1Invoking iperf3 2=============== 3 4iperf3 includes a manual page listing all of the command-line options. 5The manual page is the most up-to-date reference to the various flags and parameters. 6 7For sample command line usage, see: 8 9https://fasterdata.es.net/performance-testing/network-troubleshooting-tools/iperf/ 10 11Using the default options, iperf3 is meant to show typical well 12designed application performance. "Typical well designed application" 13means avoiding artificial enhancements that work only for testing 14(such as ``splice()``-ing the data to ``/dev/null``). iperf3 does 15also have flags for "extreme best case" optimizations but they must be 16explicitly activated. These flags include the ``-Z`` (``--zerocopy``) 17and ``-A`` (``--affinity``) options. 18 19iperf3 Manual Page 20------------------ 21 22This section contains a plaintext rendering of the iperf3 manual page. 23It is presented here only for convenience; the text here might not 24correspond to the current version of iperf3. The authoritative iperf3 25manual page is included in the source tree and installed along with 26the executable. 27 28:: 29 30 IPERF3(1) User Manuals IPERF3(1) 31 32 33 34 NAME 35 iperf3 - perform network throughput tests 36 37 SYNOPSIS 38 iperf3 -s [ options ] 39 iperf3 -c server [ options ] 40 41 42 DESCRIPTION 43 iperf3 is a tool for performing network throughput measurements. It 44 can test TCP, UDP, or SCTP throughput. To perform an iperf3 test the 45 user must establish both a server and a client. 46 47 The iperf3 executable contains both client and server functionality. 48 An iperf3 server can be started using either of the -s or --server com- 49 mand-line parameters, for example: 50 51 iperf3 -s 52 53 iperf3 --server 54 55 Note that many iperf3 parameters have both short (-s) and long 56 (--server) forms. In this section we will generally use the short form 57 of command-line flags, unless only the long form of a flag is avail- 58 able. 59 60 By default, the iperf3 server listens on TCP port 5201 for connections 61 from an iperf3 client. A custom port can be specified by using the -p 62 flag, for example: 63 64 iperf3 -s -p 5002 65 66 After the server is started, it will listen for connections from iperf3 67 clients (in other words, the iperf3 program run in client mode). The 68 client mode can be started using the -c command-line option, which also 69 requires a host to which iperf3 should connect. The host can by speci- 70 fied by hostname, IPv4 literal, or IPv6 literal: 71 72 iperf3 -c iperf3.example.com 73 74 iperf3 -c 192.0.2.1 75 76 iperf3 -c 2001:db8::1 77 78 If the iperf3 server is running on a non-default TCP port, that port 79 number needs to be specified on the client as well: 80 81 iperf3 -c iperf3.example.com -p 5002 82 83 The initial TCP connection is used to exchange test parameters, control 84 the start and end of the test, and to exchange test results. This is 85 sometimes referred to as the "control connection". The actual test 86 data is sent over a separate TCP connection, as a separate flow of UDP 87 packets, or as an independent SCTP connection, depending on what proto- 88 col was specified by the client. 89 90 Normally, the test data is sent from the client to the server, and mea- 91 sures the upload speed of the client. Measuring the download speed 92 from the server can be done by specifying the -R flag on the client. 93 This causes data to be sent from the server to the client. 94 95 iperf3 -c iperf3.example.com -p 5202 -R 96 97 Results are displayed on both the client and server. There will be at 98 least one line of output per measurement interval (by default a mea- 99 surement interval lasts for one second, but this can be changed by the 100 -i option). Each line of output includes (at least) the time since the 101 start of the test, amount of data transferred during the interval, and 102 the average bitrate over that interval. Note that the values for each 103 measurement interval are taken from the point of view of the endpoint 104 process emitting that output (in other words, the output on the client 105 shows the measurement interval data for the client. 106 107 At the end of the test is a set of statistics that shows (at least as 108 much as possible) a summary of the test as seen by both the sender and 109 the receiver, with lines tagged accordingly. Recall that by default 110 the client is the sender and the server is the receiver, although as 111 indicated above, use of the -R flag will reverse these roles. 112 113 The client can be made to retrieve the server-side output for a given 114 test by specifying the --get-server-output flag. 115 116 Either the client or the server can produce its output in a JSON struc- 117 ture, useful for integration with other programs, by passing it the -J 118 flag. Because the contents of the JSON structure are only completely 119 known after the test has finished, no JSON output will be emitted until 120 the end of the test. 121 122 iperf3 has a (overly) large set of command-line options that can be 123 used to set the parameters of a test. They are given in the "GENERAL 124 OPTIONS" section of the manual page below, as well as summarized in 125 iperf3's help output, which can be viewed by running iperf3 with the -h 126 flag. 127 128 GENERAL OPTIONS 129 -p, --port n 130 set server port to listen on/connect to to n (default 5201) 131 132 -f, --format 133 [kmgtKMGT] format to report: Kbits/Mbits/Gbits/Tbits 134 135 -i, --interval n 136 pause n seconds between periodic throughput reports; default is 137 1, use 0 to disable 138 139 -I, --pidfile file 140 write a file with the process ID, most useful when running as a 141 daemon. 142 143 -F, --file name 144 Use a file as the source (on the sender) or sink (on the 145 receiver) of data, rather than just generating random data or 146 throwing it away. This feature is used for finding whether or 147 not the storage subsystem is the bottleneck for file transfers. 148 It does not turn iperf3 into a file transfer tool. The length, 149 attributes, and in some cases contents of the received file may 150 not match those of the original file. 151 152 -A, --affinity n/n,m 153 Set the CPU affinity, if possible (Linux, FreeBSD, and Windows 154 only). On both the client and server you can set the local 155 affinity by using the n form of this argument (where n is a CPU 156 number). In addition, on the client side you can override the 157 server's affinity for just that one test, using the n,m form of 158 argument. Note that when using this feature, a process will 159 only be bound to a single CPU (as opposed to a set containing 160 potentially multiple CPUs). 161 162 -B, --bind host[%dev] 163 bind to the specific interface associated with address host. If 164 an optional interface is specified, it is treated as a shortcut 165 for --bind-dev dev. Note that a percent sign and interface 166 device name are required for IPv6 link-local address literals. 167 --bind-dev dev bind to the specified network interface. This 168 option uses SO_BINDTODEVICE, and may require root permissions. 169 (Available on Linux and possibly other systems.) 170 171 -V, --verbose 172 give more detailed output 173 174 -J, --json 175 output in JSON format 176 177 --logfile file 178 send output to a log file. 179 180 --forceflush 181 force flushing output at every interval. Used to avoid buffer- 182 ing when sending output to pipe. 183 184 --timestamps[=format] 185 prepend a timestamp at the start of each output line. By 186 default, timestamps have the format emitted by ctime(1). 187 Optionally, = followed by a format specification can be passed 188 to customize the timestamps, see strftime(3). If this optional 189 format is given, the = must immediately follow the --timestamps 190 option with no whitespace intervening. 191 192 --rcv-timeout # 193 set idle timeout for receiving data during active tests. The 194 receiver will halt a test if no data is received from the sender 195 for this number of ms (default to 12000 ms, or 2 minutes). 196 197 -d, --debug 198 emit debugging output. Primarily (perhaps exclusively) of use 199 to developers. 200 201 -v, --version 202 show version information and quit 203 204 -h, --help 205 show a help synopsis 206 207 208 SERVER SPECIFIC OPTIONS 209 -s, --server 210 run in server mode 211 212 -D, --daemon 213 run the server in background as a daemon 214 215 -1, --one-off 216 handle one client connection, then exit. If an idle time is 217 set, the server will exit after that amount of time with no con- 218 nection. 219 220 --idle-timeout n 221 restart the server after n seconds in case it gets stuck. In 222 one-off mode, this is the number of seconds the server will wait 223 before exiting. 224 225 --server-bitrate-limit n[KMGT] 226 set a limit on the server side, which will cause a test to abort 227 if the client specifies a test of more than n bits per second, 228 or if the average data sent or received by the client (including 229 all data streams) is greater than n bits per second. The 230 default limit is zero, which implies no limit. The interval 231 over which to average the data rate is 5 seconds by default, but 232 can be specified by adding a '/' and a number to the bitrate 233 specifier. 234 235 --rsa-private-key-path file 236 path to the RSA private key (not password-protected) used to 237 decrypt authentication credentials from the client (if built 238 with OpenSSL support). 239 240 --authorized-users-path file 241 path to the configuration file containing authorized users cre- 242 dentials to run iperf tests (if built with OpenSSL support). 243 The file is a comma separated list of usernames and password 244 hashes; more information on the structure of the file can be 245 found in the EXAMPLES section. 246 247 --time-skew-thresholdsecond seconds 248 time skew threshold (in seconds) between the server and client 249 during the authentication process. 250 251 CLIENT SPECIFIC OPTIONS 252 -c, --client host[%dev] 253 run in client mode, connecting to the specified server. By 254 default, a test consists of sending data from the client to the 255 server, unless the -R flag is specified. If an optional inter- 256 face is specified, it is treated as a shortcut for --bind-dev 257 dev. Note that a percent sign and interface device name are 258 required for IPv6 link-local address literals. 259 260 --sctp use SCTP rather than TCP (FreeBSD and Linux) 261 262 -u, --udp 263 use UDP rather than TCP 264 265 --connect-timeout n 266 set timeout for establishing the initial control connection to 267 the server, in milliseconds. The default behavior is the oper- 268 ating system's timeout for TCP connection establishment. Pro- 269 viding a shorter value may speed up detection of a down iperf3 270 server. 271 272 -b, --bitrate n[KMGT] 273 set target bitrate to n bits/sec (default 1 Mbit/sec for UDP, 274 unlimited for TCP/SCTP). If there are multiple streams (-P 275 flag), the throughput limit is applied separately to each 276 stream. You can also add a '/' and a number to the bitrate 277 specifier. This is called "burst mode". It will send the given 278 number of packets without pausing, even if that temporarily 279 exceeds the specified throughput limit. Setting the target 280 bitrate to 0 will disable bitrate limits (particularly useful 281 for UDP tests). This throughput limit is implemented internally 282 inside iperf3, and is available on all platforms. Compare with 283 the --fq-rate flag. This option replaces the --bandwidth flag, 284 which is now deprecated but (at least for now) still accepted. 285 286 --pacing-timer n[KMGT] 287 set pacing timer interval in microseconds (default 1000 288 microseconds, or 1 ms). This controls iperf3's internal pacing 289 timer for the -b/--bitrate option. The timer fires at the 290 interval set by this parameter. Smaller values of the pacing 291 timer parameter smooth out the traffic emitted by iperf3, but 292 potentially at the cost of performance due to more frequent 293 timer processing. 294 295 --fq-rate n[KMGT] 296 Set a rate to be used with fair-queueing based socket-level pac- 297 ing, in bits per second. This pacing (if specified) will be in 298 addition to any pacing due to iperf3's internal throughput pac- 299 ing (-b/--bitrate flag), and both can be specified for the same 300 test. Only available on platforms supporting the SO_MAX_PAC- 301 ING_RATE socket option (currently only Linux). The default is 302 no fair-queueing based pacing. 303 304 --no-fq-socket-pacing 305 This option is deprecated and will be removed. It is equivalent 306 to specifying --fq-rate=0. 307 308 -t, --time n 309 time in seconds to transmit for (default 10 secs) 310 311 -n, --bytes n[KMGT] 312 number of bytes to transmit (instead of -t) 313 314 -k, --blockcount n[KMGT] 315 number of blocks (packets) to transmit (instead of -t or -n) 316 317 -l, --length n[KMGT] 318 length of buffer to read or write. For TCP tests, the default 319 value is 128KB. In the case of UDP, iperf3 tries to dynamically 320 determine a reasonable sending size based on the path MTU; if 321 that cannot be determined it uses 1460 bytes as a sending size. 322 For SCTP tests, the default size is 64KB. 323 324 --cport port 325 bind data streams to a specific client port (for TCP and UDP 326 only, default is to use an ephemeral port) 327 328 -P, --parallel n 329 number of parallel client streams to run. Note that iperf3 is 330 single threaded, so if you are CPU bound, this will not yield 331 higher throughput. 332 333 -R, --reverse 334 reverse the direction of a test, so that the server sends data 335 to the client 336 337 --bidir 338 test in both directions (normal and reverse), with both the 339 client and server sending and receiving data simultaneously 340 341 -w, --window n[KMGT] 342 set socket buffer size / window size. This value gets sent to 343 the server and used on that side too; on both sides this option 344 sets both the sending and receiving socket buffer sizes. This 345 option can be used to set (indirectly) the maximum TCP window 346 size. Note that on Linux systems, the effective maximum window 347 size is approximately double what is specified by this option 348 (this behavior is not a bug in iperf3 but a "feature" of the 349 Linux kernel, as documented by tcp(7) and socket(7)). 350 351 -M, --set-mss n 352 set TCP/SCTP maximum segment size (MTU - 40 bytes) 353 354 -N, --no-delay 355 set TCP/SCTP no delay, disabling Nagle's Algorithm 356 357 -4, --version4 358 only use IPv4 359 360 -6, --version6 361 only use IPv6 362 363 -S, --tos n 364 set the IP type of service. The usual prefixes for octal and hex 365 can be used, i.e. 52, 064 and 0x34 all specify the same value. 366 367 --dscp dscp 368 set the IP DSCP bits. Both numeric and symbolic values are 369 accepted. Numeric values can be specified in decimal, octal and 370 hex (see --tos above). To set both the DSCP bits and the ECN 371 bits, use --tos. 372 373 -L, --flowlabel n 374 set the IPv6 flow label (currently only supported on Linux) 375 376 -X, --xbind name 377 Bind SCTP associations to a specific subset of links using 378 sctp_bindx(3). The --B flag will be ignored if this flag is 379 specified. Normally SCTP will include the protocol addresses of 380 all active links on the local host when setting up an associa- 381 tion. Specifying at least one --X name will disable this behav- 382 iour. This flag must be specified for each link to be included 383 in the association, and is supported for both iperf servers and 384 clients (the latter are supported by passing the first --X argu- 385 ment to bind(2)). Hostnames are accepted as arguments and are 386 resolved using getaddrinfo(3). If the --4 or --6 flags are 387 specified, names which do not resolve to addresses within the 388 specified protocol family will be ignored. 389 390 --nstreams n 391 Set number of SCTP streams. 392 393 -Z, --zerocopy 394 Use a "zero copy" method of sending data, such as sendfile(2), 395 instead of the usual write(2). 396 397 -O, --omit n 398 Omit the first n seconds of the test, to skip past the TCP slow- 399 start period. 400 401 -T, --title str 402 Prefix every output line with this string. 403 404 --extra-data str 405 Specify an extra data string field to be included in JSON out- 406 put. 407 408 -C, --congestion algo 409 Set the congestion control algorithm (Linux and FreeBSD only). 410 An older --linux-congestion synonym for this flag is accepted 411 but is deprecated. 412 413 --get-server-output 414 Get the output from the server. The output format is determined 415 by the server (in particular, if the server was invoked with the 416 --json flag, the output will be in JSON format, otherwise it 417 will be in human-readable format). If the client is run with 418 --json, the server output is included in a JSON object; other- 419 wise it is appended at the bottom of the human-readable output. 420 421 --udp-counters-64bit 422 Use 64-bit counters in UDP test packets. The use of this option 423 can help prevent counter overflows during long or high-bitrate 424 UDP tests. Both client and server need to be running at least 425 version 3.1 for this option to work. It may become the default 426 behavior at some point in the future. 427 428 --repeating-payload 429 Use repeating pattern in payload, instead of random bytes. The 430 same payload is used in iperf2 (ASCII '0..9' repeating). It 431 might help to test and reveal problems in networking gear with 432 hardware compression (including some WiFi access points), where 433 iperf2 and iperf3 perform differently, just based on payload 434 entropy. 435 436 --dont-fragment 437 Set the IPv4 Don't Fragment (DF) bit on outgoing packets. Only 438 applicable to tests doing UDP over IPv4. 439 440 --username username 441 username to use for authentication to the iperf server (if built 442 with OpenSSL support). The password will be prompted for inter- 443 actively when the test is run. Note, the password to use can 444 also be specified via the IPERF3_PASSWORD environment variable. 445 If this variable is present, the password prompt will be 446 skipped. 447 448 --rsa-public-key-path file 449 path to the RSA public key used to encrypt authentication cre- 450 dentials (if built with OpenSSL support) 451 452 453 EXAMPLES 454 Authentication - RSA Keypair 455 The authentication feature of iperf3 requires an RSA public keypair. 456 The public key is used to encrypt the authentication token containing 457 the user credentials, while the private key is used to decrypt the 458 authentication token. The private key must be in PEM format and addi- 459 tionally must not have a password set. The public key must be in PEM 460 format and use SubjectPrefixKeyInfo encoding. An example of a set of 461 UNIX/Linux commands using OpenSSL to generate a correctly-formed key- 462 pair follows: 463 464 > openssl genrsa -des3 -out private.pem 2048 465 > openssl rsa -in private.pem -outform PEM -pubout -out public.pem 466 > openssl rsa -in private.pem -out private_not_protected.pem -out- 467 form PEM 468 469 After these commands, the public key will be contained in the file pub- 470 lic.pem and the private key will be contained in the file pri- 471 vate_not_protected.pem. 472 473 Authentication - Authorized users configuration file 474 A simple plaintext file must be provided to the iperf3 server in order 475 to specify the authorized user credentials. The file is a simple list 476 of comma-separated pairs of a username and a corresponding password 477 hash. The password hash is a SHA256 hash of the string "{$user}$pass- 478 word". The file can also contain commented lines (starting with the # 479 character). An example of commands to generate the password hash on a 480 UNIX/Linux system is given below: 481 482 > S_USER=mario S_PASSWD=rossi 483 > echo -n "{$S_USER}$S_PASSWD" | sha256sum | awk '{ print $1 }' 484 485 An example of a password file (with an entry corresponding to the above 486 username and password) is given below: 487 > cat credentials.csv 488 # file format: username,sha256 489 mario,bf7a49a846d44b454a5d11e7acfaf13d138bbe0b7483aa3e050879700572709b 490 491 492 493 AUTHORS 494 A list of the contributors to iperf3 can be found within the documenta- 495 tion located at https://software.es.net/iperf/dev.html#authors. 496 497 498 SEE ALSO 499 libiperf(3), https://software.es.net/iperf 500 501 502 503 ESnet January 2022 IPERF3(1) 504 505The iperf3 manual page will typically be installed in manual 506section 1. 507