Comment 0 for bug 1570857

1. schema is initialized it reads servicechains from cassandra and mark them as stale
2. initial ifmap poll the result is processed and compared against the cassandra objects retrieved at step 1
3. process_stale_objects is called and remove all servicechains marked as stale
4. process_networks is called and recreate the servicechains

3. and 4. should be inverted

process_networks calls to Servicechain.create() will unset the create_staled property of servicechains that were retrieved at step 1 so that process_stale_objects will not destroy valid valid servicechains.

Observed on R2.21.x