6260 downloads (occasionally) (part)fail silently

0xFeedBeef

Left a series to download overnight, while most episodes downloaded fine, some where only partially downloaded, but were marked as OK; and while my mp4 validator would catch downloads where there's a discrepancy between the lengths of A-V tracks, some episodes were AV-duration consistent. Re-downloading the partial episodes obviously solved the problem and they downloaded fine.

The point is that it shouldn't happen- SF gets the manifest it downloads from, so it knows exactly what needs to be downloaded, if what is downloaded vs what was expected don't match, that should be flagged as a failure.

Sebastian001

For this reason, I always download each video twice and compare the file sizes.

0xFeedBeef

For this reason, I always download each video twice and compare the file sizes.
Sebastian001

You're doing too much work for yourself: basically you want to check the duration of the video track and the audio track, and the way that I catch others is (since I download subtitles into a separate file) is to check that the last timestamp in the subtitles file is not past the duration of the video track ;)

(in case you haven't figured out, my $job is in dependable and fault-tolerant computing ;) so all these sorts of (what we call) "reasonableness checks" are a second nature to me)

Sebastian001

You're doing too much work for yourself: basically you want to check the duration of the video track and the audio track, and the way that I catch others is (since I download subtitles into a separate file) is to check that the last timestamp in the subtitles file is not past the duration of the video track ;)

(in case you haven't figured out, my $job is in dependable and fault-tolerant computing ;) so all these sorts of (what we call) "reasonableness checks" are a second nature to me)
0xFeedBeef

Not much work for me, more so for my SSD. I add 100 tracks to the queue, save the system file to a different location, and then move it back after the download. The entire queue is then restored.