We had an EqualLogic controller fail-over at the weekend for some reason (fw 8.13) which resulted in some of our virtual machines getting paused (I have raised a case on techdirect about the array itself).

We use ovirt / kvm and someone at redhat has helpfully suggested that we make the following adjustments to multipathd config ( I am aware there is no support agreement for this):

You can set the timeout in multipath.conf.