No bytes for ten minutes usually means a peer has queued us, and the
stall timer could not tell position 2 from position 400: the first was
abandoned while it was about to start, the second was waited on for ten
minutes for nothing.
While a requested file is "Queued, Remotely", the grab asks slskd for
its place (GET .../downloads/{user}/{id}/position, which asks the peer)
once a minute. A place that improved counts as progress and restarts
the stall clock. A place beyond 50 twice running gives the peer up at
once, and the manager moves to the next copy; two readings because
slskd documents the figure as possibly inaccurate. Waiting in a queue
has an overall ceiling of an hour without a byte, since a queue moving
one place an hour would otherwise hold the grab all day. As with a
stall, files that already arrived still go forward.
Closes#275
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017HJiuc3ZZhxsPXz3ozTirT
