1 Frequently Asked Questions
2 ==========================
4 Open vSwitch <http://openvswitch.org>
9 - [Releases](#releases)
10 - [Terminology](#terminology)
11 - [Basic configuration](#basic-configuration)
12 - [Implementation Details](#implementation-details)
13 - [Performance](#performance)
14 - [Configuration Problems](#configuration-problems)
18 - [Using OpenFlow](#using-openflow)
19 - [Development](#development)
23 ### Q: What is Open vSwitch?
25 A: Open vSwitch is a production quality open source software switch
26 designed to be used as a vswitch in virtualized server
27 environments. A vswitch forwards traffic between different VMs on
28 the same physical host and also forwards traffic between VMs and
29 the physical network. Open vSwitch supports standard management
30 interfaces (e.g. sFlow, NetFlow, IPFIX, RSPAN, CLI), and is open to
31 programmatic extension and control using OpenFlow and the OVSDB
34 Open vSwitch as designed to be compatible with modern switching
35 chipsets. This means that it can be ported to existing high-fanout
36 switches allowing the same flexible control of the physical
37 infrastructure as the virtual infrastructure. It also means that
38 Open vSwitch will be able to take advantage of on-NIC switching
39 chipsets as their functionality matures.
41 ### Q: What virtualization platforms can use Open vSwitch?
43 A: Open vSwitch can currently run on any Linux-based virtualization
44 platform (kernel 3.10 and newer), including: KVM, VirtualBox, Xen,
45 Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the
46 mainline kernel. The bulk of the code is written in platform-
47 independent C and is easily ported to other environments. We welcome
48 inquires about integrating Open vSwitch with other virtualization
51 ### Q: How can I try Open vSwitch?
53 A: The Open vSwitch source code can be built on a Linux system. You can
54 build and experiment with Open vSwitch on any Linux machine.
55 Packages for various Linux distributions are available on many
56 platforms, including: Debian, Ubuntu, Fedora.
58 You may also download and run a virtualization platform that already
59 has Open vSwitch integrated. For example, download a recent ISO for
60 XenServer or Xen Cloud Platform. Be aware that the version
61 integrated with a particular platform may not be the most recent Open
64 ### Q: Does Open vSwitch only work on Linux?
66 A: No, Open vSwitch has been ported to a number of different operating
67 systems and hardware platforms. Most of the development work occurs
68 on Linux, but the code should be portable to any POSIX system. We've
69 seen Open vSwitch ported to a number of different platforms,
70 including FreeBSD, Windows, and even non-POSIX embedded systems.
72 By definition, the Open vSwitch Linux kernel module only works on
73 Linux and will provide the highest performance. However, a userspace
74 datapath is available that should be very portable.
76 ### Q: What's involved with porting Open vSwitch to a new platform or switching ASIC?
78 A: The [PORTING.md] document describes how one would go about
79 porting Open vSwitch to a new operating system or hardware platform.
81 ### Q: Why would I use Open vSwitch instead of the Linux bridge?
83 A: Open vSwitch is specially designed to make it easier to manage VM
84 network configuration and monitor state spread across many physical
85 hosts in dynamic virtualized environments. Please see
86 [WHY-OVS.md] for a more detailed description of how Open vSwitch
87 relates to the Linux Bridge.
89 ### Q: How is Open vSwitch related to distributed virtual switches like the VMware vNetwork distributed switch or the Cisco Nexus 1000V?
91 A: Distributed vswitch applications (e.g., VMware vNetwork distributed
92 switch, Cisco Nexus 1000V) provide a centralized way to configure and
93 monitor the network state of VMs that are spread across many physical
94 hosts. Open vSwitch is not a distributed vswitch itself, rather it
95 runs on each physical host and supports remote management in a way
96 that makes it easier for developers of virtualization/cloud
97 management platforms to offer distributed vswitch capabilities.
99 To aid in distribution, Open vSwitch provides two open protocols that
100 are specially designed for remote management in virtualized network
101 environments: OpenFlow, which exposes flow-based forwarding state,
102 and the OVSDB management protocol, which exposes switch port state.
103 In addition to the switch implementation itself, Open vSwitch
104 includes tools (ovs-ofctl, ovs-vsctl) that developers can script and
105 extend to provide distributed vswitch capabilities that are closely
106 integrated with their virtualization management platform.
108 ### Q: Why doesn't Open vSwitch support distribution?
110 A: Open vSwitch is intended to be a useful component for building
111 flexible network infrastructure. There are many different approaches
112 to distribution which balance trade-offs between simplicity,
113 scalability, hardware compatibility, convergence times, logical
114 forwarding model, etc. The goal of Open vSwitch is to be able to
115 support all as a primitive building block rather than choose a
116 particular point in the distributed design space.
118 ### Q: How can I contribute to the Open vSwitch Community?
120 A: You can start by joining the mailing lists and helping to answer
121 questions. You can also suggest improvements to documentation. If
122 you have a feature or bug you would like to work on, send a mail to
123 one of the mailing lists:
125 http://openvswitch.org/mlists/
127 ### Q: Why can I no longer connect to my OpenFlow controller or OVSDB manager?
129 A: Starting in OVS 2.4, we switched the default ports to the
130 IANA-specified port numbers for OpenFlow (6633->6653) and OVSDB
131 (6632->6640). We recommend using these port numbers, but if you
132 cannot, all the programs allow overriding the default port. See the
133 appropriate man page.
137 ### Q: What does it mean for an Open vSwitch release to be LTS (long-term support)?
139 A: All official releases have been through a comprehensive testing
140 process and are suitable for production use. Planned releases
141 occur twice a year. If a significant bug is identified in an
142 LTS release, we will provide an updated release that includes the
143 fix. Releases that are not LTS may not be fixed and may just be
144 supplanted by the next major release. The current LTS release is
147 For more information on the Open vSwitch release process, please
148 see [release-process.md].
150 ### Q: What Linux kernel versions does each Open vSwitch release work with?
152 A: The following table lists the Linux kernel versions against which the
153 given versions of the Open vSwitch kernel module will successfully
154 build. The Linux kernel versions are upstream kernel versions, so
155 Linux kernels modified from the upstream sources may not build in
156 some cases even if they are based on a supported version. This is
157 most notably true of Red Hat Enterprise Linux (RHEL) kernels, which
158 are extensively modified from upstream.
160 | Open vSwitch | Linux kernel
161 |:------------:|:-------------:
162 | 1.4.x | 2.6.18 to 3.2
163 | 1.5.x | 2.6.18 to 3.2
164 | 1.6.x | 2.6.18 to 3.2
165 | 1.7.x | 2.6.18 to 3.3
166 | 1.8.x | 2.6.18 to 3.4
167 | 1.9.x | 2.6.18 to 3.8
168 | 1.10.x | 2.6.18 to 3.8
169 | 1.11.x | 2.6.18 to 3.8
170 | 2.0.x | 2.6.32 to 3.10
171 | 2.1.x | 2.6.32 to 3.11
172 | 2.3.x | 2.6.32 to 3.14
173 | 2.4.x | 2.6.32 to 4.0
174 | 2.5.x | 2.6.32 to 4.3
175 | 2.6.x | 3.10 to 4.7
177 Open vSwitch userspace should also work with the Linux kernel module
178 built into Linux 3.3 and later.
180 Open vSwitch userspace is not sensitive to the Linux kernel version.
181 It should build against almost any kernel, certainly against 2.6.32
184 ### Q: Are all features available with all datapaths?
186 A: Open vSwitch supports different datapaths on different platforms. Each
187 datapath has a different feature set: the following tables try to summarize
192 * *Linux upstream*: The datapath implemented by the kernel module shipped
193 with Linux upstream. Since features have been gradually
194 introduced into the kernel, the table mentions the first
195 Linux release whose OVS module supports the feature.
197 * *Linux OVS tree*: The datapath implemented by the Linux kernel module
198 distributed with the OVS source tree.
200 * *Userspace*: Also known as DPDK, dpif-netdev or dummy datapath. It is the
201 only datapath that works on NetBSD, FreeBSD and Mac OSX.
203 * *Hyper-V*: Also known as the Windows datapath.
205 The following table lists the datapath supported features from
206 an Open vSwitch user's perspective.
208 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
209 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
210 NAT | 4.6 | YES | NO | NO |
211 Connection tracking | 4.3 | YES | PARTIAL | PARTIAL |
212 Tunnel - LISP | NO | YES | NO | NO |
213 Tunnel - STT | NO | YES | NO | YES |
214 Tunnel - GRE | 3.11 | YES | YES | YES |
215 Tunnel - VXLAN | 3.12 | YES | YES | YES |
216 Tunnel - Geneve | 3.18 | YES | YES | YES |
217 Tunnel - GRE-IPv6 | NO | NO | YES | NO |
218 Tunnel - VXLAN-IPv6 | 4.3 | YES | YES | NO |
219 Tunnel - Geneve-IPv6 | 4.4 | YES | YES | NO |
220 QoS - Policing | YES | YES | YES | NO |
221 QoS - Shaping | YES | YES | NO | NO |
222 sFlow | YES | YES | YES | NO |
223 IPFIX | 3.10 | YES | YES | NO |
224 Set action | YES | YES | YES | PARTIAL |
225 NIC Bonding | YES | YES | YES | NO |
226 Multiple VTEPs | YES | YES | YES | NO |
229 * Only a limited set of flow fields is modifiable via the set action by the
231 * The Hyper-V datapath only supports one physical NIC per datapath. This is
232 why bonding is not supported.
233 * The Hyper-V datapath can have at most one IP address configured as a
236 The following table lists features that do not *directly* impact an
237 Open vSwitch user, e.g. because their absence can be hidden by the ofproto
238 layer (usually this comes with a performance penalty).
240 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
241 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
242 SCTP flows | 3.12 | YES | YES | YES |
243 MPLS | 3.19 | YES | YES | YES |
244 UFID | 4.0 | YES | YES | NO |
245 Megaflows | 3.12 | YES | YES | NO |
246 Masked set action | 4.0 | YES | YES | NO |
247 Recirculation | 3.19 | YES | YES | YES |
248 TCP flags matching | 3.13 | YES | YES | NO |
249 Validate flow actions | YES | YES | N/A | NO |
250 Multiple datapaths | YES | YES | YES | NO |
251 Tunnel TSO - STT | N/A | YES | NO | YES |
253 ### Q: What DPDK version does each Open vSwitch release work with?
255 A: The following table lists the DPDK version against which the
256 given versions of Open vSwitch will successfully build.
258 | Open vSwitch | DPDK
259 |:------------:|:-----:
266 ### Q: I get an error like this when I configure Open vSwitch:
268 configure: error: Linux kernel in <dir> is version <x>, but
269 version newer than <y> is not supported (please refer to the
274 A: You have the following options:
276 - Use the Linux kernel module supplied with the kernel that you are
277 using. (See also the following FAQ.)
279 - If there is a newer released version of Open vSwitch, consider
280 building that one, because it may support the kernel that you are
281 building against. (To find out, consult the table in the
284 - The Open vSwitch "master" branch may support the kernel that you
285 are using, so consider building the kernel module from "master".
287 All versions of Open vSwitch userspace are compatible with all
288 versions of the Open vSwitch kernel module, so you do not have to
289 use the kernel module from one source along with the userspace
290 programs from the same source.
292 ### Q: What features are not available in the Open vSwitch kernel datapath that ships as part of the upstream Linux kernel?
294 A: The kernel module in upstream Linux does not include support for
295 LISP. Work is in progress to add support for LISP to the upstream
296 Linux version of the Open vSwitch kernel module. For now, if you
297 need this feature, use the kernel module from the Open vSwitch
298 distribution instead of the upstream Linux kernel module.
300 Certain features require kernel support to function or to have
301 reasonable performance. If the ovs-vswitchd log file indicates that
302 a feature is not supported, consider upgrading to a newer upstream
303 Linux release or using the kernel module paired with the userspace
306 ### Q: Why do tunnels not work when using a kernel module other than the one packaged with Open vSwitch?
308 A: Support for tunnels was added to the upstream Linux kernel module
309 after the rest of Open vSwitch. As a result, some kernels may contain
310 support for Open vSwitch but not tunnels. The minimum kernel version
311 that supports each tunnel protocol is:
313 | Protocol | Linux Kernel
314 |:--------:|:-------------:
318 | LISP | <not upstream>
319 | STT | <not upstream>
321 If you are using a version of the kernel that is older than the one
322 listed above, it is still possible to use that tunnel protocol. However,
323 you must compile and install the kernel module included with the Open
324 vSwitch distribution rather than the one on your machine. If problems
325 persist after doing this, check to make sure that the module that is
326 loaded is the one you expect.
328 ### Q: Why are UDP tunnel checksums not computed for VXLAN or Geneve?
330 A: Generating outer UDP checksums requires kernel support that was not
331 part of the initial implementation of these protocols. If using the
332 upstream Linux Open vSwitch module, you must use kernel 4.0 or
333 newer. The out-of-tree modules from Open vSwitch release 2.4 and later
334 support UDP checksums.
336 ### Q: What features are not available when using the userspace datapath?
338 A: Tunnel virtual ports are not supported, as described in the
339 previous answer. It is also not possible to use queue-related
340 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
341 may not be transmitted.
343 ### Q: Should userspace or kernel be upgraded first to minimize downtime?
345 In general, the Open vSwitch userspace should be used with the
346 kernel version included in the same release or with the version
347 from upstream Linux. However, when upgrading between two releases
348 of Open vSwitch it is best to migrate userspace first to reduce
349 the possibility of incompatibilities.
351 ### Q: What happened to the bridge compatibility feature?
353 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
354 When it was enabled, Open vSwitch imitated the interface of the
355 Linux kernel "bridge" module. This allowed users to drop Open
356 vSwitch into environments designed to use the Linux kernel bridge
357 module without adapting the environment to use Open vSwitch.
359 Open vSwitch 1.10 and later do not support bridge compatibility.
360 The feature was dropped because version 1.10 adopted a new internal
361 architecture that made bridge compatibility difficult to maintain.
362 Now that many environments use OVS directly, it would be rarely
365 To use bridge compatibility, install OVS 1.9 or earlier, including
366 the accompanying kernel modules (both the main and bridge
367 compatibility modules), following the instructions that come with
368 the release. Be sure to start the ovs-brcompatd daemon.
373 ### Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation keeps talking about bridges. What's a bridge?
375 A: In networking, the terms "bridge" and "switch" are synonyms. Open
376 vSwitch implements an Ethernet switch, which means that it is also
379 ### Q: What's a VLAN?
381 A: See the "VLAN" section below.
383 ## Basic configuration
385 ### Q: How do I configure a port as an access port?
387 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
388 the following commands configure br0 with eth0 as a trunk port (the
389 default) and tap0 as an access port for VLAN 9:
392 ovs-vsctl add-port br0 eth0
393 ovs-vsctl add-port br0 tap0 tag=9
395 If you want to configure an already added port as an access port,
396 use "ovs-vsctl set", e.g.:
398 ovs-vsctl set port tap0 tag=9
400 ### Q: How do I configure a port as a SPAN port, that is, enable mirroring of all traffic to that port?
402 A: The following commands configure br0 with eth0 and tap0 as trunk
403 ports. All traffic coming in or going out on eth0 or tap0 is also
404 mirrored to tap1; any traffic arriving on tap1 is dropped:
407 ovs-vsctl add-port br0 eth0
408 ovs-vsctl add-port br0 tap0
409 ovs-vsctl add-port br0 tap1 \
410 -- --id=@p get port tap1 \
411 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
412 -- set bridge br0 mirrors=@m
414 To later disable mirroring, run:
416 ovs-vsctl clear bridge br0 mirrors
418 ### Q: Does Open vSwitch support configuring a port in promiscuous mode?
420 A: Yes. How you configure it depends on what you mean by "promiscuous
423 - Conventionally, "promiscuous mode" is a feature of a network
424 interface card. Ordinarily, a NIC passes to the CPU only the
425 packets actually destined to its host machine. It discards
426 the rest to avoid wasting memory and CPU cycles. When
427 promiscuous mode is enabled, however, it passes every packet
428 to the CPU. On an old-style shared-media or hub-based
429 network, this allows the host to spy on all packets on the
430 network. But in the switched networks that are almost
431 everywhere these days, promiscuous mode doesn't have much
432 effect, because few packets not destined to a host are
433 delivered to the host's NIC.
435 This form of promiscuous mode is configured in the guest OS of
436 the VMs on your bridge, e.g. with "ifconfig".
438 - The VMware vSwitch uses a different definition of "promiscuous
439 mode". When you configure promiscuous mode on a VMware vNIC,
440 the vSwitch sends a copy of every packet received by the
441 vSwitch to that vNIC. That has a much bigger effect than just
442 enabling promiscuous mode in a guest OS. Rather than getting
443 a few stray packets for which the switch does not yet know the
444 correct destination, the vNIC gets every packet. The effect
445 is similar to replacing the vSwitch by a virtual hub.
447 This "promiscuous mode" is what switches normally call "port
448 mirroring" or "SPAN". For information on how to configure
449 SPAN, see "How do I configure a port as a SPAN port, that is,
450 enable mirroring of all traffic to that port?"
452 ### Q: How do I configure a DPDK port as an access port?
454 A: Firstly, you must have a DPDK-enabled version of Open vSwitch.
456 If your version is DPDK-enabled it will support the other-config:dpdk-init
457 configuration in the database and will display lines with "EAL:..."
458 during startup when other_config:dpdk-init is set to 'true'.
460 Secondly, when adding a DPDK port, unlike a system port, the
461 type for the interface must be specified. For example;
464 ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
466 Finally, it is required that DPDK port names begin with 'dpdk'.
468 See [INSTALL.DPDK.md] for more information on enabling and using DPDK with
471 ### Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all traffic to that VLAN?
473 A: The following commands configure br0 with eth0 as a trunk port and
474 tap0 as an access port for VLAN 10. All traffic coming in or going
475 out on tap0, as well as traffic coming in or going out on eth0 in
476 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
477 VLAN 10, in cases where one is present, is dropped as part of
481 ovs-vsctl add-port br0 eth0
482 ovs-vsctl add-port br0 tap0 tag=10
484 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
486 -- set bridge br0 mirrors=@m
488 To later disable mirroring, run:
490 ovs-vsctl clear bridge br0 mirrors
492 Mirroring to a VLAN can disrupt a network that contains unmanaged
493 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
494 GRE tunnel has fewer caveats than mirroring to a VLAN and should
495 generally be preferred.
497 ### Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
499 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
500 of the specified output-vlan. This loss of information may make
501 the mirrored traffic too hard to interpret.
503 To mirror multiple VLANs, use the commands above, but specify a
504 comma-separated list of VLANs as the value for select-vlan. To
505 mirror every VLAN, use the commands above, but omit select-vlan and
508 When a packet arrives on a VLAN that is used as a mirror output
509 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
510 floods the packet across all the ports for which the mirror output
511 VLAN is configured. (If an OpenFlow controller is in use, then it
512 can override this behavior through the flow table.) If OVS is used
513 as an intermediate switch, rather than an edge switch, this ensures
514 that the RSPAN traffic is distributed through the network.
516 Mirroring to a VLAN can disrupt a network that contains unmanaged
517 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
518 GRE tunnel has fewer caveats than mirroring to a VLAN and should
519 generally be preferred.
521 ### Q: How do I configure mirroring of all traffic to a GRE tunnel?
523 A: The following commands configure br0 with eth0 and tap0 as trunk
524 ports. All traffic coming in or going out on eth0 or tap0 is also
525 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
526 traffic arriving on gre0 is dropped:
529 ovs-vsctl add-port br0 eth0
530 ovs-vsctl add-port br0 tap0
531 ovs-vsctl add-port br0 gre0 \
532 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
533 -- --id=@p get port gre0 \
534 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
535 -- set bridge br0 mirrors=@m
537 To later disable mirroring and destroy the GRE tunnel:
539 ovs-vsctl clear bridge br0 mirrors
540 ovs-vsctl del-port br0 gre0
542 ### Q: Does Open vSwitch support ERSPAN?
544 A: No. ERSPAN is an undocumented proprietary protocol. As an
545 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
548 ### Q: How do I connect two bridges?
550 A: First, why do you want to do this? Two connected bridges are not
551 much different from a single bridge, so you might as well just have
552 a single bridge with all your ports on it.
554 If you still want to connect two bridges, you can use a pair of
555 patch ports. The following example creates bridges br0 and br1,
556 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
557 and br1 with a pair of patch ports.
560 ovs-vsctl add-port br0 eth0
561 ovs-vsctl add-port br0 tap0
563 ovs-vsctl add-port br1 tap1
565 -- add-port br0 patch0 \
566 -- set interface patch0 type=patch options:peer=patch1 \
567 -- add-port br1 patch1 \
568 -- set interface patch1 type=patch options:peer=patch0
570 Bridges connected with patch ports are much like a single bridge.
571 For instance, if the example above also added eth1 to br1, and both
572 eth0 and eth1 happened to be connected to the same next-hop switch,
573 then you could loop your network just as you would if you added
574 eth0 and eth1 to the same bridge (see the "Configuration Problems"
575 section below for more information).
577 If you are using Open vSwitch 1.9 or an earlier version, then you
578 need to be using the kernel module bundled with Open vSwitch rather
579 than the one that is integrated into Linux 3.3 and later, because
580 Open vSwitch 1.9 and earlier versions need kernel support for patch
581 ports. This also means that in Open vSwitch 1.9 and earlier, patch
582 ports will not work with the userspace datapath, only with the
585 ### Q: How do I configure a bridge without an OpenFlow local port? (Local port in the sense of OFPP_LOCAL)
587 A: Open vSwitch does not support such a configuration.
588 Bridges always have their local ports.
590 ## Implementation Details
592 ### Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
594 A: Open vSwitch uses different kinds of flows for different purposes:
596 - OpenFlow flows are the most important kind of flow. OpenFlow
597 controllers use these flows to define a switch's policy.
598 OpenFlow flows support wildcards, priorities, and multiple
601 When in-band control is in use, Open vSwitch sets up a few
602 "hidden" flows, with priority higher than a controller or the
603 user can configure, that are not visible via OpenFlow. (See
604 the "Controller" section of the FAQ for more information
607 - The Open vSwitch software switch implementation uses a second
608 kind of flow internally. These flows, called "datapath" or
609 "kernel" flows, do not support priorities and comprise only a
610 single table, which makes them suitable for caching. (Like
611 OpenFlow flows, datapath flows do support wildcarding, in Open
612 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
613 also support different actions and number ports differently.
615 Datapath flows are an implementation detail that is subject to
616 change in future versions of Open vSwitch. Even with the
617 current version of Open vSwitch, hardware switch
618 implementations do not necessarily use this architecture.
620 Users and controllers directly control only the OpenFlow flow
621 table. Open vSwitch manages the datapath flow table itself, so
622 users should not normally be concerned with it.
624 ### Q: Why are there so many different ways to dump flows?
626 A: Open vSwitch has two kinds of flows (see the previous question), so
627 it has commands with different purposes for dumping each kind of
630 - `ovs-ofctl dump-flows <br>` dumps OpenFlow flows, excluding
631 hidden flows. This is the most commonly useful form of flow
632 dump. (Unlike the other commands, this should work with any
633 OpenFlow switch, not just Open vSwitch.)
635 - `ovs-appctl bridge/dump-flows <br>` dumps OpenFlow flows,
636 including hidden flows. This is occasionally useful for
637 troubleshooting suspected issues with in-band control.
639 - `ovs-dpctl dump-flows [dp]` dumps the datapath flow table
640 entries for a Linux kernel-based datapath. In Open vSwitch
641 1.10 and later, ovs-vswitchd merges multiple switches into a
642 single datapath, so it will show all the flows on all your
643 kernel-based switches. This command can occasionally be
644 useful for debugging.
646 - `ovs-appctl dpif/dump-flows <br>`, new in Open vSwitch 1.10,
647 dumps datapath flows for only the specified bridge, regardless
650 ### Q: How does multicast snooping works with VLANs?
652 A: Open vSwitch maintains snooping tables for each VLAN.
654 ### Q: Can OVS populate the kernel flow table in advance instead of in reaction to packets?
656 A: No. There are several reasons:
658 - Kernel flows are not as sophisticated as OpenFlow flows, which
659 means that some OpenFlow policies could require a large number of
660 kernel flows. The "conjunctive match" feature is an extreme
661 example: the number of kernel flows it requires is the product of
662 the number of flows in each dimension.
664 - With multiple OpenFlow flow tables and simple sets of actions, the
665 number of kernel flows required can be as large as the product of
666 the number of flows in each dimension. With more sophisticated
667 actions, the number of kernel flows could be even larger.
669 - Open vSwitch is designed so that any version of OVS userspace
670 interoperates with any version of the OVS kernel module. This
671 forward and backward compatibility requires that userspace observe
672 how the kernel module parses received packets. This is only
673 possible in a straightforward way when userspace adds kernel flows
674 in reaction to received packets.
676 For more relevant information on the architecture of Open vSwitch,
677 please read "The Design and Implementation of Open vSwitch",
678 published in USENIX NSDI 2015.
682 ### Q: I just upgraded and I see a performance drop. Why?
684 A: The OVS kernel datapath may have been updated to a newer version than
685 the OVS userspace components. Sometimes new versions of OVS kernel
686 module add functionality that is backwards compatible with older
687 userspace components but may cause a drop in performance with them.
688 Especially, if a kernel module from OVS 2.1 or newer is paired with
689 OVS userspace 1.10 or older, there will be a performance drop for
692 Updating the OVS userspace components to the latest released
693 version should fix the performance degradation.
695 To get the best possible performance and functionality, it is
696 recommended to pair the same versions of the kernel module and OVS
700 ## Configuration Problems
702 ### Q: I created a bridge and added my Ethernet port to it, using commands
706 ovs-vsctl add-port br0 eth0
708 and as soon as I ran the "add-port" command I lost all connectivity
711 A: A physical Ethernet device that is part of an Open vSwitch bridge
712 should not have an IP address. If one does, then that IP address
713 will not be fully functional.
715 You can restore functionality by moving the IP address to an Open
716 vSwitch "internal" device, such as the network device named after
717 the bridge itself. For example, assuming that eth0's IP address is
718 192.168.128.5, you could run the commands below to fix up the
721 ifconfig eth0 0.0.0.0
722 ifconfig br0 192.168.128.5
724 (If your only connection to the machine running OVS is through the
725 IP address in question, then you would want to run all of these
726 commands on a single command line, or put them into a script.) If
727 there were any additional routes assigned to eth0, then you would
728 also want to use commands to adjust these routes to go through br0.
730 If you use DHCP to obtain an IP address, then you should kill the
731 DHCP client that was listening on the physical Ethernet interface
732 (e.g. eth0) and start one listening on the internal interface
733 (e.g. br0). You might still need to manually clear the IP address
734 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
736 There is no compelling reason why Open vSwitch must work this way.
737 However, this is the way that the Linux kernel bridge module has
738 always worked, so it's a model that those accustomed to Linux
739 bridging are already used to. Also, the model that most people
740 expect is not implementable without kernel changes on all the
741 versions of Linux that Open vSwitch supports.
743 By the way, this issue is not specific to physical Ethernet
744 devices. It applies to all network devices except Open vSwitch
747 ### Q: I created a bridge and added a couple of Ethernet ports to it,
748 ### using commands like these:
751 ovs-vsctl add-port br0 eth0
752 ovs-vsctl add-port br0 eth1
754 and now my network seems to have melted: connectivity is unreliable
755 (even connectivity that doesn't go through Open vSwitch), all the
756 LEDs on my physical switches are blinking, wireshark shows
757 duplicated packets, and CPU usage is very high.
759 A: More than likely, you've looped your network. Probably, eth0 and
760 eth1 are connected to the same physical Ethernet switch. This
761 yields a scenario where OVS receives a broadcast packet on eth0 and
762 sends it out on eth1, then the physical switch connected to eth1
763 sends the packet back on eth0, and so on forever. More complicated
764 scenarios, involving a loop through multiple switches, are possible
767 The solution depends on what you are trying to do:
769 - If you added eth0 and eth1 to get higher bandwidth or higher
770 reliability between OVS and your physical Ethernet switch,
771 use a bond. The following commands create br0 and then add
772 eth0 and eth1 as a bond:
775 ovs-vsctl add-bond br0 bond0 eth0 eth1
777 Bonds have tons of configuration options. Please read the
778 documentation on the Port table in ovs-vswitchd.conf.db(5)
781 Configuration for DPDK-enabled interfaces is slightly less
782 straightforward: see [INSTALL.DPDK.md].
784 - Perhaps you don't actually need eth0 and eth1 to be on the
785 same bridge. For example, if you simply want to be able to
786 connect each of them to virtual machines, then you can put
787 each of them on a bridge of its own:
790 ovs-vsctl add-port br0 eth0
793 ovs-vsctl add-port br1 eth1
795 and then connect VMs to br0 and br1. (A potential
796 disadvantage is that traffic cannot directly pass between br0
797 and br1. Instead, it will go out eth0 and come back in eth1,
800 - If you have a redundant or complex network topology and you
801 want to prevent loops, turn on spanning tree protocol (STP).
802 The following commands create br0, enable STP, and add eth0
803 and eth1 to the bridge. The order is important because you
804 don't want have to have a loop in your network even
808 ovs-vsctl set bridge br0 stp_enable=true
809 ovs-vsctl add-port br0 eth0
810 ovs-vsctl add-port br0 eth1
812 The Open vSwitch implementation of STP is not well tested.
813 Please report any bugs you observe, but if you'd rather avoid
814 acting as a beta tester then another option might be your
817 ### Q: I can't seem to use Open vSwitch in a wireless network.
819 A: Wireless base stations generally only allow packets with the source
820 MAC address of NIC that completed the initial handshake.
821 Therefore, without MAC rewriting, only a single device can
822 communicate over a single wireless link.
824 This isn't specific to Open vSwitch, it's enforced by the access
825 point, so the same problems will show up with the Linux bridge or
826 any other way to do bridging.
828 ### Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
830 A: PPP most commonly carries IP packets, but Open vSwitch works only
831 with Ethernet frames. The correct way to interface PPP to an
832 Ethernet network is usually to use routing instead of switching.
834 ### Q: Is there any documentation on the database tables and fields?
836 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
838 ### Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
839 I only see a datapath called "ovs-system". How can I see datapath
840 information about a particular bridge?
842 A: In version 1.9.0, OVS switched to using a single datapath that is
843 shared by all bridges of that type. The "ovs-appctl dpif/*"
844 commands provide similar functionality that is scoped by the bridge.
846 ### Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
847 see the port in the datapath?
849 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
850 cannot be loaded at the same time. It is likely that on your system the
851 Linux GRE module is already loaded and blocking OVS (to confirm, check
852 dmesg for errors regarding GRE registration). To fix this, unload all
853 GRE modules that appear in lsmod as well as the OVS kernel module. You
854 can then reload the OVS module following the directions in
855 [INSTALL.md], which will ensure that dependencies are satisfied.
857 ### Q: Open vSwitch does not seem to obey my packet filter rules.
859 A: It depends on mechanisms and configurations you want to use.
861 You cannot usefully use typical packet filters, like iptables, on
862 physical Ethernet ports that you add to an Open vSwitch bridge.
863 This is because Open vSwitch captures packets from the interface at
864 a layer lower below where typical packet-filter implementations
865 install their hooks. (This actually applies to any interface of
866 type "system" that you might add to an Open vSwitch bridge.)
868 You can usefully use typical packet filters on Open vSwitch
869 internal ports as they are mostly ordinary interfaces from the point
870 of view of packet filters.
872 For example, suppose you create a bridge br0 and add Ethernet port
873 eth0 to it. Then you can usefully add iptables rules to affect the
874 internal interface br0, but not the physical interface eth0. (br0
875 is also where you would add an IP address, as discussed elsewhere
878 For simple filtering rules, it might be possible to achieve similar
879 results by installing appropriate OpenFlow flows instead.
881 If the use of a particular packet filter setup is essential, Open
882 vSwitch might not be the best choice for you. On Linux, you might
883 want to consider using the Linux Bridge. (This is the only choice if
884 you want to use ebtables rules.) On NetBSD, you might want to
885 consider using the bridge(4) with BRIDGE_IPF option.
887 ### Q: It seems that Open vSwitch does nothing when I removed a port and
888 then immediately put it back. For example, consider that p1 is
889 a port of type=internal:
891 ovs-vsctl del-port br0 p1 -- \
893 set interface p1 type=internal
895 A: It's an expected behaviour.
897 If del-port and add-port happen in a single OVSDB transaction as
898 your example, Open vSwitch always "skips" the intermediate steps.
899 Even if they are done in multiple transactions, it's still allowed
900 for Open vSwitch to skip the intermediate steps and just implement
901 the overall effect. In both cases, your example would be turned
904 If you want to make Open vSwitch actually destroy and then re-create
905 the port for some side effects like resetting kernel setting for the
906 corresponding interface, you need to separate operations into multiple
907 OVSDB transactions and ensure that at least the first one does not have
908 --no-wait. In the following example, the first ovs-vsctl will block
909 until Open vSwitch reloads the new configuration and removes the port:
911 ovs-vsctl del-port br0 p1
912 ovs-vsctl add-port br0 p1 -- \
913 set interface p1 type=internal
915 ### Q: I want to add thousands of ports to an Open vSwitch bridge, but
916 it takes too long (minutes or hours) to do it with ovs-vsctl. How
919 A: If you add them one at a time with ovs-vsctl, it can take a long
920 time to add thousands of ports to an Open vSwitch bridge. This is
921 because every invocation of ovs-vsctl first reads the current
922 configuration from OVSDB. As the number of ports grows, this
923 starts to take an appreciable amount of time, and when it is
924 repeated thousands of times the total time becomes significant.
926 The solution is to add the ports in one invocation of ovs-vsctl (or
927 a small number of them). For example, using bash:
930 cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done
933 takes seconds, not minutes or hours, in the OVS sandbox environment.
935 ### Q: I created a bridge named br0. My bridge shows up in "ovs-vsctl
936 show", but "ovs-ofctl show br0" just prints "br0 is not a bridge
939 A: Open vSwitch wasn't able to create the bridge. Check the
940 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
941 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log).
943 In general, the Open vSwitch database reflects the desired
944 configuration state. ovs-vswitchd monitors the database and, when
945 it changes, reconfigures the system to reflect the new desired
946 state. This normally happens very quickly. Thus, a discrepancy
947 between the database and the actual state indicates that
948 ovs-vswitchd could not implement the configuration, and so one
949 should check the log to find out why. (Another possible cause is
950 that ovs-vswitchd is not running. This will make "ovs-vsctl"
951 commands hang, if they change the configuration, unless one
952 specifies "--no-wait".)
954 ### Q: I have a bridge br0. I added a new port vif1.0, and it shows
955 up in "ovs-vsctl show", but "ovs-vsctl list port" says that it has
956 OpenFlow port ("ofport") -1, and "ovs-ofctl show br0" doesn't show
959 A: Open vSwitch wasn't able to create the port. Check the
960 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
961 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log). Please
962 see the previous question for more information.
964 You may want to upgrade to Open vSwitch 2.3 (or later), in which
965 ovs-vsctl will immediately report when there is an issue creating a
968 ### Q: I created a tap device tap0, configured an IP address on it, and
969 added it to a bridge, like this:
972 ifconfig tap0 192.168.0.123
974 ovs-vsctl add-port br0 tap0
976 I expected that I could then use this IP address to contact other
977 hosts on the network, but it doesn't work. Why not?
979 A: The short answer is that this is a misuse of a "tap" device. Use
980 an "internal" device implemented by Open vSwitch, which works
981 differently and is designed for this use. To solve this problem
982 with an internal device, instead run:
985 ovs-vsctl add-port br0 int0 -- set Interface int0 type=internal
986 ifconfig int0 192.168.0.123
988 Even more simply, you can take advantage of the internal port that
989 every bridge has under the name of the bridge:
992 ifconfig br0 192.168.0.123
994 In more detail, a "tap" device is an interface between the Linux
995 (or *BSD) network stack and a user program that opens it as a
996 socket. When the "tap" device transmits a packet, it appears in
997 the socket opened by the userspace program. Conversely, when the
998 userspace program writes to the "tap" socket, the kernel TCP/IP
999 stack processes the packet as if it had been received by the "tap"
1002 Consider the configuration above. Given this configuration, if you
1003 "ping" an IP address in the 192.168.0.x subnet, the Linux kernel
1004 routing stack will transmit an ARP on the tap0 device. Open
1005 vSwitch userspace treats "tap" devices just like any other network
1006 device; that is, it doesn't open them as "tap" sockets. That means
1007 that the ARP packet will simply get dropped.
1009 You might wonder why the Open vSwitch kernel module doesn't
1010 intercept the ARP packet and bridge it. After all, Open vSwitch
1011 intercepts packets on other devices. The answer is that Open
1012 vSwitch only intercepts *received* packets, but this is a packet
1013 being transmitted. The same thing happens for all other types of
1014 network devices, except for Open vSwitch "internal" ports. If you,
1015 for example, add a physical Ethernet port to an OVS bridge,
1016 configure an IP address on a physical Ethernet port, and then issue
1017 a "ping" to an address in that subnet, the same thing happens: an
1018 ARP gets transmitted on the physical Ethernet port and Open vSwitch
1019 never sees it. (You should not do that, as documented at the
1020 beginning of this section.)
1022 It can make sense to add a "tap" device to an Open vSwitch bridge,
1023 if some userspace program (other than Open vSwitch) has opened the
1024 tap socket. This is the case, for example, if the "tap" device was
1025 created by KVM (or QEMU) to simulate a virtual NIC. In such a
1026 case, when OVS bridges a packet to the "tap" device, the kernel
1027 forwards that packet to KVM in userspace, which passes it along to
1028 the VM, and in the other direction, when the VM sends a packet, KVM
1029 writes it to the "tap" socket, which causes OVS to receive it and
1030 bridge it to the other OVS ports. Please note that in such a case
1031 no IP address is configured on the "tap" device (there is normally
1032 an IP address configured in the virtual NIC inside the VM, but this
1033 is not visible to the host Linux kernel or to Open vSwitch).
1035 There is one special case in which Open vSwitch does directly read
1036 and write "tap" sockets. This is an implementation detail of the
1037 Open vSwitch userspace switch, which implements its "internal"
1038 ports as Linux (or *BSD) "tap" sockets. In such a userspace
1039 switch, OVS receives packets sent on the "tap" device used to
1040 implement an "internal" port by reading the associated "tap"
1041 socket, and bridges them to the rest of the switch. In the other
1042 direction, OVS transmits packets bridged to the "internal" port by
1043 writing them to the "tap" socket, causing them to be processed by
1044 the kernel TCP/IP stack as if they had been received on the "tap"
1045 device. Users should not need to be concerned with this
1046 implementation detail.
1048 Open vSwitch has a network device type called "tap". This is
1049 intended only for implementing "internal" ports in the OVS
1050 userspace switch and should not be used otherwise. In particular,
1051 users should not configure KVM "tap" devices as type "tap" (use
1052 type "system", the default, instead).
1054 ### Q: I observe packet loss at the beginning of RFC2544 tests on a
1055 server running few hundred container apps bridged to OVS with traffic
1056 generated by HW traffic generator. How can I fix this?
1058 A: This is expected behavior on virtual switches. RFC2544 tests were
1059 designed for hardware switches, which don't have caches on the fastpath
1060 that need to be heated. Traffic generators in order to prime the switch
1061 use learning phase to heat the caches before sending the actual traffic
1062 in test phase. In case of OVS the cache is flushed quickly and to
1063 accommodate the traffic generator's delay between learning and test phase,
1064 the max-idle timeout settings should be changed to 50000 ms.
1066 ovs-vsctl --no-wait set Open_vSwitch . other_config:max-idle=50000
1070 ### Q: Does OVS support Quality of Service (QoS)?
1072 A: Yes. For traffic that egresses from a switch, OVS supports traffic
1073 shaping; for traffic that ingresses into a switch, OVS support
1074 policing. Policing is a simple form of quality-of-service that
1075 simply drops packets received in excess of the configured rate. Due
1076 to its simplicity, policing is usually less accurate and less
1077 effective than egress traffic shaping, which queues packets.
1079 Keep in mind that ingress and egress are from the perspective of the
1080 switch. That means that egress shaping limits the rate at which
1081 traffic is allowed to transmit from a physical interface, but not the
1082 rate at which traffic will be received on a virtual machine's VIF.
1083 For ingress policing, the behavior is the opposite.
1085 ### Q: How do I configure egress traffic shaping?
1087 A: Suppose that you want to set up bridge br0 connected to physical
1088 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
1089 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
1090 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
1091 could configure the bridge this way:
1095 add-port br0 eth0 -- \
1096 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
1097 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
1098 set port eth0 qos=@newqos -- \
1099 --id=@newqos create qos type=linux-htb \
1100 other-config:max-rate=1000000000 \
1101 queues:123=@vif10queue \
1102 queues:234=@vif20queue -- \
1103 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
1104 --id=@vif20queue create queue other-config:max-rate=20000000
1106 At this point, bridge br0 is configured with the ports and eth0 is
1107 configured with the queues that you need for QoS, but nothing is
1108 actually directing packets from vif1.0 or vif2.0 to the queues that
1109 we have set up for them. That means that all of the packets to
1110 eth0 are going to the "default queue", which is not what we want.
1112 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
1113 queues reserved for them:
1115 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
1116 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
1118 Each of the above flows matches on the input port, sets up the
1119 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
1120 executes the "normal" action, which performs the same switching
1121 that Open vSwitch would have done without any OpenFlow flows being
1122 present. (We know that vif1.0 and vif2.0 have OpenFlow port
1123 numbers 5 and 6, respectively, because we set their ofport_request
1124 columns above. If we had not done that, then we would have needed
1125 to find out their port numbers before setting up these flows.)
1127 Now traffic going from vif1.0 or vif2.0 to eth0 should be
1130 By the way, if you delete the bridge created by the above commands,
1133 ovs-vsctl del-br br0
1135 then that will leave one unreferenced QoS record and two
1136 unreferenced Queue records in the Open vSwich database. One way to
1137 clear them out, assuming you don't have other QoS or Queue records
1138 that you want to keep, is:
1140 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
1142 If you do want to keep some QoS or Queue records, or the Open
1143 vSwitch you are using is older than version 1.8 (which added the
1144 --all option), then you will have to destroy QoS and Queue records
1147 ### Q: How do I configure ingress policing?
1149 A: A policing policy can be configured on an interface to drop packets
1150 that arrive at a higher rate than the configured value. For example,
1151 the following commands will rate-limit traffic that vif1.0 may
1154 ovs-vsctl set interface vif1.0 ingress_policing_rate=10000
1155 ovs-vsctl set interface vif1.0 ingress_policing_burst=8000
1157 Traffic policing can interact poorly with some network protocols and
1158 can have surprising results. The "Ingress Policing" section of
1159 ovs-vswitchd.conf.db(5) discusses the issues in greater detail.
1161 ### Q: I configured Quality of Service (QoS) in my OpenFlow network by
1162 adding records to the QoS and Queue table, but the results aren't
1165 A: Did you install OpenFlow flows that use your queues? This is the
1166 primary way to tell Open vSwitch which queues you want to use. If
1167 you don't do this, then the default queue will be used, which will
1168 probably not have the effect you want.
1170 Refer to the previous question for an example.
1172 ### Q: I'd like to take advantage of some QoS feature that Open vSwitch
1173 doesn't yet support. How do I do that?
1175 A: Open vSwitch does not implement QoS itself. Instead, it can
1176 configure some, but not all, of the QoS features built into the
1177 Linux kernel. If you need some QoS feature that OVS cannot
1178 configure itself, then the first step is to figure out whether
1179 Linux QoS supports that feature. If it does, then you can submit a
1180 patch to support Open vSwitch configuration for that feature, or
1181 you can use "tc" directly to configure the feature in Linux. (If
1182 Linux QoS doesn't support the feature you want, then first you have
1183 to add that support to Linux.)
1185 ### Q: I configured QoS, correctly, but my measurements show that it isn't
1186 working as well as I expect.
1188 A: With the Linux kernel, the Open vSwitch implementation of QoS has
1191 - Open vSwitch configures a subset of Linux kernel QoS
1192 features, according to what is in OVSDB. It is possible that
1193 this code has bugs. If you believe that this is so, then you
1194 can configure the Linux traffic control (QoS) stack directly
1195 with the "tc" program. If you get better results that way,
1196 you can send a detailed bug report to bugs@openvswitch.org.
1198 It is certain that Open vSwitch cannot configure every Linux
1199 kernel QoS feature. If you need some feature that OVS cannot
1200 configure, then you can also use "tc" directly (or add that
1203 - The Open vSwitch implementation of OpenFlow allows flows to
1204 be directed to particular queues. This is pretty simple and
1205 unlikely to have serious bugs at this point.
1207 However, most problems with QoS on Linux are not bugs in Open
1208 vSwitch at all. They tend to be either configuration errors
1209 (please see the earlier questions in this section) or issues with
1210 the traffic control (QoS) stack in Linux. The Open vSwitch
1211 developers are not experts on Linux traffic control. We suggest
1212 that, if you believe you are encountering a problem with Linux
1213 traffic control, that you consult the tc manpages (e.g. tc(8),
1214 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
1215 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
1217 ### Q: Does Open vSwitch support OpenFlow meters?
1219 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
1220 OpenFlow meters. There is no implementation of meters in the Open
1221 vSwitch software switch (neither the kernel-based nor userspace
1226 ### Q: What's a VLAN?
1228 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
1229 partition a single switch into multiple switches. Suppose, for
1230 example, that you have two groups of machines, group A and group B.
1231 You want the machines in group A to be able to talk to each other,
1232 and you want the machine in group B to be able to talk to each
1233 other, but you don't want the machines in group A to be able to
1234 talk to the machines in group B. You can do this with two
1235 switches, by plugging the machines in group A into one switch and
1236 the machines in group B into the other switch.
1238 If you only have one switch, then you can use VLANs to do the same
1239 thing, by configuring the ports for machines in group A as VLAN
1240 "access ports" for one VLAN and the ports for group B as "access
1241 ports" for a different VLAN. The switch will only forward packets
1242 between ports that are assigned to the same VLAN, so this
1243 effectively subdivides your single switch into two independent
1244 switches, one for each group of machines.
1246 So far we haven't said anything about VLAN headers. With access
1247 ports, like we've described so far, no VLAN header is present in
1248 the Ethernet frame. This means that the machines (or switches)
1249 connected to access ports need not be aware that VLANs are
1250 involved, just like in the case where we use two different physical
1253 Now suppose that you have a whole bunch of switches in your
1254 network, instead of just one, and that some machines in group A are
1255 connected directly to both switches 1 and 2. To allow these
1256 machines to talk to each other, you could add an access port for
1257 group A's VLAN to switch 1 and another to switch 2, and then
1258 connect an Ethernet cable between those ports. That works fine,
1259 but it doesn't scale well as the number of switches and the number
1260 of VLANs increases, because you use up a lot of valuable switch
1261 ports just connecting together your VLANs.
1263 This is where VLAN headers come in. Instead of using one cable and
1264 two ports per VLAN to connect a pair of switches, we configure a
1265 port on each switch as a VLAN "trunk port". Packets sent and
1266 received on a trunk port carry a VLAN header that says what VLAN
1267 the packet belongs to, so that only two ports total are required to
1268 connect the switches, regardless of the number of VLANs in use.
1269 Normally, only switches (either physical or virtual) are connected
1270 to a trunk port, not individual hosts, because individual hosts
1271 don't expect to see a VLAN header in the traffic that they receive.
1273 None of the above discussion says anything about particular VLAN
1274 numbers. This is because VLAN numbers are completely arbitrary.
1275 One must only ensure that a given VLAN is numbered consistently
1276 throughout a network and that different VLANs are given different
1277 numbers. (That said, VLAN 0 is usually synonymous with a packet
1278 that has no VLAN header, and VLAN 4095 is reserved.)
1280 ### Q: VLANs don't work.
1282 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
1283 bugs. If you are having problems with VLANs that you suspect to be
1284 driver related, then you have several options:
1286 - Upgrade to Linux 3.3 or later.
1288 - Build and install a fixed version of the particular driver
1289 that is causing trouble, if one is available.
1291 - Use a NIC whose driver does not have VLAN problems.
1293 - Use "VLAN splinters", a feature in Open vSwitch 1.4 upto 2.5
1294 that works around bugs in kernel drivers. To enable VLAN
1295 splinters on interface eth0, use the command:
1297 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
1299 For VLAN splinters to be effective, Open vSwitch must know
1300 which VLANs are in use. See the "VLAN splinters" section in
1301 the Interface table in ovs-vswitchd.conf.db(5) for details on
1302 how Open vSwitch infers in-use VLANs.
1304 VLAN splinters increase memory use and reduce performance, so
1305 use them only if needed.
1307 - Apply the "vlan workaround" patch from the XenServer kernel
1308 patch queue, build Open vSwitch against this patched kernel,
1309 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
1310 workaround for each interface whose driver is buggy.
1312 (This is a nontrivial exercise, so this option is included
1313 only for completeness.)
1315 It is not always easy to tell whether a Linux kernel driver has
1316 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
1317 can help you test. See their manpages for details. Of the two
1318 utilities, ovs-test(8) is newer and more thorough, but
1319 ovs-vlan-test(8) may be easier to use.
1321 ### Q: VLANs still don't work. I've tested the driver so I know that it's OK.
1323 A: Do you have VLANs enabled on the physical switch that OVS is
1324 attached to? Make sure that the port is configured to trunk the
1325 VLAN or VLANs that you are using with OVS.
1327 ### Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
1328 and to its destination host, but OVS seems to drop incoming return
1331 A: It's possible that you have the VLAN configured on your physical
1332 switch as the "native" VLAN. In this mode, the switch treats
1333 incoming packets either tagged with the native VLAN or untagged as
1334 part of the native VLAN. It may also send outgoing packets in the
1335 native VLAN without a VLAN tag.
1337 If this is the case, you have two choices:
1339 - Change the physical switch port configuration to tag packets
1340 it forwards to OVS with the native VLAN instead of forwarding
1343 - Change the OVS configuration for the physical port to a
1344 native VLAN mode. For example, the following sets up a
1345 bridge with port eth0 in "native-tagged" mode in VLAN 9:
1347 ovs-vsctl add-br br0
1348 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
1350 In this situation, "native-untagged" mode will probably work
1351 equally well. Refer to the documentation for the Port table
1352 in ovs-vswitchd.conf.db(5) for more information.
1354 ### Q: I added a pair of VMs on different VLANs, like this:
1356 ovs-vsctl add-br br0
1357 ovs-vsctl add-port br0 eth0
1358 ovs-vsctl add-port br0 tap0 tag=9
1359 ovs-vsctl add-port br0 tap1 tag=10
1361 but the VMs can't access each other, the external network, or the
1364 A: It is to be expected that the VMs can't access each other. VLANs
1365 are a means to partition a network. When you configured tap0 and
1366 tap1 as access ports for different VLANs, you indicated that they
1367 should be isolated from each other.
1369 As for the external network and the Internet, it seems likely that
1370 the machines you are trying to access are not on VLAN 9 (or 10) and
1371 that the Internet is not available on VLAN 9 (or 10).
1373 ### Q: I added a pair of VMs on the same VLAN, like this:
1375 ovs-vsctl add-br br0
1376 ovs-vsctl add-port br0 eth0
1377 ovs-vsctl add-port br0 tap0 tag=9
1378 ovs-vsctl add-port br0 tap1 tag=9
1380 The VMs can access each other, but not the external network or the
1383 A: It seems likely that the machines you are trying to access in the
1384 external network are not on VLAN 9 and that the Internet is not
1385 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
1386 trunk VLAN on the upstream switch port to which eth0 is connected.
1388 ### Q: Can I configure an IP address on a VLAN?
1390 A: Yes. Use an "internal port" configured as an access port. For
1391 example, the following configures IP address 192.168.0.7 on VLAN 9.
1392 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
1393 they have an 802.1Q header with VLAN 9. Conversely, traffic
1394 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
1397 ovs-vsctl add-br br0
1398 ovs-vsctl add-port br0 eth0
1399 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1400 ifconfig vlan9 192.168.0.7
1402 See also the following question.
1404 ### Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
1407 ovs-vsctl add-br br0
1408 ovs-vsctl add-port br0 eth0
1409 ifconfig br0 192.168.0.5
1410 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1411 ifconfig vlan9 192.168.0.9
1413 but other hosts that are only on VLAN 0 can reach the IP address
1414 configured on VLAN 9. What's going on?
1416 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
1417 approaches to IP address handling in Internet hosts:
1419 - In the "Strong ES Model", where an ES is a host ("End
1420 System"), an IP address is primarily associated with a
1421 particular interface. The host discards packets that arrive
1422 on interface A if they are destined for an IP address that is
1423 configured on interface B. The host never sends packets from
1424 interface A using a source address configured on interface B.
1426 - In the "Weak ES Model", an IP address is primarily associated
1427 with a host. The host accepts packets that arrive on any
1428 interface if they are destined for any of the host's IP
1429 addresses, even if the address is configured on some
1430 interface other than the one on which it arrived. The host
1431 does not restrict itself to sending packets from an IP
1432 address associated with the originating interface.
1434 Linux uses the weak ES model. That means that when packets
1435 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
1436 br0, the kernel IP stack accepts them there for the VLAN 9 IP
1437 address, even though they were not received on vlan9, the network
1440 To simulate the strong ES model on Linux, one may add iptables rule
1441 to filter packets based on source and destination address and
1442 adjust ARP configuration with sysctls.
1444 BSD uses the strong ES model.
1446 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1448 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1449 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1450 implementation of the OpenFlow "normal switching" action. By
1451 default, when Open vSwitch isn't connected to a controller and
1452 nothing has been manually configured in the flow table, all traffic
1453 goes through the "normal switching" action. But, if you set up
1454 OpenFlow flows on your own, through a controller or using ovs-ofctl
1455 or through other means, then you have to implement VLAN handling
1458 You can use "normal switching" as a component of your OpenFlow
1459 actions, e.g. by putting "normal" into the lists of actions on
1460 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1461 controller. In situations where this is not suitable, you can
1462 implement VLAN handling yourself, e.g.:
1464 - If a packet comes in on an access port, and the flow table
1465 needs to send it out on a trunk port, then the flow can add
1466 the appropriate VLAN tag with the "mod_vlan_vid" action.
1468 - If a packet comes in on a trunk port, and the flow table
1469 needs to send it out on an access port, then the flow can
1470 strip the VLAN tag with the "strip_vlan" action.
1472 ### Q: I configured ports on a bridge as access ports with different VLAN
1475 ovs-vsctl add-br br0
1476 ovs-vsctl set-controller br0 tcp:192.168.0.10:6653
1477 ovs-vsctl add-port br0 eth0
1478 ovs-vsctl add-port br0 tap0 tag=9
1479 ovs-vsctl add-port br0 tap1 tag=10
1481 but the VMs running behind tap0 and tap1 can still communicate,
1482 that is, they are not isolated from each other even though they are
1485 A: Do you have a controller configured on br0 (as the commands above
1486 do)? If so, then this is a variant on the previous question, "My
1487 OpenFlow controller doesn't see the VLANs that I expect," and you
1488 can refer to the answer there for more information.
1490 ### Q: How MAC learning works with VLANs?
1492 A: Open vSwitch implements Independent VLAN Learning (IVL) for
1493 OFPP_NORMAL action. I.e. it logically has separate learning tables
1499 ### Q: What's a VXLAN?
1501 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1502 to solve the scaling challenges of VLAN networks in a multi-tenant
1503 environment. VXLAN is an overlay network which transports an L2 network
1504 over an existing L3 network. For more information on VXLAN, please see
1507 http://tools.ietf.org/html/rfc7348
1509 ### Q: How much of the VXLAN protocol does Open vSwitch currently support?
1511 A: Open vSwitch currently supports the framing format for packets on the
1512 wire. There is currently no support for the multicast aspects of VXLAN.
1513 To get around the lack of multicast support, it is possible to
1514 pre-provision MAC to IP address mappings either manually or from a
1517 ### Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1520 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1521 is 4789. However, it is possible to configure the destination UDP port
1522 manually on a per-VXLAN tunnel basis. An example of this configuration is
1525 ovs-vsctl add-br br0
1526 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1527 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1528 options:dst_port=8472
1533 ### Q: What versions of OpenFlow does Open vSwitch support?
1535 A: The following table lists the versions of OpenFlow supported by
1536 each version of Open vSwitch:
1538 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5 OF1.6
1539 ###============ ===== ===== ===== ===== ===== ===== =====
1540 1.9 and earlier yes --- --- --- --- --- ---
1541 1.10 yes --- [*] [*] --- --- ---
1542 1.11 yes --- [*] [*] --- --- ---
1543 2.0 yes [*] [*] [*] --- --- ---
1544 2.1 yes [*] [*] [*] --- --- ---
1545 2.2 yes [*] [*] [*] [%] [*] ---
1546 2.3 yes yes yes yes [*] [*] ---
1547 2.4 yes yes yes yes [*] [*] ---
1548 2.5 yes yes yes yes [*] [*] [*]
1550 [*] Supported, with one or more missing features.
1551 [%] Experimental, unsafe implementation.
1553 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default
1554 in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1,
1555 1.2, and 1.3 must be enabled manually in ovs-vswitchd.
1557 Some versions of OpenFlow are supported with missing features and
1558 therefore not enabled by default: OpenFlow 1.4 and 1.5, in Open
1559 vSwitch 2.3 and later, as well as OpenFlow 1.6 in Open vSwitch 2.5
1560 and later. Also, the OpenFlow 1.6 specification is still under
1561 development and thus subject to change.
1563 In any case, the user may override the default:
1565 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0:
1567 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1569 - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0:
1571 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15
1573 - To enable only OpenFlow 1.0 on bridge br0:
1575 ovs-vsctl set bridge br0 protocols=OpenFlow10
1577 All current versions of ovs-ofctl enable only OpenFlow 1.0 by
1578 default. Use the -O option to enable support for later versions of
1579 OpenFlow in ovs-ofctl. For example:
1581 ovs-ofctl -O OpenFlow13 dump-flows br0
1583 (Open vSwitch 2.2 had an experimental implementation of OpenFlow
1584 1.4 that could cause crashes. We don't recommend enabling it.)
1586 [OPENFLOW-1.1+.md] in the Open vSwitch source tree tracks support for
1587 OpenFlow 1.1 and later features. When support for OpenFlow 1.4 and
1588 1.5 is solidly implemented, Open vSwitch will enable those version
1591 ### Q: Does Open vSwitch support MPLS?
1593 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1594 these versions can match on MPLS Ethernet types, but they cannot
1595 match, push, or pop MPLS labels, nor can they look past MPLS labels
1596 into the encapsulated packet.
1598 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1599 for MPLS. With the userspace datapath only, these versions can
1600 match, push, or pop a single MPLS label, but they still cannot look
1601 past MPLS labels (even after popping them) into the encapsulated
1602 packet. Kernel datapath support is unchanged from earlier
1605 Open vSwitch version 2.3 can match, push, or pop a single MPLS
1606 label and look past the MPLS label into the encapsulated packet.
1607 Both userspace and kernel datapaths will be supported, but MPLS
1608 processing always happens in userspace either way, so kernel
1609 datapath performance will be disappointing.
1611 Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS
1612 labels and look past the MPLS label into the encapsulated packet.
1613 It will have kernel support for MPLS, yielding improved
1616 ### Q: I'm getting "error type 45250 code 0". What's that?
1618 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1619 vSwitch uses this extension when it must report an error to an
1620 OpenFlow controller but no standard OpenFlow error code is
1623 Open vSwitch logs the errors that it sends to controllers, so the
1624 easiest thing to do is probably to look at the ovs-vswitchd log to
1625 find out what the error was.
1627 If you want to dissect the extended error message yourself, the
1628 format is documented in include/openflow/nicira-ext.h in the Open
1629 vSwitch source distribution. The extended error codes are
1630 documented in include/openvswitch/ofp-errors.h.
1632 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1633 doesn't actually appear through the OpenFlow connection, even
1634 though I know that it's going through.
1635 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1636 to apply to certain traffic, especially traffic between OVS and
1637 the controller itself.
1639 A: By default, Open vSwitch assumes that OpenFlow controllers are
1640 connected "in-band", that is, that the controllers are actually
1641 part of the network that is being controlled. In in-band mode,
1642 Open vSwitch sets up special "hidden" flows to make sure that
1643 traffic can make it back and forth between OVS and the controllers.
1644 These hidden flows are higher priority than any flows that can be
1645 set up through OpenFlow, and they are not visible through normal
1646 OpenFlow flow table dumps.
1648 Usually, the hidden flows are desirable and helpful, but
1649 occasionally they can cause unexpected behavior. You can view the
1650 full OpenFlow flow table, including hidden flows, on bridge br0
1653 ovs-appctl bridge/dump-flows br0
1655 to help you debug. The hidden flows are those with priorities
1656 greater than 65535 (the maximum priority that can be set with
1659 The DESIGN file at the top level of the Open vSwitch source
1660 distribution describes the in-band model in detail.
1662 If your controllers are not actually in-band (e.g. they are on
1663 localhost via 127.0.0.1, or on a separate network), then you should
1664 configure your controllers in "out-of-band" mode. If you have one
1665 controller on bridge br0, then you can configure out-of-band mode
1668 ovs-vsctl set controller br0 connection-mode=out-of-band
1670 ### Q: I configured all my controllers for out-of-band control mode but
1671 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1673 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1674 set-manager"). By default, Open vSwitch assumes that managers need
1675 in-band rules set up on every bridge. You can disable these rules
1678 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1680 This actually disables in-band control entirely for the bridge, as
1681 if all the bridge's controllers were configured for out-of-band
1684 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1686 A: See answer under "VLANs", above.
1688 ### Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1689 but I got a funny message like this:
1691 ofp_util|INFO|normalization changed ofp_match, details:
1692 ofp_util|INFO| pre: nw_dst=192.168.0.1
1695 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1696 match had disappeared, so that the flow ends up matching every
1699 A: The term "normalization" in the log message means that a flow
1700 cannot match on an L3 field without saying what L3 protocol is in
1701 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1702 so the L3 field match was dropped.
1704 In this case, the L3 protocol could be IP or ARP. A correct
1705 command for each possibility is, respectively:
1707 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1711 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1713 Similarly, a flow cannot match on an L4 field without saying what
1714 L4 protocol is in use. For example, the flow match "tp_src=1234"
1715 is, by itself, meaningless and will be ignored. Instead, to match
1716 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1717 source port 1234, write "udp,tp_src=1234".
1719 ### Q: How can I figure out the OpenFlow port number for a given port?
1721 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1722 respond with an OFPT_FEATURES_REPLY that, among other information,
1723 includes a mapping between OpenFlow port names and numbers. From a
1724 command prompt, "ovs-ofctl show br0" makes such a request and
1725 prints the response for switch br0.
1727 The Interface table in the Open vSwitch database also maps OpenFlow
1728 port names to numbers. To print the OpenFlow port number
1729 associated with interface eth0, run:
1731 ovs-vsctl get Interface eth0 ofport
1733 You can print the entire mapping with:
1735 ovs-vsctl -- --columns=name,ofport list Interface
1737 but the output mixes together interfaces from all bridges in the
1738 database, so it may be confusing if more than one bridge exists.
1740 In the Open vSwitch database, ofport value -1 means that the
1741 interface could not be created due to an error. (The Open vSwitch
1742 log should indicate the reason.) ofport value [] (the empty set)
1743 means that the interface hasn't been created yet. The latter is
1744 normally an intermittent condition (unless ovs-vswitchd is not
1747 ### Q: I added some flows with my controller or with ovs-ofctl, but when I
1748 run "ovs-dpctl dump-flows" I don't see them.
1750 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1751 won't display the information that you want. You want to use
1752 "ovs-ofctl dump-flows" instead.
1754 ### Q: It looks like each of the interfaces in my bonded port shows up
1755 as an individual OpenFlow port. Is that right?
1757 A: Yes, Open vSwitch makes individual bond interfaces visible as
1758 OpenFlow ports, rather than the bond as a whole. The interfaces
1759 are treated together as a bond for only a few purposes:
1761 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1762 controller is not configured, this happens implicitly to
1765 - Mirrors configured for output to a bonded port.
1767 It would make a lot of sense for Open vSwitch to present a bond as
1768 a single OpenFlow port. If you want to contribute an
1769 implementation of such a feature, please bring it up on the Open
1770 vSwitch development mailing list at dev@openvswitch.org.
1772 ### Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1773 multiple hosts, and other components. The behavior isn't what I
1776 A: To debug network behavior problems, trace the path of a packet,
1777 hop-by-hop, from its origin in one host to a remote host. If
1778 that's correct, then trace the path of the response packet back to
1781 The open source tool called "plotnetcfg" can help to understand the
1782 relationship between the networking devices on a single host.
1784 Usually a simple ICMP echo request and reply ("ping") packet is
1785 good enough. Start by initiating an ongoing "ping" from the origin
1786 host to a remote host. If you are tracking down a connectivity
1787 problem, the "ping" will not display any successful output, but
1788 packets are still being sent. (In this case the packets being sent
1789 are likely ARP rather than ICMP.)
1791 Tools available for tracing include the following:
1793 - "tcpdump" and "wireshark" for observing hops across network
1794 devices, such as Open vSwitch internal devices and physical
1797 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1798 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1799 These tools allow one to observe the actions being taken on
1800 packets in ongoing flows.
1802 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1803 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1804 documentation, and "Why are there so many different ways to
1805 dump flows?" above for some background.
1807 - "ovs-appctl ofproto/trace" to observe the logic behind how
1808 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1809 documentation. You can out more details about a given flow
1810 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1811 a flow from the output into an "ovs-appctl ofproto/trace"
1814 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1815 observe what goes on at these physical hops.
1817 Starting at the origin of a given packet, observe the packet at
1818 each hop in turn. For example, in one plausible scenario, you
1821 1. "tcpdump" the "eth" interface through which an ARP egresses
1822 a VM, from inside the VM.
1824 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1825 ingresses the host machine.
1827 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1828 the host interface through which the ARP egresses the
1829 physical machine. You may need to use "ovs-dpctl show" to
1830 interpret the port numbers. If the output seems surprising,
1831 you can use "ovs-appctl ofproto/trace" to observe details of
1832 how ovs-vswitchd determined the actions in the "ovs-dpctl
1835 4. "tcpdump" the "eth" interface through which the ARP egresses
1836 the physical machine.
1838 5. "tcpdump" the "eth" interface through which the ARP
1839 ingresses the physical machine, at the remote host that
1842 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1843 remote host that receives the ARP and observe the VM "vif"
1844 or "tap" interface to which the flow is directed. Again,
1845 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1847 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1850 8. "tcpdump" the "eth" interface through which the ARP
1851 ingresses a VM, from inside the VM.
1853 It is likely that during one of these steps you will figure out the
1854 problem. If not, then follow the ARP reply back to the origin, in
1857 ### Q: How do I make a flow drop packets?
1859 A: To drop a packet is to receive it without forwarding it. OpenFlow
1860 explicitly specifies forwarding actions. Thus, a flow with an
1861 empty set of actions does not forward packets anywhere, causing
1862 them to be dropped. You can specify an empty set of actions with
1863 "actions=" on the ovs-ofctl command line. For example:
1865 ovs-ofctl add-flow br0 priority=65535,actions=
1867 would cause every packet entering switch br0 to be dropped.
1869 You can write "drop" explicitly if you like. The effect is the
1870 same. Thus, the following command also causes every packet
1871 entering switch br0 to be dropped:
1873 ovs-ofctl add-flow br0 priority=65535,actions=drop
1875 "drop" is not an action, either in OpenFlow or Open vSwitch.
1876 Rather, it is only a way to say that there are no actions.
1878 ### Q: I added a flow to send packets out the ingress port, like this:
1880 ovs-ofctl add-flow br0 in_port=2,actions=2
1882 but OVS drops the packets instead.
1884 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1885 out its ingress port. The rationale is that dropping these packets
1886 makes it harder to loop the network. Sometimes this behavior can
1887 even be convenient, e.g. it is often the desired behavior in a flow
1888 that forwards a packet to several ports ("floods" the packet).
1890 Sometimes one really needs to send a packet out its ingress port
1891 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1892 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1894 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1896 This also works in some circumstances where the flow doesn't match
1897 on the input port. For example, if you know that your switch has
1898 five ports numbered 2 through 6, then the following will send every
1899 received packet out every port, even its ingress port:
1901 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1905 ovs-ofctl add-flow br0 actions=all,in_port
1907 Sometimes, in complicated flow tables with multiple levels of
1908 "resubmit" actions, a flow needs to output to a particular port
1909 that may or may not be the ingress port. It's difficult to take
1910 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1911 provides, as an OpenFlow extension, the ability to modify the
1912 in_port field. Whatever value is currently in the in_port field is
1913 the port to which outputs will be dropped, as well as the
1914 destination for OFPP_IN_PORT. This means that the following will
1915 reliably output to port 2 or to ports 2 through 6, respectively:
1917 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1918 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1920 If the input port is important, then one may save and restore it on
1923 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1924 load:0->NXM_OF_IN_PORT[],\
1926 pop:NXM_OF_IN_PORT[]
1928 ### Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1929 on port 2. I set up flows to forward only traffic destined to the
1930 other host and drop other traffic, like this:
1932 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1933 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1934 priority=0,actions=drop
1936 But it doesn't work--I don't get any connectivity when I do this.
1939 A: These flows drop the ARP packets that IP hosts use to establish IP
1940 connectivity over Ethernet. To solve the problem, add flows to
1941 allow ARP to pass between the hosts:
1943 priority=5,in_port=1,arp,actions=2
1944 priority=5,in_port=2,arp,actions=1
1946 This issue can manifest other ways, too. The following flows that
1947 match on Ethernet addresses instead of IP addresses will also drop
1948 ARP packets, because ARP requests are broadcast instead of being
1949 directed to a specific host:
1951 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1952 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1953 priority=0,actions=drop
1955 The solution already described above will also work in this case.
1956 It may be better to add flows to allow all multicast and broadcast
1959 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1960 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1962 ### Q: My bridge disconnects from my controller on add-port/del-port.
1964 A: Reconfiguring your bridge can change your bridge's datapath-id because
1965 Open vSwitch generates datapath-id from the MAC address of one of its ports.
1966 In that case, Open vSwitch disconnects from controllers because there's
1967 no graceful way to notify controllers about the change of datapath-id.
1969 To avoid the behaviour, you can configure datapath-id manually.
1971 ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef
1973 ### Q: My controller complains that OVS is not buffering packets.
1976 A: "Packet buffering" is an optional OpenFlow feature, and controllers
1977 should detect how many "buffers" an OpenFlow switch implements. It
1978 was recently noticed that OVS implementation of the buffering
1979 feature was not compliant to OpenFlow specifications. Rather than
1980 fix it and risk controller incompatibility, the buffering feature
1981 is removed as of OVS 2.7. Controllers are already expected to work
1982 properly in cases where the switch can not buffer packets, but
1983 sends full packets in "packet-in" messages instead, so this change
1984 should not affect existing users. After the change OVS always
1985 sends the 'buffer_id' as 0xffffffff in "packet-in" messages and
1986 will send an error response if any other value of this field is
1987 included in a "packet-out" or a "flow mod" sent by a controller.
1989 ### Q: How does OVS divide flows among buckets in an OpenFlow "select" group?
1991 A: In Open vSwitch 2.3 and earlier, Open vSwitch used the destination
1992 Ethernet address to choose a bucket in a select group.
1994 Open vSwitch 2.4 and later by default hashes the source and
1995 destination Ethernet address, VLAN ID, Ethernet type, IPv4/v6
1996 source and destination address and protocol, and for TCP and SCTP
1997 only, the source and destination ports. The hash is "symmetric",
1998 meaning that exchanging source and destination addresses does not
1999 change the bucket selection.
2001 Select groups in Open vSwitch 2.4 and later can be configured to
2002 use a different hash function, using a Netronome extension to the
2003 OpenFlow 1.5+ group_mod message. For more information, see
2004 Documentation/group-selection-method-property.txt in the Open
2005 vSwitch source tree. (OpenFlow 1.5 support in Open vSwitch is still
2008 ### Q: I added a flow to accept packets on VLAN 123 and output them on
2011 ovs-ofctl add-flow br0 dl_vlan=123,actions=output:1,mod_vlan_vid:456
2013 but the packets are actually being output in VLAN 123. Why?
2015 A: OpenFlow actions are executed in the order specified. Thus, the
2016 actions above first output the packet, then change its VLAN. Since
2017 the output occurs before changing the VLAN, the change in VLAN will
2018 have no visible effect.
2020 To solve this and similar problems, order actions so that changes
2021 to headers happen before output, e.g.:
2023 ovs-ofctl add-flow br0 dl_vlan=123,actions=mod_vlan_vid:456,output:1
2025 ### Q: The "learn" action can't learn the action I want, can you improve it?
2027 A: By itself, the "learn" action can only put two kinds of actions
2028 into the flows that it creates: "load" and "output" actions. If
2029 "learn" is used in isolation, these are severe limits.
2031 However, "learn" is not meant to be used in isolation. It is a
2032 primitive meant to be used together with other Open vSwitch
2033 features to accomplish a task. Its existing features are enough to
2034 accomplish most tasks.
2036 Here is an outline of a typical pipeline structure that allows for
2037 versatile behavior using "learn":
2039 - Flows in table A contain a "learn" action, that populates flows
2040 in table L, that use a "load" action to populate register R
2041 with information about what was learned.
2043 - Flows in table B contain two sequential resubmit actions: one
2044 to table L and another one to table B+1.
2046 - Flows in table B+1 match on register R and act differently
2047 depending on what the flows in table L loaded into it.
2049 This approach can be used to implement many "learn"-based features.
2052 - Resubmit to a table selected based on learned information, e.g. see:
2053 http://openvswitch.org/pipermail/discuss/2016-June/021694.html
2055 - MAC learning in the middle of a pipeline, as described in
2058 - TCP state based firewalling, by learning outgoing connections
2059 based on SYN packets and matching them up with incoming
2062 - At least some of the features described in T. A. Hoff,
2063 "Extending Open vSwitch to Facilitate Creation of Stateful SDN
2069 ### Q: How do I implement a new OpenFlow message?
2071 A: Add your new message to "enum ofpraw" and "enum ofptype" in
2072 lib/ofp-msgs.h, following the existing pattern. Then recompile and
2073 fix all of the new warnings, implementing new functionality for the
2074 new message as needed. (If you configure with --enable-Werror, as
2075 described in [INSTALL.md], then it is impossible to miss any warnings.)
2077 If you need to add an OpenFlow vendor extension message for a
2078 vendor that doesn't yet have any extension messages, then you will
2079 also need to edit build-aux/extract-ofp-msgs.
2081 ### Q: How do I add support for a new field or header?
2083 A: Add new members for your field to "struct flow" in lib/flow.h, and
2084 add new enumerations for your new field to "enum mf_field_id" in
2085 lib/meta-flow.h, following the existing pattern. Also, add support
2086 to miniflow_extract() in lib/flow.c for extracting your new field
2087 from a packet into struct miniflow, and to nx_put_raw() in
2088 lib/nx-match.c to output your new field in OXM matches. Then
2089 recompile and fix all of the new warnings, implementing new
2090 functionality for the new field or header as needed. (If you
2091 configure with --enable-Werror, as described in [INSTALL.md], then
2092 it is impossible to miss any warnings.)
2094 If you want kernel datapath support for your new field, you also
2095 need to modify the kernel module for the operating systems you are
2096 interested in. This isn't mandatory, since fields understood only
2097 by userspace work too (with a performance penalty), so it's
2098 reasonable to start development without it. If you implement
2099 kernel module support for Linux, then the Linux kernel "netdev"
2100 mailing list is the place to submit that support first; please read
2101 up on the Linux kernel development process separately. The Windows
2102 datapath kernel module support, on the other hand, is maintained
2103 within the OVS tree, so patches for that can go directly to
2106 ### Q: How do I add support for a new OpenFlow action?
2108 A: Add your new action to "enum ofp_raw_action_type" in
2109 lib/ofp-actions.c, following the existing pattern. Then recompile
2110 and fix all of the new warnings, implementing new functionality for
2111 the new action as needed. (If you configure with --enable-Werror,
2112 as described in [INSTALL.md], then it is impossible to miss any
2115 If you need to add an OpenFlow vendor extension action for a vendor
2116 that doesn't yet have any extension actions, then you will also
2117 need to edit build-aux/extract-ofp-actions.
2123 bugs@openvswitch.org
2124 http://openvswitch.org/
2126 [PORTING.md]:PORTING.md
2127 [WHY-OVS.md]:WHY-OVS.md
2128 [INSTALL.md]:INSTALL.md
2129 [OPENFLOW-1.1+.md]:OPENFLOW-1.1+.md
2130 [INSTALL.DPDK.md]:INSTALL.DPDK.md
2131 [Tutorial.md]:tutorial/Tutorial.md
2132 [release-process.md]:Documentation/release-process.md
projects / ovs.git / blob