throughput: the upstream direction, counted by the only party that can

The client generates the traffic and the server counts it. No grant is involved
- the client is sending its own packets, so there is nothing to amplify - but it
does need the server's tally, because only the far end knows how much arrived.
Without that number a sender measures how fast it can transmit, which is usually
just the speed of the local NIC and is not the question being asked.

A new wire type the server counts and deliberately never answers: a reply would
double the traffic and drag the return path into a measurement that is
specifically about the outbound one.

The tally is a counter, not a list, and short-circuits before the observation
log. A five-second run at 20 Mbps is around ten thousand packets; one struct
each would turn a measurement into an allocation storm on a shared server, and
nothing needs the per-packet detail since the client holds the send-side record.
The gap between the two counts is the loss.

direction=up on the throughput action sends nothing - it zeroes the counter, so
a second run in one session measures itself instead of inheriting the first.

Same honesty rule as downstream: measures_network is false when what arrived
matches what was offered, because then the path was never the constraint.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

echolot-app/core-protocol/src/main/kotlin/app/echo_lot/protocol/ProbeSession.kt

@@ -109,6 +109,49 @@ class ProbeSession(
         return out
     }
 
+    /**
+     * Sends paced upstream traffic for [durationMs] and reports what was put on the wire.
+     *
+     * Paced rather than flat out, for the same reason the server paces: an unpaced burst measures
+     * the local NIC and the first queue it meets, then collapses into loss that reads as a network
+     * fault. The schedule is absolute rather than sleep-per-packet, which accumulates the
+     * scheduler's error and drifts the achieved rate below target over a multi-second run.
+     *
+     * Nothing comes back — the server counts and stays silent — so the result here is only the
+     * send side. The measurement is the gap between this and the server's tally.
+     */
+    fun sendThroughput(durationMs: Long, kbps: Int, sizeBytes: Int = 1200): Sent {
+        val size = sizeBytes.coerceIn(Wire.HEADER_SIZE + 16, 1472)
+        val payload = ByteArray(size - Wire.HEADER_SIZE)
+        val perPacketNs = (size.toLong() * 8 * 1_000_000 / kbps.coerceAtLeast(1)).coerceAtLeast(1_000)
+
+        val start = System.nanoTime()
+        val deadline = start + durationMs * 1_000_000
+        var next = start
+        var packets = 0
+        var bytes = 0L
+        while (System.nanoTime() < deadline) {
+            val pkt = Wire.build(Wire.TYPE_THROUGHPUT_UP, prefix, ++seq, nowNs(), key, payload)
+            try {
+                socket.send(DatagramPacket(pkt, pkt.size, server))
+            } catch (e: java.io.IOException) {
+                // A local send failure is our condition, not the path's. Stop and report what
+                // actually left, rather than counting the remainder as loss on the network.
+                break
+            }
+            packets++
+            bytes += pkt.size
+            next += perPacketNs
+            val sleepNs = next - System.nanoTime()
+            if (sleepNs > 0) Thread.sleep(sleepNs / 1_000_000, (sleepNs % 1_000_000).toInt())
+        }
+        val elapsedMs = (System.nanoTime() - start) / 1_000_000
+        return Sent(packets, bytes, elapsedMs, if (elapsedMs > 0) (bytes * 8 / elapsedMs).toInt() else 0)
+    }
+
+    /** What one upstream run put on the wire locally. */
+    data class Sent(val packets: Int, val bytes: Long, val durationMs: Long, val kbps: Int)
+
     /** One packet received from the server, with the wire size actually delivered. */
     data class Received(val type: Int, val seq: Int, val sizeBytes: Int, val tRxNs: Long)
 

echolot-app/core-protocol/src/main/kotlin/app/echo_lot/protocol/Wire.kt

@@ -41,6 +41,13 @@ object Wire {
     /** One packet of a sustained-rate downstream run. */
     const val TYPE_THROUGHPUT_DATA: Int = 0x0E
 
+    /**
+     * One packet of a client-driven upstream run. The server counts it and does not answer:
+     * a reply would double the traffic and drag the return path into a measurement that is
+     * specifically about the outbound one.
+     */
+    const val TYPE_THROUGHPUT_UP: Int = 0x0F
+
     /** The 8-byte on-the-wire prefix = first 16 hex chars of the session id, decoded. */
     fun wirePrefix(sessionId: String): ByteArray {
         require(sessionId.length >= 16) { "session id too short" }