1 Frequently Asked Questions
2 ==========================
4 Open vSwitch <http://openvswitch.org>
9 ### Q: What is Open vSwitch?
11 A: Open vSwitch is a production quality open source software switch
12 designed to be used as a vswitch in virtualized server
13 environments. A vswitch forwards traffic between different VMs on
14 the same physical host and also forwards traffic between VMs and
15 the physical network. Open vSwitch supports standard management
16 interfaces (e.g. sFlow, NetFlow, IPFIX, RSPAN, CLI), and is open to
17 programmatic extension and control using OpenFlow and the OVSDB
20 Open vSwitch as designed to be compatible with modern switching
21 chipsets. This means that it can be ported to existing high-fanout
22 switches allowing the same flexible control of the physical
23 infrastructure as the virtual infrastructure. It also means that
24 Open vSwitch will be able to take advantage of on-NIC switching
25 chipsets as their functionality matures.
27 ### Q: What virtualization platforms can use Open vSwitch?
29 A: Open vSwitch can currently run on any Linux-based virtualization
30 platform (kernel 2.6.32 and newer), including: KVM, VirtualBox, Xen,
31 Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the
32 mainline kernel. The bulk of the code is written in platform-
33 independent C and is easily ported to other environments. We welcome
34 inquires about integrating Open vSwitch with other virtualization
37 ### Q: How can I try Open vSwitch?
39 A: The Open vSwitch source code can be built on a Linux system. You can
40 build and experiment with Open vSwitch on any Linux machine.
41 Packages for various Linux distributions are available on many
42 platforms, including: Debian, Ubuntu, Fedora.
44 You may also download and run a virtualization platform that already
45 has Open vSwitch integrated. For example, download a recent ISO for
46 XenServer or Xen Cloud Platform. Be aware that the version
47 integrated with a particular platform may not be the most recent Open
50 ### Q: Does Open vSwitch only work on Linux?
52 A: No, Open vSwitch has been ported to a number of different operating
53 systems and hardware platforms. Most of the development work occurs
54 on Linux, but the code should be portable to any POSIX system. We've
55 seen Open vSwitch ported to a number of different platforms,
56 including FreeBSD, Windows, and even non-POSIX embedded systems.
58 By definition, the Open vSwitch Linux kernel module only works on
59 Linux and will provide the highest performance. However, a userspace
60 datapath is available that should be very portable.
62 ### Q: What's involved with porting Open vSwitch to a new platform or switching ASIC?
64 A: The [PORTING.md] document describes how one would go about
65 porting Open vSwitch to a new operating system or hardware platform.
67 ### Q: Why would I use Open vSwitch instead of the Linux bridge?
69 A: Open vSwitch is specially designed to make it easier to manage VM
70 network configuration and monitor state spread across many physical
71 hosts in dynamic virtualized environments. Please see
72 [WHY-OVS.md] for a more detailed description of how Open vSwitch
73 relates to the Linux Bridge.
75 ### Q: How is Open vSwitch related to distributed virtual switches like the VMware vNetwork distributed switch or the Cisco Nexus 1000V?
77 A: Distributed vswitch applications (e.g., VMware vNetwork distributed
78 switch, Cisco Nexus 1000V) provide a centralized way to configure and
79 monitor the network state of VMs that are spread across many physical
80 hosts. Open vSwitch is not a distributed vswitch itself, rather it
81 runs on each physical host and supports remote management in a way
82 that makes it easier for developers of virtualization/cloud
83 management platforms to offer distributed vswitch capabilities.
85 To aid in distribution, Open vSwitch provides two open protocols that
86 are specially designed for remote management in virtualized network
87 environments: OpenFlow, which exposes flow-based forwarding state,
88 and the OVSDB management protocol, which exposes switch port state.
89 In addition to the switch implementation itself, Open vSwitch
90 includes tools (ovs-ofctl, ovs-vsctl) that developers can script and
91 extend to provide distributed vswitch capabilities that are closely
92 integrated with their virtualization management platform.
94 ### Q: Why doesn't Open vSwitch support distribution?
96 A: Open vSwitch is intended to be a useful component for building
97 flexible network infrastructure. There are many different approaches
98 to distribution which balance trade-offs between simplicity,
99 scalability, hardware compatibility, convergence times, logical
100 forwarding model, etc. The goal of Open vSwitch is to be able to
101 support all as a primitive building block rather than choose a
102 particular point in the distributed design space.
104 ### Q: How can I contribute to the Open vSwitch Community?
106 A: You can start by joining the mailing lists and helping to answer
107 questions. You can also suggest improvements to documentation. If
108 you have a feature or bug you would like to work on, send a mail to
109 one of the mailing lists:
111 http://openvswitch.org/mlists/
113 ### Q: Why can I no longer connect to my OpenFlow controller or OVSDB manager?
115 A: Starting in OVS 2.4, we switched the default ports to the
116 IANA-specified port numbers for OpenFlow (6633->6653) and OVSDB
117 (6632->6640). We recommend using these port numbers, but if you
118 cannot, all the programs allow overriding the default port. See the
119 appropriate man page.
125 ### Q: What does it mean for an Open vSwitch release to be LTS (long-term support)?
127 A: All official releases have been through a comprehensive testing
128 process and are suitable for production use. Planned releases will
129 occur several times a year. If a significant bug is identified in an
130 LTS release, we will provide an updated release that includes the
131 fix. Releases that are not LTS may not be fixed and may just be
132 supplanted by the next major release. The current LTS release is
135 ### Q: What Linux kernel versions does each Open vSwitch release work with?
137 A: The following table lists the Linux kernel versions against which the
138 given versions of the Open vSwitch kernel module will successfully
139 build. The Linux kernel versions are upstream kernel versions, so
140 Linux kernels modified from the upstream sources may not build in
141 some cases even if they are based on a supported version. This is
142 most notably true of Red Hat Enterprise Linux (RHEL) kernels, which
143 are extensively modified from upstream.
145 | Open vSwitch | Linux kernel
146 |:------------:|:-------------:
147 | 1.4.x | 2.6.18 to 3.2
148 | 1.5.x | 2.6.18 to 3.2
149 | 1.6.x | 2.6.18 to 3.2
150 | 1.7.x | 2.6.18 to 3.3
151 | 1.8.x | 2.6.18 to 3.4
152 | 1.9.x | 2.6.18 to 3.8
153 | 1.10.x | 2.6.18 to 3.8
154 | 1.11.x | 2.6.18 to 3.8
155 | 2.0.x | 2.6.32 to 3.10
156 | 2.1.x | 2.6.32 to 3.11
157 | 2.3.x | 2.6.32 to 3.14
158 | 2.4.x | 2.6.32 to 4.0
159 | 2.5.x | 2.6.32 to 4.3
161 Open vSwitch userspace should also work with the Linux kernel module
162 built into Linux 3.3 and later.
164 Open vSwitch userspace is not sensitive to the Linux kernel version.
165 It should build against almost any kernel, certainly against 2.6.32
168 ### Q: Are all features available with all datapaths?
170 A: Open vSwitch supports different datapaths on different platforms. Each
171 datapath has a different feature set: the following tables try to summarize
176 * *Linux upstream*: The datapath implemented by the kernel module shipped
177 with Linux upstream. Since features have been gradually
178 introduced into the kernel, the table mentions the first
179 Linux release whose OVS module supports the feature.
181 * *Linux OVS tree*: The datapath implemented by the Linux kernel module
182 distributed with the OVS source tree. Some features of
183 this module rely on functionality not available in older
184 kernels: in this case the minumum Linux version (against
185 which the feature can be compiled) is listed.
187 * *Userspace*: Also known as DPDK, dpif-netdev or dummy datapath. It is the
188 only datapath that works on NetBSD and FreeBSD.
190 * *Hyper-V*: Also known as the Windows datapath.
192 The following table lists the datapath supported features from
193 an Open vSwitch user's perspective.
195 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
196 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
197 Connection tracking | 4.3 | 3.10 | NO | NO |
198 Tunnel - LISP | NO | YES | NO | NO |
199 Tunnel - STT | NO | 3.5 | NO | YES |
200 Tunnel - GRE | 3.11 | YES | YES | YES |
201 Tunnel - VXLAN | 3.12 | YES | YES | YES |
202 Tunnel - Geneve | 3.18 | YES | YES | NO |
203 QoS - Policing | YES | YES | NO | NO |
204 QoS - Shaping | YES | YES | NO | NO |
205 sFlow | YES | YES | YES | NO |
206 Set action | YES | YES | YES | PARTIAL |
207 NIC Bonding | YES | YES | YES | NO |
208 Multiple VTEPs | YES | YES | YES | NO |
211 * Only a limited set of flow fields is modifiable via the set action by the
213 * The Hyper-V datapath only supports one physical NIC per datapath. This is
214 why bonding is not supported.
215 * The Hyper-V datapath can have at most one IP address configured as a
218 The following table lists features that do not *directly* impact an
219 Open vSwitch user, e.g. because their absence can be hidden by the ofproto
220 layer (usually this comes with a performance penalty).
222 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
223 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
224 SCTP flows | 3.12 | YES | YES | YES |
225 MPLS | 3.19 | YES | YES | NO |
226 UFID | 4.0 | YES | YES | NO |
227 Megaflows | 3.12 | YES | YES | NO |
228 Masked set action | 4.0 | YES | YES | NO |
229 Recirculation | 3.19 | YES | YES | NO |
230 TCP flags matching | 3.13 | YES | YES | NO |
231 Validate flow actions | YES | YES | N/A | NO |
232 Multiple datapaths | YES | YES | YES | NO |
233 Tunnel TSO - STT | N/A | YES | NO | YES |
235 ### Q: I get an error like this when I configure Open vSwitch:
237 configure: error: Linux kernel in <dir> is version <x>, but
238 version newer than <y> is not supported (please refer to the
243 A: You have the following options:
245 - Use the Linux kernel module supplied with the kernel that you are
246 using. (See also the following FAQ.)
248 - If there is a newer released version of Open vSwitch, consider
249 building that one, because it may support the kernel that you are
250 building against. (To find out, consult the table in the
253 - The Open vSwitch "master" branch may support the kernel that you
254 are using, so consider building the kernel module from "master".
256 All versions of Open vSwitch userspace are compatible with all
257 versions of the Open vSwitch kernel module, so you do not have to
258 use the kernel module from one source along with the userspace
259 programs from the same source.
261 ### Q: What features are not available in the Open vSwitch kernel datapath that ships as part of the upstream Linux kernel?
263 A: The kernel module in upstream Linux does not include support for
264 LISP. Work is in progress to add support for LISP to the upstream
265 Linux version of the Open vSwitch kernel module. For now, if you
266 need this feature, use the kernel module from the Open vSwitch
267 distribution instead of the upstream Linux kernel module.
269 Certain features require kernel support to function or to have
270 reasonable performance. If the ovs-vswitchd log file indicates that
271 a feature is not supported, consider upgrading to a newer upstream
272 Linux release or using the kernel module paired with the userspace
275 ### Q: Why do tunnels not work when using a kernel module other than the one packaged with Open vSwitch?
277 A: Support for tunnels was added to the upstream Linux kernel module
278 after the rest of Open vSwitch. As a result, some kernels may contain
279 support for Open vSwitch but not tunnels. The minimum kernel version
280 that supports each tunnel protocol is:
282 | Protocol | Linux Kernel
283 |:--------:|:-------------:
287 | LISP | <not upstream>
288 | STT | <not upstream>
290 If you are using a version of the kernel that is older than the one
291 listed above, it is still possible to use that tunnel protocol. However,
292 you must compile and install the kernel module included with the Open
293 vSwitch distribution rather than the one on your machine. If problems
294 persist after doing this, check to make sure that the module that is
295 loaded is the one you expect.
297 ### Q: Why are UDP tunnel checksums not computed for VXLAN or Geneve?
299 A: Generating outer UDP checksums requires kernel support that was not
300 part of the initial implementation of these protocols. If using the
301 upstream Linux Open vSwitch module, you must use kernel 4.0 or
302 newer. The out-of-tree modules from Open vSwitch release 2.4 and later
303 support UDP checksums.
305 ### Q: What features are not available when using the userspace datapath?
307 A: Tunnel virtual ports are not supported, as described in the
308 previous answer. It is also not possible to use queue-related
309 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
310 may not be transmitted.
312 ### Q: What Linux kernel versions does IPFIX flow monitoring work with?
314 A: IPFIX flow monitoring requires the Linux kernel module from Linux
315 3.10 or later, or the out-of-tree module from Open vSwitch version
318 ### Q: Should userspace or kernel be upgraded first to minimize downtime?
320 In general, the Open vSwitch userspace should be used with the
321 kernel version included in the same release or with the version
322 from upstream Linux. However, when upgrading between two releases
323 of Open vSwitch it is best to migrate userspace first to reduce
324 the possibility of incompatibilities.
326 ### Q: What happened to the bridge compatibility feature?
328 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
329 When it was enabled, Open vSwitch imitated the interface of the
330 Linux kernel "bridge" module. This allowed users to drop Open
331 vSwitch into environments designed to use the Linux kernel bridge
332 module without adapting the environment to use Open vSwitch.
334 Open vSwitch 1.10 and later do not support bridge compatibility.
335 The feature was dropped because version 1.10 adopted a new internal
336 architecture that made bridge compatibility difficult to maintain.
337 Now that many environments use OVS directly, it would be rarely
340 To use bridge compatibility, install OVS 1.9 or earlier, including
341 the accompanying kernel modules (both the main and bridge
342 compatibility modules), following the instructions that come with
343 the release. Be sure to start the ovs-brcompatd daemon.
349 ### Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation keeps talking about bridges. What's a bridge?
351 A: In networking, the terms "bridge" and "switch" are synonyms. Open
352 vSwitch implements an Ethernet switch, which means that it is also
355 ### Q: What's a VLAN?
357 A: See the "VLAN" section below.
363 ### Q: How do I configure a port as an access port?
365 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
366 the following commands configure br0 with eth0 as a trunk port (the
367 default) and tap0 as an access port for VLAN 9:
370 ovs-vsctl add-port br0 eth0
371 ovs-vsctl add-port br0 tap0 tag=9
373 If you want to configure an already added port as an access port,
374 use "ovs-vsctl set", e.g.:
376 ovs-vsctl set port tap0 tag=9
378 ### Q: How do I configure a port as a SPAN port, that is, enable mirroring of all traffic to that port?
380 A: The following commands configure br0 with eth0 and tap0 as trunk
381 ports. All traffic coming in or going out on eth0 or tap0 is also
382 mirrored to tap1; any traffic arriving on tap1 is dropped:
385 ovs-vsctl add-port br0 eth0
386 ovs-vsctl add-port br0 tap0
387 ovs-vsctl add-port br0 tap1 \
388 -- --id=@p get port tap1 \
389 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
390 -- set bridge br0 mirrors=@m
392 To later disable mirroring, run:
394 ovs-vsctl clear bridge br0 mirrors
396 ### Q: Does Open vSwitch support configuring a port in promiscuous mode?
398 A: Yes. How you configure it depends on what you mean by "promiscuous
401 - Conventionally, "promiscuous mode" is a feature of a network
402 interface card. Ordinarily, a NIC passes to the CPU only the
403 packets actually destined to its host machine. It discards
404 the rest to avoid wasting memory and CPU cycles. When
405 promiscuous mode is enabled, however, it passes every packet
406 to the CPU. On an old-style shared-media or hub-based
407 network, this allows the host to spy on all packets on the
408 network. But in the switched networks that are almost
409 everywhere these days, promiscuous mode doesn't have much
410 effect, because few packets not destined to a host are
411 delivered to the host's NIC.
413 This form of promiscuous mode is configured in the guest OS of
414 the VMs on your bridge, e.g. with "ifconfig".
416 - The VMware vSwitch uses a different definition of "promiscuous
417 mode". When you configure promiscuous mode on a VMware vNIC,
418 the vSwitch sends a copy of every packet received by the
419 vSwitch to that vNIC. That has a much bigger effect than just
420 enabling promiscuous mode in a guest OS. Rather than getting
421 a few stray packets for which the switch does not yet know the
422 correct destination, the vNIC gets every packet. The effect
423 is similar to replacing the vSwitch by a virtual hub.
425 This "promiscuous mode" is what switches normally call "port
426 mirroring" or "SPAN". For information on how to configure
427 SPAN, see "How do I configure a port as a SPAN port, that is,
428 enable mirroring of all traffic to that port?"
430 ### Q: How do I configure a DPDK port as an access port?
432 A: Firstly, you must have a DPDK-enabled version of Open vSwitch.
434 If your version is DPDK-enabled it will support the --dpdk
435 argument on the command line and will display lines with
436 "EAL:..." during startup when --dpdk is supplied.
438 Secondly, when adding a DPDK port, unlike a system port, the
439 type for the interface must be specified. For example;
442 ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
444 Finally, it is required that DPDK port names begin with 'dpdk'.
446 See [INSTALL.DPDK.md] for more information on enabling and using DPDK with
449 ### Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all traffic to that VLAN?
451 A: The following commands configure br0 with eth0 as a trunk port and
452 tap0 as an access port for VLAN 10. All traffic coming in or going
453 out on tap0, as well as traffic coming in or going out on eth0 in
454 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
455 VLAN 10, in cases where one is present, is dropped as part of
459 ovs-vsctl add-port br0 eth0
460 ovs-vsctl add-port br0 tap0 tag=10
462 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
464 -- set bridge br0 mirrors=@m
466 To later disable mirroring, run:
468 ovs-vsctl clear bridge br0 mirrors
470 Mirroring to a VLAN can disrupt a network that contains unmanaged
471 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
472 GRE tunnel has fewer caveats than mirroring to a VLAN and should
473 generally be preferred.
475 ### Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
477 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
478 of the specified output-vlan. This loss of information may make
479 the mirrored traffic too hard to interpret.
481 To mirror multiple VLANs, use the commands above, but specify a
482 comma-separated list of VLANs as the value for select-vlan. To
483 mirror every VLAN, use the commands above, but omit select-vlan and
486 When a packet arrives on a VLAN that is used as a mirror output
487 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
488 floods the packet across all the ports for which the mirror output
489 VLAN is configured. (If an OpenFlow controller is in use, then it
490 can override this behavior through the flow table.) If OVS is used
491 as an intermediate switch, rather than an edge switch, this ensures
492 that the RSPAN traffic is distributed through the network.
494 Mirroring to a VLAN can disrupt a network that contains unmanaged
495 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
496 GRE tunnel has fewer caveats than mirroring to a VLAN and should
497 generally be preferred.
499 ### Q: How do I configure mirroring of all traffic to a GRE tunnel?
501 A: The following commands configure br0 with eth0 and tap0 as trunk
502 ports. All traffic coming in or going out on eth0 or tap0 is also
503 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
504 traffic arriving on gre0 is dropped:
507 ovs-vsctl add-port br0 eth0
508 ovs-vsctl add-port br0 tap0
509 ovs-vsctl add-port br0 gre0 \
510 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
511 -- --id=@p get port gre0 \
512 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
513 -- set bridge br0 mirrors=@m
515 To later disable mirroring and destroy the GRE tunnel:
517 ovs-vsctl clear bridge br0 mirrors
518 ovs-vsctl del-port br0 gre0
520 ### Q: Does Open vSwitch support ERSPAN?
522 A: No. ERSPAN is an undocumented proprietary protocol. As an
523 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
526 ### Q: How do I connect two bridges?
528 A: First, why do you want to do this? Two connected bridges are not
529 much different from a single bridge, so you might as well just have
530 a single bridge with all your ports on it.
532 If you still want to connect two bridges, you can use a pair of
533 patch ports. The following example creates bridges br0 and br1,
534 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
535 and br1 with a pair of patch ports.
538 ovs-vsctl add-port br0 eth0
539 ovs-vsctl add-port br0 tap0
541 ovs-vsctl add-port br1 tap1
543 -- add-port br0 patch0 \
544 -- set interface patch0 type=patch options:peer=patch1 \
545 -- add-port br1 patch1 \
546 -- set interface patch1 type=patch options:peer=patch0
548 Bridges connected with patch ports are much like a single bridge.
549 For instance, if the example above also added eth1 to br1, and both
550 eth0 and eth1 happened to be connected to the same next-hop switch,
551 then you could loop your network just as you would if you added
552 eth0 and eth1 to the same bridge (see the "Configuration Problems"
553 section below for more information).
555 If you are using Open vSwitch 1.9 or an earlier version, then you
556 need to be using the kernel module bundled with Open vSwitch rather
557 than the one that is integrated into Linux 3.3 and later, because
558 Open vSwitch 1.9 and earlier versions need kernel support for patch
559 ports. This also means that in Open vSwitch 1.9 and earlier, patch
560 ports will not work with the userspace datapath, only with the
563 ### Q: How do I configure a bridge without an OpenFlow local port? (Local port in the sense of OFPP_LOCAL)
565 A: Open vSwitch does not support such a configuration.
566 Bridges always have their local ports.
569 Implementation Details
570 ----------------------
572 ### Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
574 A: Open vSwitch uses different kinds of flows for different purposes:
576 - OpenFlow flows are the most important kind of flow. OpenFlow
577 controllers use these flows to define a switch's policy.
578 OpenFlow flows support wildcards, priorities, and multiple
581 When in-band control is in use, Open vSwitch sets up a few
582 "hidden" flows, with priority higher than a controller or the
583 user can configure, that are not visible via OpenFlow. (See
584 the "Controller" section of the FAQ for more information
587 - The Open vSwitch software switch implementation uses a second
588 kind of flow internally. These flows, called "datapath" or
589 "kernel" flows, do not support priorities and comprise only a
590 single table, which makes them suitable for caching. (Like
591 OpenFlow flows, datapath flows do support wildcarding, in Open
592 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
593 also support different actions and number ports differently.
595 Datapath flows are an implementation detail that is subject to
596 change in future versions of Open vSwitch. Even with the
597 current version of Open vSwitch, hardware switch
598 implementations do not necessarily use this architecture.
600 Users and controllers directly control only the OpenFlow flow
601 table. Open vSwitch manages the datapath flow table itself, so
602 users should not normally be concerned with it.
604 ### Q: Why are there so many different ways to dump flows?
606 A: Open vSwitch has two kinds of flows (see the previous question), so
607 it has commands with different purposes for dumping each kind of
610 - `ovs-ofctl dump-flows <br>` dumps OpenFlow flows, excluding
611 hidden flows. This is the most commonly useful form of flow
612 dump. (Unlike the other commands, this should work with any
613 OpenFlow switch, not just Open vSwitch.)
615 - `ovs-appctl bridge/dump-flows <br>` dumps OpenFlow flows,
616 including hidden flows. This is occasionally useful for
617 troubleshooting suspected issues with in-band control.
619 - `ovs-dpctl dump-flows [dp]` dumps the datapath flow table
620 entries for a Linux kernel-based datapath. In Open vSwitch
621 1.10 and later, ovs-vswitchd merges multiple switches into a
622 single datapath, so it will show all the flows on all your
623 kernel-based switches. This command can occasionally be
624 useful for debugging.
626 - `ovs-appctl dpif/dump-flows <br>`, new in Open vSwitch 1.10,
627 dumps datapath flows for only the specified bridge, regardless
630 ### Q: How does multicast snooping works with VLANs?
632 A: Open vSwitch maintains snooping tables for each VLAN.
634 ### Q: Can OVS populate the kernel flow table in advance instead of in reaction to packets?
636 A: No. There are several reasons:
638 - Kernel flows are not as sophisticated as OpenFlow flows, which
639 means that some OpenFlow policies could require a large number of
640 kernel flows. The "conjunctive match" feature is an extreme
641 example: the number of kernel flows it requires is the product of
642 the number of flows in each dimension.
644 - With multiple OpenFlow flow tables and simple sets of actions, the
645 number of kernel flows required can be as large as the product of
646 the number of flows in each dimension. With more sophisticated
647 actions, the number of kernel flows could be even larger.
649 - Open vSwitch is designed so that any version of OVS userspace
650 interoperates with any version of the OVS kernel module. This
651 forward and backward compatibility requires that userspace observe
652 how the kernel module parses received packets. This is only
653 possible in a straightforward way when userspace adds kernel flows
654 in reaction to received packets.
656 For more relevant information on the architecture of Open vSwitch,
657 please read "The Design and Implementation of Open vSwitch",
658 published in USENIX NSDI 2015.
664 ### Q: I just upgraded and I see a performance drop. Why?
666 A: The OVS kernel datapath may have been updated to a newer version than
667 the OVS userspace components. Sometimes new versions of OVS kernel
668 module add functionality that is backwards compatible with older
669 userspace components but may cause a drop in performance with them.
670 Especially, if a kernel module from OVS 2.1 or newer is paired with
671 OVS userspace 1.10 or older, there will be a performance drop for
674 Updating the OVS userspace components to the latest released
675 version should fix the performance degradation.
677 To get the best possible performance and functionality, it is
678 recommended to pair the same versions of the kernel module and OVS
682 Configuration Problems
683 ----------------------
685 ### Q: I created a bridge and added my Ethernet port to it, using commands
689 ovs-vsctl add-port br0 eth0
691 and as soon as I ran the "add-port" command I lost all connectivity
694 A: A physical Ethernet device that is part of an Open vSwitch bridge
695 should not have an IP address. If one does, then that IP address
696 will not be fully functional.
698 You can restore functionality by moving the IP address to an Open
699 vSwitch "internal" device, such as the network device named after
700 the bridge itself. For example, assuming that eth0's IP address is
701 192.168.128.5, you could run the commands below to fix up the
704 ifconfig eth0 0.0.0.0
705 ifconfig br0 192.168.128.5
707 (If your only connection to the machine running OVS is through the
708 IP address in question, then you would want to run all of these
709 commands on a single command line, or put them into a script.) If
710 there were any additional routes assigned to eth0, then you would
711 also want to use commands to adjust these routes to go through br0.
713 If you use DHCP to obtain an IP address, then you should kill the
714 DHCP client that was listening on the physical Ethernet interface
715 (e.g. eth0) and start one listening on the internal interface
716 (e.g. br0). You might still need to manually clear the IP address
717 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
719 There is no compelling reason why Open vSwitch must work this way.
720 However, this is the way that the Linux kernel bridge module has
721 always worked, so it's a model that those accustomed to Linux
722 bridging are already used to. Also, the model that most people
723 expect is not implementable without kernel changes on all the
724 versions of Linux that Open vSwitch supports.
726 By the way, this issue is not specific to physical Ethernet
727 devices. It applies to all network devices except Open vSwitch
730 ### Q: I created a bridge and added a couple of Ethernet ports to it,
731 ### using commands like these:
734 ovs-vsctl add-port br0 eth0
735 ovs-vsctl add-port br0 eth1
737 and now my network seems to have melted: connectivity is unreliable
738 (even connectivity that doesn't go through Open vSwitch), all the
739 LEDs on my physical switches are blinking, wireshark shows
740 duplicated packets, and CPU usage is very high.
742 A: More than likely, you've looped your network. Probably, eth0 and
743 eth1 are connected to the same physical Ethernet switch. This
744 yields a scenario where OVS receives a broadcast packet on eth0 and
745 sends it out on eth1, then the physical switch connected to eth1
746 sends the packet back on eth0, and so on forever. More complicated
747 scenarios, involving a loop through multiple switches, are possible
750 The solution depends on what you are trying to do:
752 - If you added eth0 and eth1 to get higher bandwidth or higher
753 reliability between OVS and your physical Ethernet switch,
754 use a bond. The following commands create br0 and then add
755 eth0 and eth1 as a bond:
758 ovs-vsctl add-bond br0 bond0 eth0 eth1
760 Bonds have tons of configuration options. Please read the
761 documentation on the Port table in ovs-vswitchd.conf.db(5)
764 Configuration for DPDK-enabled interfaces is slightly less
765 straightforward: see [INSTALL.DPDK.md].
767 - Perhaps you don't actually need eth0 and eth1 to be on the
768 same bridge. For example, if you simply want to be able to
769 connect each of them to virtual machines, then you can put
770 each of them on a bridge of its own:
773 ovs-vsctl add-port br0 eth0
776 ovs-vsctl add-port br1 eth1
778 and then connect VMs to br0 and br1. (A potential
779 disadvantage is that traffic cannot directly pass between br0
780 and br1. Instead, it will go out eth0 and come back in eth1,
783 - If you have a redundant or complex network topology and you
784 want to prevent loops, turn on spanning tree protocol (STP).
785 The following commands create br0, enable STP, and add eth0
786 and eth1 to the bridge. The order is important because you
787 don't want have to have a loop in your network even
791 ovs-vsctl set bridge br0 stp_enable=true
792 ovs-vsctl add-port br0 eth0
793 ovs-vsctl add-port br0 eth1
795 The Open vSwitch implementation of STP is not well tested.
796 Please report any bugs you observe, but if you'd rather avoid
797 acting as a beta tester then another option might be your
800 ### Q: I can't seem to use Open vSwitch in a wireless network.
802 A: Wireless base stations generally only allow packets with the source
803 MAC address of NIC that completed the initial handshake.
804 Therefore, without MAC rewriting, only a single device can
805 communicate over a single wireless link.
807 This isn't specific to Open vSwitch, it's enforced by the access
808 point, so the same problems will show up with the Linux bridge or
809 any other way to do bridging.
811 ### Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
813 A: PPP most commonly carries IP packets, but Open vSwitch works only
814 with Ethernet frames. The correct way to interface PPP to an
815 Ethernet network is usually to use routing instead of switching.
817 ### Q: Is there any documentation on the database tables and fields?
819 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
821 ### Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
822 I only see a datapath called "ovs-system". How can I see datapath
823 information about a particular bridge?
825 A: In version 1.9.0, OVS switched to using a single datapath that is
826 shared by all bridges of that type. The "ovs-appctl dpif/*"
827 commands provide similar functionality that is scoped by the bridge.
829 ### Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
830 see the port in the datapath?
832 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
833 cannot be loaded at the same time. It is likely that on your system the
834 Linux GRE module is already loaded and blocking OVS (to confirm, check
835 dmesg for errors regarding GRE registration). To fix this, unload all
836 GRE modules that appear in lsmod as well as the OVS kernel module. You
837 can then reload the OVS module following the directions in
838 [INSTALL.md], which will ensure that dependencies are satisfied.
840 ### Q: Open vSwitch does not seem to obey my packet filter rules.
842 A: It depends on mechanisms and configurations you want to use.
844 You cannot usefully use typical packet filters, like iptables, on
845 physical Ethernet ports that you add to an Open vSwitch bridge.
846 This is because Open vSwitch captures packets from the interface at
847 a layer lower below where typical packet-filter implementations
848 install their hooks. (This actually applies to any interface of
849 type "system" that you might add to an Open vSwitch bridge.)
851 You can usefully use typical packet filters on Open vSwitch
852 internal ports as they are mostly ordinary interfaces from the point
853 of view of packet filters.
855 For example, suppose you create a bridge br0 and add Ethernet port
856 eth0 to it. Then you can usefully add iptables rules to affect the
857 internal interface br0, but not the physical interface eth0. (br0
858 is also where you would add an IP address, as discussed elsewhere
861 For simple filtering rules, it might be possible to achieve similar
862 results by installing appropriate OpenFlow flows instead.
864 If the use of a particular packet filter setup is essential, Open
865 vSwitch might not be the best choice for you. On Linux, you might
866 want to consider using the Linux Bridge. (This is the only choice if
867 you want to use ebtables rules.) On NetBSD, you might want to
868 consider using the bridge(4) with BRIDGE_IPF option.
870 ### Q: It seems that Open vSwitch does nothing when I removed a port and
871 then immediately put it back. For example, consider that p1 is
872 a port of type=internal:
874 ovs-vsctl del-port br0 p1 -- \
876 set interface p1 type=internal
878 A: It's an expected behaviour.
880 If del-port and add-port happen in a single OVSDB transaction as
881 your example, Open vSwitch always "skips" the intermediate steps.
882 Even if they are done in multiple transactions, it's still allowed
883 for Open vSwitch to skip the intermediate steps and just implement
884 the overall effect. In both cases, your example would be turned
887 If you want to make Open vSwitch actually destroy and then re-create
888 the port for some side effects like resetting kernel setting for the
889 corresponding interface, you need to separate operations into multiple
890 OVSDB transactions and ensure that at least the first one does not have
891 --no-wait. In the following example, the first ovs-vsctl will block
892 until Open vSwitch reloads the new configuration and removes the port:
894 ovs-vsctl del-port br0 p1
895 ovs-vsctl add-port br0 p1 -- \
896 set interface p1 type=internal
898 ### Q: I want to add thousands of ports to an Open vSwitch bridge, but
899 it takes too long (minutes or hours) to do it with ovs-vsctl. How
902 A: If you add them one at a time with ovs-vsctl, it can take a long
903 time to add thousands of ports to an Open vSwitch bridge. This is
904 because every invocation of ovs-vsctl first reads the current
905 configuration from OVSDB. As the number of ports grows, this
906 starts to take an appreciable amount of time, and when it is
907 repeated thousands of times the total time becomes significant.
909 The solution is to add the ports in one invocation of ovs-vsctl (or
910 a small number of them). For example, using bash:
913 cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done
916 takes seconds, not minutes or hours, in the OVS sandbox environment.
918 ### Q: I created a bridge named br0. My bridge shows up in "ovs-vsctl
919 show", but "ovs-ofctl show br0" just prints "br0 is not a bridge
922 A: Open vSwitch wasn't able to create the bridge. Check the
923 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
924 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log).
926 In general, the Open vSwitch database reflects the desired
927 configuration state. ovs-vswitchd monitors the database and, when
928 it changes, reconfigures the system to reflect the new desired
929 state. This normally happens very quickly. Thus, a discrepancy
930 between the database and the actual state indicates that
931 ovs-vswitchd could not implement the configuration, and so one
932 should check the log to find out why. (Another possible cause is
933 that ovs-vswitchd is not running. This will make "ovs-vsctl"
934 commands hang, if they change the configuration, unless one
935 specifies "--no-wait".)
937 ### Q: I have a bridge br0. I added a new port vif1.0, and it shows
938 up in "ovs-vsctl show", but "ovs-vsctl list port" says that it has
939 OpenFlow port ("ofport") -1, and "ovs-ofctl show br0" doesn't show
942 A: Open vSwitch wasn't able to create the port. Check the
943 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
944 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log). Please
945 see the previous question for more information.
947 You may want to upgrade to Open vSwitch 2.3 (or later), in which
948 ovs-vsctl will immediately report when there is an issue creating a
951 ### Q: I created a tap device tap0, configured an IP address on it, and
952 added it to a bridge, like this:
955 ifconfig tap0 192.168.0.123
957 ovs-vsctl add-port br0 tap0
959 I expected that I could then use this IP address to contact other
960 hosts on the network, but it doesn't work. Why not?
962 A: The short answer is that this is a misuse of a "tap" device. Use
963 an "internal" device implemented by Open vSwitch, which works
964 differently and is designed for this use. To solve this problem
965 with an internal device, instead run:
968 ovs-vsctl add-port br0 int0 -- set Interface int0 type=internal
969 ifconfig int0 192.168.0.123
971 Even more simply, you can take advantage of the internal port that
972 every bridge has under the name of the bridge:
975 ifconfig br0 192.168.0.123
977 In more detail, a "tap" device is an interface between the Linux
978 (or *BSD) network stack and a user program that opens it as a
979 socket. When the "tap" device transmits a packet, it appears in
980 the socket opened by the userspace program. Conversely, when the
981 userspace program writes to the "tap" socket, the kernel TCP/IP
982 stack processes the packet as if it had been received by the "tap"
985 Consider the configuration above. Given this configuration, if you
986 "ping" an IP address in the 192.168.0.x subnet, the Linux kernel
987 routing stack will transmit an ARP on the tap0 device. Open
988 vSwitch userspace treats "tap" devices just like any other network
989 device; that is, it doesn't open them as "tap" sockets. That means
990 that the ARP packet will simply get dropped.
992 You might wonder why the Open vSwitch kernel module doesn't
993 intercept the ARP packet and bridge it. After all, Open vSwitch
994 intercepts packets on other devices. The answer is that Open
995 vSwitch only intercepts *received* packets, but this is a packet
996 being transmitted. The same thing happens for all other types of
997 network devices, except for Open vSwitch "internal" ports. If you,
998 for example, add a physical Ethernet port to an OVS bridge,
999 configure an IP address on a physical Ethernet port, and then issue
1000 a "ping" to an address in that subnet, the same thing happens: an
1001 ARP gets transmitted on the physical Ethernet port and Open vSwitch
1002 never sees it. (You should not do that, as documented at the
1003 beginning of this section.)
1005 It can make sense to add a "tap" device to an Open vSwitch bridge,
1006 if some userspace program (other than Open vSwitch) has opened the
1007 tap socket. This is the case, for example, if the "tap" device was
1008 created by KVM (or QEMU) to simulate a virtual NIC. In such a
1009 case, when OVS bridges a packet to the "tap" device, the kernel
1010 forwards that packet to KVM in userspace, which passes it along to
1011 the VM, and in the other direction, when the VM sends a packet, KVM
1012 writes it to the "tap" socket, which causes OVS to receive it and
1013 bridge it to the other OVS ports. Please note that in such a case
1014 no IP address is configured on the "tap" device (there is normally
1015 an IP address configured in the virtual NIC inside the VM, but this
1016 is not visible to the host Linux kernel or to Open vSwitch).
1018 There is one special case in which Open vSwitch does directly read
1019 and write "tap" sockets. This is an implementation detail of the
1020 Open vSwitch userspace switch, which implements its "internal"
1021 ports as Linux (or *BSD) "tap" sockets. In such a userspace
1022 switch, OVS receives packets sent on the "tap" device used to
1023 implement an "internal" port by reading the associated "tap"
1024 socket, and bridges them to the rest of the switch. In the other
1025 direction, OVS transmits packets bridged to the "internal" port by
1026 writing them to the "tap" socket, causing them to be processed by
1027 the kernel TCP/IP stack as if they had been received on the "tap"
1028 device. Users should not need to be concerned with this
1029 implementation detail.
1031 Open vSwitch has a network device type called "tap". This is
1032 intended only for implementing "internal" ports in the OVS
1033 userspace switch and should not be used otherwise. In particular,
1034 users should not configure KVM "tap" devices as type "tap" (use
1035 type "system", the default, instead).
1038 Quality of Service (QoS)
1039 ------------------------
1041 ### Q: Does OVS support Quality of Service (QoS)?
1043 A: Yes. For traffic that egresses from a switch, OVS supports traffic
1044 shaping; for traffic that ingresses into a switch, OVS support
1045 policing. Policing is a simple form of quality-of-service that
1046 simply drops packets received in excess of the configured rate. Due
1047 to its simplicity, policing is usually less accurate and less
1048 effective than egress traffic shaping, which queues packets.
1050 Keep in mind that ingress and egress are from the perspective of the
1051 switch. That means that egress shaping limits the rate at which
1052 traffic is allowed to transmit from a physical interface, but the
1053 rate at which traffic will be received on a virtual machine's VIF.
1054 For ingress policing, the behavior is the opposite.
1056 ### Q: How do I configure egress traffic shaping?
1058 A: Suppose that you want to set up bridge br0 connected to physical
1059 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
1060 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
1061 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
1062 could configure the bridge this way:
1066 add-port br0 eth0 -- \
1067 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
1068 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
1069 set port eth0 qos=@newqos -- \
1070 --id=@newqos create qos type=linux-htb \
1071 other-config:max-rate=1000000000 \
1072 queues:123=@vif10queue \
1073 queues:234=@vif20queue -- \
1074 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
1075 --id=@vif20queue create queue other-config:max-rate=20000000
1077 At this point, bridge br0 is configured with the ports and eth0 is
1078 configured with the queues that you need for QoS, but nothing is
1079 actually directing packets from vif1.0 or vif2.0 to the queues that
1080 we have set up for them. That means that all of the packets to
1081 eth0 are going to the "default queue", which is not what we want.
1083 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
1084 queues reserved for them:
1086 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
1087 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
1089 Each of the above flows matches on the input port, sets up the
1090 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
1091 executes the "normal" action, which performs the same switching
1092 that Open vSwitch would have done without any OpenFlow flows being
1093 present. (We know that vif1.0 and vif2.0 have OpenFlow port
1094 numbers 5 and 6, respectively, because we set their ofport_request
1095 columns above. If we had not done that, then we would have needed
1096 to find out their port numbers before setting up these flows.)
1098 Now traffic going from vif1.0 or vif2.0 to eth0 should be
1101 By the way, if you delete the bridge created by the above commands,
1104 ovs-vsctl del-br br0
1106 then that will leave one unreferenced QoS record and two
1107 unreferenced Queue records in the Open vSwich database. One way to
1108 clear them out, assuming you don't have other QoS or Queue records
1109 that you want to keep, is:
1111 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
1113 If you do want to keep some QoS or Queue records, or the Open
1114 vSwitch you are using is older than version 1.8 (which added the
1115 --all option), then you will have to destroy QoS and Queue records
1118 ### Q: How do I configure ingress policing?
1120 A: A policing policy can be configured on an interface to drop packets
1121 that arrive at a higher rate than the configured value. For example,
1122 the following commands will rate-limit traffic that vif1.0 may
1125 ovs-vsctl set interface vif1.0 ingress_policing_rate=10000
1126 ovs-vsctl set interface vif1.0 ingress_policing_burst=1000
1128 Traffic policing can interact poorly with some network protocols and
1129 can have surprising results. The "Ingress Policing" section of
1130 ovs-vswitchd.conf.db(5) discusses the issues in greater detail.
1132 ### Q: I configured Quality of Service (QoS) in my OpenFlow network by
1133 adding records to the QoS and Queue table, but the results aren't
1136 A: Did you install OpenFlow flows that use your queues? This is the
1137 primary way to tell Open vSwitch which queues you want to use. If
1138 you don't do this, then the default queue will be used, which will
1139 probably not have the effect you want.
1141 Refer to the previous question for an example.
1143 ### Q: I'd like to take advantage of some QoS feature that Open vSwitch
1144 doesn't yet support. How do I do that?
1146 A: Open vSwitch does not implement QoS itself. Instead, it can
1147 configure some, but not all, of the QoS features built into the
1148 Linux kernel. If you need some QoS feature that OVS cannot
1149 configure itself, then the first step is to figure out whether
1150 Linux QoS supports that feature. If it does, then you can submit a
1151 patch to support Open vSwitch configuration for that feature, or
1152 you can use "tc" directly to configure the feature in Linux. (If
1153 Linux QoS doesn't support the feature you want, then first you have
1154 to add that support to Linux.)
1156 ### Q: I configured QoS, correctly, but my measurements show that it isn't
1157 working as well as I expect.
1159 A: With the Linux kernel, the Open vSwitch implementation of QoS has
1162 - Open vSwitch configures a subset of Linux kernel QoS
1163 features, according to what is in OVSDB. It is possible that
1164 this code has bugs. If you believe that this is so, then you
1165 can configure the Linux traffic control (QoS) stack directly
1166 with the "tc" program. If you get better results that way,
1167 you can send a detailed bug report to bugs@openvswitch.org.
1169 It is certain that Open vSwitch cannot configure every Linux
1170 kernel QoS feature. If you need some feature that OVS cannot
1171 configure, then you can also use "tc" directly (or add that
1174 - The Open vSwitch implementation of OpenFlow allows flows to
1175 be directed to particular queues. This is pretty simple and
1176 unlikely to have serious bugs at this point.
1178 However, most problems with QoS on Linux are not bugs in Open
1179 vSwitch at all. They tend to be either configuration errors
1180 (please see the earlier questions in this section) or issues with
1181 the traffic control (QoS) stack in Linux. The Open vSwitch
1182 developers are not experts on Linux traffic control. We suggest
1183 that, if you believe you are encountering a problem with Linux
1184 traffic control, that you consult the tc manpages (e.g. tc(8),
1185 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
1186 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
1188 ### Q: Does Open vSwitch support OpenFlow meters?
1190 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
1191 OpenFlow meters. There is no implementation of meters in the Open
1192 vSwitch software switch (neither the kernel-based nor userspace
1199 ### Q: What's a VLAN?
1201 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
1202 partition a single switch into multiple switches. Suppose, for
1203 example, that you have two groups of machines, group A and group B.
1204 You want the machines in group A to be able to talk to each other,
1205 and you want the machine in group B to be able to talk to each
1206 other, but you don't want the machines in group A to be able to
1207 talk to the machines in group B. You can do this with two
1208 switches, by plugging the machines in group A into one switch and
1209 the machines in group B into the other switch.
1211 If you only have one switch, then you can use VLANs to do the same
1212 thing, by configuring the ports for machines in group A as VLAN
1213 "access ports" for one VLAN and the ports for group B as "access
1214 ports" for a different VLAN. The switch will only forward packets
1215 between ports that are assigned to the same VLAN, so this
1216 effectively subdivides your single switch into two independent
1217 switches, one for each group of machines.
1219 So far we haven't said anything about VLAN headers. With access
1220 ports, like we've described so far, no VLAN header is present in
1221 the Ethernet frame. This means that the machines (or switches)
1222 connected to access ports need not be aware that VLANs are
1223 involved, just like in the case where we use two different physical
1226 Now suppose that you have a whole bunch of switches in your
1227 network, instead of just one, and that some machines in group A are
1228 connected directly to both switches 1 and 2. To allow these
1229 machines to talk to each other, you could add an access port for
1230 group A's VLAN to switch 1 and another to switch 2, and then
1231 connect an Ethernet cable between those ports. That works fine,
1232 but it doesn't scale well as the number of switches and the number
1233 of VLANs increases, because you use up a lot of valuable switch
1234 ports just connecting together your VLANs.
1236 This is where VLAN headers come in. Instead of using one cable and
1237 two ports per VLAN to connect a pair of switches, we configure a
1238 port on each switch as a VLAN "trunk port". Packets sent and
1239 received on a trunk port carry a VLAN header that says what VLAN
1240 the packet belongs to, so that only two ports total are required to
1241 connect the switches, regardless of the number of VLANs in use.
1242 Normally, only switches (either physical or virtual) are connected
1243 to a trunk port, not individual hosts, because individual hosts
1244 don't expect to see a VLAN header in the traffic that they receive.
1246 None of the above discussion says anything about particular VLAN
1247 numbers. This is because VLAN numbers are completely arbitrary.
1248 One must only ensure that a given VLAN is numbered consistently
1249 throughout a network and that different VLANs are given different
1250 numbers. (That said, VLAN 0 is usually synonymous with a packet
1251 that has no VLAN header, and VLAN 4095 is reserved.)
1253 ### Q: VLANs don't work.
1255 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
1256 bugs. If you are having problems with VLANs that you suspect to be
1257 driver related, then you have several options:
1259 - Upgrade to Linux 3.3 or later.
1261 - Build and install a fixed version of the particular driver
1262 that is causing trouble, if one is available.
1264 - Use a NIC whose driver does not have VLAN problems.
1266 - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later
1267 that works around bugs in kernel drivers. To enable VLAN
1268 splinters on interface eth0, use the command:
1270 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
1272 For VLAN splinters to be effective, Open vSwitch must know
1273 which VLANs are in use. See the "VLAN splinters" section in
1274 the Interface table in ovs-vswitchd.conf.db(5) for details on
1275 how Open vSwitch infers in-use VLANs.
1277 VLAN splinters increase memory use and reduce performance, so
1278 use them only if needed.
1280 - Apply the "vlan workaround" patch from the XenServer kernel
1281 patch queue, build Open vSwitch against this patched kernel,
1282 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
1283 workaround for each interface whose driver is buggy.
1285 (This is a nontrivial exercise, so this option is included
1286 only for completeness.)
1288 It is not always easy to tell whether a Linux kernel driver has
1289 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
1290 can help you test. See their manpages for details. Of the two
1291 utilities, ovs-test(8) is newer and more thorough, but
1292 ovs-vlan-test(8) may be easier to use.
1294 ### Q: VLANs still don't work. I've tested the driver so I know that it's OK.
1296 A: Do you have VLANs enabled on the physical switch that OVS is
1297 attached to? Make sure that the port is configured to trunk the
1298 VLAN or VLANs that you are using with OVS.
1300 ### Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
1301 and to its destination host, but OVS seems to drop incoming return
1304 A: It's possible that you have the VLAN configured on your physical
1305 switch as the "native" VLAN. In this mode, the switch treats
1306 incoming packets either tagged with the native VLAN or untagged as
1307 part of the native VLAN. It may also send outgoing packets in the
1308 native VLAN without a VLAN tag.
1310 If this is the case, you have two choices:
1312 - Change the physical switch port configuration to tag packets
1313 it forwards to OVS with the native VLAN instead of forwarding
1316 - Change the OVS configuration for the physical port to a
1317 native VLAN mode. For example, the following sets up a
1318 bridge with port eth0 in "native-tagged" mode in VLAN 9:
1320 ovs-vsctl add-br br0
1321 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
1323 In this situation, "native-untagged" mode will probably work
1324 equally well. Refer to the documentation for the Port table
1325 in ovs-vswitchd.conf.db(5) for more information.
1327 ### Q: I added a pair of VMs on different VLANs, like this:
1329 ovs-vsctl add-br br0
1330 ovs-vsctl add-port br0 eth0
1331 ovs-vsctl add-port br0 tap0 tag=9
1332 ovs-vsctl add-port br0 tap1 tag=10
1334 but the VMs can't access each other, the external network, or the
1337 A: It is to be expected that the VMs can't access each other. VLANs
1338 are a means to partition a network. When you configured tap0 and
1339 tap1 as access ports for different VLANs, you indicated that they
1340 should be isolated from each other.
1342 As for the external network and the Internet, it seems likely that
1343 the machines you are trying to access are not on VLAN 9 (or 10) and
1344 that the Internet is not available on VLAN 9 (or 10).
1346 ### Q: I added a pair of VMs on the same VLAN, like this:
1348 ovs-vsctl add-br br0
1349 ovs-vsctl add-port br0 eth0
1350 ovs-vsctl add-port br0 tap0 tag=9
1351 ovs-vsctl add-port br0 tap1 tag=9
1353 The VMs can access each other, but not the external network or the
1356 A: It seems likely that the machines you are trying to access in the
1357 external network are not on VLAN 9 and that the Internet is not
1358 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
1359 trunk VLAN on the upstream switch port to which eth0 is connected.
1361 ### Q: Can I configure an IP address on a VLAN?
1363 A: Yes. Use an "internal port" configured as an access port. For
1364 example, the following configures IP address 192.168.0.7 on VLAN 9.
1365 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
1366 they have an 802.1Q header with VLAN 9. Conversely, traffic
1367 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
1370 ovs-vsctl add-br br0
1371 ovs-vsctl add-port br0 eth0
1372 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1373 ifconfig vlan9 192.168.0.7
1375 See also the following question.
1377 ### Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
1380 ovs-vsctl add-br br0
1381 ovs-vsctl add-port br0 eth0
1382 ifconfig br0 192.168.0.5
1383 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1384 ifconfig vlan9 192.168.0.9
1386 but other hosts that are only on VLAN 0 can reach the IP address
1387 configured on VLAN 9. What's going on?
1389 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
1390 approaches to IP address handling in Internet hosts:
1392 - In the "Strong ES Model", where an ES is a host ("End
1393 System"), an IP address is primarily associated with a
1394 particular interface. The host discards packets that arrive
1395 on interface A if they are destined for an IP address that is
1396 configured on interface B. The host never sends packets from
1397 interface A using a source address configured on interface B.
1399 - In the "Weak ES Model", an IP address is primarily associated
1400 with a host. The host accepts packets that arrive on any
1401 interface if they are destined for any of the host's IP
1402 addresses, even if the address is configured on some
1403 interface other than the one on which it arrived. The host
1404 does not restrict itself to sending packets from an IP
1405 address associated with the originating interface.
1407 Linux uses the weak ES model. That means that when packets
1408 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
1409 br0, the kernel IP stack accepts them there for the VLAN 9 IP
1410 address, even though they were not received on vlan9, the network
1413 To simulate the strong ES model on Linux, one may add iptables rule
1414 to filter packets based on source and destination address and
1415 adjust ARP configuration with sysctls.
1417 BSD uses the strong ES model.
1419 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1421 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1422 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1423 implementation of the OpenFlow "normal switching" action. By
1424 default, when Open vSwitch isn't connected to a controller and
1425 nothing has been manually configured in the flow table, all traffic
1426 goes through the "normal switching" action. But, if you set up
1427 OpenFlow flows on your own, through a controller or using ovs-ofctl
1428 or through other means, then you have to implement VLAN handling
1431 You can use "normal switching" as a component of your OpenFlow
1432 actions, e.g. by putting "normal" into the lists of actions on
1433 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1434 controller. In situations where this is not suitable, you can
1435 implement VLAN handling yourself, e.g.:
1437 - If a packet comes in on an access port, and the flow table
1438 needs to send it out on a trunk port, then the flow can add
1439 the appropriate VLAN tag with the "mod_vlan_vid" action.
1441 - If a packet comes in on a trunk port, and the flow table
1442 needs to send it out on an access port, then the flow can
1443 strip the VLAN tag with the "strip_vlan" action.
1445 ### Q: I configured ports on a bridge as access ports with different VLAN
1448 ovs-vsctl add-br br0
1449 ovs-vsctl set-controller br0 tcp:192.168.0.10:6653
1450 ovs-vsctl add-port br0 eth0
1451 ovs-vsctl add-port br0 tap0 tag=9
1452 ovs-vsctl add-port br0 tap1 tag=10
1454 but the VMs running behind tap0 and tap1 can still communicate,
1455 that is, they are not isolated from each other even though they are
1458 A: Do you have a controller configured on br0 (as the commands above
1459 do)? If so, then this is a variant on the previous question, "My
1460 OpenFlow controller doesn't see the VLANs that I expect," and you
1461 can refer to the answer there for more information.
1463 ### Q: How MAC learning works with VLANs?
1465 A: Open vSwitch implements Independent VLAN Learning (IVL) for
1466 OFPP_NORMAL action. I.e. it logically has separate learning tables
1473 ### Q: What's a VXLAN?
1475 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1476 to solve the scaling challenges of VLAN networks in a multi-tenant
1477 environment. VXLAN is an overlay network which transports an L2 network
1478 over an existing L3 network. For more information on VXLAN, please see
1481 http://tools.ietf.org/html/rfc7348
1483 ### Q: How much of the VXLAN protocol does Open vSwitch currently support?
1485 A: Open vSwitch currently supports the framing format for packets on the
1486 wire. There is currently no support for the multicast aspects of VXLAN.
1487 To get around the lack of multicast support, it is possible to
1488 pre-provision MAC to IP address mappings either manually or from a
1491 ### Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1494 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1495 is 4789. However, it is possible to configure the destination UDP port
1496 manually on a per-VXLAN tunnel basis. An example of this configuration is
1499 ovs-vsctl add-br br0
1500 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1501 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1502 options:dst_port=8472
1505 Using OpenFlow (Manually or Via Controller)
1506 -------------------------------------------
1508 ### Q: What versions of OpenFlow does Open vSwitch support?
1510 A: The following table lists the versions of OpenFlow supported by
1511 each version of Open vSwitch:
1513 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5
1514 ###============ ===== ===== ===== ===== ===== =====
1515 1.9 and earlier yes --- --- --- --- ---
1516 1.10 yes --- [*] [*] --- ---
1517 1.11 yes --- [*] [*] --- ---
1518 2.0 yes [*] [*] [*] --- ---
1519 2.1 yes [*] [*] [*] --- ---
1520 2.2 yes [*] [*] [*] [%] [*]
1521 2.3 yes yes yes yes [*] [*]
1523 [*] Supported, with one or more missing features.
1524 [%] Experimental, unsafe implementation.
1526 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default
1527 in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1,
1528 1.2, and 1.3 must be enabled manually in ovs-vswitchd. OpenFlow
1529 1.4 and 1.5 are also supported, with missing features, in Open
1530 vSwitch 2.3 and later, but not enabled by default. In any case,
1531 the user may override the default:
1533 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0:
1535 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1537 - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0:
1539 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15
1541 - To enable only OpenFlow 1.0 on bridge br0:
1543 ovs-vsctl set bridge br0 protocols=OpenFlow10
1545 All current versions of ovs-ofctl enable only OpenFlow 1.0 by
1546 default. Use the -O option to enable support for later versions of
1547 OpenFlow in ovs-ofctl. For example:
1549 ovs-ofctl -O OpenFlow13 dump-flows br0
1551 (Open vSwitch 2.2 had an experimental implementation of OpenFlow
1552 1.4 that could cause crashes. We don't recommend enabling it.)
1554 [OPENFLOW-1.1+.md] in the Open vSwitch source tree tracks support for
1555 OpenFlow 1.1 and later features. When support for OpenFlow 1.4 and
1556 1.5 is solidly implemented, Open vSwitch will enable those version
1559 ### Q: Does Open vSwitch support MPLS?
1561 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1562 these versions can match on MPLS Ethernet types, but they cannot
1563 match, push, or pop MPLS labels, nor can they look past MPLS labels
1564 into the encapsulated packet.
1566 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1567 for MPLS. With the userspace datapath only, these versions can
1568 match, push, or pop a single MPLS label, but they still cannot look
1569 past MPLS labels (even after popping them) into the encapsulated
1570 packet. Kernel datapath support is unchanged from earlier
1573 Open vSwitch version 2.3 can match, push, or pop a single MPLS
1574 label and look past the MPLS label into the encapsulated packet.
1575 Both userspace and kernel datapaths will be supported, but MPLS
1576 processing always happens in userspace either way, so kernel
1577 datapath performance will be disappointing.
1579 Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS
1580 labels and look past the MPLS label into the encapsulated packet.
1581 It will have kernel support for MPLS, yielding improved
1584 ### Q: I'm getting "error type 45250 code 0". What's that?
1586 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1587 vSwitch uses this extension when it must report an error to an
1588 OpenFlow controller but no standard OpenFlow error code is
1591 Open vSwitch logs the errors that it sends to controllers, so the
1592 easiest thing to do is probably to look at the ovs-vswitchd log to
1593 find out what the error was.
1595 If you want to dissect the extended error message yourself, the
1596 format is documented in include/openflow/nicira-ext.h in the Open
1597 vSwitch source distribution. The extended error codes are
1598 documented in lib/ofp-errors.h.
1600 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1601 doesn't actually appear through the OpenFlow connection, even
1602 though I know that it's going through.
1603 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1604 to apply to certain traffic, especially traffic between OVS and
1605 the controller itself.
1607 A: By default, Open vSwitch assumes that OpenFlow controllers are
1608 connected "in-band", that is, that the controllers are actually
1609 part of the network that is being controlled. In in-band mode,
1610 Open vSwitch sets up special "hidden" flows to make sure that
1611 traffic can make it back and forth between OVS and the controllers.
1612 These hidden flows are higher priority than any flows that can be
1613 set up through OpenFlow, and they are not visible through normal
1614 OpenFlow flow table dumps.
1616 Usually, the hidden flows are desirable and helpful, but
1617 occasionally they can cause unexpected behavior. You can view the
1618 full OpenFlow flow table, including hidden flows, on bridge br0
1621 ovs-appctl bridge/dump-flows br0
1623 to help you debug. The hidden flows are those with priorities
1624 greater than 65535 (the maximum priority that can be set with
1627 The DESIGN file at the top level of the Open vSwitch source
1628 distribution describes the in-band model in detail.
1630 If your controllers are not actually in-band (e.g. they are on
1631 localhost via 127.0.0.1, or on a separate network), then you should
1632 configure your controllers in "out-of-band" mode. If you have one
1633 controller on bridge br0, then you can configure out-of-band mode
1636 ovs-vsctl set controller br0 connection-mode=out-of-band
1638 ### Q: I configured all my controllers for out-of-band control mode but
1639 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1641 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1642 set-manager"). By default, Open vSwitch assumes that managers need
1643 in-band rules set up on every bridge. You can disable these rules
1646 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1648 This actually disables in-band control entirely for the bridge, as
1649 if all the bridge's controllers were configured for out-of-band
1652 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1654 A: See answer under "VLANs", above.
1656 ### Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1657 but I got a funny message like this:
1659 ofp_util|INFO|normalization changed ofp_match, details:
1660 ofp_util|INFO| pre: nw_dst=192.168.0.1
1663 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1664 match had disappeared, so that the flow ends up matching every
1667 A: The term "normalization" in the log message means that a flow
1668 cannot match on an L3 field without saying what L3 protocol is in
1669 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1670 so the L3 field match was dropped.
1672 In this case, the L3 protocol could be IP or ARP. A correct
1673 command for each possibility is, respectively:
1675 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1679 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1681 Similarly, a flow cannot match on an L4 field without saying what
1682 L4 protocol is in use. For example, the flow match "tp_src=1234"
1683 is, by itself, meaningless and will be ignored. Instead, to match
1684 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1685 source port 1234, write "udp,tp_src=1234".
1687 ### Q: How can I figure out the OpenFlow port number for a given port?
1689 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1690 respond with an OFPT_FEATURES_REPLY that, among other information,
1691 includes a mapping between OpenFlow port names and numbers. From a
1692 command prompt, "ovs-ofctl show br0" makes such a request and
1693 prints the response for switch br0.
1695 The Interface table in the Open vSwitch database also maps OpenFlow
1696 port names to numbers. To print the OpenFlow port number
1697 associated with interface eth0, run:
1699 ovs-vsctl get Interface eth0 ofport
1701 You can print the entire mapping with:
1703 ovs-vsctl -- --columns=name,ofport list Interface
1705 but the output mixes together interfaces from all bridges in the
1706 database, so it may be confusing if more than one bridge exists.
1708 In the Open vSwitch database, ofport value -1 means that the
1709 interface could not be created due to an error. (The Open vSwitch
1710 log should indicate the reason.) ofport value [] (the empty set)
1711 means that the interface hasn't been created yet. The latter is
1712 normally an intermittent condition (unless ovs-vswitchd is not
1715 ### Q: I added some flows with my controller or with ovs-ofctl, but when I
1716 run "ovs-dpctl dump-flows" I don't see them.
1718 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1719 won't display the information that you want. You want to use
1720 "ovs-ofctl dump-flows" instead.
1722 ### Q: It looks like each of the interfaces in my bonded port shows up
1723 as an individual OpenFlow port. Is that right?
1725 A: Yes, Open vSwitch makes individual bond interfaces visible as
1726 OpenFlow ports, rather than the bond as a whole. The interfaces
1727 are treated together as a bond for only a few purposes:
1729 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1730 controller is not configured, this happens implicitly to
1733 - Mirrors configured for output to a bonded port.
1735 It would make a lot of sense for Open vSwitch to present a bond as
1736 a single OpenFlow port. If you want to contribute an
1737 implementation of such a feature, please bring it up on the Open
1738 vSwitch development mailing list at dev@openvswitch.org.
1740 ### Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1741 multiple hosts, and other components. The behavior isn't what I
1744 A: To debug network behavior problems, trace the path of a packet,
1745 hop-by-hop, from its origin in one host to a remote host. If
1746 that's correct, then trace the path of the response packet back to
1749 The open source tool called "plotnetcfg" can help to understand the
1750 relationship between the networking devices on a single host.
1752 Usually a simple ICMP echo request and reply ("ping") packet is
1753 good enough. Start by initiating an ongoing "ping" from the origin
1754 host to a remote host. If you are tracking down a connectivity
1755 problem, the "ping" will not display any successful output, but
1756 packets are still being sent. (In this case the packets being sent
1757 are likely ARP rather than ICMP.)
1759 Tools available for tracing include the following:
1761 - "tcpdump" and "wireshark" for observing hops across network
1762 devices, such as Open vSwitch internal devices and physical
1765 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1766 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1767 These tools allow one to observe the actions being taken on
1768 packets in ongoing flows.
1770 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1771 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1772 documentation, and "Why are there so many different ways to
1773 dump flows?" above for some background.
1775 - "ovs-appctl ofproto/trace" to observe the logic behind how
1776 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1777 documentation. You can out more details about a given flow
1778 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1779 a flow from the output into an "ovs-appctl ofproto/trace"
1782 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1783 observe what goes on at these physical hops.
1785 Starting at the origin of a given packet, observe the packet at
1786 each hop in turn. For example, in one plausible scenario, you
1789 1. "tcpdump" the "eth" interface through which an ARP egresses
1790 a VM, from inside the VM.
1792 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1793 ingresses the host machine.
1795 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1796 the host interface through which the ARP egresses the
1797 physical machine. You may need to use "ovs-dpctl show" to
1798 interpret the port numbers. If the output seems surprising,
1799 you can use "ovs-appctl ofproto/trace" to observe details of
1800 how ovs-vswitchd determined the actions in the "ovs-dpctl
1803 4. "tcpdump" the "eth" interface through which the ARP egresses
1804 the physical machine.
1806 5. "tcpdump" the "eth" interface through which the ARP
1807 ingresses the physical machine, at the remote host that
1810 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1811 remote host that receives the ARP and observe the VM "vif"
1812 or "tap" interface to which the flow is directed. Again,
1813 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1815 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1818 8. "tcpdump" the "eth" interface through which the ARP
1819 ingresses a VM, from inside the VM.
1821 It is likely that during one of these steps you will figure out the
1822 problem. If not, then follow the ARP reply back to the origin, in
1825 ### Q: How do I make a flow drop packets?
1827 A: To drop a packet is to receive it without forwarding it. OpenFlow
1828 explicitly specifies forwarding actions. Thus, a flow with an
1829 empty set of actions does not forward packets anywhere, causing
1830 them to be dropped. You can specify an empty set of actions with
1831 "actions=" on the ovs-ofctl command line. For example:
1833 ovs-ofctl add-flow br0 priority=65535,actions=
1835 would cause every packet entering switch br0 to be dropped.
1837 You can write "drop" explicitly if you like. The effect is the
1838 same. Thus, the following command also causes every packet
1839 entering switch br0 to be dropped:
1841 ovs-ofctl add-flow br0 priority=65535,actions=drop
1843 "drop" is not an action, either in OpenFlow or Open vSwitch.
1844 Rather, it is only a way to say that there are no actions.
1846 ### Q: I added a flow to send packets out the ingress port, like this:
1848 ovs-ofctl add-flow br0 in_port=2,actions=2
1850 but OVS drops the packets instead.
1852 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1853 out its ingress port. The rationale is that dropping these packets
1854 makes it harder to loop the network. Sometimes this behavior can
1855 even be convenient, e.g. it is often the desired behavior in a flow
1856 that forwards a packet to several ports ("floods" the packet).
1858 Sometimes one really needs to send a packet out its ingress port
1859 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1860 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1862 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1864 This also works in some circumstances where the flow doesn't match
1865 on the input port. For example, if you know that your switch has
1866 five ports numbered 2 through 6, then the following will send every
1867 received packet out every port, even its ingress port:
1869 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1873 ovs-ofctl add-flow br0 actions=all,in_port
1875 Sometimes, in complicated flow tables with multiple levels of
1876 "resubmit" actions, a flow needs to output to a particular port
1877 that may or may not be the ingress port. It's difficult to take
1878 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1879 provides, as an OpenFlow extension, the ability to modify the
1880 in_port field. Whatever value is currently in the in_port field is
1881 the port to which outputs will be dropped, as well as the
1882 destination for OFPP_IN_PORT. This means that the following will
1883 reliably output to port 2 or to ports 2 through 6, respectively:
1885 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1886 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1888 If the input port is important, then one may save and restore it on
1891 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1892 load:0->NXM_OF_IN_PORT[],\
1894 pop:NXM_OF_IN_PORT[]
1896 ### Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1897 on port 2. I set up flows to forward only traffic destined to the
1898 other host and drop other traffic, like this:
1900 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1901 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1902 priority=0,actions=drop
1904 But it doesn't work--I don't get any connectivity when I do this.
1907 A: These flows drop the ARP packets that IP hosts use to establish IP
1908 connectivity over Ethernet. To solve the problem, add flows to
1909 allow ARP to pass between the hosts:
1911 priority=5,in_port=1,arp,actions=2
1912 priority=5,in_port=2,arp,actions=1
1914 This issue can manifest other ways, too. The following flows that
1915 match on Ethernet addresses instead of IP addresses will also drop
1916 ARP packets, because ARP requests are broadcast instead of being
1917 directed to a specific host:
1919 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1920 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1921 priority=0,actions=drop
1923 The solution already described above will also work in this case.
1924 It may be better to add flows to allow all multicast and broadcast
1927 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1928 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1930 ### Q: My bridge disconnects from my controller on add-port/del-port.
1932 A: Reconfiguring your bridge can change your bridge's datapath-id because
1933 Open vSwitch generates datapath-id from the MAC address of one of its ports.
1934 In that case, Open vSwitch disconnects from controllers because there's
1935 no graceful way to notify controllers about the change of datapath-id.
1937 To avoid the behaviour, you can configure datapath-id manually.
1939 ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef
1941 ### Q: My controller is getting errors about "buffers". What's going on?
1943 A: When a switch sends a packet to an OpenFlow controller using a
1944 "packet-in" message, it can also keep a copy of that packet in a
1945 "buffer", identified by a 32-bit integer "buffer_id". There are
1946 two advantages to buffering. First, when the controller wants to
1947 tell the switch to do something with the buffered packet (with a
1948 "packet-out" OpenFlow request), it does not need to send another
1949 copy of the packet back across the OpenFlow connection, which
1950 reduces the bandwidth cost of the connection and improves latency.
1951 This enables the second advantage: the switch can optionally send
1952 only the first part of the packet to the controller (assuming that
1953 the switch only needs to look at the first few bytes of the
1954 packet), further reducing bandwidth and improving latency.
1956 However, buffering introduces some issues of its own. First, any
1957 switch has limited resources, so if the controller does not use a
1958 buffered packet, the switch has to decide how long to keep it
1959 buffered. When many packets are sent to a controller and buffered,
1960 Open vSwitch can discard buffered packets that the controller has
1961 not used after as little as 5 seconds. This means that
1962 controllers, if they make use of packet buffering, should use the
1963 buffered packets promptly. (This includes sending a "packet-out"
1964 with no actions if the controller does not want to do anything with
1965 a buffered packet, to clear the packet buffer and effectively
1968 Second, packet buffers are one-time-use, meaning that a controller
1969 cannot use a single packet buffer in two or more "packet-out"
1970 commands. Open vSwitch will respond with an error to the second
1971 and subsequent "packet-out"s in such a case.
1973 Finally, a common error early in controller development is to try
1974 to use buffer_id 0 in a "packet-out" message as if 0 represented
1975 "no buffered packet". This is incorrect usage: the buffer_id with
1976 this meaning is actually 0xffffffff.
1978 ovs-vswitchd(8) describes some details of Open vSwitch packet
1979 buffering that the OpenFlow specification requires implementations
1986 ### Q: How do I implement a new OpenFlow message?
1988 A: Add your new message to "enum ofpraw" and "enum ofptype" in
1989 lib/ofp-msgs.h, following the existing pattern. Then recompile and
1990 fix all of the new warnings, implementing new functionality for the
1991 new message as needed. (If you configure with --enable-Werror, as
1992 described in [INSTALL.md], then it is impossible to miss any warnings.)
1994 If you need to add an OpenFlow vendor extension message for a
1995 vendor that doesn't yet have any extension messages, then you will
1996 also need to edit build-aux/extract-ofp-msgs.
1998 ### Q: How do I add support for a new field or header?
2000 A: Add new members for your field to "struct flow" in lib/flow.h, and
2001 add new enumerations for your new field to "enum mf_field_id" in
2002 lib/meta-flow.h, following the existing pattern. Also, add support
2003 to miniflow_extract() in lib/flow.c for extracting your new field
2004 from a packet into struct miniflow. Then recompile and fix all of
2005 the new warnings, implementing new functionality for the new field
2006 or header as needed. (If you configure with --enable-Werror, as
2007 described in [INSTALL.md], then it is impossible to miss any
2010 If you want kernel datapath support for your new field, you also
2011 need to modify the kernel module for the operating systems you are
2012 interested in. This isn't mandatory, since fields understood only
2013 by userspace work too (with a performance penalty), so it's
2014 reasonable to start development without it. If you implement
2015 kernel module support for Linux, then the Linux kernel "netdev"
2016 mailing list is the place to submit that support first; please read
2017 up on the Linux kernel development process separately. The Windows
2018 datapath kernel module support, on the other hand, is maintained
2019 within the OVS tree, so patches for that can go directly to
2022 ### Q: How do I add support for a new OpenFlow action?
2024 A: Add your new action to "enum ofp_raw_action_type" in
2025 lib/ofp-actions.c, following the existing pattern. Then recompile
2026 and fix all of the new warnings, implementing new functionality for
2027 the new action as needed. (If you configure with --enable-Werror,
2028 as described in [INSTALL.md], then it is impossible to miss any
2031 If you need to add an OpenFlow vendor extension action for a vendor
2032 that doesn't yet have any extension actions, then you will also
2033 need to edit build-aux/extract-ofp-actions.
2039 bugs@openvswitch.org
2040 http://openvswitch.org/
2042 [PORTING.md]:PORTING.md
2043 [WHY-OVS.md]:WHY-OVS.md
2044 [INSTALL.md]:INSTALL.md
2045 [OPENFLOW-1.1+.md]:OPENFLOW-1.1+.md
2046 [INSTALL.DPDK.md]:INSTALL.DPDK.md
projects / ovs.git / blob