Create a DICOMweb transfer endpoint ("heal") that is able to compare a resource stored locally to that stored on a remote and transfer the failed instances
It is not uncommon for a DICOMweb transfer in Orthanc to fail because of network instability, timeouts, or other issues. When this happens, there will be a partially complete series that is missing instances. How to fix (heal) these failed transfers is not always straightforward.
One potential solution might be to create a "heal" endpoint within the Orthanc plugin that is able to retrieve the instance UIDs from the remote server, compare those to the instances stored locally, create a list, and then queue a transfer job for only those instances. This would allow for the device to repair a damaged instance with the fewest number of HTTP calls required.