NAME
rtpmon - A Third Party RTCP Monitor
SYNOPSIS
rtpmon [ -C conferenceName ] [ -d display ] [ -t threshold ]
[ -u script ] [ -X resource=value ] addr/port
DESCRIPTION
Rtpmon is a third-party RTCP monitor. It can be used to
monitor the control information exchanged between applica-
tions that implement RTP, the Real-Time Transport Protocol.
Feedback from receivers, including the loss rate and jitter,
are displayed in a table that can be sorted in various ways
to help isolate and diagnose multicast distribution prob-
lems.
OPTIONS
-C Use conferenceName as the title for the main rtpmon
window. If the -C flag is ommitted, the destination
address and port are used as the window title.
-d Connect to the X server indicated by the display argu-
ment.
-t Make threshold the default threshold for loss rate in
the main display window as described below.
-u Source script, in addition to the compiled-in script,
to build the user interface. This is only useful dur-
ing development.
-X Override the X resource Rtpmon.resource with value.
OPERATION
Upon startup, rtpmon starts listening for RTCP packets sent
to the RTP session on the multicast address addr and the
appropriate RTPv2 control port corresponding to port. (that
is, if port is odd, it is used and if port is even, port + 1
is used).
The main window is a table with current senders listed
across the top and listeners along the left side. Each row
in the table displays that listener's statistics about its
reception from each sender. The default display shows loss
rate measured over the last 4 RTCP packets received from the
listener. Other statistics can be displayed as described
below.
Clicking on a name (either a sender or a listener) will
bring up a window that includes the latest contents of all
RTCP SDES packets seen from that source as well as their RTP
synchronization source id, IP address, and the time of their
most recent control packet.
Clicking on a statistic in the main display will bring up a
window with stripcharts that display the recent history of
all the statistics available. Three statistics are
currently displayed. Packet Loss is calculated by comparing
the number of dropped packets to the number of expected
packets over the last 4 RTCP packets from a listener. Fil-
tered Packet Loss is computed with a Moving Weighted Average
over the ``fraction lost'' field of all RTCP packets from a
listener. Delay Jitter is computed with a Moving Weighted
Average over the ``interarrival jitter'' field of all RTCP
packets from a listener.
If mtrace(8) is installed, the stripchart window will have a
button at the bottom that will run mtrace from the sender to
the listener via the multicast group used for the session.
The button labeled ``Menu'' at the bottom of the main window
brings up a new menu window. The statistic displayed in the
main window can be selected from this dialog. Various other
parameters (as described below) can also be changed from the
menu window.
THRESHOLDS
In a large conference, there may be many listeners with very
low loss rates and jitter. To keep the display from getting
too cluttered and to focus on interesting statistics, it is
possible to display only those listeners for which the loss
rate (or another statistic) is above some threshold. The
default threshold comes from the Rtpmon.threshold X resource
or the -t command line switch. The threshold associated
with each display parameter can also be changed from the
menu window.
To keep the display reasonably stable, if a listener is
displayed in the main window and their loss rate then falls
below the threshold, they are not immediately removed from
the display. The button labled ``Clean'' on the main window
will remove those listeners that have loss rates below the
treshold. Pressing the `c' key in the main window does the
same thing.
The display can be cleaned periodically by clicking the
``Autoclean'' button in the menu and selecting the interval
(in seconds) at which the display should be automatically
cleaned. Autoclean can be turned on by default by setting
the X resource Rtpmon.autoclean to 1 (the default is 0 which
turns autoclean off). The default autoclean interval can be
set by the X resource Rtpmon.autocleanInterval.
SORTING
The main display can be sorted by any of several keys. The
key to be used can be chosen from the ``Sorting'' section of
the menu. Maximum Loss sorts the rows by the maximum loss
seen in all columns. Average Loss sorts the rows by the
loss rate averaged over all columns. IP Address sorts the
rows by the IP address from which RTCP packets are received.
IP addresses can be used for crude topographical grouping.
A single column can be chosen by selecting Sender and then
choosing a particular sender.
By default, the display is not continuously re-sorted. It
can be explicitly sorted by pressing the ``Sort'' button on
the main window or pressing the `s' key. The display can be
sorted periodically using the ``Autosort'' button in the
menu which is analogous to ``Autoclean''.
X RESOURCES
The following are the names and default values of X
resources used by rtpmon. Any of these resources can be
overridden by the -X command switch, which may be used mul-
tiple times on the command line. For example, "-
Xautosort=1" overrides Rtpmon.autosort with 1.
Rtpmon.threshold (1)
the default threshold for loss rate in the main display
window
Rtpmon.autoclean (0)
turns on or off autocleaning.
Rtpmon.autocleanInterval (5)
interval (in seconds) at which the display should be
cleaned if autocleaning is on
Rtpmon.autosort (0)
turns on or off autosorting
Rtpmon.autosortInterval (5)
interval (in seconds) at which the display should be
sorted if autosorting is on
Rtpmon.refreshInterval (.5)
interval (in seconds) at which values in the main
display should be refreshed
Rtpmon.sessbw (64.)
the estimated data bandwidth for the session; used to
estimate the frequency with which RTCP packets should
be expected.
SEE ALSO
vic(1), vat(1), mtrace(8),
Schulzrinne, Casner, Frederick, Jacobson, ``RTP: A Transport
Protocol for Real-Time Applications'', Internet Request For
Comments 1889, available via anonymous ftp at
ftp://ds.internic.net/rfc/rfc1889.txt.
ACKNOWLEDGMENTS
The original design for rtpmon was suggested by Van Jacob-
son. The software architecture was inspired by, and bits of
code have been borrowed from, vic and vat written by Van
Jacobson and Steve McCanne. Bill Fenner has also provided
many useful comments, suggestions, and bug fixes.
AUTHORS
David Bacher (drbacher@cs.berkeley.edu) and Andrew Swan
(aswan@cs.berkeley.edu)
BUGS
This is an alpha release of rtpmon so there are likely to be
many bugs. A conference with multiple simultaneous senders
is not handled very well right now. Older MBone software
may not conform to the RTP specification, leading to odd
reports for packet loss and jitter from receivers using
those programs. Please report any other bugs to
rtpmon@bmrc.berkeley.edu