Copying the description I just wrote into github, as it explains this behavior better:
A small correction - if it fails to find the DHCP record of a deleted host in the subnet under evaluation, it will not attempt to delete it from the subnet under evaluation. This allows it to carry on, and actually delete the record that is intended.
The problem arises due to the evaluation of DHCP records with a status of 'deleted'. These records are evaluated against the subnet that contains the actual host we want to delete, and if they are found, they are removed prior to executing the target deletion.
In the case that triggered this bug, the record with the DHCP deleted status was originally in a different subnet before it was removed. On a subsequent delete run for a second host, the previously deleted host could not be found in the subnet under evaluation, and line 198 never returned a value. This left the return value of find_record_by_hostname as the result of subnet.records.each (an array), which then caused the subnet.delete call to throw an error.
A better fix might be to track the subnets of hosts with a DHCP status of deleted, but this gets the job done without added complexity.
HTH.
Updated by 
Updated by 
Updated by Anonymous 
Updated by