[pve-devel] Default cache mode for VM hard drives

[Dietmar Maurer]

> Again, migration code flush all changes to disk, so there are no "out of sync"
> blocks
> after migration. What do I miss?
> 
> I'll try to explain within more details.
> When write cache is enabled then KVM process write, write and write. It doesn't
> care about what really happen to this data after it goes to buffer.

Sorry, but we use "cache=none", so cache is not enabled (what cache do you talk exactly?).

> What DRBD does it is writing data from buffer to both nodes but DRBD can't do
> this simultaneously. So it writes data to local node first and then to another
> node. Between these 'writes' data in buffer can be changed and nobody know it
> was changed. So every time data is written DRBD can't be sure that data written
> locally and data written remotely are identical.
> Why barriers usually helps? Because OS inside VM doesn't write anything until
> 'data is committed' message received from DRBD. 'Data
>  is committed' comes from DRBD when real data committed to both nodes, local
> and remote.

Right. Both, the VM kernel, and the KVM migration code issue a flush when required.
So you never end up in inconsistent state.

Migration would never work if above theory is correct. But it works perfectly
with iSCSI, NFS, glusterfs, ceph, ...