"), reports file system errors or a journal that needs recovering. Not sure where/if I am doing something wrong. I suspect the NFS-backed disk file to be the issue?! Any pointers where to look for further information appreciated! Thanks, Thomas. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
On Tue, Feb 1, 2011 at 1:03 AM, Thomas Knauth <thomas.knauth@googlemail.com> wrote:> For some reason, the virtual disk seems to get corrupted very > frequently. Running e2fsck on the disk file (after powering down the > VM with "xm shutdown vm0"), reports file system errors or a journal > that needs recovering. Not sure where/if I am doing something wrong. > > I suspect the NFS-backed disk file to be the issue?!Possibly.> Any pointers > where to look for further information appreciated!I suggest you do some tests: - replace "file:/" with "tap:aio:/" - copy the disk image to local disk and use that -- Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Hi Thomas, 2011/1/31 Thomas Knauth <thomas.knauth@googlemail.com>:> I have a simple setup where VMs are created using the following command line > file:/home/thomas/euca-debian-5.0-x86_64/vm1.img is an NFS mount > > For some reason, the virtual disk seems to get corrupted very > frequently. Running e2fsck on the disk file (after powering down theCan you try to "drop caches" via /proc after the vm shutdown? Then try hard and see if you can reproduce the issue. I think I remember loop devices are being buffered by the kernel, it will not care that they reside on a "sync" filesystem like nfs. Good luck, Florian -- the purpose of libvirt is to provide an abstraction layer hiding all xen features added since 2006 until they were finally understood and copied by the kvm devs. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
> Can you try to "drop caches" via /proc after the vm shutdown? Then try > hard and see if you can reproduce the issue. > I think I remember loop devices are being buffered by the kernel, it > will not care that they reside on a "sync" filesystem like nfs.I resorted to calling /bin/sync after shutting down the VM. This seems to help. _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users