Document column headers for proto_load

diff --git a/raddb/sites-available/load b/raddb/sites-available/load
index e9d0f4784b1c1e3b41b4e58d1edb94e7068ba3c0..f55be4dae0c233ae91598377890a78be3fb0061b 100644 (file)
--- a/raddb/sites-available/load
+++ b/raddb/sites-available/load
@@ -46,6 +46,44 @@ server load {
                        #
                        #  Where the statistics file goes, in CSV format.
                        #
+                       #  One line will be written to the file for each
+                       #  second of the load generation run.
+                       #
+                       #  Each line will contain the following columns:
+                       #  - time - since the test started running
+                       #  - last_packet - The last time we added a request
+                       #    to the backlog.
+                       #  - rtt - Round Trip Time i.e. the average delay
+                       #    between a request and response.  This is a
+                       #    moving average in nanoseconds.
+                       #  - rttvar - Round Trip Time variance.  Moving
+                       #    average of the range between the smallest RTT
+                       #    and largest RTT.  The value is in nanoseconds.
+                       #  - pps - Packets per second.  The maximum packet
+                       #    rate we're aiming for with this "step".  The
+                       #    load generator increases the load periodically
+                       #    in "steps".
+                       #  - pps_accepted - Packets per second. Rate of
+                       #    response packets received.
+                       #  - sent - How many packets have been generated
+                       #    by the load generation module from the start of
+                       #    the run.
+                       #  - received - How many packets received since the
+                       #    start of the run.
+                       #  - backlog - How many requests are awaiting
+                       #    responses.
+                       #  - backlog_max - The largest the backlog has been
+                       #    since the start of the run.
+                       #  - <usec - Absolute counter.  Responses received
+                       #    in less that a microsecond since the start
+                       #    of the run.
+                       #  - us - Responses received in >= 1us < 10us since
+                       #    the start of the run.
+                       #  - 100us, ms, 10ms, 100ms, s are all similar
+                       #    latency bins.
+                       #  - blocked - 1 = true, 0 = false.   We're refusing
+                       #    to enqueue more packets until we get responses
+                       #    to the outstanding requests.
                        csv = ${confdir}/stats.csv
 
                        #