This makes sense given the generic nature of block jobs.  If mirroring
was only for live migration, for example, then we could avoid all this
by choosing a single policy.  As a generic operation it's nice to have
control over error policy.

Stefan


Reply via email to