[Qemu-discuss] Disk integrity problems

[Scott Hazelhurst]

Dear all

I’ve started to run into some problems with several of my machines — as it’s 
happened on several of my physical machines acquired at different times I don’t 
think it’s a hardware problem and running physical diagnostics doesn’t pick up 
problems.

Every week or so, one of my VMs’ root file system becomes read only. If I 
reboot it all seems to run OK.

However, 
1. On the VM, if I do an fsck -n on root file system of the VM it reports that 
there are errors. 
2. If I shutdown the VM, and on the host machine do a kpartx -a on the 
underlying partition, and then do an fsck -f on the VM’s disk, no errors are 
reported.
3. If I boot the VM with a livedisk using the same OS as the VM guest and do an 
fsck -n then no error is reported. (the version of fsck is the same as (1))

But if I reboot up the VM and do an fsck it still reports errors.
        
One variant of this that when I do (2) initially some errors are picked up and 
fixed by the fsck on the host, but after that (2) and (3) report no further 
errors but (1) does.

My set up is 
        • The VM hosts run Ubuntu 16.04 with qemu-vm  1:2.5+dfsg-5ubuntu10.9
        • My VM guests run Scientific Linux 6.8
        • The VM’s disk images are LVM2 logical volumes

Any help would be gratefully received

Many thanks

Scott