Comment 5 for bug 1488608

I would like to have core members weigh in on this bug - how much of a value is there when someone fixes this bug for a case in which the cluster has two EC storage policies running? I don't know if someone out there in the field would like to have the cluster running jerasure policy with 10+4 scheme and then liberasurecode policy with a different scheme - it just seems unlikely to me at the point. I could be wrong, but, thus, core members' input is desired here.

Basically, the reconstructor iterates each policy and checks the devices associated with the policy ring and assigns them to self.device_count instead of increments them.

The quickest way to remediate this bug would be the following in reconstructor.py, collect_parts method, but I haven't tested much yet:

Lines 800 thru 803:
if override_devices:
   self.device_count += len(override_devices)
else:
    self.device_count += len(local_devices)