Blob mirror download does not resume after all blobs sent to reflector

[tzarebczan]

The Issue

I've seen this scenario happen twice on spee.ch over the weekend and was able to confirm locally. The other scenario where this will happen is with YouTube Sync - we may not be done with mirroring all blobs before users try to download them (via subscriptions).

Scenario: I publish a 1GB file to LBRY, it confirms on the blockchain, and continues to upload to reflector in the background. During this time, I try to access it on spee.ch or send the URL to a friend. spee.ch/friend accesses the file and starts downloading from reflector. They are able to grab 500MB, which is where the upload progress left off on the reflector.

Result: The download will be stuck at 50% and not retry even though more blobs have been written to reflector. Calling a get does not help - lbrynet.daemon.Daemon:1378: Already have a file for testreflector002. File_set_status start doesn't help either. file_set_status stop results in: ERROR: [Error 32] The process cannot access the file because it is being used by another process: 'D:\\LBRY Downloads Main PC\\RobotMonster_weboptimized.mp4'

The file will resume on daemon restart or a P2P node announces it (did not test this locally, but seen it happen on spee.ch)

Potential solution:
Keep checking for new blobs via blob_mirror at some interval

System Configuration

• LBRY Daemon version:
• LBRY App version:
• LBRY Installation ID:
• Operating system:

Anything Else

Screenshots

[tzarebczan]

I mentioned to @jackrobison that this would be good to get into a patch release, with any other smaller fixes we may have. The app team is planning to get a release out on Thursday, so we'd need to ready by Wednesday afternoon the latest.