diff options
Diffstat (limited to 'docs')
| -rw-r--r-- | docs/MESHBAY_DESIGN.md | 17 |
1 files changed, 17 insertions, 0 deletions
diff --git a/docs/MESHBAY_DESIGN.md b/docs/MESHBAY_DESIGN.md index 8d14e37..3820587 100644 --- a/docs/MESHBAY_DESIGN.md +++ b/docs/MESHBAY_DESIGN.md @@ -1385,6 +1385,23 @@ for the CLI and the Node page, and a periodic debug line. When someone reports a transfer stuck at "waiting", that line is the only thing that will say whether the node ever had them in a queue at all. +**A transfer that ends says what it moved.** One INFO line per lease, however it +ended (closed by the client, the connection gone, reclaimed as idle): + +``` +transfer: end download 1a2b3c4d done 213.0 MB in 35.2 s (6.05 MB/s) sctp cwnd=180 KB rtt=22 ms loss=31 t3=0 +``` + +The bytes and the duration are the lease's own, from the first request under it +to the last byte handed to the channel. The `sctp` figures are the association's +totals since the connection opened, not the transfer's share: a connection +carries several transfers and they share one congestion window. `loss` counts +entries into fast recovery, each of which halves the window, and `t3` counts +retransmission timeouts, each of which drops it to one packet (aiortc records +neither; `transport/webrtc/sctp_stats.py` counts them). A window that stays at a +few hundred KB with `loss` climbing is a transfer held back by the path, and +before this line the journal could not say so. + The lease module is free of asyncio and of the transport: it decides, and the caller does the I/O. A queue that reveals itself only through a DataChannel is a queue nobody can prove things about. |