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 3.10 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
160 | 2.6.x | 3.10 to 4.3
162 Open vSwitch userspace should also work with the Linux kernel module
163 built into Linux 3.3 and later.
165 Open vSwitch userspace is not sensitive to the Linux kernel version.
166 It should build against almost any kernel, certainly against 2.6.32
169 ### Q: Are all features available with all datapaths?
171 A: Open vSwitch supports different datapaths on different platforms. Each
172 datapath has a different feature set: the following tables try to summarize
177 * *Linux upstream*: The datapath implemented by the kernel module shipped
178 with Linux upstream. Since features have been gradually
179 introduced into the kernel, the table mentions the first
180 Linux release whose OVS module supports the feature.
182 * *Linux OVS tree*: The datapath implemented by the Linux kernel module
183 distributed with the OVS source tree.
185 * *Userspace*: Also known as DPDK, dpif-netdev or dummy datapath. It is the
186 only datapath that works on NetBSD, FreeBSD and Mac OSX.
188 * *Hyper-V*: Also known as the Windows datapath.
190 The following table lists the datapath supported features from
191 an Open vSwitch user's perspective.
193 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
194 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
195 NAT | 4.6 | YES | NO | NO |
196 Connection tracking | 4.3 | YES | NO | PARTIAL |
197 Tunnel - LISP | NO | YES | NO | NO |
198 Tunnel - STT | NO | YES | NO | YES |
199 Tunnel - GRE | 3.11 | YES | YES | YES |
200 Tunnel - VXLAN | 3.12 | YES | YES | YES |
201 Tunnel - Geneve | 3.18 | YES | YES | YES |
202 Tunnel - GRE-IPv6 | NO | NO | YES | NO |
203 Tunnel - VXLAN-IPv6 | 4.3 | YES | YES | NO |
204 Tunnel - Geneve-IPv6 | 4.4 | YES | YES | NO |
205 QoS - Policing | YES | YES | NO | NO |
206 QoS - Shaping | YES | YES | NO | NO |
207 sFlow | YES | YES | YES | NO |
208 IPFIX | 3.10 | YES | YES | NO |
209 Set action | YES | YES | YES | PARTIAL |
210 NIC Bonding | YES | YES | YES | NO |
211 Multiple VTEPs | YES | YES | YES | NO |
214 * Only a limited set of flow fields is modifiable via the set action by the
216 * The Hyper-V datapath only supports one physical NIC per datapath. This is
217 why bonding is not supported.
218 * The Hyper-V datapath can have at most one IP address configured as a
221 The following table lists features that do not *directly* impact an
222 Open vSwitch user, e.g. because their absence can be hidden by the ofproto
223 layer (usually this comes with a performance penalty).
225 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
226 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
227 SCTP flows | 3.12 | YES | YES | YES |
228 MPLS | 3.19 | YES | YES | YES |
229 UFID | 4.0 | YES | YES | NO |
230 Megaflows | 3.12 | YES | YES | NO |
231 Masked set action | 4.0 | YES | YES | NO |
232 Recirculation | 3.19 | YES | YES | YES |
233 TCP flags matching | 3.13 | YES | YES | NO |
234 Validate flow actions | YES | YES | N/A | NO |
235 Multiple datapaths | YES | YES | YES | NO |
236 Tunnel TSO - STT | N/A | YES | NO | YES |
238 ### Q: What DPDK version does each Open vSwitch release work with?
240 A: The following table lists the DPDK version against which the
241 given versions of Open vSwitch will successfully build.
243 | Open vSwitch | DPDK
244 |:------------:|:-----:
251 ### Q: I get an error like this when I configure Open vSwitch:
253 configure: error: Linux kernel in <dir> is version <x>, but
254 version newer than <y> is not supported (please refer to the
259 A: You have the following options:
261 - Use the Linux kernel module supplied with the kernel that you are
262 using. (See also the following FAQ.)
264 - If there is a newer released version of Open vSwitch, consider
265 building that one, because it may support the kernel that you are
266 building against. (To find out, consult the table in the
269 - The Open vSwitch "master" branch may support the kernel that you
270 are using, so consider building the kernel module from "master".
272 All versions of Open vSwitch userspace are compatible with all
273 versions of the Open vSwitch kernel module, so you do not have to
274 use the kernel module from one source along with the userspace
275 programs from the same source.
277 ### Q: What features are not available in the Open vSwitch kernel datapath that ships as part of the upstream Linux kernel?
279 A: The kernel module in upstream Linux does not include support for
280 LISP. Work is in progress to add support for LISP to the upstream
281 Linux version of the Open vSwitch kernel module. For now, if you
282 need this feature, use the kernel module from the Open vSwitch
283 distribution instead of the upstream Linux kernel module.
285 Certain features require kernel support to function or to have
286 reasonable performance. If the ovs-vswitchd log file indicates that
287 a feature is not supported, consider upgrading to a newer upstream
288 Linux release or using the kernel module paired with the userspace
291 ### Q: Why do tunnels not work when using a kernel module other than the one packaged with Open vSwitch?
293 A: Support for tunnels was added to the upstream Linux kernel module
294 after the rest of Open vSwitch. As a result, some kernels may contain
295 support for Open vSwitch but not tunnels. The minimum kernel version
296 that supports each tunnel protocol is:
298 | Protocol | Linux Kernel
299 |:--------:|:-------------:
303 | LISP | <not upstream>
304 | STT | <not upstream>
306 If you are using a version of the kernel that is older than the one
307 listed above, it is still possible to use that tunnel protocol. However,
308 you must compile and install the kernel module included with the Open
309 vSwitch distribution rather than the one on your machine. If problems
310 persist after doing this, check to make sure that the module that is
311 loaded is the one you expect.
313 ### Q: Why are UDP tunnel checksums not computed for VXLAN or Geneve?
315 A: Generating outer UDP checksums requires kernel support that was not
316 part of the initial implementation of these protocols. If using the
317 upstream Linux Open vSwitch module, you must use kernel 4.0 or
318 newer. The out-of-tree modules from Open vSwitch release 2.4 and later
319 support UDP checksums.
321 ### Q: What features are not available when using the userspace datapath?
323 A: Tunnel virtual ports are not supported, as described in the
324 previous answer. It is also not possible to use queue-related
325 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
326 may not be transmitted.
328 ### Q: Should userspace or kernel be upgraded first to minimize downtime?
330 In general, the Open vSwitch userspace should be used with the
331 kernel version included in the same release or with the version
332 from upstream Linux. However, when upgrading between two releases
333 of Open vSwitch it is best to migrate userspace first to reduce
334 the possibility of incompatibilities.
336 ### Q: What happened to the bridge compatibility feature?
338 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
339 When it was enabled, Open vSwitch imitated the interface of the
340 Linux kernel "bridge" module. This allowed users to drop Open
341 vSwitch into environments designed to use the Linux kernel bridge
342 module without adapting the environment to use Open vSwitch.
344 Open vSwitch 1.10 and later do not support bridge compatibility.
345 The feature was dropped because version 1.10 adopted a new internal
346 architecture that made bridge compatibility difficult to maintain.
347 Now that many environments use OVS directly, it would be rarely
350 To use bridge compatibility, install OVS 1.9 or earlier, including
351 the accompanying kernel modules (both the main and bridge
352 compatibility modules), following the instructions that come with
353 the release. Be sure to start the ovs-brcompatd daemon.
359 ### Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation keeps talking about bridges. What's a bridge?
361 A: In networking, the terms "bridge" and "switch" are synonyms. Open
362 vSwitch implements an Ethernet switch, which means that it is also
365 ### Q: What's a VLAN?
367 A: See the "VLAN" section below.
373 ### Q: How do I configure a port as an access port?
375 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
376 the following commands configure br0 with eth0 as a trunk port (the
377 default) and tap0 as an access port for VLAN 9:
380 ovs-vsctl add-port br0 eth0
381 ovs-vsctl add-port br0 tap0 tag=9
383 If you want to configure an already added port as an access port,
384 use "ovs-vsctl set", e.g.:
386 ovs-vsctl set port tap0 tag=9
388 ### Q: How do I configure a port as a SPAN port, that is, enable mirroring of all traffic to that port?
390 A: The following commands configure br0 with eth0 and tap0 as trunk
391 ports. All traffic coming in or going out on eth0 or tap0 is also
392 mirrored to tap1; any traffic arriving on tap1 is dropped:
395 ovs-vsctl add-port br0 eth0
396 ovs-vsctl add-port br0 tap0
397 ovs-vsctl add-port br0 tap1 \
398 -- --id=@p get port tap1 \
399 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
400 -- set bridge br0 mirrors=@m
402 To later disable mirroring, run:
404 ovs-vsctl clear bridge br0 mirrors
406 ### Q: Does Open vSwitch support configuring a port in promiscuous mode?
408 A: Yes. How you configure it depends on what you mean by "promiscuous
411 - Conventionally, "promiscuous mode" is a feature of a network
412 interface card. Ordinarily, a NIC passes to the CPU only the
413 packets actually destined to its host machine. It discards
414 the rest to avoid wasting memory and CPU cycles. When
415 promiscuous mode is enabled, however, it passes every packet
416 to the CPU. On an old-style shared-media or hub-based
417 network, this allows the host to spy on all packets on the
418 network. But in the switched networks that are almost
419 everywhere these days, promiscuous mode doesn't have much
420 effect, because few packets not destined to a host are
421 delivered to the host's NIC.
423 This form of promiscuous mode is configured in the guest OS of
424 the VMs on your bridge, e.g. with "ifconfig".
426 - The VMware vSwitch uses a different definition of "promiscuous
427 mode". When you configure promiscuous mode on a VMware vNIC,
428 the vSwitch sends a copy of every packet received by the
429 vSwitch to that vNIC. That has a much bigger effect than just
430 enabling promiscuous mode in a guest OS. Rather than getting
431 a few stray packets for which the switch does not yet know the
432 correct destination, the vNIC gets every packet. The effect
433 is similar to replacing the vSwitch by a virtual hub.
435 This "promiscuous mode" is what switches normally call "port
436 mirroring" or "SPAN". For information on how to configure
437 SPAN, see "How do I configure a port as a SPAN port, that is,
438 enable mirroring of all traffic to that port?"
440 ### Q: How do I configure a DPDK port as an access port?
442 A: Firstly, you must have a DPDK-enabled version of Open vSwitch.
444 If your version is DPDK-enabled it will support the other-config:dpdk-init
445 configuration in the database and will display lines with "EAL:..."
446 during startup when other_config:dpdk-init is set to 'true'.
448 Secondly, when adding a DPDK port, unlike a system port, the
449 type for the interface must be specified. For example;
452 ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
454 Finally, it is required that DPDK port names begin with 'dpdk'.
456 See [INSTALL.DPDK.md] for more information on enabling and using DPDK with
459 ### Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all traffic to that VLAN?
461 A: The following commands configure br0 with eth0 as a trunk port and
462 tap0 as an access port for VLAN 10. All traffic coming in or going
463 out on tap0, as well as traffic coming in or going out on eth0 in
464 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
465 VLAN 10, in cases where one is present, is dropped as part of
469 ovs-vsctl add-port br0 eth0
470 ovs-vsctl add-port br0 tap0 tag=10
472 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
474 -- set bridge br0 mirrors=@m
476 To later disable mirroring, run:
478 ovs-vsctl clear bridge br0 mirrors
480 Mirroring to a VLAN can disrupt a network that contains unmanaged
481 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
482 GRE tunnel has fewer caveats than mirroring to a VLAN and should
483 generally be preferred.
485 ### Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
487 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
488 of the specified output-vlan. This loss of information may make
489 the mirrored traffic too hard to interpret.
491 To mirror multiple VLANs, use the commands above, but specify a
492 comma-separated list of VLANs as the value for select-vlan. To
493 mirror every VLAN, use the commands above, but omit select-vlan and
496 When a packet arrives on a VLAN that is used as a mirror output
497 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
498 floods the packet across all the ports for which the mirror output
499 VLAN is configured. (If an OpenFlow controller is in use, then it
500 can override this behavior through the flow table.) If OVS is used
501 as an intermediate switch, rather than an edge switch, this ensures
502 that the RSPAN traffic is distributed through the network.
504 Mirroring to a VLAN can disrupt a network that contains unmanaged
505 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
506 GRE tunnel has fewer caveats than mirroring to a VLAN and should
507 generally be preferred.
509 ### Q: How do I configure mirroring of all traffic to a GRE tunnel?
511 A: The following commands configure br0 with eth0 and tap0 as trunk
512 ports. All traffic coming in or going out on eth0 or tap0 is also
513 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
514 traffic arriving on gre0 is dropped:
517 ovs-vsctl add-port br0 eth0
518 ovs-vsctl add-port br0 tap0
519 ovs-vsctl add-port br0 gre0 \
520 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
521 -- --id=@p get port gre0 \
522 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
523 -- set bridge br0 mirrors=@m
525 To later disable mirroring and destroy the GRE tunnel:
527 ovs-vsctl clear bridge br0 mirrors
528 ovs-vsctl del-port br0 gre0
530 ### Q: Does Open vSwitch support ERSPAN?
532 A: No. ERSPAN is an undocumented proprietary protocol. As an
533 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
536 ### Q: How do I connect two bridges?
538 A: First, why do you want to do this? Two connected bridges are not
539 much different from a single bridge, so you might as well just have
540 a single bridge with all your ports on it.
542 If you still want to connect two bridges, you can use a pair of
543 patch ports. The following example creates bridges br0 and br1,
544 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
545 and br1 with a pair of patch ports.
548 ovs-vsctl add-port br0 eth0
549 ovs-vsctl add-port br0 tap0
551 ovs-vsctl add-port br1 tap1
553 -- add-port br0 patch0 \
554 -- set interface patch0 type=patch options:peer=patch1 \
555 -- add-port br1 patch1 \
556 -- set interface patch1 type=patch options:peer=patch0
558 Bridges connected with patch ports are much like a single bridge.
559 For instance, if the example above also added eth1 to br1, and both
560 eth0 and eth1 happened to be connected to the same next-hop switch,
561 then you could loop your network just as you would if you added
562 eth0 and eth1 to the same bridge (see the "Configuration Problems"
563 section below for more information).
565 If you are using Open vSwitch 1.9 or an earlier version, then you
566 need to be using the kernel module bundled with Open vSwitch rather
567 than the one that is integrated into Linux 3.3 and later, because
568 Open vSwitch 1.9 and earlier versions need kernel support for patch
569 ports. This also means that in Open vSwitch 1.9 and earlier, patch
570 ports will not work with the userspace datapath, only with the
573 ### Q: How do I configure a bridge without an OpenFlow local port? (Local port in the sense of OFPP_LOCAL)
575 A: Open vSwitch does not support such a configuration.
576 Bridges always have their local ports.
579 Implementation Details
580 ----------------------
582 ### Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
584 A: Open vSwitch uses different kinds of flows for different purposes:
586 - OpenFlow flows are the most important kind of flow. OpenFlow
587 controllers use these flows to define a switch's policy.
588 OpenFlow flows support wildcards, priorities, and multiple
591 When in-band control is in use, Open vSwitch sets up a few
592 "hidden" flows, with priority higher than a controller or the
593 user can configure, that are not visible via OpenFlow. (See
594 the "Controller" section of the FAQ for more information
597 - The Open vSwitch software switch implementation uses a second
598 kind of flow internally. These flows, called "datapath" or
599 "kernel" flows, do not support priorities and comprise only a
600 single table, which makes them suitable for caching. (Like
601 OpenFlow flows, datapath flows do support wildcarding, in Open
602 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
603 also support different actions and number ports differently.
605 Datapath flows are an implementation detail that is subject to
606 change in future versions of Open vSwitch. Even with the
607 current version of Open vSwitch, hardware switch
608 implementations do not necessarily use this architecture.
610 Users and controllers directly control only the OpenFlow flow
611 table. Open vSwitch manages the datapath flow table itself, so
612 users should not normally be concerned with it.
614 ### Q: Why are there so many different ways to dump flows?
616 A: Open vSwitch has two kinds of flows (see the previous question), so
617 it has commands with different purposes for dumping each kind of
620 - `ovs-ofctl dump-flows <br>` dumps OpenFlow flows, excluding
621 hidden flows. This is the most commonly useful form of flow
622 dump. (Unlike the other commands, this should work with any
623 OpenFlow switch, not just Open vSwitch.)
625 - `ovs-appctl bridge/dump-flows <br>` dumps OpenFlow flows,
626 including hidden flows. This is occasionally useful for
627 troubleshooting suspected issues with in-band control.
629 - `ovs-dpctl dump-flows [dp]` dumps the datapath flow table
630 entries for a Linux kernel-based datapath. In Open vSwitch
631 1.10 and later, ovs-vswitchd merges multiple switches into a
632 single datapath, so it will show all the flows on all your
633 kernel-based switches. This command can occasionally be
634 useful for debugging.
636 - `ovs-appctl dpif/dump-flows <br>`, new in Open vSwitch 1.10,
637 dumps datapath flows for only the specified bridge, regardless
640 ### Q: How does multicast snooping works with VLANs?
642 A: Open vSwitch maintains snooping tables for each VLAN.
644 ### Q: Can OVS populate the kernel flow table in advance instead of in reaction to packets?
646 A: No. There are several reasons:
648 - Kernel flows are not as sophisticated as OpenFlow flows, which
649 means that some OpenFlow policies could require a large number of
650 kernel flows. The "conjunctive match" feature is an extreme
651 example: the number of kernel flows it requires is the product of
652 the number of flows in each dimension.
654 - With multiple OpenFlow flow tables and simple sets of actions, the
655 number of kernel flows required can be as large as the product of
656 the number of flows in each dimension. With more sophisticated
657 actions, the number of kernel flows could be even larger.
659 - Open vSwitch is designed so that any version of OVS userspace
660 interoperates with any version of the OVS kernel module. This
661 forward and backward compatibility requires that userspace observe
662 how the kernel module parses received packets. This is only
663 possible in a straightforward way when userspace adds kernel flows
664 in reaction to received packets.
666 For more relevant information on the architecture of Open vSwitch,
667 please read "The Design and Implementation of Open vSwitch",
668 published in USENIX NSDI 2015.
674 ### Q: I just upgraded and I see a performance drop. Why?
676 A: The OVS kernel datapath may have been updated to a newer version than
677 the OVS userspace components. Sometimes new versions of OVS kernel
678 module add functionality that is backwards compatible with older
679 userspace components but may cause a drop in performance with them.
680 Especially, if a kernel module from OVS 2.1 or newer is paired with
681 OVS userspace 1.10 or older, there will be a performance drop for
684 Updating the OVS userspace components to the latest released
685 version should fix the performance degradation.
687 To get the best possible performance and functionality, it is
688 recommended to pair the same versions of the kernel module and OVS
692 Configuration Problems
693 ----------------------
695 ### Q: I created a bridge and added my Ethernet port to it, using commands
699 ovs-vsctl add-port br0 eth0
701 and as soon as I ran the "add-port" command I lost all connectivity
704 A: A physical Ethernet device that is part of an Open vSwitch bridge
705 should not have an IP address. If one does, then that IP address
706 will not be fully functional.
708 You can restore functionality by moving the IP address to an Open
709 vSwitch "internal" device, such as the network device named after
710 the bridge itself. For example, assuming that eth0's IP address is
711 192.168.128.5, you could run the commands below to fix up the
714 ifconfig eth0 0.0.0.0
715 ifconfig br0 192.168.128.5
717 (If your only connection to the machine running OVS is through the
718 IP address in question, then you would want to run all of these
719 commands on a single command line, or put them into a script.) If
720 there were any additional routes assigned to eth0, then you would
721 also want to use commands to adjust these routes to go through br0.
723 If you use DHCP to obtain an IP address, then you should kill the
724 DHCP client that was listening on the physical Ethernet interface
725 (e.g. eth0) and start one listening on the internal interface
726 (e.g. br0). You might still need to manually clear the IP address
727 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
729 There is no compelling reason why Open vSwitch must work this way.
730 However, this is the way that the Linux kernel bridge module has
731 always worked, so it's a model that those accustomed to Linux
732 bridging are already used to. Also, the model that most people
733 expect is not implementable without kernel changes on all the
734 versions of Linux that Open vSwitch supports.
736 By the way, this issue is not specific to physical Ethernet
737 devices. It applies to all network devices except Open vSwitch
740 ### Q: I created a bridge and added a couple of Ethernet ports to it,
741 ### using commands like these:
744 ovs-vsctl add-port br0 eth0
745 ovs-vsctl add-port br0 eth1
747 and now my network seems to have melted: connectivity is unreliable
748 (even connectivity that doesn't go through Open vSwitch), all the
749 LEDs on my physical switches are blinking, wireshark shows
750 duplicated packets, and CPU usage is very high.
752 A: More than likely, you've looped your network. Probably, eth0 and
753 eth1 are connected to the same physical Ethernet switch. This
754 yields a scenario where OVS receives a broadcast packet on eth0 and
755 sends it out on eth1, then the physical switch connected to eth1
756 sends the packet back on eth0, and so on forever. More complicated
757 scenarios, involving a loop through multiple switches, are possible
760 The solution depends on what you are trying to do:
762 - If you added eth0 and eth1 to get higher bandwidth or higher
763 reliability between OVS and your physical Ethernet switch,
764 use a bond. The following commands create br0 and then add
765 eth0 and eth1 as a bond:
768 ovs-vsctl add-bond br0 bond0 eth0 eth1
770 Bonds have tons of configuration options. Please read the
771 documentation on the Port table in ovs-vswitchd.conf.db(5)
774 Configuration for DPDK-enabled interfaces is slightly less
775 straightforward: see [INSTALL.DPDK.md].
777 - Perhaps you don't actually need eth0 and eth1 to be on the
778 same bridge. For example, if you simply want to be able to
779 connect each of them to virtual machines, then you can put
780 each of them on a bridge of its own:
783 ovs-vsctl add-port br0 eth0
786 ovs-vsctl add-port br1 eth1
788 and then connect VMs to br0 and br1. (A potential
789 disadvantage is that traffic cannot directly pass between br0
790 and br1. Instead, it will go out eth0 and come back in eth1,
793 - If you have a redundant or complex network topology and you
794 want to prevent loops, turn on spanning tree protocol (STP).
795 The following commands create br0, enable STP, and add eth0
796 and eth1 to the bridge. The order is important because you
797 don't want have to have a loop in your network even
801 ovs-vsctl set bridge br0 stp_enable=true
802 ovs-vsctl add-port br0 eth0
803 ovs-vsctl add-port br0 eth1
805 The Open vSwitch implementation of STP is not well tested.
806 Please report any bugs you observe, but if you'd rather avoid
807 acting as a beta tester then another option might be your
810 ### Q: I can't seem to use Open vSwitch in a wireless network.
812 A: Wireless base stations generally only allow packets with the source
813 MAC address of NIC that completed the initial handshake.
814 Therefore, without MAC rewriting, only a single device can
815 communicate over a single wireless link.
817 This isn't specific to Open vSwitch, it's enforced by the access
818 point, so the same problems will show up with the Linux bridge or
819 any other way to do bridging.
821 ### Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
823 A: PPP most commonly carries IP packets, but Open vSwitch works only
824 with Ethernet frames. The correct way to interface PPP to an
825 Ethernet network is usually to use routing instead of switching.
827 ### Q: Is there any documentation on the database tables and fields?
829 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
831 ### Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
832 I only see a datapath called "ovs-system". How can I see datapath
833 information about a particular bridge?
835 A: In version 1.9.0, OVS switched to using a single datapath that is
836 shared by all bridges of that type. The "ovs-appctl dpif/*"
837 commands provide similar functionality that is scoped by the bridge.
839 ### Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
840 see the port in the datapath?
842 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
843 cannot be loaded at the same time. It is likely that on your system the
844 Linux GRE module is already loaded and blocking OVS (to confirm, check
845 dmesg for errors regarding GRE registration). To fix this, unload all
846 GRE modules that appear in lsmod as well as the OVS kernel module. You
847 can then reload the OVS module following the directions in
848 [INSTALL.md], which will ensure that dependencies are satisfied.
850 ### Q: Open vSwitch does not seem to obey my packet filter rules.
852 A: It depends on mechanisms and configurations you want to use.
854 You cannot usefully use typical packet filters, like iptables, on
855 physical Ethernet ports that you add to an Open vSwitch bridge.
856 This is because Open vSwitch captures packets from the interface at
857 a layer lower below where typical packet-filter implementations
858 install their hooks. (This actually applies to any interface of
859 type "system" that you might add to an Open vSwitch bridge.)
861 You can usefully use typical packet filters on Open vSwitch
862 internal ports as they are mostly ordinary interfaces from the point
863 of view of packet filters.
865 For example, suppose you create a bridge br0 and add Ethernet port
866 eth0 to it. Then you can usefully add iptables rules to affect the
867 internal interface br0, but not the physical interface eth0. (br0
868 is also where you would add an IP address, as discussed elsewhere
871 For simple filtering rules, it might be possible to achieve similar
872 results by installing appropriate OpenFlow flows instead.
874 If the use of a particular packet filter setup is essential, Open
875 vSwitch might not be the best choice for you. On Linux, you might
876 want to consider using the Linux Bridge. (This is the only choice if
877 you want to use ebtables rules.) On NetBSD, you might want to
878 consider using the bridge(4) with BRIDGE_IPF option.
880 ### Q: It seems that Open vSwitch does nothing when I removed a port and
881 then immediately put it back. For example, consider that p1 is
882 a port of type=internal:
884 ovs-vsctl del-port br0 p1 -- \
886 set interface p1 type=internal
888 A: It's an expected behaviour.
890 If del-port and add-port happen in a single OVSDB transaction as
891 your example, Open vSwitch always "skips" the intermediate steps.
892 Even if they are done in multiple transactions, it's still allowed
893 for Open vSwitch to skip the intermediate steps and just implement
894 the overall effect. In both cases, your example would be turned
897 If you want to make Open vSwitch actually destroy and then re-create
898 the port for some side effects like resetting kernel setting for the
899 corresponding interface, you need to separate operations into multiple
900 OVSDB transactions and ensure that at least the first one does not have
901 --no-wait. In the following example, the first ovs-vsctl will block
902 until Open vSwitch reloads the new configuration and removes the port:
904 ovs-vsctl del-port br0 p1
905 ovs-vsctl add-port br0 p1 -- \
906 set interface p1 type=internal
908 ### Q: I want to add thousands of ports to an Open vSwitch bridge, but
909 it takes too long (minutes or hours) to do it with ovs-vsctl. How
912 A: If you add them one at a time with ovs-vsctl, it can take a long
913 time to add thousands of ports to an Open vSwitch bridge. This is
914 because every invocation of ovs-vsctl first reads the current
915 configuration from OVSDB. As the number of ports grows, this
916 starts to take an appreciable amount of time, and when it is
917 repeated thousands of times the total time becomes significant.
919 The solution is to add the ports in one invocation of ovs-vsctl (or
920 a small number of them). For example, using bash:
923 cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done
926 takes seconds, not minutes or hours, in the OVS sandbox environment.
928 ### Q: I created a bridge named br0. My bridge shows up in "ovs-vsctl
929 show", but "ovs-ofctl show br0" just prints "br0 is not a bridge
932 A: Open vSwitch wasn't able to create the bridge. Check the
933 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
934 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log).
936 In general, the Open vSwitch database reflects the desired
937 configuration state. ovs-vswitchd monitors the database and, when
938 it changes, reconfigures the system to reflect the new desired
939 state. This normally happens very quickly. Thus, a discrepancy
940 between the database and the actual state indicates that
941 ovs-vswitchd could not implement the configuration, and so one
942 should check the log to find out why. (Another possible cause is
943 that ovs-vswitchd is not running. This will make "ovs-vsctl"
944 commands hang, if they change the configuration, unless one
945 specifies "--no-wait".)
947 ### Q: I have a bridge br0. I added a new port vif1.0, and it shows
948 up in "ovs-vsctl show", but "ovs-vsctl list port" says that it has
949 OpenFlow port ("ofport") -1, and "ovs-ofctl show br0" doesn't show
952 A: Open vSwitch wasn't able to create the port. Check the
953 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
954 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log). Please
955 see the previous question for more information.
957 You may want to upgrade to Open vSwitch 2.3 (or later), in which
958 ovs-vsctl will immediately report when there is an issue creating a
961 ### Q: I created a tap device tap0, configured an IP address on it, and
962 added it to a bridge, like this:
965 ifconfig tap0 192.168.0.123
967 ovs-vsctl add-port br0 tap0
969 I expected that I could then use this IP address to contact other
970 hosts on the network, but it doesn't work. Why not?
972 A: The short answer is that this is a misuse of a "tap" device. Use
973 an "internal" device implemented by Open vSwitch, which works
974 differently and is designed for this use. To solve this problem
975 with an internal device, instead run:
978 ovs-vsctl add-port br0 int0 -- set Interface int0 type=internal
979 ifconfig int0 192.168.0.123
981 Even more simply, you can take advantage of the internal port that
982 every bridge has under the name of the bridge:
985 ifconfig br0 192.168.0.123
987 In more detail, a "tap" device is an interface between the Linux
988 (or *BSD) network stack and a user program that opens it as a
989 socket. When the "tap" device transmits a packet, it appears in
990 the socket opened by the userspace program. Conversely, when the
991 userspace program writes to the "tap" socket, the kernel TCP/IP
992 stack processes the packet as if it had been received by the "tap"
995 Consider the configuration above. Given this configuration, if you
996 "ping" an IP address in the 192.168.0.x subnet, the Linux kernel
997 routing stack will transmit an ARP on the tap0 device. Open
998 vSwitch userspace treats "tap" devices just like any other network
999 device; that is, it doesn't open them as "tap" sockets. That means
1000 that the ARP packet will simply get dropped.
1002 You might wonder why the Open vSwitch kernel module doesn't
1003 intercept the ARP packet and bridge it. After all, Open vSwitch
1004 intercepts packets on other devices. The answer is that Open
1005 vSwitch only intercepts *received* packets, but this is a packet
1006 being transmitted. The same thing happens for all other types of
1007 network devices, except for Open vSwitch "internal" ports. If you,
1008 for example, add a physical Ethernet port to an OVS bridge,
1009 configure an IP address on a physical Ethernet port, and then issue
1010 a "ping" to an address in that subnet, the same thing happens: an
1011 ARP gets transmitted on the physical Ethernet port and Open vSwitch
1012 never sees it. (You should not do that, as documented at the
1013 beginning of this section.)
1015 It can make sense to add a "tap" device to an Open vSwitch bridge,
1016 if some userspace program (other than Open vSwitch) has opened the
1017 tap socket. This is the case, for example, if the "tap" device was
1018 created by KVM (or QEMU) to simulate a virtual NIC. In such a
1019 case, when OVS bridges a packet to the "tap" device, the kernel
1020 forwards that packet to KVM in userspace, which passes it along to
1021 the VM, and in the other direction, when the VM sends a packet, KVM
1022 writes it to the "tap" socket, which causes OVS to receive it and
1023 bridge it to the other OVS ports. Please note that in such a case
1024 no IP address is configured on the "tap" device (there is normally
1025 an IP address configured in the virtual NIC inside the VM, but this
1026 is not visible to the host Linux kernel or to Open vSwitch).
1028 There is one special case in which Open vSwitch does directly read
1029 and write "tap" sockets. This is an implementation detail of the
1030 Open vSwitch userspace switch, which implements its "internal"
1031 ports as Linux (or *BSD) "tap" sockets. In such a userspace
1032 switch, OVS receives packets sent on the "tap" device used to
1033 implement an "internal" port by reading the associated "tap"
1034 socket, and bridges them to the rest of the switch. In the other
1035 direction, OVS transmits packets bridged to the "internal" port by
1036 writing them to the "tap" socket, causing them to be processed by
1037 the kernel TCP/IP stack as if they had been received on the "tap"
1038 device. Users should not need to be concerned with this
1039 implementation detail.
1041 Open vSwitch has a network device type called "tap". This is
1042 intended only for implementing "internal" ports in the OVS
1043 userspace switch and should not be used otherwise. In particular,
1044 users should not configure KVM "tap" devices as type "tap" (use
1045 type "system", the default, instead).
1048 Quality of Service (QoS)
1049 ------------------------
1051 ### Q: Does OVS support Quality of Service (QoS)?
1053 A: Yes. For traffic that egresses from a switch, OVS supports traffic
1054 shaping; for traffic that ingresses into a switch, OVS support
1055 policing. Policing is a simple form of quality-of-service that
1056 simply drops packets received in excess of the configured rate. Due
1057 to its simplicity, policing is usually less accurate and less
1058 effective than egress traffic shaping, which queues packets.
1060 Keep in mind that ingress and egress are from the perspective of the
1061 switch. That means that egress shaping limits the rate at which
1062 traffic is allowed to transmit from a physical interface, but not the
1063 rate at which traffic will be received on a virtual machine's VIF.
1064 For ingress policing, the behavior is the opposite.
1066 ### Q: How do I configure egress traffic shaping?
1068 A: Suppose that you want to set up bridge br0 connected to physical
1069 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
1070 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
1071 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
1072 could configure the bridge this way:
1076 add-port br0 eth0 -- \
1077 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
1078 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
1079 set port eth0 qos=@newqos -- \
1080 --id=@newqos create qos type=linux-htb \
1081 other-config:max-rate=1000000000 \
1082 queues:123=@vif10queue \
1083 queues:234=@vif20queue -- \
1084 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
1085 --id=@vif20queue create queue other-config:max-rate=20000000
1087 At this point, bridge br0 is configured with the ports and eth0 is
1088 configured with the queues that you need for QoS, but nothing is
1089 actually directing packets from vif1.0 or vif2.0 to the queues that
1090 we have set up for them. That means that all of the packets to
1091 eth0 are going to the "default queue", which is not what we want.
1093 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
1094 queues reserved for them:
1096 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
1097 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
1099 Each of the above flows matches on the input port, sets up the
1100 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
1101 executes the "normal" action, which performs the same switching
1102 that Open vSwitch would have done without any OpenFlow flows being
1103 present. (We know that vif1.0 and vif2.0 have OpenFlow port
1104 numbers 5 and 6, respectively, because we set their ofport_request
1105 columns above. If we had not done that, then we would have needed
1106 to find out their port numbers before setting up these flows.)
1108 Now traffic going from vif1.0 or vif2.0 to eth0 should be
1111 By the way, if you delete the bridge created by the above commands,
1114 ovs-vsctl del-br br0
1116 then that will leave one unreferenced QoS record and two
1117 unreferenced Queue records in the Open vSwich database. One way to
1118 clear them out, assuming you don't have other QoS or Queue records
1119 that you want to keep, is:
1121 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
1123 If you do want to keep some QoS or Queue records, or the Open
1124 vSwitch you are using is older than version 1.8 (which added the
1125 --all option), then you will have to destroy QoS and Queue records
1128 ### Q: How do I configure ingress policing?
1130 A: A policing policy can be configured on an interface to drop packets
1131 that arrive at a higher rate than the configured value. For example,
1132 the following commands will rate-limit traffic that vif1.0 may
1135 ovs-vsctl set interface vif1.0 ingress_policing_rate=10000
1136 ovs-vsctl set interface vif1.0 ingress_policing_burst=8000
1138 Traffic policing can interact poorly with some network protocols and
1139 can have surprising results. The "Ingress Policing" section of
1140 ovs-vswitchd.conf.db(5) discusses the issues in greater detail.
1142 ### Q: I configured Quality of Service (QoS) in my OpenFlow network by
1143 adding records to the QoS and Queue table, but the results aren't
1146 A: Did you install OpenFlow flows that use your queues? This is the
1147 primary way to tell Open vSwitch which queues you want to use. If
1148 you don't do this, then the default queue will be used, which will
1149 probably not have the effect you want.
1151 Refer to the previous question for an example.
1153 ### Q: I'd like to take advantage of some QoS feature that Open vSwitch
1154 doesn't yet support. How do I do that?
1156 A: Open vSwitch does not implement QoS itself. Instead, it can
1157 configure some, but not all, of the QoS features built into the
1158 Linux kernel. If you need some QoS feature that OVS cannot
1159 configure itself, then the first step is to figure out whether
1160 Linux QoS supports that feature. If it does, then you can submit a
1161 patch to support Open vSwitch configuration for that feature, or
1162 you can use "tc" directly to configure the feature in Linux. (If
1163 Linux QoS doesn't support the feature you want, then first you have
1164 to add that support to Linux.)
1166 ### Q: I configured QoS, correctly, but my measurements show that it isn't
1167 working as well as I expect.
1169 A: With the Linux kernel, the Open vSwitch implementation of QoS has
1172 - Open vSwitch configures a subset of Linux kernel QoS
1173 features, according to what is in OVSDB. It is possible that
1174 this code has bugs. If you believe that this is so, then you
1175 can configure the Linux traffic control (QoS) stack directly
1176 with the "tc" program. If you get better results that way,
1177 you can send a detailed bug report to bugs@openvswitch.org.
1179 It is certain that Open vSwitch cannot configure every Linux
1180 kernel QoS feature. If you need some feature that OVS cannot
1181 configure, then you can also use "tc" directly (or add that
1184 - The Open vSwitch implementation of OpenFlow allows flows to
1185 be directed to particular queues. This is pretty simple and
1186 unlikely to have serious bugs at this point.
1188 However, most problems with QoS on Linux are not bugs in Open
1189 vSwitch at all. They tend to be either configuration errors
1190 (please see the earlier questions in this section) or issues with
1191 the traffic control (QoS) stack in Linux. The Open vSwitch
1192 developers are not experts on Linux traffic control. We suggest
1193 that, if you believe you are encountering a problem with Linux
1194 traffic control, that you consult the tc manpages (e.g. tc(8),
1195 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
1196 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
1198 ### Q: Does Open vSwitch support OpenFlow meters?
1200 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
1201 OpenFlow meters. There is no implementation of meters in the Open
1202 vSwitch software switch (neither the kernel-based nor userspace
1209 ### Q: What's a VLAN?
1211 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
1212 partition a single switch into multiple switches. Suppose, for
1213 example, that you have two groups of machines, group A and group B.
1214 You want the machines in group A to be able to talk to each other,
1215 and you want the machine in group B to be able to talk to each
1216 other, but you don't want the machines in group A to be able to
1217 talk to the machines in group B. You can do this with two
1218 switches, by plugging the machines in group A into one switch and
1219 the machines in group B into the other switch.
1221 If you only have one switch, then you can use VLANs to do the same
1222 thing, by configuring the ports for machines in group A as VLAN
1223 "access ports" for one VLAN and the ports for group B as "access
1224 ports" for a different VLAN. The switch will only forward packets
1225 between ports that are assigned to the same VLAN, so this
1226 effectively subdivides your single switch into two independent
1227 switches, one for each group of machines.
1229 So far we haven't said anything about VLAN headers. With access
1230 ports, like we've described so far, no VLAN header is present in
1231 the Ethernet frame. This means that the machines (or switches)
1232 connected to access ports need not be aware that VLANs are
1233 involved, just like in the case where we use two different physical
1236 Now suppose that you have a whole bunch of switches in your
1237 network, instead of just one, and that some machines in group A are
1238 connected directly to both switches 1 and 2. To allow these
1239 machines to talk to each other, you could add an access port for
1240 group A's VLAN to switch 1 and another to switch 2, and then
1241 connect an Ethernet cable between those ports. That works fine,
1242 but it doesn't scale well as the number of switches and the number
1243 of VLANs increases, because you use up a lot of valuable switch
1244 ports just connecting together your VLANs.
1246 This is where VLAN headers come in. Instead of using one cable and
1247 two ports per VLAN to connect a pair of switches, we configure a
1248 port on each switch as a VLAN "trunk port". Packets sent and
1249 received on a trunk port carry a VLAN header that says what VLAN
1250 the packet belongs to, so that only two ports total are required to
1251 connect the switches, regardless of the number of VLANs in use.
1252 Normally, only switches (either physical or virtual) are connected
1253 to a trunk port, not individual hosts, because individual hosts
1254 don't expect to see a VLAN header in the traffic that they receive.
1256 None of the above discussion says anything about particular VLAN
1257 numbers. This is because VLAN numbers are completely arbitrary.
1258 One must only ensure that a given VLAN is numbered consistently
1259 throughout a network and that different VLANs are given different
1260 numbers. (That said, VLAN 0 is usually synonymous with a packet
1261 that has no VLAN header, and VLAN 4095 is reserved.)
1263 ### Q: VLANs don't work.
1265 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
1266 bugs. If you are having problems with VLANs that you suspect to be
1267 driver related, then you have several options:
1269 - Upgrade to Linux 3.3 or later.
1271 - Build and install a fixed version of the particular driver
1272 that is causing trouble, if one is available.
1274 - Use a NIC whose driver does not have VLAN problems.
1276 - Use "VLAN splinters", a feature in Open vSwitch 1.4 upto 2.5
1277 that works around bugs in kernel drivers. To enable VLAN
1278 splinters on interface eth0, use the command:
1280 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
1282 For VLAN splinters to be effective, Open vSwitch must know
1283 which VLANs are in use. See the "VLAN splinters" section in
1284 the Interface table in ovs-vswitchd.conf.db(5) for details on
1285 how Open vSwitch infers in-use VLANs.
1287 VLAN splinters increase memory use and reduce performance, so
1288 use them only if needed.
1290 - Apply the "vlan workaround" patch from the XenServer kernel
1291 patch queue, build Open vSwitch against this patched kernel,
1292 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
1293 workaround for each interface whose driver is buggy.
1295 (This is a nontrivial exercise, so this option is included
1296 only for completeness.)
1298 It is not always easy to tell whether a Linux kernel driver has
1299 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
1300 can help you test. See their manpages for details. Of the two
1301 utilities, ovs-test(8) is newer and more thorough, but
1302 ovs-vlan-test(8) may be easier to use.
1304 ### Q: VLANs still don't work. I've tested the driver so I know that it's OK.
1306 A: Do you have VLANs enabled on the physical switch that OVS is
1307 attached to? Make sure that the port is configured to trunk the
1308 VLAN or VLANs that you are using with OVS.
1310 ### Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
1311 and to its destination host, but OVS seems to drop incoming return
1314 A: It's possible that you have the VLAN configured on your physical
1315 switch as the "native" VLAN. In this mode, the switch treats
1316 incoming packets either tagged with the native VLAN or untagged as
1317 part of the native VLAN. It may also send outgoing packets in the
1318 native VLAN without a VLAN tag.
1320 If this is the case, you have two choices:
1322 - Change the physical switch port configuration to tag packets
1323 it forwards to OVS with the native VLAN instead of forwarding
1326 - Change the OVS configuration for the physical port to a
1327 native VLAN mode. For example, the following sets up a
1328 bridge with port eth0 in "native-tagged" mode in VLAN 9:
1330 ovs-vsctl add-br br0
1331 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
1333 In this situation, "native-untagged" mode will probably work
1334 equally well. Refer to the documentation for the Port table
1335 in ovs-vswitchd.conf.db(5) for more information.
1337 ### Q: I added a pair of VMs on different VLANs, like this:
1339 ovs-vsctl add-br br0
1340 ovs-vsctl add-port br0 eth0
1341 ovs-vsctl add-port br0 tap0 tag=9
1342 ovs-vsctl add-port br0 tap1 tag=10
1344 but the VMs can't access each other, the external network, or the
1347 A: It is to be expected that the VMs can't access each other. VLANs
1348 are a means to partition a network. When you configured tap0 and
1349 tap1 as access ports for different VLANs, you indicated that they
1350 should be isolated from each other.
1352 As for the external network and the Internet, it seems likely that
1353 the machines you are trying to access are not on VLAN 9 (or 10) and
1354 that the Internet is not available on VLAN 9 (or 10).
1356 ### Q: I added a pair of VMs on the same VLAN, like this:
1358 ovs-vsctl add-br br0
1359 ovs-vsctl add-port br0 eth0
1360 ovs-vsctl add-port br0 tap0 tag=9
1361 ovs-vsctl add-port br0 tap1 tag=9
1363 The VMs can access each other, but not the external network or the
1366 A: It seems likely that the machines you are trying to access in the
1367 external network are not on VLAN 9 and that the Internet is not
1368 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
1369 trunk VLAN on the upstream switch port to which eth0 is connected.
1371 ### Q: Can I configure an IP address on a VLAN?
1373 A: Yes. Use an "internal port" configured as an access port. For
1374 example, the following configures IP address 192.168.0.7 on VLAN 9.
1375 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
1376 they have an 802.1Q header with VLAN 9. Conversely, traffic
1377 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
1380 ovs-vsctl add-br br0
1381 ovs-vsctl add-port br0 eth0
1382 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1383 ifconfig vlan9 192.168.0.7
1385 See also the following question.
1387 ### Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
1390 ovs-vsctl add-br br0
1391 ovs-vsctl add-port br0 eth0
1392 ifconfig br0 192.168.0.5
1393 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1394 ifconfig vlan9 192.168.0.9
1396 but other hosts that are only on VLAN 0 can reach the IP address
1397 configured on VLAN 9. What's going on?
1399 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
1400 approaches to IP address handling in Internet hosts:
1402 - In the "Strong ES Model", where an ES is a host ("End
1403 System"), an IP address is primarily associated with a
1404 particular interface. The host discards packets that arrive
1405 on interface A if they are destined for an IP address that is
1406 configured on interface B. The host never sends packets from
1407 interface A using a source address configured on interface B.
1409 - In the "Weak ES Model", an IP address is primarily associated
1410 with a host. The host accepts packets that arrive on any
1411 interface if they are destined for any of the host's IP
1412 addresses, even if the address is configured on some
1413 interface other than the one on which it arrived. The host
1414 does not restrict itself to sending packets from an IP
1415 address associated with the originating interface.
1417 Linux uses the weak ES model. That means that when packets
1418 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
1419 br0, the kernel IP stack accepts them there for the VLAN 9 IP
1420 address, even though they were not received on vlan9, the network
1423 To simulate the strong ES model on Linux, one may add iptables rule
1424 to filter packets based on source and destination address and
1425 adjust ARP configuration with sysctls.
1427 BSD uses the strong ES model.
1429 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1431 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1432 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1433 implementation of the OpenFlow "normal switching" action. By
1434 default, when Open vSwitch isn't connected to a controller and
1435 nothing has been manually configured in the flow table, all traffic
1436 goes through the "normal switching" action. But, if you set up
1437 OpenFlow flows on your own, through a controller or using ovs-ofctl
1438 or through other means, then you have to implement VLAN handling
1441 You can use "normal switching" as a component of your OpenFlow
1442 actions, e.g. by putting "normal" into the lists of actions on
1443 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1444 controller. In situations where this is not suitable, you can
1445 implement VLAN handling yourself, e.g.:
1447 - If a packet comes in on an access port, and the flow table
1448 needs to send it out on a trunk port, then the flow can add
1449 the appropriate VLAN tag with the "mod_vlan_vid" action.
1451 - If a packet comes in on a trunk port, and the flow table
1452 needs to send it out on an access port, then the flow can
1453 strip the VLAN tag with the "strip_vlan" action.
1455 ### Q: I configured ports on a bridge as access ports with different VLAN
1458 ovs-vsctl add-br br0
1459 ovs-vsctl set-controller br0 tcp:192.168.0.10:6653
1460 ovs-vsctl add-port br0 eth0
1461 ovs-vsctl add-port br0 tap0 tag=9
1462 ovs-vsctl add-port br0 tap1 tag=10
1464 but the VMs running behind tap0 and tap1 can still communicate,
1465 that is, they are not isolated from each other even though they are
1468 A: Do you have a controller configured on br0 (as the commands above
1469 do)? If so, then this is a variant on the previous question, "My
1470 OpenFlow controller doesn't see the VLANs that I expect," and you
1471 can refer to the answer there for more information.
1473 ### Q: How MAC learning works with VLANs?
1475 A: Open vSwitch implements Independent VLAN Learning (IVL) for
1476 OFPP_NORMAL action. I.e. it logically has separate learning tables
1483 ### Q: What's a VXLAN?
1485 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1486 to solve the scaling challenges of VLAN networks in a multi-tenant
1487 environment. VXLAN is an overlay network which transports an L2 network
1488 over an existing L3 network. For more information on VXLAN, please see
1491 http://tools.ietf.org/html/rfc7348
1493 ### Q: How much of the VXLAN protocol does Open vSwitch currently support?
1495 A: Open vSwitch currently supports the framing format for packets on the
1496 wire. There is currently no support for the multicast aspects of VXLAN.
1497 To get around the lack of multicast support, it is possible to
1498 pre-provision MAC to IP address mappings either manually or from a
1501 ### Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1504 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1505 is 4789. However, it is possible to configure the destination UDP port
1506 manually on a per-VXLAN tunnel basis. An example of this configuration is
1509 ovs-vsctl add-br br0
1510 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1511 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1512 options:dst_port=8472
1515 Using OpenFlow (Manually or Via Controller)
1516 -------------------------------------------
1518 ### Q: What versions of OpenFlow does Open vSwitch support?
1520 A: The following table lists the versions of OpenFlow supported by
1521 each version of Open vSwitch:
1523 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5 OF1.6
1524 ###============ ===== ===== ===== ===== ===== ===== =====
1525 1.9 and earlier yes --- --- --- --- --- ---
1526 1.10 yes --- [*] [*] --- --- ---
1527 1.11 yes --- [*] [*] --- --- ---
1528 2.0 yes [*] [*] [*] --- --- ---
1529 2.1 yes [*] [*] [*] --- --- ---
1530 2.2 yes [*] [*] [*] [%] [*] ---
1531 2.3 yes yes yes yes [*] [*] ---
1532 2.4 yes yes yes yes [*] [*] ---
1533 2.5 yes yes yes yes [*] [*] [*]
1535 [*] Supported, with one or more missing features.
1536 [%] Experimental, unsafe implementation.
1538 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default
1539 in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1,
1540 1.2, and 1.3 must be enabled manually in ovs-vswitchd.
1542 Some versions of OpenFlow are supported with missing features and
1543 therefore not enabled by default: OpenFlow 1.4 and 1.5, in Open
1544 vSwitch 2.3 and later, as well as OpenFlow 1.6 in Open vSwitch 2.5
1545 and later. Also, the OpenFlow 1.6 specification is still under
1546 development and thus subject to change.
1548 In any case, the user may override the default:
1550 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0:
1552 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1554 - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0:
1556 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15
1558 - To enable only OpenFlow 1.0 on bridge br0:
1560 ovs-vsctl set bridge br0 protocols=OpenFlow10
1562 All current versions of ovs-ofctl enable only OpenFlow 1.0 by
1563 default. Use the -O option to enable support for later versions of
1564 OpenFlow in ovs-ofctl. For example:
1566 ovs-ofctl -O OpenFlow13 dump-flows br0
1568 (Open vSwitch 2.2 had an experimental implementation of OpenFlow
1569 1.4 that could cause crashes. We don't recommend enabling it.)
1571 [OPENFLOW-1.1+.md] in the Open vSwitch source tree tracks support for
1572 OpenFlow 1.1 and later features. When support for OpenFlow 1.4 and
1573 1.5 is solidly implemented, Open vSwitch will enable those version
1576 ### Q: Does Open vSwitch support MPLS?
1578 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1579 these versions can match on MPLS Ethernet types, but they cannot
1580 match, push, or pop MPLS labels, nor can they look past MPLS labels
1581 into the encapsulated packet.
1583 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1584 for MPLS. With the userspace datapath only, these versions can
1585 match, push, or pop a single MPLS label, but they still cannot look
1586 past MPLS labels (even after popping them) into the encapsulated
1587 packet. Kernel datapath support is unchanged from earlier
1590 Open vSwitch version 2.3 can match, push, or pop a single MPLS
1591 label and look past the MPLS label into the encapsulated packet.
1592 Both userspace and kernel datapaths will be supported, but MPLS
1593 processing always happens in userspace either way, so kernel
1594 datapath performance will be disappointing.
1596 Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS
1597 labels and look past the MPLS label into the encapsulated packet.
1598 It will have kernel support for MPLS, yielding improved
1601 ### Q: I'm getting "error type 45250 code 0". What's that?
1603 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1604 vSwitch uses this extension when it must report an error to an
1605 OpenFlow controller but no standard OpenFlow error code is
1608 Open vSwitch logs the errors that it sends to controllers, so the
1609 easiest thing to do is probably to look at the ovs-vswitchd log to
1610 find out what the error was.
1612 If you want to dissect the extended error message yourself, the
1613 format is documented in include/openflow/nicira-ext.h in the Open
1614 vSwitch source distribution. The extended error codes are
1615 documented in include/openvswitch/ofp-errors.h.
1617 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1618 doesn't actually appear through the OpenFlow connection, even
1619 though I know that it's going through.
1620 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1621 to apply to certain traffic, especially traffic between OVS and
1622 the controller itself.
1624 A: By default, Open vSwitch assumes that OpenFlow controllers are
1625 connected "in-band", that is, that the controllers are actually
1626 part of the network that is being controlled. In in-band mode,
1627 Open vSwitch sets up special "hidden" flows to make sure that
1628 traffic can make it back and forth between OVS and the controllers.
1629 These hidden flows are higher priority than any flows that can be
1630 set up through OpenFlow, and they are not visible through normal
1631 OpenFlow flow table dumps.
1633 Usually, the hidden flows are desirable and helpful, but
1634 occasionally they can cause unexpected behavior. You can view the
1635 full OpenFlow flow table, including hidden flows, on bridge br0
1638 ovs-appctl bridge/dump-flows br0
1640 to help you debug. The hidden flows are those with priorities
1641 greater than 65535 (the maximum priority that can be set with
1644 The DESIGN file at the top level of the Open vSwitch source
1645 distribution describes the in-band model in detail.
1647 If your controllers are not actually in-band (e.g. they are on
1648 localhost via 127.0.0.1, or on a separate network), then you should
1649 configure your controllers in "out-of-band" mode. If you have one
1650 controller on bridge br0, then you can configure out-of-band mode
1653 ovs-vsctl set controller br0 connection-mode=out-of-band
1655 ### Q: I configured all my controllers for out-of-band control mode but
1656 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1658 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1659 set-manager"). By default, Open vSwitch assumes that managers need
1660 in-band rules set up on every bridge. You can disable these rules
1663 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1665 This actually disables in-band control entirely for the bridge, as
1666 if all the bridge's controllers were configured for out-of-band
1669 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1671 A: See answer under "VLANs", above.
1673 ### Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1674 but I got a funny message like this:
1676 ofp_util|INFO|normalization changed ofp_match, details:
1677 ofp_util|INFO| pre: nw_dst=192.168.0.1
1680 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1681 match had disappeared, so that the flow ends up matching every
1684 A: The term "normalization" in the log message means that a flow
1685 cannot match on an L3 field without saying what L3 protocol is in
1686 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1687 so the L3 field match was dropped.
1689 In this case, the L3 protocol could be IP or ARP. A correct
1690 command for each possibility is, respectively:
1692 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1696 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1698 Similarly, a flow cannot match on an L4 field without saying what
1699 L4 protocol is in use. For example, the flow match "tp_src=1234"
1700 is, by itself, meaningless and will be ignored. Instead, to match
1701 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1702 source port 1234, write "udp,tp_src=1234".
1704 ### Q: How can I figure out the OpenFlow port number for a given port?
1706 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1707 respond with an OFPT_FEATURES_REPLY that, among other information,
1708 includes a mapping between OpenFlow port names and numbers. From a
1709 command prompt, "ovs-ofctl show br0" makes such a request and
1710 prints the response for switch br0.
1712 The Interface table in the Open vSwitch database also maps OpenFlow
1713 port names to numbers. To print the OpenFlow port number
1714 associated with interface eth0, run:
1716 ovs-vsctl get Interface eth0 ofport
1718 You can print the entire mapping with:
1720 ovs-vsctl -- --columns=name,ofport list Interface
1722 but the output mixes together interfaces from all bridges in the
1723 database, so it may be confusing if more than one bridge exists.
1725 In the Open vSwitch database, ofport value -1 means that the
1726 interface could not be created due to an error. (The Open vSwitch
1727 log should indicate the reason.) ofport value [] (the empty set)
1728 means that the interface hasn't been created yet. The latter is
1729 normally an intermittent condition (unless ovs-vswitchd is not
1732 ### Q: I added some flows with my controller or with ovs-ofctl, but when I
1733 run "ovs-dpctl dump-flows" I don't see them.
1735 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1736 won't display the information that you want. You want to use
1737 "ovs-ofctl dump-flows" instead.
1739 ### Q: It looks like each of the interfaces in my bonded port shows up
1740 as an individual OpenFlow port. Is that right?
1742 A: Yes, Open vSwitch makes individual bond interfaces visible as
1743 OpenFlow ports, rather than the bond as a whole. The interfaces
1744 are treated together as a bond for only a few purposes:
1746 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1747 controller is not configured, this happens implicitly to
1750 - Mirrors configured for output to a bonded port.
1752 It would make a lot of sense for Open vSwitch to present a bond as
1753 a single OpenFlow port. If you want to contribute an
1754 implementation of such a feature, please bring it up on the Open
1755 vSwitch development mailing list at dev@openvswitch.org.
1757 ### Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1758 multiple hosts, and other components. The behavior isn't what I
1761 A: To debug network behavior problems, trace the path of a packet,
1762 hop-by-hop, from its origin in one host to a remote host. If
1763 that's correct, then trace the path of the response packet back to
1766 The open source tool called "plotnetcfg" can help to understand the
1767 relationship between the networking devices on a single host.
1769 Usually a simple ICMP echo request and reply ("ping") packet is
1770 good enough. Start by initiating an ongoing "ping" from the origin
1771 host to a remote host. If you are tracking down a connectivity
1772 problem, the "ping" will not display any successful output, but
1773 packets are still being sent. (In this case the packets being sent
1774 are likely ARP rather than ICMP.)
1776 Tools available for tracing include the following:
1778 - "tcpdump" and "wireshark" for observing hops across network
1779 devices, such as Open vSwitch internal devices and physical
1782 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1783 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1784 These tools allow one to observe the actions being taken on
1785 packets in ongoing flows.
1787 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1788 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1789 documentation, and "Why are there so many different ways to
1790 dump flows?" above for some background.
1792 - "ovs-appctl ofproto/trace" to observe the logic behind how
1793 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1794 documentation. You can out more details about a given flow
1795 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1796 a flow from the output into an "ovs-appctl ofproto/trace"
1799 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1800 observe what goes on at these physical hops.
1802 Starting at the origin of a given packet, observe the packet at
1803 each hop in turn. For example, in one plausible scenario, you
1806 1. "tcpdump" the "eth" interface through which an ARP egresses
1807 a VM, from inside the VM.
1809 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1810 ingresses the host machine.
1812 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1813 the host interface through which the ARP egresses the
1814 physical machine. You may need to use "ovs-dpctl show" to
1815 interpret the port numbers. If the output seems surprising,
1816 you can use "ovs-appctl ofproto/trace" to observe details of
1817 how ovs-vswitchd determined the actions in the "ovs-dpctl
1820 4. "tcpdump" the "eth" interface through which the ARP egresses
1821 the physical machine.
1823 5. "tcpdump" the "eth" interface through which the ARP
1824 ingresses the physical machine, at the remote host that
1827 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1828 remote host that receives the ARP and observe the VM "vif"
1829 or "tap" interface to which the flow is directed. Again,
1830 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1832 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1835 8. "tcpdump" the "eth" interface through which the ARP
1836 ingresses a VM, from inside the VM.
1838 It is likely that during one of these steps you will figure out the
1839 problem. If not, then follow the ARP reply back to the origin, in
1842 ### Q: How do I make a flow drop packets?
1844 A: To drop a packet is to receive it without forwarding it. OpenFlow
1845 explicitly specifies forwarding actions. Thus, a flow with an
1846 empty set of actions does not forward packets anywhere, causing
1847 them to be dropped. You can specify an empty set of actions with
1848 "actions=" on the ovs-ofctl command line. For example:
1850 ovs-ofctl add-flow br0 priority=65535,actions=
1852 would cause every packet entering switch br0 to be dropped.
1854 You can write "drop" explicitly if you like. The effect is the
1855 same. Thus, the following command also causes every packet
1856 entering switch br0 to be dropped:
1858 ovs-ofctl add-flow br0 priority=65535,actions=drop
1860 "drop" is not an action, either in OpenFlow or Open vSwitch.
1861 Rather, it is only a way to say that there are no actions.
1863 ### Q: I added a flow to send packets out the ingress port, like this:
1865 ovs-ofctl add-flow br0 in_port=2,actions=2
1867 but OVS drops the packets instead.
1869 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1870 out its ingress port. The rationale is that dropping these packets
1871 makes it harder to loop the network. Sometimes this behavior can
1872 even be convenient, e.g. it is often the desired behavior in a flow
1873 that forwards a packet to several ports ("floods" the packet).
1875 Sometimes one really needs to send a packet out its ingress port
1876 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1877 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1879 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1881 This also works in some circumstances where the flow doesn't match
1882 on the input port. For example, if you know that your switch has
1883 five ports numbered 2 through 6, then the following will send every
1884 received packet out every port, even its ingress port:
1886 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1890 ovs-ofctl add-flow br0 actions=all,in_port
1892 Sometimes, in complicated flow tables with multiple levels of
1893 "resubmit" actions, a flow needs to output to a particular port
1894 that may or may not be the ingress port. It's difficult to take
1895 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1896 provides, as an OpenFlow extension, the ability to modify the
1897 in_port field. Whatever value is currently in the in_port field is
1898 the port to which outputs will be dropped, as well as the
1899 destination for OFPP_IN_PORT. This means that the following will
1900 reliably output to port 2 or to ports 2 through 6, respectively:
1902 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1903 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1905 If the input port is important, then one may save and restore it on
1908 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1909 load:0->NXM_OF_IN_PORT[],\
1911 pop:NXM_OF_IN_PORT[]
1913 ### Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1914 on port 2. I set up flows to forward only traffic destined to the
1915 other host and drop other traffic, like this:
1917 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1918 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1919 priority=0,actions=drop
1921 But it doesn't work--I don't get any connectivity when I do this.
1924 A: These flows drop the ARP packets that IP hosts use to establish IP
1925 connectivity over Ethernet. To solve the problem, add flows to
1926 allow ARP to pass between the hosts:
1928 priority=5,in_port=1,arp,actions=2
1929 priority=5,in_port=2,arp,actions=1
1931 This issue can manifest other ways, too. The following flows that
1932 match on Ethernet addresses instead of IP addresses will also drop
1933 ARP packets, because ARP requests are broadcast instead of being
1934 directed to a specific host:
1936 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1937 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1938 priority=0,actions=drop
1940 The solution already described above will also work in this case.
1941 It may be better to add flows to allow all multicast and broadcast
1944 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1945 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1947 ### Q: My bridge disconnects from my controller on add-port/del-port.
1949 A: Reconfiguring your bridge can change your bridge's datapath-id because
1950 Open vSwitch generates datapath-id from the MAC address of one of its ports.
1951 In that case, Open vSwitch disconnects from controllers because there's
1952 no graceful way to notify controllers about the change of datapath-id.
1954 To avoid the behaviour, you can configure datapath-id manually.
1956 ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef
1958 ### Q: My controller is getting errors about "buffers". What's going on?
1960 A: When a switch sends a packet to an OpenFlow controller using a
1961 "packet-in" message, it can also keep a copy of that packet in a
1962 "buffer", identified by a 32-bit integer "buffer_id". There are
1963 two advantages to buffering. First, when the controller wants to
1964 tell the switch to do something with the buffered packet (with a
1965 "packet-out" OpenFlow request), it does not need to send another
1966 copy of the packet back across the OpenFlow connection, which
1967 reduces the bandwidth cost of the connection and improves latency.
1968 This enables the second advantage: the switch can optionally send
1969 only the first part of the packet to the controller (assuming that
1970 the switch only needs to look at the first few bytes of the
1971 packet), further reducing bandwidth and improving latency.
1973 However, buffering introduces some issues of its own. First, any
1974 switch has limited resources, so if the controller does not use a
1975 buffered packet, the switch has to decide how long to keep it
1976 buffered. When many packets are sent to a controller and buffered,
1977 Open vSwitch can discard buffered packets that the controller has
1978 not used after as little as 5 seconds. This means that
1979 controllers, if they make use of packet buffering, should use the
1980 buffered packets promptly. (This includes sending a "packet-out"
1981 with no actions if the controller does not want to do anything with
1982 a buffered packet, to clear the packet buffer and effectively
1985 Second, packet buffers are one-time-use, meaning that a controller
1986 cannot use a single packet buffer in two or more "packet-out"
1987 commands. Open vSwitch will respond with an error to the second
1988 and subsequent "packet-out"s in such a case.
1990 Finally, a common error early in controller development is to try
1991 to use buffer_id 0 in a "packet-out" message as if 0 represented
1992 "no buffered packet". This is incorrect usage: the buffer_id with
1993 this meaning is actually 0xffffffff.
1995 ovs-vswitchd(8) describes some details of Open vSwitch packet
1996 buffering that the OpenFlow specification requires implementations
1999 ### Q: How does OVS divide flows among buckets in an OpenFlow "select" group?
2001 A: In Open vSwitch 2.3 and earlier, Open vSwitch used the destination
2002 Ethernet address to choose a bucket in a select group.
2004 Open vSwitch 2.4 and later by default hashes the source and
2005 destination Ethernet address, VLAN ID, Ethernet type, IPv4/v6
2006 source and destination address and protocol, and for TCP and SCTP
2007 only, the source and destination ports. The hash is "symmetric",
2008 meaning that exchanging source and destination addresses does not
2009 change the bucket selection.
2011 Select groups in Open vSwitch 2.4 and later can be configured to
2012 use a different hash function, using a Netronome extension to the
2013 OpenFlow 1.5+ group_mod message. For more information, see
2014 Documentation/group-selection-method-property.txt in the Open
2015 vSwitch source tree. (OpenFlow 1.5 support in Open vSwitch is still
2018 ### Q: I added a flow to accept packets on VLAN 123 and output them on
2021 ovs-ofctl add-flow br0 dl_vlan=123,actions=output:1,mod_vlan_vid:456
2023 but the packets are actually being output in VLAN 123. Why?
2025 A: OpenFlow actions are executed in the order specified. Thus, the
2026 actions above first output the packet, then change its VLAN. Since
2027 the output occurs before changing the VLAN, the change in VLAN will
2028 have no visible effect.
2030 To solve this and similar problems, order actions so that changes
2031 to headers happen before output, e.g.:
2033 ovs-ofctl add-flow br0 dl_vlan=123,actions=mod_vlan_vid:456,output:1
2035 ### Q: The "learn" action can't learn the action I want, can you improve it?
2037 A: By itself, the "learn" action can only put two kinds of actions
2038 into the flows that it creates: "load" and "output" actions. If
2039 "learn" is used in isolation, these are severe limits.
2041 However, "learn" is not meant to be used in isolation. It is a
2042 primitive meant to be used together with other Open vSwitch
2043 features to accomplish a task. Its existing features are enough to
2044 accomplish most tasks.
2046 Here is an outline of a typical pipeline structure that allows for
2047 versatile behavior using "learn":
2049 - Flows in table A contain a "learn" action, that populates flows
2050 in table L, that use a "load" action to populate register R
2051 with information about what was learned.
2053 - Flows in table B contain two sequential resubmit actions: one
2054 to table L and another one to table B+1.
2056 - Flows in table B+1 match on register R and act differently
2057 depending on what the flows in table L loaded into it.
2059 This approach can be used to implement many "learn"-based features.
2062 - Resubmit to a table selected based on learned information, e.g. see:
2063 http://openvswitch.org/pipermail/discuss/2016-June/021694.html
2065 - MAC learning in the middle of a pipeline, as described in
2068 - TCP state based firewalling, by learning outgoing connections
2069 based on SYN packets and matching them up with incoming
2072 - At least some of the features described in T. A. Hoff,
2073 "Extending Open vSwitch to Facilitate Creation of Stateful SDN
2080 ### Q: How do I implement a new OpenFlow message?
2082 A: Add your new message to "enum ofpraw" and "enum ofptype" in
2083 lib/ofp-msgs.h, following the existing pattern. Then recompile and
2084 fix all of the new warnings, implementing new functionality for the
2085 new message as needed. (If you configure with --enable-Werror, as
2086 described in [INSTALL.md], then it is impossible to miss any warnings.)
2088 If you need to add an OpenFlow vendor extension message for a
2089 vendor that doesn't yet have any extension messages, then you will
2090 also need to edit build-aux/extract-ofp-msgs.
2092 ### Q: How do I add support for a new field or header?
2094 A: Add new members for your field to "struct flow" in lib/flow.h, and
2095 add new enumerations for your new field to "enum mf_field_id" in
2096 lib/meta-flow.h, following the existing pattern. Also, add support
2097 to miniflow_extract() in lib/flow.c for extracting your new field
2098 from a packet into struct miniflow, and to nx_put_raw() in
2099 lib/nx-match.c to output your new field in OXM matches. Then
2100 recompile and fix all of the new warnings, implementing new
2101 functionality for the new field or header as needed. (If you
2102 configure with --enable-Werror, as described in [INSTALL.md], then
2103 it is impossible to miss any warnings.)
2105 If you want kernel datapath support for your new field, you also
2106 need to modify the kernel module for the operating systems you are
2107 interested in. This isn't mandatory, since fields understood only
2108 by userspace work too (with a performance penalty), so it's
2109 reasonable to start development without it. If you implement
2110 kernel module support for Linux, then the Linux kernel "netdev"
2111 mailing list is the place to submit that support first; please read
2112 up on the Linux kernel development process separately. The Windows
2113 datapath kernel module support, on the other hand, is maintained
2114 within the OVS tree, so patches for that can go directly to
2117 ### Q: How do I add support for a new OpenFlow action?
2119 A: Add your new action to "enum ofp_raw_action_type" in
2120 lib/ofp-actions.c, following the existing pattern. Then recompile
2121 and fix all of the new warnings, implementing new functionality for
2122 the new action as needed. (If you configure with --enable-Werror,
2123 as described in [INSTALL.md], then it is impossible to miss any
2126 If you need to add an OpenFlow vendor extension action for a vendor
2127 that doesn't yet have any extension actions, then you will also
2128 need to edit build-aux/extract-ofp-actions.
2134 bugs@openvswitch.org
2135 http://openvswitch.org/
2137 [PORTING.md]:PORTING.md
2138 [WHY-OVS.md]:WHY-OVS.md
2139 [INSTALL.md]:INSTALL.md
2140 [OPENFLOW-1.1+.md]:OPENFLOW-1.1+.md
2141 [INSTALL.DPDK.md]:INSTALL.DPDK.md
2142 [Tutorial.md]:tutorial/Tutorial.md