1 How to Install Open vSwitch on Linux, FreeBSD and NetBSD
2 ========================================================
4 This document describes how to build and install Open vSwitch on a
5 generic Linux, FreeBSD, or NetBSD host. For specifics around installation
6 on a specific platform, please see one of these files:
18 To compile the userspace programs in the Open vSwitch distribution,
19 you will need the following software:
23 - A C compiler, such as:
27 * Clang. Clang 3.4 and later provide useful static semantic
28 analysis and thread-safety checks. For Ubuntu, there are
29 nightly built packages available on clang's website.
31 While OVS may be compatible with other compilers, optimal
32 support for atomic operations may be missing, making OVS very
33 slow (see lib/ovs-atomic.h).
35 - libssl, from OpenSSL, is optional but recommended if you plan to
36 connect the Open vSwitch to an OpenFlow controller. libssl is
37 required to establish confidentiality and authenticity in the
38 connections from an Open vSwitch to an OpenFlow controller. If
39 libssl is installed, then Open vSwitch will automatically build
42 - Python 2.x, for x >= 4.
44 On Linux, you may choose to compile the kernel module that comes with
45 the Open vSwitch distribution or to use the kernel module built into
46 the Linux kernel (version 3.3 or later). See the FAQ question "What
47 features are not available in the Open vSwitch kernel datapath that
48 ships as part of the upstream Linux kernel?" for more information on
49 this trade-off. You may also use the userspace-only implementation,
50 at some cost in features and performance (see INSTALL.userspace for
51 details). To compile the kernel module on Linux, you must also
52 install the following:
54 - A supported Linux kernel version. Please refer to README for a
55 list of supported versions.
57 The Open vSwitch datapath requires bridging support
58 (CONFIG_BRIDGE) to be built as a kernel module. (This is common
59 in kernels provided by Linux distributions.) The bridge module
60 must not be loaded or in use. If the bridge module is running
61 (check with "lsmod | grep bridge"), you must remove it ("rmmod
62 bridge") before starting the datapath.
64 For optional support of ingress policing, you must enable kernel
65 configuration options NET_CLS_BASIC, NET_SCH_INGRESS, and
66 NET_ACT_POLICE, either built-in or as modules. (NET_CLS_POLICE is
67 obsolete and not needed.)
69 To use GRE tunneling on Linux 2.6.37 or newer, kernel support
70 for GRE must be compiled in or available as a module
71 (CONFIG_NET_IPGRE_DEMUX).
73 To configure HTB or HFSC quality of service with Open vSwitch,
74 you must enable the respective configuration options.
76 To use Open vSwitch support for TAP devices, you must enable
79 - To build a kernel module, you need the same version of GCC that
80 was used to build that kernel.
82 - A kernel build directory corresponding to the Linux kernel image
83 the module is to run on. Under Debian and Ubuntu, for example,
84 each linux-image package containing a kernel binary has a
85 corresponding linux-headers package with the required build
88 If you are working from a Git tree or snapshot (instead of from a
89 distribution tarball), or if you modify the Open vSwitch build system
90 or the database schema, you will also need the following software:
92 - Autoconf version 2.64 or later.
94 - Automake version 1.10 or later.
96 - libtool version 2.4 or later. (Older versions might work too.)
98 To run the unit tests, you also need:
100 - Perl. Version 5.10.1 is known to work. Earlier versions should
103 The ovs-vswitchd.conf.db(5) manpage will include an E-R diagram, in
104 formats other than plain text, only if you have the following:
106 - "dot" from graphviz (http://www.graphviz.org/).
108 - Perl. Version 5.10.1 is known to work. Earlier versions should
111 - Python 2.x, for x >= 4.
113 If you are going to extensively modify Open vSwitch, please consider
114 installing the following to obtain better warnings:
116 - "sparse" version 0.4.4 or later
117 (http://www.kernel.org/pub/software/devel/sparse/dist/).
121 - clang, version 3.4 or later
123 Also, you may find the ovs-dev script found in utilities/ovs-dev.py useful.
125 Installation Requirements
126 -------------------------
128 The machine on which Open vSwitch is to be installed must have the
131 - libc compatible with the libc used for build.
133 - libssl compatible with the libssl used for build, if OpenSSL was
136 - On Linux, the same kernel version configured as part of the build.
138 - For optional support of ingress policing on Linux, the "tc" program
139 from iproute2 (part of all major distributions and available at
140 http://www.linux-foundation.org/en/Net:Iproute2).
142 On Linux you should ensure that /dev/urandom exists. To support TAP
143 devices, you must also ensure that /dev/net/tun exists.
145 Building and Installing Open vSwitch for Linux, FreeBSD or NetBSD
146 =================================================================
148 Once you have installed all the prerequisites listed above in the Base
149 Prerequisites section, follow the procedure below to build.
151 1. If you pulled the sources directly from an Open vSwitch Git tree,
152 run boot.sh in the top source directory:
156 2. Configure the package by running the configure script. You can
157 usually invoke configure without any arguments. For example:
161 By default all files are installed under /usr/local. If you want
162 to install into, e.g., /usr and /var instead of /usr/local and
163 /usr/local/var, add options as shown here:
165 % ./configure --prefix=/usr --localstatedir=/var
167 To use a specific C compiler for compiling Open vSwitch user
168 programs, also specify it on the configure command line, like so:
170 % ./configure CC=gcc-4.2
172 To use 'clang' compiler:
174 % ./configure CC=clang
176 To build the Linux kernel module, so that you can run the
177 kernel-based switch, pass the location of the kernel build
178 directory on --with-linux. For example, to build for a running
181 % ./configure --with-linux=/lib/modules/`uname -r`/build
183 If --with-linux requests building for an unsupported version of
184 Linux, then "configure" will fail with an error message. Please
185 refer to the FAQ for advice in that case.
187 If you wish to build the kernel module for an architecture other
188 than the architecture of the machine used for the build, you may
189 specify the kernel architecture string using the KARCH variable
190 when invoking the configure script. For example, to build for MIPS
193 % ./configure --with-linux=/path/to/linux KARCH=mips
195 If you plan to do much Open vSwitch development, you might want to
196 add --enable-Werror, which adds the -Werror option to the compiler
197 command line, turning warnings into errors. That makes it
198 impossible to miss warnings generated by the build.
200 To build with gcov code coverage support, add --enable-coverage,
203 % ./configure --enable-coverage
205 The configure script accepts a number of other options and honors
206 additional environment variables. For a full list, invoke
207 configure with the --help option.
209 You can also run configure from a separate build directory. This
210 is helpful if you want to build Open vSwitch in more than one way
211 from a single source directory, e.g. to try out both GCC and Clang
212 builds, or to build kernel modules for more than one Linux version.
215 % mkdir _gcc && (cd _gcc && ../configure CC=gcc)
216 % mkdir _clang && (cd _clang && ../configure CC=clang)
218 3. Run GNU make in the build directory, e.g.:
222 or if GNU make is installed as "gmake":
226 If you used a separate build directory, run make or gmake from that
232 For improved warnings if you installed "sparse" (see
233 "Prerequisites"), add C=1 to the command line.
235 4. Consider running the testsuite. Refer to "Running the Testsuite"
236 below, for instructions.
238 5. Become root by running "su" or another program.
240 6. Run "make install" to install the executables and manpages into the
241 running system, by default under /usr/local.
243 7. If you built kernel modules, you may install and load them, e.g.:
245 % make modules_install
246 % /sbin/modprobe openvswitch
248 To verify that the modules have been loaded, run "/sbin/lsmod" and
249 check that openvswitch is listed.
251 If the "modprobe" operation fails, look at the last few kernel log
252 messages (e.g. with "dmesg | tail"):
254 - The message "openvswitch: exports duplicate symbol
255 br_should_route_hook (owned by bridge)" means that the bridge
256 module is loaded. Run "/sbin/rmmod bridge" to remove it.
258 If "/sbin/rmmod bridge" fails with "ERROR: Module bridge does
259 not exist in /proc/modules", then the bridge is compiled into
260 the kernel, rather than as a module. Open vSwitch does not
261 support this configuration (see "Build Requirements", above).
263 - The message "openvswitch: exports duplicate symbol
264 dp_ioctl_hook (owned by ofdatapath)" means that the ofdatapath
265 module from the OpenFlow reference implementation is loaded.
266 Run "/sbin/rmmod ofdatapath" to remove it. (You might have to
267 delete any existing datapaths beforehand, using the "dpctl"
268 program included with the OpenFlow reference implementation.
269 "ovs-dpctl" will not work.)
271 - Otherwise, the most likely problem is that Open vSwitch was
272 built for a kernel different from the one into which you are
273 trying to load it. Run "modinfo" on openvswitch.ko and on
274 a module built for the running kernel, e.g.:
276 % /sbin/modinfo openvswitch.ko
277 % /sbin/modinfo /lib/modules/`uname -r`/kernel/net/bridge/bridge.ko
279 Compare the "vermagic" lines output by the two commands. If
280 they differ, then Open vSwitch was built for the wrong kernel.
282 - If you decide to report a bug or ask a question related to
283 module loading, please include the output from the "dmesg" and
284 "modinfo" commands mentioned above.
286 There is an optional module parameter to openvswitch.ko called
287 vlan_tso that enables TCP segmentation offload over VLANs on NICs
288 that support it. Many drivers do not expose support for TSO on VLANs
289 in a way that Open vSwitch can use but there is no way to detect
290 whether this is the case. If you know that your particular driver can
291 handle it (for example by testing sending large TCP packets over VLANs)
292 then passing in a value of 1 may improve performance. Modules built for
293 Linux kernels 2.6.37 and later, as well as specially patched versions
294 of earlier kernels, do not need this and do not have this parameter. If
295 you do not understand what this means or do not know if your driver
296 will work, do not set this.
298 8. Initialize the configuration database using ovsdb-tool, e.g.:
300 % mkdir -p /usr/local/etc/openvswitch
301 % ovsdb-tool create /usr/local/etc/openvswitch/conf.db vswitchd/vswitch.ovsschema
306 Before starting ovs-vswitchd itself, you need to start its
307 configuration database, ovsdb-server. Each machine on which Open
308 vSwitch is installed should run its own copy of ovsdb-server.
309 Configure it to use the database you created during installation (as
310 explained above), to listen on a Unix domain socket, to connect to any
311 managers specified in the database itself, and to use the SSL
312 configuration in the database:
314 % ovsdb-server --remote=punix:/usr/local/var/run/openvswitch/db.sock \
315 --remote=db:Open_vSwitch,Open_vSwitch,manager_options \
316 --private-key=db:Open_vSwitch,SSL,private_key \
317 --certificate=db:Open_vSwitch,SSL,certificate \
318 --bootstrap-ca-cert=db:Open_vSwitch,SSL,ca_cert \
321 (If you built Open vSwitch without SSL support, then omit
322 --private-key, --certificate, and --bootstrap-ca-cert.)
324 Then initialize the database using ovs-vsctl. This is only
325 necessary the first time after you create the database with
326 ovsdb-tool (but running it at any time is harmless):
328 % ovs-vsctl --no-wait init
330 Then start the main Open vSwitch daemon, telling it to connect to the
331 same Unix domain socket:
333 % ovs-vswitchd --pidfile --detach
335 Now you may use ovs-vsctl to set up bridges and other Open vSwitch
336 features. For example, to create a bridge named br0 and add ports
337 eth0 and vif1.0 to it:
339 % ovs-vsctl add-br br0
340 % ovs-vsctl add-port br0 eth0
341 % ovs-vsctl add-port br0 vif1.0
343 Please refer to ovs-vsctl(8) for more details.
348 When you upgrade Open vSwitch from one version to another, you should
349 also upgrade the database schema:
351 1. Stop the Open vSwitch daemons, e.g.:
353 % kill `cd /usr/local/var/run/openvswitch && cat ovsdb-server.pid ovs-vswitchd.pid`
355 2. Install the new Open vSwitch release.
357 3. Upgrade the database, in one of the following two ways:
359 - If there is no important data in your database, then you may
360 delete the database file and recreate it with ovsdb-tool,
361 following the instructions under "Building and Installing Open
362 vSwitch for Linux, FreeBSD or NetBSD".
364 - If you want to preserve the contents of your database, back it
365 up first, then use "ovsdb-tool convert" to upgrade it, e.g.:
367 % ovsdb-tool convert /usr/local/etc/openvswitch/conf.db vswitchd/vswitch.ovsschema
369 4. Start the Open vSwitch daemons as described under "Building and
370 Installing Open vSwitch for Linux, FreeBSD or NetBSD" above.
374 Upgrading Open vSwitch from one version to the next version with minimum
375 disruption of traffic going through the system that is using that Open vSwitch
376 needs some considerations:
378 1. If the upgrade only involves upgrading the userspace utilities and daemons
379 of Open vSwitch, make sure that the new userspace version is compatible with
380 the previously loaded kernel module.
382 2. An upgrade of userspace daemons means that they have to be restarted.
383 Restarting the daemons means that the Openflow flows in the ovs-vswitchd daemon
384 will be lost. One way to restore the flows is to let the controller
385 re-populate it. Another way is to save the previous flows using a utility
386 like ovs-ofctl and then re-add them after the restart. Restoring the old flows
387 is accurate only if the new Open vSwitch interfaces retain the old 'ofport'
390 3. When the new userspace daemons get restarted, they automatically flush
391 the old flows setup in the kernel. This can be expensive if there are hundreds
392 of new flows that are entering the kernel but userspace daemons are busy
393 setting up new userspace flows from either the controller or an utility like
394 ovs-ofctl. Open vSwitch database provides an option to solve this problem
395 through the other_config:flow-restore-wait column of the Open_vSwitch table.
396 Refer to the ovs-vswitchd.conf.db(5) manpage for details.
398 4. If the upgrade also involves upgrading the kernel module, the old kernel
399 module needs to be unloaded and the new kernel module should be loaded. This
400 means that the kernel network devices belonging to Open vSwitch is recreated
401 and the kernel flows are lost. The downtime of the traffic can be reduced
402 if the userspace daemons are restarted immediately and the userspace flows
403 are restored as soon as possible.
405 The ovs-ctl utility's "restart" function only restarts the userspace daemons,
406 makes sure that the 'ofport' values remain consistent across restarts, restores
407 userspace flows using the ovs-ofctl utility and also uses the
408 other_config:flow-restore-wait column to keep the traffic downtime to the
409 minimum. The ovs-ctl utility's "force-reload-kmod" function does all of the
410 above, but also replaces the old kernel module with the new one. Open vSwitch
411 startup scripts for Debian, XenServer and RHEL use ovs-ctl's functions and it
412 is recommended that these functions be used for other software platforms too.
417 This section describe Open vSwitch's built-in support for various test
418 suites. You must configure and build Open vSwitch (steps 1 through 3
419 in "Building and Installing Open vSwitch for Linux, FreeBSD or NetBSD"
420 above) before you run the tests described here. You do not need to
421 install Open vSwitch or to build or load the kernel module to run
422 these test suites. You do not need supervisor privilege to run these
428 Open vSwitch includes a suite of self-tests. Before you submit patches
429 upstream, we advise that you run the tests and ensure that they pass.
430 If you add new features to Open vSwitch, then adding tests for those
431 features will ensure your features don't break as developers modify
432 other areas of Open vSwitch.
434 Refer to "Testsuites" above for prerequisites.
436 To run all the unit tests in Open vSwitch, one at a time:
438 This takes under 5 minutes on a modern desktop system.
440 To run all the unit tests in Open vSwitch, up to 8 in parallel:
441 make check TESTSUITEFLAGS=-j8
442 This takes under a minute on a modern 4-core desktop system.
444 To see a list of all the available tests, run:
445 make check TESTSUITEFLAGS=--list
447 To run only a subset of tests, e.g. test 123 and tests 477 through 484:
448 make check TESTSUITEFLAGS='123 477-484'
449 (Tests do not have inter-dependencies, so you may run any subset.)
451 To run tests matching a keyword, e.g. "ovsdb":
452 make check TESTSUITEFLAGS='-k ovsdb'
454 To see a complete list of test options:
455 make check TESTSUITEFLAGS=--help
457 The results of a testing run are reported in tests/testsuite.log.
458 Please report test failures as bugs and include the testsuite.log in
461 If you have "valgrind" installed, then you can also run the testsuite
462 under valgrind by using "make check-valgrind" in place of "make
463 check". All the same options are available via TESTSUITEFLAGS. When
464 you do this, the "valgrind" results for test <N> are reported in files
465 named tests/testsuite.dir/<N>/valgrind.*. You may find that the
466 valgrind results are easier to interpret if you put "-q" in
467 ~/.valgrindrc, since that reduces the amount of output.
469 Sometimes a few tests may fail on some runs but not others. This is
470 usually a bug in the testsuite, not a bug in Open vSwitch itself. If
471 you find that a test fails intermittently, please report it, since the
472 developers may not have noticed.
477 OFTest is an OpenFlow protocol testing suite. Open vSwitch includes a
478 Makefile target to run OFTest with Open vSwitch in "dummy mode". In
479 this mode of testing, no packets travel across physical or virtual
480 networks. Instead, Unix domain sockets stand in as simulated
481 networks. This simulation is imperfect, but it is much easier to set
482 up, does not require extra physical or virtual hardware, and does not
483 require supervisor privileges.
485 To run OFTest with Open vSwitch, first read and follow the
486 instructions under "Testsuites" above. Second, obtain a copy of
487 OFTest and install its prerequisites. You need a copy of OFTest that
488 includes commit 406614846c5 (make ovs-dummy platform work again).
489 This commit was merged into the OFTest repository on Feb 1, 2013, so
490 any copy of OFTest more recent than that should work. Testing OVS in
491 dummy mode does not require root privilege, so you may ignore that
494 Optionally, add the top-level OFTest directory (containing the "oft"
495 program) to your $PATH. This slightly simplifies running OFTest later.
497 To run OFTest in dummy mode, run the following command from your Open
498 vSwitch build directory:
499 make check-oftest OFT=<oft-binary>
500 where <oft-binary> is the absolute path to the "oft" program in
503 If you added "oft" to your $PATH, you may omit the OFT variable
506 By default, "check-oftest" passes "oft" just enough options to enable
507 dummy mode. You can use OFTFLAGS to pass additional options. For
508 example, to run just the basic.Echo test instead of all tests (the
509 default) and enable verbose logging:
510 make check-oftest OFT=<oft-binary> OFTFLAGS='--verbose -T basic.Echo'
512 If you use OFTest that does not include commit 4d1f3eb2c792 (oft:
513 change default port to 6653), merged into the OFTest repository in
514 October 2013, then you need to add an option to use the IETF-assigned
516 make check-oftest OFT=<oft-binary> OFTFLAGS='--port=6653'
518 Please interpret OFTest results cautiously. Open vSwitch can fail a
519 given test in OFTest for many reasons, including bugs in Open vSwitch,
520 bugs in OFTest, bugs in the "dummy mode" integration, and differing
521 interpretations of the OpenFlow standard and other standards.
523 Open vSwitch has not been validated against OFTest. Please do report
524 test failures that you believe to represent bugs in Open vSwitch.
525 Include the precise versions of Open vSwitch and OFTest in your bug
526 report, plus any other information needed to reproduce the problem.
531 Ryu is an OpenFlow controller written in Python that includes an
532 extensive OpenFlow testsuite. Open vSwitch includes a Makefile target
533 to run Ryu in "dummy mode". See "OFTest" above for an explanation of
536 To run Ryu tests with Open vSwitch, first read and follow the
537 instructions under "Testsuites" above. Second, obtain a copy of Ryu,
538 install its prerequisites, and build it. You do not need to install
539 Ryu (some of the tests do not get installed, so it does not help).
541 To run Ryu tests, run the following command from your Open vSwitch
543 make check-ryu RYUDIR=<ryu-source-dir>
544 where <ryu-source-dir> is the absolute path to the root of the Ryu
545 source distribution. The default <ryu-source-dir> is $srcdir/../ryu
546 where $srcdir is your Open vSwitch source directory, so if this
547 default is correct then you make simply run "make check-ryu".
549 Open vSwitch has not been validated against Ryu. Please do report
550 test failures that you believe to represent bugs in Open vSwitch.
551 Include the precise versions of Open vSwitch and Ryu in your bug
552 report, plus any other information needed to reproduce the problem.
557 Please report problems to bugs@openvswitch.org.