Fixed a bug where the server died on client interrupt. Also updated README, TODO, and man page.

2010-07-08 22:41:22 +00:00 · 2010-07-08 22:41:22 +00:00 · 1ac522ec39
--- a/35
+++ b/35
@ -1,7 +1,15 @@
+Summary
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

+iperf3 is a tool for measuring Internet bandwidth performance.
 This is iperf3, a tool for measuring Internet bandwidth performance.

-iperf3 should work on Linux, FreeBSD, and Mac OSX
+Supported platforms:
+   iperf3 should work on Linux, FreeBSD, and Mac OSX
+
+
+Building iperf3
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 Prerequisites:
   libuuid: this is not installed by default for Debian/Ubuntu Systems
@ -9,7 +17,28 @@ Prerequisites:

 To build, just type 'make'  (or 'gmake' on FreeBSD)

+
+Release Notes
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+iperf - Version 3, beta 1
+
+New in this version:
+  * TCP control socket now manages messages between client and server
+  * Dynamic server (gets test parameters from client)
+      - Server can now set test options dynamically without having to restart.
+             Currently supported options: -l, -t, -n, -P, -R
+             Future options: -u, -b, -w, -M, -N, -I, -T, -Z, -6
+  * Results exchange
+      - Client can now see server results (and vice versa)
+  * Reverse mode (-R)
+      - Server sends, client receives
+
+Note: iperf3 is NOT backwards compatible with iperf2. 
+
+
+Documentation
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
 For more information see: http://code.google.com/p/iperf/

-Note that iperf3 is NOT backwards compatible with iperf2. 
-
--- a/12
+++ b/12
@ -3,18 +3,16 @@ Current list of things to fix/add to iperf3

 -  readd -Z option (linux tcp congestion control algoritm)
 -  readd flag to set IP_TOS (-S)
-  finish adding verbose and debug options
-  separate iperf_api.c into iperf_client.c and iperf_utils.c 
+-  support for setting TCP options (MSS, no delay, etc.)
+-  add verbose and debug options
+-  separate iperf_api.c into iperf_client.c and iperf_utils.c
+-  add full iperf API support and documentation 
 -  look for 'XXX' in code and address
 -  much better/standard error handling throughout
 -  better packaging/makefile, README, LICENCE, etc files
-  finish/fix receive_result_from_server()
-      - should this be called for TCP too, or only UDP (currently its both,
-              but I think it should be UDP only, or maybe a command line option for TCP
 -  document and verify the 'state machine'. Is it an error to send messages in the wrong order?
     - e.g.: what is "STREAM_RUNNING" vs "TEST_RUNNING"??
-  cleanup/fix/test UDP mode
-       add TCP control socket?
+-  add/test UDP mode
       use clock_nanosleep() for more accurate timing in Linux?
 	(http://www.kernel.org/doc/man-pages/online/pages/man2/clock_nanosleep.2.html)
       fix server to handle both TCP and UDP
--- a/src/iperf.1
+++ b/src/iperf.1
@ -1,4 +1,4 @@
-.TH IPERF 1 "APRIL 2008" NLANR/DAST "User Manuals"
+.TH IPERF 1 "July 2010" NLANR/DAST "User Manuals"
 .SH NAME
 iperf \- perform network throughput tests
 .SH SYNOPSIS
@ -12,129 +12,79 @@ iperf \- perform network throughput tests
 .I options
 .B ]

-.B iperf -u -s [
-.I options
-.B ]

-.B iperf -u -c 
-.I server
-.B [
-.I options
-.B ]
 .SH DESCRIPTION
 iperf is a tool for performing network throughput measurements.  It can test
 either TCP or UDP throughput.  To perform an iperf test the user must
-establish both a server (to discard traffic) and a client (to generate
-traffic).  
+establish both a server and a client.
+
+
 .SH "GENERAL OPTIONS"
 .TP
+.BR -p ", " --port " \fIn\fR"
+set server port to listen on/connect to to \fIn\fR (default 5201)
+.TP
 .BR -f ", " --format " "
 [kmKM]   format to report: Kbits, Mbits, KBytes, MBytes
 .TP
-.BR -h ", " --help " "
-print a help synopsis
-.TP
 .BR -i ", " --interval " \fIn\fR"
 pause \fIn\fR seconds between periodic bandwidth reports
 .TP
-.BR -l ", " --len " \fIn\fR[KM]"
-set length read/write buffer to \fIn\fR (default 8 KB)
-.TP
 .BR -m ", " --print_mss " "
 print TCP maximum segment size (MTU - TCP/IP header)
 .TP
-.BR -o ", " --output " <filename>"
-output the report or error message to this specified file
+.BR -h ", " --help " "
+print a help synopsis
 .TP
-.BR -p ", " --port " \fIn\fR"
-set server port to listen on/connect to to \fIn\fR (default 5001)
+.BR -v ", " --version " "
+print version information and quit
+
+
+.SH "SERVER SPECIFIC OPTIONS"
 .TP
-.BR -u ", " --udp " "
-use UDP rather than TCP
+.BR -s ", " --server " "
+run in server mode
+
+
+.SH "CLIENT SPECIFIC OPTIONS"
+.TP
+.BR -c ", " --client " <host>"
+run in client mode, connecting to <host>
+.TP
+.BR -t ", " --time " \fIn\fR"
+time in seconds to transmit for (default 10 secs)
+.TP
+.BR -n ", " --num " \fIn\fR[KM]"
+number of bytes to transmit (instead of -t)
+.TP
+.BR -l ", " --len " \fIn\fR[KM]"
+set length read/write buffer to \fIn\fR (default 128 KB)
+.TP
+.BR -P ", " --parallel " \fIn\fR"
+number of parallel client streams to run
+.TP
+.BR -R ", " --reverse
+run in reverse mode (server sends, client receives)
 .TP
 .BR -w ", " --window " \fIn\fR[KM]"
 TCP window size (socket buffer size)
 .TP
-.BR -B ", " --bind " <host>"
-bind to <host>, an interface or multicast address
-.TP
-.BR -C ", " --compatibility " "
-for use with older versions does not sent extra msgs
-.TP
 .BR -M ", " --mss " \fIn\fR"
 set TCP maximum segment size (MTU - 40 bytes)
 .TP
 .BR -N ", " --nodelay " "
 set TCP no delay, disabling Nagle's Algorithm
-.TP
-.BR -v ", " --version " "
-print version information and quit
-.TP
-.BR -V ", " --IPv6Version " "
-Set the domain to IPv6
-.TP
-.BR -x ", " --reportexclude " "
-[CDMSV]   exclude C(connection) D(data) M(multicast) S(settings) V(server) reports
-.TP
-.BR -y ", " --reportstyle " C|c"
-if set to C or c report results as CSV (comma separated values)
-.SH "SERVER SPECIFIC OPTIONS"
-.TP
-.BR -s ", " --server " "
-run in server mode
-.TP
-.BR -U ", " --single_udp " "
-run in single threaded UDP mode
-.TP
-.BR -D ", " --daemon " "
-run the server as a daemon
-.SH "CLIENT SPECIFIC OPTIONS"
-.TP
-.BR -b ", " --bandwidth " \fIn\fR[KM]"
-set target bandwidth to \fIn\fR bits/sec (default 1 Mbit/sec).
-This setting requires UDP (-u).
-.TP
-.BR -c ", " --client " <host>"
-run in client mode, connecting to <host>
-.TP
-.BR -d ", " --dualtest " "
-Do a bidirectional test simultaneously
-.TP
-.BR -n ", " --num " \fIn\fR[KM]"
-number of bytes to transmit (instead of -t)
-.TP
-.BR -r ", " --tradeoff " "
-Do a bidirectional test individually
-.TP
-.BR -t ", " --time " \fIn\fR"
-time in seconds to transmit for (default 10 secs)
-.TP
-.BR -F ", " --fileinput " <name>"
-input the data to be transmitted from a file
-.TP
-.BR -I ", " --stdin " "
-input the data to be transmitted from stdin
-.TP
-.BR -L ", " --listenport " \fIn\fR"
-port to recieve bidirectional tests back on
-.TP
-.BR -P ", " --parallel " \fIn\fR"
-number of parallel client threads to run
-.TP
-.BR -T ", " --ttl " \fIn\fR"
-time-to-live, for multicast (default 1)
-.TP
-.BR -Z ", " --linux-congestion " <algo>"
-set TCP congestion control algorithm (Linux only)
-.SH ENVIRONMENT
-.TP
-.BR TCP_WINDOW_SIZE
-Controls the size of TCP buffers.
+
+
 .SH DIAGNOSTICS
 This section needs to be filled in.
+
+
 .SH BUGS
 Exit statuses are inconsistent.
-The threading implementation is rather heinous.
+Could possibly block indefinitely under some circumstances.
+
+
 .SH AUTHORS
 Iperf was originally written by Mark Gates and Alex Warshavsky.
 Man page and maintence by Jon Dugan <jdugan at x1024 dot net>.
@ -144,5 +94,7 @@ Kevin Gibbs,
 John Estabrook <jestabro at ncsa.uiuc.edu>,
 Andrew Gallatin <gallatin at gmail.com>,
 Stephen Hemminger <shemminger at linux-foundation.org>
+
+
 .SH "SEE ALSO"
-http://iperf.sourceforge.net/
+http://code.google.com/p/iperf/
--- a/src/iperf_api.c
+++ b/src/iperf_api.c
@ -1151,7 +1151,7 @@ iperf_run_client(struct iperf_test * test)

    signal(SIGINT, sig_handler);
    if (setjmp(env)) {
-        fprintf(stderr, "Interrupt received. Exiting...\n");
+        fprintf(stderr, "Exiting...\n");
        test->state = CLIENT_TERMINATE;
        if (Nwrite(test->ctrl_sck, &test->state, sizeof(char), Ptcp) < 0) {
            fprintf(stderr, "Unable to send CLIENT_TERMINATE message to serer\n");
--- a/src/iperf_server_api.c
+++ b/src/iperf_server_api.c
@ -267,8 +267,14 @@ iperf_handle_message_server(struct iperf_test *test)
        case IPERF_DONE:
            break;
        case CLIENT_TERMINATE:
-            fprintf(stderr, "The client has terminated. Exiting...\n");
-            exit(1);
+            fprintf(stderr, "The client has terminated.\n");
+            for (sp = test->streams; sp; sp = sp->next) {
+                FD_CLR(sp->socket, &test->read_set);
+                FD_CLR(sp->socket, &test->write_set);
+                close(sp->socket);
+            }
+            test->state = IPERF_DONE;
+            break;
        default:
            // XXX: This needs to be replaced by actual error handling
            fprintf(stderr, "Unrecognized state: %d\n", test->state);
@ -343,7 +349,7 @@ iperf_run_server(struct iperf_test *test)

    signal(SIGINT, sig_handler);
    if (setjmp(env)) {
-        fprintf(stderr, "Interrupt received. Exiting...\n");
+        fprintf(stderr, "Exiting...\n");
        test->state = SERVER_TERMINATE;
        if (test->ctrl_sck >= 0) {
            if (Nwrite(test->ctrl_sck, &test->state, sizeof(char), Ptcp) < 0) {