[BitBucket user: Derek Merck]
[BitBucket date: 2018-07-28.18:21:19]

Using `jodogne/orthanc:1.4` docker image on a recent Debian Stretch Vagrant box.

1. Run 2 orthanc containers, source and target, that register one another as peers
2. Load a study into source, send it to target peer through UI -> success
3. Resend study to target peer -> failure after about 15 slices are sent
4. Send study from target back to source peer -> failure after about 15 slices are sent
5. Load another study into source (or create a new one by anonymizing the first), send it to target peer -> success
6. Resend new study to target peer -> failure after about 15 slices are sent

Sender reports an "HTTP status 0" in logs (which appears to actually be a 500 in the page headers?) No reference to the failure in the receivers logs, just the 15ish successful receipts.

Same sequence using DICOM modality sends succeeds.

Same sequence with docker image using recent build from source fails.

Same sequence with `jodogne/orthanc:1.3.2` image succeeds.

My suspicion based on narrowing it down to resends on the new threaded version is that the worker threads are able to keep up with new studies, but not with resends, which appear in the logs to also require file deletes.  So eventually the scheduler runs out of workers, makes the sender wait, and the sender immediately gives up.

Let me know if it would be helpful to have more details about my setup to reproduce.

Thanks again for your continuing work on this -- it is an amazing piece of open source software!