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. Some features of
184 this module rely on functionality not available in older
185 kernels: in this case the minumum Linux version (against
186 which the feature can be compiled) is listed.
188 * *Userspace*: Also known as DPDK, dpif-netdev or dummy datapath. It is the
189 only datapath that works on NetBSD and FreeBSD.
191 * *Hyper-V*: Also known as the Windows datapath.
193 The following table lists the datapath supported features from
194 an Open vSwitch user's perspective.
196 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
197 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
198 Connection tracking | 4.3 | 3.10 | NO | NO |
199 Tunnel - LISP | NO | YES | NO | NO |
200 Tunnel - STT | NO | 3.5 | NO | YES |
201 Tunnel - GRE | 3.11 | YES | YES | YES |
202 Tunnel - VXLAN | 3.12 | YES | YES | YES |
203 Tunnel - Geneve | 3.18 | YES | YES | NO |
204 QoS - Policing | YES | YES | NO | NO |
205 QoS - Shaping | YES | YES | NO | NO |
206 sFlow | YES | YES | YES | NO |
207 Set action | YES | YES | YES | PARTIAL |
208 NIC Bonding | YES | YES | YES | NO |
209 Multiple VTEPs | YES | YES | YES | NO |
212 * Only a limited set of flow fields is modifiable via the set action by the
214 * The Hyper-V datapath only supports one physical NIC per datapath. This is
215 why bonding is not supported.
216 * The Hyper-V datapath can have at most one IP address configured as a
219 The following table lists features that do not *directly* impact an
220 Open vSwitch user, e.g. because their absence can be hidden by the ofproto
221 layer (usually this comes with a performance penalty).
223 Feature | Linux upstream | Linux OVS tree | Userspace | Hyper-V |
224 ----------------------|:--------------:|:--------------:|:---------:|:-------:|
225 SCTP flows | 3.12 | YES | YES | YES |
226 MPLS | 3.19 | YES | YES | YES |
227 UFID | 4.0 | YES | YES | NO |
228 Megaflows | 3.12 | YES | YES | NO |
229 Masked set action | 4.0 | YES | YES | NO |
230 Recirculation | 3.19 | YES | YES | YES |
231 TCP flags matching | 3.13 | YES | YES | NO |
232 Validate flow actions | YES | YES | N/A | NO |
233 Multiple datapaths | YES | YES | YES | NO |
234 Tunnel TSO - STT | N/A | YES | NO | YES |
236 ### Q: What DPDK version does each Open vSwitch release work with?
238 A: The following table lists the DPDK version against which the
239 given versions of Open vSwitch will successfully build.
241 | Open vSwitch | DPDK
242 |:------------:|:-----:
249 ### Q: I get an error like this when I configure Open vSwitch:
251 configure: error: Linux kernel in <dir> is version <x>, but
252 version newer than <y> is not supported (please refer to the
257 A: You have the following options:
259 - Use the Linux kernel module supplied with the kernel that you are
260 using. (See also the following FAQ.)
262 - If there is a newer released version of Open vSwitch, consider
263 building that one, because it may support the kernel that you are
264 building against. (To find out, consult the table in the
267 - The Open vSwitch "master" branch may support the kernel that you
268 are using, so consider building the kernel module from "master".
270 All versions of Open vSwitch userspace are compatible with all
271 versions of the Open vSwitch kernel module, so you do not have to
272 use the kernel module from one source along with the userspace
273 programs from the same source.
275 ### Q: What features are not available in the Open vSwitch kernel datapath that ships as part of the upstream Linux kernel?
277 A: The kernel module in upstream Linux does not include support for
278 LISP. Work is in progress to add support for LISP to the upstream
279 Linux version of the Open vSwitch kernel module. For now, if you
280 need this feature, use the kernel module from the Open vSwitch
281 distribution instead of the upstream Linux kernel module.
283 Certain features require kernel support to function or to have
284 reasonable performance. If the ovs-vswitchd log file indicates that
285 a feature is not supported, consider upgrading to a newer upstream
286 Linux release or using the kernel module paired with the userspace
289 ### Q: Why do tunnels not work when using a kernel module other than the one packaged with Open vSwitch?
291 A: Support for tunnels was added to the upstream Linux kernel module
292 after the rest of Open vSwitch. As a result, some kernels may contain
293 support for Open vSwitch but not tunnels. The minimum kernel version
294 that supports each tunnel protocol is:
296 | Protocol | Linux Kernel
297 |:--------:|:-------------:
301 | LISP | <not upstream>
302 | STT | <not upstream>
304 If you are using a version of the kernel that is older than the one
305 listed above, it is still possible to use that tunnel protocol. However,
306 you must compile and install the kernel module included with the Open
307 vSwitch distribution rather than the one on your machine. If problems
308 persist after doing this, check to make sure that the module that is
309 loaded is the one you expect.
311 ### Q: Why are UDP tunnel checksums not computed for VXLAN or Geneve?
313 A: Generating outer UDP checksums requires kernel support that was not
314 part of the initial implementation of these protocols. If using the
315 upstream Linux Open vSwitch module, you must use kernel 4.0 or
316 newer. The out-of-tree modules from Open vSwitch release 2.4 and later
317 support UDP checksums.
319 ### Q: What features are not available when using the userspace datapath?
321 A: Tunnel virtual ports are not supported, as described in the
322 previous answer. It is also not possible to use queue-related
323 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
324 may not be transmitted.
326 ### Q: What Linux kernel versions does IPFIX flow monitoring work with?
328 A: IPFIX flow monitoring requires the Linux kernel module from Linux
329 3.10 or later, or the out-of-tree module from Open vSwitch version
332 ### Q: Should userspace or kernel be upgraded first to minimize downtime?
334 In general, the Open vSwitch userspace should be used with the
335 kernel version included in the same release or with the version
336 from upstream Linux. However, when upgrading between two releases
337 of Open vSwitch it is best to migrate userspace first to reduce
338 the possibility of incompatibilities.
340 ### Q: What happened to the bridge compatibility feature?
342 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
343 When it was enabled, Open vSwitch imitated the interface of the
344 Linux kernel "bridge" module. This allowed users to drop Open
345 vSwitch into environments designed to use the Linux kernel bridge
346 module without adapting the environment to use Open vSwitch.
348 Open vSwitch 1.10 and later do not support bridge compatibility.
349 The feature was dropped because version 1.10 adopted a new internal
350 architecture that made bridge compatibility difficult to maintain.
351 Now that many environments use OVS directly, it would be rarely
354 To use bridge compatibility, install OVS 1.9 or earlier, including
355 the accompanying kernel modules (both the main and bridge
356 compatibility modules), following the instructions that come with
357 the release. Be sure to start the ovs-brcompatd daemon.
363 ### Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation keeps talking about bridges. What's a bridge?
365 A: In networking, the terms "bridge" and "switch" are synonyms. Open
366 vSwitch implements an Ethernet switch, which means that it is also
369 ### Q: What's a VLAN?
371 A: See the "VLAN" section below.
377 ### Q: How do I configure a port as an access port?
379 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
380 the following commands configure br0 with eth0 as a trunk port (the
381 default) and tap0 as an access port for VLAN 9:
384 ovs-vsctl add-port br0 eth0
385 ovs-vsctl add-port br0 tap0 tag=9
387 If you want to configure an already added port as an access port,
388 use "ovs-vsctl set", e.g.:
390 ovs-vsctl set port tap0 tag=9
392 ### Q: How do I configure a port as a SPAN port, that is, enable mirroring of all traffic to that port?
394 A: The following commands configure br0 with eth0 and tap0 as trunk
395 ports. All traffic coming in or going out on eth0 or tap0 is also
396 mirrored to tap1; any traffic arriving on tap1 is dropped:
399 ovs-vsctl add-port br0 eth0
400 ovs-vsctl add-port br0 tap0
401 ovs-vsctl add-port br0 tap1 \
402 -- --id=@p get port tap1 \
403 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
404 -- set bridge br0 mirrors=@m
406 To later disable mirroring, run:
408 ovs-vsctl clear bridge br0 mirrors
410 ### Q: Does Open vSwitch support configuring a port in promiscuous mode?
412 A: Yes. How you configure it depends on what you mean by "promiscuous
415 - Conventionally, "promiscuous mode" is a feature of a network
416 interface card. Ordinarily, a NIC passes to the CPU only the
417 packets actually destined to its host machine. It discards
418 the rest to avoid wasting memory and CPU cycles. When
419 promiscuous mode is enabled, however, it passes every packet
420 to the CPU. On an old-style shared-media or hub-based
421 network, this allows the host to spy on all packets on the
422 network. But in the switched networks that are almost
423 everywhere these days, promiscuous mode doesn't have much
424 effect, because few packets not destined to a host are
425 delivered to the host's NIC.
427 This form of promiscuous mode is configured in the guest OS of
428 the VMs on your bridge, e.g. with "ifconfig".
430 - The VMware vSwitch uses a different definition of "promiscuous
431 mode". When you configure promiscuous mode on a VMware vNIC,
432 the vSwitch sends a copy of every packet received by the
433 vSwitch to that vNIC. That has a much bigger effect than just
434 enabling promiscuous mode in a guest OS. Rather than getting
435 a few stray packets for which the switch does not yet know the
436 correct destination, the vNIC gets every packet. The effect
437 is similar to replacing the vSwitch by a virtual hub.
439 This "promiscuous mode" is what switches normally call "port
440 mirroring" or "SPAN". For information on how to configure
441 SPAN, see "How do I configure a port as a SPAN port, that is,
442 enable mirroring of all traffic to that port?"
444 ### Q: How do I configure a DPDK port as an access port?
446 A: Firstly, you must have a DPDK-enabled version of Open vSwitch.
448 If your version is DPDK-enabled it will support the --dpdk
449 argument on the command line and will display lines with
450 "EAL:..." during startup when --dpdk is supplied.
452 Secondly, when adding a DPDK port, unlike a system port, the
453 type for the interface must be specified. For example;
456 ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
458 Finally, it is required that DPDK port names begin with 'dpdk'.
460 See [INSTALL.DPDK.md] for more information on enabling and using DPDK with
463 ### Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all traffic to that VLAN?
465 A: The following commands configure br0 with eth0 as a trunk port and
466 tap0 as an access port for VLAN 10. All traffic coming in or going
467 out on tap0, as well as traffic coming in or going out on eth0 in
468 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
469 VLAN 10, in cases where one is present, is dropped as part of
473 ovs-vsctl add-port br0 eth0
474 ovs-vsctl add-port br0 tap0 tag=10
476 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
478 -- set bridge br0 mirrors=@m
480 To later disable mirroring, run:
482 ovs-vsctl clear bridge br0 mirrors
484 Mirroring to a VLAN can disrupt a network that contains unmanaged
485 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
486 GRE tunnel has fewer caveats than mirroring to a VLAN and should
487 generally be preferred.
489 ### Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
491 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
492 of the specified output-vlan. This loss of information may make
493 the mirrored traffic too hard to interpret.
495 To mirror multiple VLANs, use the commands above, but specify a
496 comma-separated list of VLANs as the value for select-vlan. To
497 mirror every VLAN, use the commands above, but omit select-vlan and
500 When a packet arrives on a VLAN that is used as a mirror output
501 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
502 floods the packet across all the ports for which the mirror output
503 VLAN is configured. (If an OpenFlow controller is in use, then it
504 can override this behavior through the flow table.) If OVS is used
505 as an intermediate switch, rather than an edge switch, this ensures
506 that the RSPAN traffic is distributed through the network.
508 Mirroring to a VLAN can disrupt a network that contains unmanaged
509 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
510 GRE tunnel has fewer caveats than mirroring to a VLAN and should
511 generally be preferred.
513 ### Q: How do I configure mirroring of all traffic to a GRE tunnel?
515 A: The following commands configure br0 with eth0 and tap0 as trunk
516 ports. All traffic coming in or going out on eth0 or tap0 is also
517 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
518 traffic arriving on gre0 is dropped:
521 ovs-vsctl add-port br0 eth0
522 ovs-vsctl add-port br0 tap0
523 ovs-vsctl add-port br0 gre0 \
524 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
525 -- --id=@p get port gre0 \
526 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
527 -- set bridge br0 mirrors=@m
529 To later disable mirroring and destroy the GRE tunnel:
531 ovs-vsctl clear bridge br0 mirrors
532 ovs-vsctl del-port br0 gre0
534 ### Q: Does Open vSwitch support ERSPAN?
536 A: No. ERSPAN is an undocumented proprietary protocol. As an
537 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
540 ### Q: How do I connect two bridges?
542 A: First, why do you want to do this? Two connected bridges are not
543 much different from a single bridge, so you might as well just have
544 a single bridge with all your ports on it.
546 If you still want to connect two bridges, you can use a pair of
547 patch ports. The following example creates bridges br0 and br1,
548 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
549 and br1 with a pair of patch ports.
552 ovs-vsctl add-port br0 eth0
553 ovs-vsctl add-port br0 tap0
555 ovs-vsctl add-port br1 tap1
557 -- add-port br0 patch0 \
558 -- set interface patch0 type=patch options:peer=patch1 \
559 -- add-port br1 patch1 \
560 -- set interface patch1 type=patch options:peer=patch0
562 Bridges connected with patch ports are much like a single bridge.
563 For instance, if the example above also added eth1 to br1, and both
564 eth0 and eth1 happened to be connected to the same next-hop switch,
565 then you could loop your network just as you would if you added
566 eth0 and eth1 to the same bridge (see the "Configuration Problems"
567 section below for more information).
569 If you are using Open vSwitch 1.9 or an earlier version, then you
570 need to be using the kernel module bundled with Open vSwitch rather
571 than the one that is integrated into Linux 3.3 and later, because
572 Open vSwitch 1.9 and earlier versions need kernel support for patch
573 ports. This also means that in Open vSwitch 1.9 and earlier, patch
574 ports will not work with the userspace datapath, only with the
577 ### Q: How do I configure a bridge without an OpenFlow local port? (Local port in the sense of OFPP_LOCAL)
579 A: Open vSwitch does not support such a configuration.
580 Bridges always have their local ports.
583 Implementation Details
584 ----------------------
586 ### Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
588 A: Open vSwitch uses different kinds of flows for different purposes:
590 - OpenFlow flows are the most important kind of flow. OpenFlow
591 controllers use these flows to define a switch's policy.
592 OpenFlow flows support wildcards, priorities, and multiple
595 When in-band control is in use, Open vSwitch sets up a few
596 "hidden" flows, with priority higher than a controller or the
597 user can configure, that are not visible via OpenFlow. (See
598 the "Controller" section of the FAQ for more information
601 - The Open vSwitch software switch implementation uses a second
602 kind of flow internally. These flows, called "datapath" or
603 "kernel" flows, do not support priorities and comprise only a
604 single table, which makes them suitable for caching. (Like
605 OpenFlow flows, datapath flows do support wildcarding, in Open
606 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
607 also support different actions and number ports differently.
609 Datapath flows are an implementation detail that is subject to
610 change in future versions of Open vSwitch. Even with the
611 current version of Open vSwitch, hardware switch
612 implementations do not necessarily use this architecture.
614 Users and controllers directly control only the OpenFlow flow
615 table. Open vSwitch manages the datapath flow table itself, so
616 users should not normally be concerned with it.
618 ### Q: Why are there so many different ways to dump flows?
620 A: Open vSwitch has two kinds of flows (see the previous question), so
621 it has commands with different purposes for dumping each kind of
624 - `ovs-ofctl dump-flows <br>` dumps OpenFlow flows, excluding
625 hidden flows. This is the most commonly useful form of flow
626 dump. (Unlike the other commands, this should work with any
627 OpenFlow switch, not just Open vSwitch.)
629 - `ovs-appctl bridge/dump-flows <br>` dumps OpenFlow flows,
630 including hidden flows. This is occasionally useful for
631 troubleshooting suspected issues with in-band control.
633 - `ovs-dpctl dump-flows [dp]` dumps the datapath flow table
634 entries for a Linux kernel-based datapath. In Open vSwitch
635 1.10 and later, ovs-vswitchd merges multiple switches into a
636 single datapath, so it will show all the flows on all your
637 kernel-based switches. This command can occasionally be
638 useful for debugging.
640 - `ovs-appctl dpif/dump-flows <br>`, new in Open vSwitch 1.10,
641 dumps datapath flows for only the specified bridge, regardless
644 ### Q: How does multicast snooping works with VLANs?
646 A: Open vSwitch maintains snooping tables for each VLAN.
648 ### Q: Can OVS populate the kernel flow table in advance instead of in reaction to packets?
650 A: No. There are several reasons:
652 - Kernel flows are not as sophisticated as OpenFlow flows, which
653 means that some OpenFlow policies could require a large number of
654 kernel flows. The "conjunctive match" feature is an extreme
655 example: the number of kernel flows it requires is the product of
656 the number of flows in each dimension.
658 - With multiple OpenFlow flow tables and simple sets of actions, the
659 number of kernel flows required can be as large as the product of
660 the number of flows in each dimension. With more sophisticated
661 actions, the number of kernel flows could be even larger.
663 - Open vSwitch is designed so that any version of OVS userspace
664 interoperates with any version of the OVS kernel module. This
665 forward and backward compatibility requires that userspace observe
666 how the kernel module parses received packets. This is only
667 possible in a straightforward way when userspace adds kernel flows
668 in reaction to received packets.
670 For more relevant information on the architecture of Open vSwitch,
671 please read "The Design and Implementation of Open vSwitch",
672 published in USENIX NSDI 2015.
678 ### Q: I just upgraded and I see a performance drop. Why?
680 A: The OVS kernel datapath may have been updated to a newer version than
681 the OVS userspace components. Sometimes new versions of OVS kernel
682 module add functionality that is backwards compatible with older
683 userspace components but may cause a drop in performance with them.
684 Especially, if a kernel module from OVS 2.1 or newer is paired with
685 OVS userspace 1.10 or older, there will be a performance drop for
688 Updating the OVS userspace components to the latest released
689 version should fix the performance degradation.
691 To get the best possible performance and functionality, it is
692 recommended to pair the same versions of the kernel module and OVS
696 Configuration Problems
697 ----------------------
699 ### Q: I created a bridge and added my Ethernet port to it, using commands
703 ovs-vsctl add-port br0 eth0
705 and as soon as I ran the "add-port" command I lost all connectivity
708 A: A physical Ethernet device that is part of an Open vSwitch bridge
709 should not have an IP address. If one does, then that IP address
710 will not be fully functional.
712 You can restore functionality by moving the IP address to an Open
713 vSwitch "internal" device, such as the network device named after
714 the bridge itself. For example, assuming that eth0's IP address is
715 192.168.128.5, you could run the commands below to fix up the
718 ifconfig eth0 0.0.0.0
719 ifconfig br0 192.168.128.5
721 (If your only connection to the machine running OVS is through the
722 IP address in question, then you would want to run all of these
723 commands on a single command line, or put them into a script.) If
724 there were any additional routes assigned to eth0, then you would
725 also want to use commands to adjust these routes to go through br0.
727 If you use DHCP to obtain an IP address, then you should kill the
728 DHCP client that was listening on the physical Ethernet interface
729 (e.g. eth0) and start one listening on the internal interface
730 (e.g. br0). You might still need to manually clear the IP address
731 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
733 There is no compelling reason why Open vSwitch must work this way.
734 However, this is the way that the Linux kernel bridge module has
735 always worked, so it's a model that those accustomed to Linux
736 bridging are already used to. Also, the model that most people
737 expect is not implementable without kernel changes on all the
738 versions of Linux that Open vSwitch supports.
740 By the way, this issue is not specific to physical Ethernet
741 devices. It applies to all network devices except Open vSwitch
744 ### Q: I created a bridge and added a couple of Ethernet ports to it,
745 ### using commands like these:
748 ovs-vsctl add-port br0 eth0
749 ovs-vsctl add-port br0 eth1
751 and now my network seems to have melted: connectivity is unreliable
752 (even connectivity that doesn't go through Open vSwitch), all the
753 LEDs on my physical switches are blinking, wireshark shows
754 duplicated packets, and CPU usage is very high.
756 A: More than likely, you've looped your network. Probably, eth0 and
757 eth1 are connected to the same physical Ethernet switch. This
758 yields a scenario where OVS receives a broadcast packet on eth0 and
759 sends it out on eth1, then the physical switch connected to eth1
760 sends the packet back on eth0, and so on forever. More complicated
761 scenarios, involving a loop through multiple switches, are possible
764 The solution depends on what you are trying to do:
766 - If you added eth0 and eth1 to get higher bandwidth or higher
767 reliability between OVS and your physical Ethernet switch,
768 use a bond. The following commands create br0 and then add
769 eth0 and eth1 as a bond:
772 ovs-vsctl add-bond br0 bond0 eth0 eth1
774 Bonds have tons of configuration options. Please read the
775 documentation on the Port table in ovs-vswitchd.conf.db(5)
778 Configuration for DPDK-enabled interfaces is slightly less
779 straightforward: see [INSTALL.DPDK.md].
781 - Perhaps you don't actually need eth0 and eth1 to be on the
782 same bridge. For example, if you simply want to be able to
783 connect each of them to virtual machines, then you can put
784 each of them on a bridge of its own:
787 ovs-vsctl add-port br0 eth0
790 ovs-vsctl add-port br1 eth1
792 and then connect VMs to br0 and br1. (A potential
793 disadvantage is that traffic cannot directly pass between br0
794 and br1. Instead, it will go out eth0 and come back in eth1,
797 - If you have a redundant or complex network topology and you
798 want to prevent loops, turn on spanning tree protocol (STP).
799 The following commands create br0, enable STP, and add eth0
800 and eth1 to the bridge. The order is important because you
801 don't want have to have a loop in your network even
805 ovs-vsctl set bridge br0 stp_enable=true
806 ovs-vsctl add-port br0 eth0
807 ovs-vsctl add-port br0 eth1
809 The Open vSwitch implementation of STP is not well tested.
810 Please report any bugs you observe, but if you'd rather avoid
811 acting as a beta tester then another option might be your
814 ### Q: I can't seem to use Open vSwitch in a wireless network.
816 A: Wireless base stations generally only allow packets with the source
817 MAC address of NIC that completed the initial handshake.
818 Therefore, without MAC rewriting, only a single device can
819 communicate over a single wireless link.
821 This isn't specific to Open vSwitch, it's enforced by the access
822 point, so the same problems will show up with the Linux bridge or
823 any other way to do bridging.
825 ### Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
827 A: PPP most commonly carries IP packets, but Open vSwitch works only
828 with Ethernet frames. The correct way to interface PPP to an
829 Ethernet network is usually to use routing instead of switching.
831 ### Q: Is there any documentation on the database tables and fields?
833 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
835 ### Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
836 I only see a datapath called "ovs-system". How can I see datapath
837 information about a particular bridge?
839 A: In version 1.9.0, OVS switched to using a single datapath that is
840 shared by all bridges of that type. The "ovs-appctl dpif/*"
841 commands provide similar functionality that is scoped by the bridge.
843 ### Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
844 see the port in the datapath?
846 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
847 cannot be loaded at the same time. It is likely that on your system the
848 Linux GRE module is already loaded and blocking OVS (to confirm, check
849 dmesg for errors regarding GRE registration). To fix this, unload all
850 GRE modules that appear in lsmod as well as the OVS kernel module. You
851 can then reload the OVS module following the directions in
852 [INSTALL.md], which will ensure that dependencies are satisfied.
854 ### Q: Open vSwitch does not seem to obey my packet filter rules.
856 A: It depends on mechanisms and configurations you want to use.
858 You cannot usefully use typical packet filters, like iptables, on
859 physical Ethernet ports that you add to an Open vSwitch bridge.
860 This is because Open vSwitch captures packets from the interface at
861 a layer lower below where typical packet-filter implementations
862 install their hooks. (This actually applies to any interface of
863 type "system" that you might add to an Open vSwitch bridge.)
865 You can usefully use typical packet filters on Open vSwitch
866 internal ports as they are mostly ordinary interfaces from the point
867 of view of packet filters.
869 For example, suppose you create a bridge br0 and add Ethernet port
870 eth0 to it. Then you can usefully add iptables rules to affect the
871 internal interface br0, but not the physical interface eth0. (br0
872 is also where you would add an IP address, as discussed elsewhere
875 For simple filtering rules, it might be possible to achieve similar
876 results by installing appropriate OpenFlow flows instead.
878 If the use of a particular packet filter setup is essential, Open
879 vSwitch might not be the best choice for you. On Linux, you might
880 want to consider using the Linux Bridge. (This is the only choice if
881 you want to use ebtables rules.) On NetBSD, you might want to
882 consider using the bridge(4) with BRIDGE_IPF option.
884 ### Q: It seems that Open vSwitch does nothing when I removed a port and
885 then immediately put it back. For example, consider that p1 is
886 a port of type=internal:
888 ovs-vsctl del-port br0 p1 -- \
890 set interface p1 type=internal
892 A: It's an expected behaviour.
894 If del-port and add-port happen in a single OVSDB transaction as
895 your example, Open vSwitch always "skips" the intermediate steps.
896 Even if they are done in multiple transactions, it's still allowed
897 for Open vSwitch to skip the intermediate steps and just implement
898 the overall effect. In both cases, your example would be turned
901 If you want to make Open vSwitch actually destroy and then re-create
902 the port for some side effects like resetting kernel setting for the
903 corresponding interface, you need to separate operations into multiple
904 OVSDB transactions and ensure that at least the first one does not have
905 --no-wait. In the following example, the first ovs-vsctl will block
906 until Open vSwitch reloads the new configuration and removes the port:
908 ovs-vsctl del-port br0 p1
909 ovs-vsctl add-port br0 p1 -- \
910 set interface p1 type=internal
912 ### Q: I want to add thousands of ports to an Open vSwitch bridge, but
913 it takes too long (minutes or hours) to do it with ovs-vsctl. How
916 A: If you add them one at a time with ovs-vsctl, it can take a long
917 time to add thousands of ports to an Open vSwitch bridge. This is
918 because every invocation of ovs-vsctl first reads the current
919 configuration from OVSDB. As the number of ports grows, this
920 starts to take an appreciable amount of time, and when it is
921 repeated thousands of times the total time becomes significant.
923 The solution is to add the ports in one invocation of ovs-vsctl (or
924 a small number of them). For example, using bash:
927 cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done
930 takes seconds, not minutes or hours, in the OVS sandbox environment.
932 ### Q: I created a bridge named br0. My bridge shows up in "ovs-vsctl
933 show", but "ovs-ofctl show br0" just prints "br0 is not a bridge
936 A: Open vSwitch wasn't able to create the bridge. Check the
937 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
938 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log).
940 In general, the Open vSwitch database reflects the desired
941 configuration state. ovs-vswitchd monitors the database and, when
942 it changes, reconfigures the system to reflect the new desired
943 state. This normally happens very quickly. Thus, a discrepancy
944 between the database and the actual state indicates that
945 ovs-vswitchd could not implement the configuration, and so one
946 should check the log to find out why. (Another possible cause is
947 that ovs-vswitchd is not running. This will make "ovs-vsctl"
948 commands hang, if they change the configuration, unless one
949 specifies "--no-wait".)
951 ### Q: I have a bridge br0. I added a new port vif1.0, and it shows
952 up in "ovs-vsctl show", but "ovs-vsctl list port" says that it has
953 OpenFlow port ("ofport") -1, and "ovs-ofctl show br0" doesn't show
956 A: Open vSwitch wasn't able to create the port. Check the
957 ovs-vswitchd log for details (Debian and Red Hat packaging for Open
958 vSwitch put it in /var/log/openvswitch/ovs-vswitchd.log). Please
959 see the previous question for more information.
961 You may want to upgrade to Open vSwitch 2.3 (or later), in which
962 ovs-vsctl will immediately report when there is an issue creating a
965 ### Q: I created a tap device tap0, configured an IP address on it, and
966 added it to a bridge, like this:
969 ifconfig tap0 192.168.0.123
971 ovs-vsctl add-port br0 tap0
973 I expected that I could then use this IP address to contact other
974 hosts on the network, but it doesn't work. Why not?
976 A: The short answer is that this is a misuse of a "tap" device. Use
977 an "internal" device implemented by Open vSwitch, which works
978 differently and is designed for this use. To solve this problem
979 with an internal device, instead run:
982 ovs-vsctl add-port br0 int0 -- set Interface int0 type=internal
983 ifconfig int0 192.168.0.123
985 Even more simply, you can take advantage of the internal port that
986 every bridge has under the name of the bridge:
989 ifconfig br0 192.168.0.123
991 In more detail, a "tap" device is an interface between the Linux
992 (or *BSD) network stack and a user program that opens it as a
993 socket. When the "tap" device transmits a packet, it appears in
994 the socket opened by the userspace program. Conversely, when the
995 userspace program writes to the "tap" socket, the kernel TCP/IP
996 stack processes the packet as if it had been received by the "tap"
999 Consider the configuration above. Given this configuration, if you
1000 "ping" an IP address in the 192.168.0.x subnet, the Linux kernel
1001 routing stack will transmit an ARP on the tap0 device. Open
1002 vSwitch userspace treats "tap" devices just like any other network
1003 device; that is, it doesn't open them as "tap" sockets. That means
1004 that the ARP packet will simply get dropped.
1006 You might wonder why the Open vSwitch kernel module doesn't
1007 intercept the ARP packet and bridge it. After all, Open vSwitch
1008 intercepts packets on other devices. The answer is that Open
1009 vSwitch only intercepts *received* packets, but this is a packet
1010 being transmitted. The same thing happens for all other types of
1011 network devices, except for Open vSwitch "internal" ports. If you,
1012 for example, add a physical Ethernet port to an OVS bridge,
1013 configure an IP address on a physical Ethernet port, and then issue
1014 a "ping" to an address in that subnet, the same thing happens: an
1015 ARP gets transmitted on the physical Ethernet port and Open vSwitch
1016 never sees it. (You should not do that, as documented at the
1017 beginning of this section.)
1019 It can make sense to add a "tap" device to an Open vSwitch bridge,
1020 if some userspace program (other than Open vSwitch) has opened the
1021 tap socket. This is the case, for example, if the "tap" device was
1022 created by KVM (or QEMU) to simulate a virtual NIC. In such a
1023 case, when OVS bridges a packet to the "tap" device, the kernel
1024 forwards that packet to KVM in userspace, which passes it along to
1025 the VM, and in the other direction, when the VM sends a packet, KVM
1026 writes it to the "tap" socket, which causes OVS to receive it and
1027 bridge it to the other OVS ports. Please note that in such a case
1028 no IP address is configured on the "tap" device (there is normally
1029 an IP address configured in the virtual NIC inside the VM, but this
1030 is not visible to the host Linux kernel or to Open vSwitch).
1032 There is one special case in which Open vSwitch does directly read
1033 and write "tap" sockets. This is an implementation detail of the
1034 Open vSwitch userspace switch, which implements its "internal"
1035 ports as Linux (or *BSD) "tap" sockets. In such a userspace
1036 switch, OVS receives packets sent on the "tap" device used to
1037 implement an "internal" port by reading the associated "tap"
1038 socket, and bridges them to the rest of the switch. In the other
1039 direction, OVS transmits packets bridged to the "internal" port by
1040 writing them to the "tap" socket, causing them to be processed by
1041 the kernel TCP/IP stack as if they had been received on the "tap"
1042 device. Users should not need to be concerned with this
1043 implementation detail.
1045 Open vSwitch has a network device type called "tap". This is
1046 intended only for implementing "internal" ports in the OVS
1047 userspace switch and should not be used otherwise. In particular,
1048 users should not configure KVM "tap" devices as type "tap" (use
1049 type "system", the default, instead).
1052 Quality of Service (QoS)
1053 ------------------------
1055 ### Q: Does OVS support Quality of Service (QoS)?
1057 A: Yes. For traffic that egresses from a switch, OVS supports traffic
1058 shaping; for traffic that ingresses into a switch, OVS support
1059 policing. Policing is a simple form of quality-of-service that
1060 simply drops packets received in excess of the configured rate. Due
1061 to its simplicity, policing is usually less accurate and less
1062 effective than egress traffic shaping, which queues packets.
1064 Keep in mind that ingress and egress are from the perspective of the
1065 switch. That means that egress shaping limits the rate at which
1066 traffic is allowed to transmit from a physical interface, but the
1067 rate at which traffic will be received on a virtual machine's VIF.
1068 For ingress policing, the behavior is the opposite.
1070 ### Q: How do I configure egress traffic shaping?
1072 A: Suppose that you want to set up bridge br0 connected to physical
1073 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
1074 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
1075 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
1076 could configure the bridge this way:
1080 add-port br0 eth0 -- \
1081 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
1082 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
1083 set port eth0 qos=@newqos -- \
1084 --id=@newqos create qos type=linux-htb \
1085 other-config:max-rate=1000000000 \
1086 queues:123=@vif10queue \
1087 queues:234=@vif20queue -- \
1088 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
1089 --id=@vif20queue create queue other-config:max-rate=20000000
1091 At this point, bridge br0 is configured with the ports and eth0 is
1092 configured with the queues that you need for QoS, but nothing is
1093 actually directing packets from vif1.0 or vif2.0 to the queues that
1094 we have set up for them. That means that all of the packets to
1095 eth0 are going to the "default queue", which is not what we want.
1097 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
1098 queues reserved for them:
1100 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
1101 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
1103 Each of the above flows matches on the input port, sets up the
1104 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
1105 executes the "normal" action, which performs the same switching
1106 that Open vSwitch would have done without any OpenFlow flows being
1107 present. (We know that vif1.0 and vif2.0 have OpenFlow port
1108 numbers 5 and 6, respectively, because we set their ofport_request
1109 columns above. If we had not done that, then we would have needed
1110 to find out their port numbers before setting up these flows.)
1112 Now traffic going from vif1.0 or vif2.0 to eth0 should be
1115 By the way, if you delete the bridge created by the above commands,
1118 ovs-vsctl del-br br0
1120 then that will leave one unreferenced QoS record and two
1121 unreferenced Queue records in the Open vSwich database. One way to
1122 clear them out, assuming you don't have other QoS or Queue records
1123 that you want to keep, is:
1125 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
1127 If you do want to keep some QoS or Queue records, or the Open
1128 vSwitch you are using is older than version 1.8 (which added the
1129 --all option), then you will have to destroy QoS and Queue records
1132 ### Q: How do I configure ingress policing?
1134 A: A policing policy can be configured on an interface to drop packets
1135 that arrive at a higher rate than the configured value. For example,
1136 the following commands will rate-limit traffic that vif1.0 may
1139 ovs-vsctl set interface vif1.0 ingress_policing_rate=10000
1140 ovs-vsctl set interface vif1.0 ingress_policing_burst=8000
1142 Traffic policing can interact poorly with some network protocols and
1143 can have surprising results. The "Ingress Policing" section of
1144 ovs-vswitchd.conf.db(5) discusses the issues in greater detail.
1146 ### Q: I configured Quality of Service (QoS) in my OpenFlow network by
1147 adding records to the QoS and Queue table, but the results aren't
1150 A: Did you install OpenFlow flows that use your queues? This is the
1151 primary way to tell Open vSwitch which queues you want to use. If
1152 you don't do this, then the default queue will be used, which will
1153 probably not have the effect you want.
1155 Refer to the previous question for an example.
1157 ### Q: I'd like to take advantage of some QoS feature that Open vSwitch
1158 doesn't yet support. How do I do that?
1160 A: Open vSwitch does not implement QoS itself. Instead, it can
1161 configure some, but not all, of the QoS features built into the
1162 Linux kernel. If you need some QoS feature that OVS cannot
1163 configure itself, then the first step is to figure out whether
1164 Linux QoS supports that feature. If it does, then you can submit a
1165 patch to support Open vSwitch configuration for that feature, or
1166 you can use "tc" directly to configure the feature in Linux. (If
1167 Linux QoS doesn't support the feature you want, then first you have
1168 to add that support to Linux.)
1170 ### Q: I configured QoS, correctly, but my measurements show that it isn't
1171 working as well as I expect.
1173 A: With the Linux kernel, the Open vSwitch implementation of QoS has
1176 - Open vSwitch configures a subset of Linux kernel QoS
1177 features, according to what is in OVSDB. It is possible that
1178 this code has bugs. If you believe that this is so, then you
1179 can configure the Linux traffic control (QoS) stack directly
1180 with the "tc" program. If you get better results that way,
1181 you can send a detailed bug report to bugs@openvswitch.org.
1183 It is certain that Open vSwitch cannot configure every Linux
1184 kernel QoS feature. If you need some feature that OVS cannot
1185 configure, then you can also use "tc" directly (or add that
1188 - The Open vSwitch implementation of OpenFlow allows flows to
1189 be directed to particular queues. This is pretty simple and
1190 unlikely to have serious bugs at this point.
1192 However, most problems with QoS on Linux are not bugs in Open
1193 vSwitch at all. They tend to be either configuration errors
1194 (please see the earlier questions in this section) or issues with
1195 the traffic control (QoS) stack in Linux. The Open vSwitch
1196 developers are not experts on Linux traffic control. We suggest
1197 that, if you believe you are encountering a problem with Linux
1198 traffic control, that you consult the tc manpages (e.g. tc(8),
1199 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
1200 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
1202 ### Q: Does Open vSwitch support OpenFlow meters?
1204 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
1205 OpenFlow meters. There is no implementation of meters in the Open
1206 vSwitch software switch (neither the kernel-based nor userspace
1213 ### Q: What's a VLAN?
1215 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
1216 partition a single switch into multiple switches. Suppose, for
1217 example, that you have two groups of machines, group A and group B.
1218 You want the machines in group A to be able to talk to each other,
1219 and you want the machine in group B to be able to talk to each
1220 other, but you don't want the machines in group A to be able to
1221 talk to the machines in group B. You can do this with two
1222 switches, by plugging the machines in group A into one switch and
1223 the machines in group B into the other switch.
1225 If you only have one switch, then you can use VLANs to do the same
1226 thing, by configuring the ports for machines in group A as VLAN
1227 "access ports" for one VLAN and the ports for group B as "access
1228 ports" for a different VLAN. The switch will only forward packets
1229 between ports that are assigned to the same VLAN, so this
1230 effectively subdivides your single switch into two independent
1231 switches, one for each group of machines.
1233 So far we haven't said anything about VLAN headers. With access
1234 ports, like we've described so far, no VLAN header is present in
1235 the Ethernet frame. This means that the machines (or switches)
1236 connected to access ports need not be aware that VLANs are
1237 involved, just like in the case where we use two different physical
1240 Now suppose that you have a whole bunch of switches in your
1241 network, instead of just one, and that some machines in group A are
1242 connected directly to both switches 1 and 2. To allow these
1243 machines to talk to each other, you could add an access port for
1244 group A's VLAN to switch 1 and another to switch 2, and then
1245 connect an Ethernet cable between those ports. That works fine,
1246 but it doesn't scale well as the number of switches and the number
1247 of VLANs increases, because you use up a lot of valuable switch
1248 ports just connecting together your VLANs.
1250 This is where VLAN headers come in. Instead of using one cable and
1251 two ports per VLAN to connect a pair of switches, we configure a
1252 port on each switch as a VLAN "trunk port". Packets sent and
1253 received on a trunk port carry a VLAN header that says what VLAN
1254 the packet belongs to, so that only two ports total are required to
1255 connect the switches, regardless of the number of VLANs in use.
1256 Normally, only switches (either physical or virtual) are connected
1257 to a trunk port, not individual hosts, because individual hosts
1258 don't expect to see a VLAN header in the traffic that they receive.
1260 None of the above discussion says anything about particular VLAN
1261 numbers. This is because VLAN numbers are completely arbitrary.
1262 One must only ensure that a given VLAN is numbered consistently
1263 throughout a network and that different VLANs are given different
1264 numbers. (That said, VLAN 0 is usually synonymous with a packet
1265 that has no VLAN header, and VLAN 4095 is reserved.)
1267 ### Q: VLANs don't work.
1269 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
1270 bugs. If you are having problems with VLANs that you suspect to be
1271 driver related, then you have several options:
1273 - Upgrade to Linux 3.3 or later.
1275 - Build and install a fixed version of the particular driver
1276 that is causing trouble, if one is available.
1278 - Use a NIC whose driver does not have VLAN problems.
1280 - Use "VLAN splinters", a feature in Open vSwitch 1.4 upto 2.5
1281 that works around bugs in kernel drivers. To enable VLAN
1282 splinters on interface eth0, use the command:
1284 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
1286 For VLAN splinters to be effective, Open vSwitch must know
1287 which VLANs are in use. See the "VLAN splinters" section in
1288 the Interface table in ovs-vswitchd.conf.db(5) for details on
1289 how Open vSwitch infers in-use VLANs.
1291 VLAN splinters increase memory use and reduce performance, so
1292 use them only if needed.
1294 - Apply the "vlan workaround" patch from the XenServer kernel
1295 patch queue, build Open vSwitch against this patched kernel,
1296 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
1297 workaround for each interface whose driver is buggy.
1299 (This is a nontrivial exercise, so this option is included
1300 only for completeness.)
1302 It is not always easy to tell whether a Linux kernel driver has
1303 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
1304 can help you test. See their manpages for details. Of the two
1305 utilities, ovs-test(8) is newer and more thorough, but
1306 ovs-vlan-test(8) may be easier to use.
1308 ### Q: VLANs still don't work. I've tested the driver so I know that it's OK.
1310 A: Do you have VLANs enabled on the physical switch that OVS is
1311 attached to? Make sure that the port is configured to trunk the
1312 VLAN or VLANs that you are using with OVS.
1314 ### Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
1315 and to its destination host, but OVS seems to drop incoming return
1318 A: It's possible that you have the VLAN configured on your physical
1319 switch as the "native" VLAN. In this mode, the switch treats
1320 incoming packets either tagged with the native VLAN or untagged as
1321 part of the native VLAN. It may also send outgoing packets in the
1322 native VLAN without a VLAN tag.
1324 If this is the case, you have two choices:
1326 - Change the physical switch port configuration to tag packets
1327 it forwards to OVS with the native VLAN instead of forwarding
1330 - Change the OVS configuration for the physical port to a
1331 native VLAN mode. For example, the following sets up a
1332 bridge with port eth0 in "native-tagged" mode in VLAN 9:
1334 ovs-vsctl add-br br0
1335 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
1337 In this situation, "native-untagged" mode will probably work
1338 equally well. Refer to the documentation for the Port table
1339 in ovs-vswitchd.conf.db(5) for more information.
1341 ### Q: I added a pair of VMs on different VLANs, like this:
1343 ovs-vsctl add-br br0
1344 ovs-vsctl add-port br0 eth0
1345 ovs-vsctl add-port br0 tap0 tag=9
1346 ovs-vsctl add-port br0 tap1 tag=10
1348 but the VMs can't access each other, the external network, or the
1351 A: It is to be expected that the VMs can't access each other. VLANs
1352 are a means to partition a network. When you configured tap0 and
1353 tap1 as access ports for different VLANs, you indicated that they
1354 should be isolated from each other.
1356 As for the external network and the Internet, it seems likely that
1357 the machines you are trying to access are not on VLAN 9 (or 10) and
1358 that the Internet is not available on VLAN 9 (or 10).
1360 ### Q: I added a pair of VMs on the same VLAN, like this:
1362 ovs-vsctl add-br br0
1363 ovs-vsctl add-port br0 eth0
1364 ovs-vsctl add-port br0 tap0 tag=9
1365 ovs-vsctl add-port br0 tap1 tag=9
1367 The VMs can access each other, but not the external network or the
1370 A: It seems likely that the machines you are trying to access in the
1371 external network are not on VLAN 9 and that the Internet is not
1372 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
1373 trunk VLAN on the upstream switch port to which eth0 is connected.
1375 ### Q: Can I configure an IP address on a VLAN?
1377 A: Yes. Use an "internal port" configured as an access port. For
1378 example, the following configures IP address 192.168.0.7 on VLAN 9.
1379 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
1380 they have an 802.1Q header with VLAN 9. Conversely, traffic
1381 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
1384 ovs-vsctl add-br br0
1385 ovs-vsctl add-port br0 eth0
1386 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1387 ifconfig vlan9 192.168.0.7
1389 See also the following question.
1391 ### Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
1394 ovs-vsctl add-br br0
1395 ovs-vsctl add-port br0 eth0
1396 ifconfig br0 192.168.0.5
1397 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1398 ifconfig vlan9 192.168.0.9
1400 but other hosts that are only on VLAN 0 can reach the IP address
1401 configured on VLAN 9. What's going on?
1403 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
1404 approaches to IP address handling in Internet hosts:
1406 - In the "Strong ES Model", where an ES is a host ("End
1407 System"), an IP address is primarily associated with a
1408 particular interface. The host discards packets that arrive
1409 on interface A if they are destined for an IP address that is
1410 configured on interface B. The host never sends packets from
1411 interface A using a source address configured on interface B.
1413 - In the "Weak ES Model", an IP address is primarily associated
1414 with a host. The host accepts packets that arrive on any
1415 interface if they are destined for any of the host's IP
1416 addresses, even if the address is configured on some
1417 interface other than the one on which it arrived. The host
1418 does not restrict itself to sending packets from an IP
1419 address associated with the originating interface.
1421 Linux uses the weak ES model. That means that when packets
1422 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
1423 br0, the kernel IP stack accepts them there for the VLAN 9 IP
1424 address, even though they were not received on vlan9, the network
1427 To simulate the strong ES model on Linux, one may add iptables rule
1428 to filter packets based on source and destination address and
1429 adjust ARP configuration with sysctls.
1431 BSD uses the strong ES model.
1433 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1435 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1436 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1437 implementation of the OpenFlow "normal switching" action. By
1438 default, when Open vSwitch isn't connected to a controller and
1439 nothing has been manually configured in the flow table, all traffic
1440 goes through the "normal switching" action. But, if you set up
1441 OpenFlow flows on your own, through a controller or using ovs-ofctl
1442 or through other means, then you have to implement VLAN handling
1445 You can use "normal switching" as a component of your OpenFlow
1446 actions, e.g. by putting "normal" into the lists of actions on
1447 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1448 controller. In situations where this is not suitable, you can
1449 implement VLAN handling yourself, e.g.:
1451 - If a packet comes in on an access port, and the flow table
1452 needs to send it out on a trunk port, then the flow can add
1453 the appropriate VLAN tag with the "mod_vlan_vid" action.
1455 - If a packet comes in on a trunk port, and the flow table
1456 needs to send it out on an access port, then the flow can
1457 strip the VLAN tag with the "strip_vlan" action.
1459 ### Q: I configured ports on a bridge as access ports with different VLAN
1462 ovs-vsctl add-br br0
1463 ovs-vsctl set-controller br0 tcp:192.168.0.10:6653
1464 ovs-vsctl add-port br0 eth0
1465 ovs-vsctl add-port br0 tap0 tag=9
1466 ovs-vsctl add-port br0 tap1 tag=10
1468 but the VMs running behind tap0 and tap1 can still communicate,
1469 that is, they are not isolated from each other even though they are
1472 A: Do you have a controller configured on br0 (as the commands above
1473 do)? If so, then this is a variant on the previous question, "My
1474 OpenFlow controller doesn't see the VLANs that I expect," and you
1475 can refer to the answer there for more information.
1477 ### Q: How MAC learning works with VLANs?
1479 A: Open vSwitch implements Independent VLAN Learning (IVL) for
1480 OFPP_NORMAL action. I.e. it logically has separate learning tables
1487 ### Q: What's a VXLAN?
1489 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1490 to solve the scaling challenges of VLAN networks in a multi-tenant
1491 environment. VXLAN is an overlay network which transports an L2 network
1492 over an existing L3 network. For more information on VXLAN, please see
1495 http://tools.ietf.org/html/rfc7348
1497 ### Q: How much of the VXLAN protocol does Open vSwitch currently support?
1499 A: Open vSwitch currently supports the framing format for packets on the
1500 wire. There is currently no support for the multicast aspects of VXLAN.
1501 To get around the lack of multicast support, it is possible to
1502 pre-provision MAC to IP address mappings either manually or from a
1505 ### Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1508 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1509 is 4789. However, it is possible to configure the destination UDP port
1510 manually on a per-VXLAN tunnel basis. An example of this configuration is
1513 ovs-vsctl add-br br0
1514 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1515 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1516 options:dst_port=8472
1519 Using OpenFlow (Manually or Via Controller)
1520 -------------------------------------------
1522 ### Q: What versions of OpenFlow does Open vSwitch support?
1524 A: The following table lists the versions of OpenFlow supported by
1525 each version of Open vSwitch:
1527 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5 OF1.6
1528 ###============ ===== ===== ===== ===== ===== ===== =====
1529 1.9 and earlier yes --- --- --- --- --- ---
1530 1.10 yes --- [*] [*] --- --- ---
1531 1.11 yes --- [*] [*] --- --- ---
1532 2.0 yes [*] [*] [*] --- --- ---
1533 2.1 yes [*] [*] [*] --- --- ---
1534 2.2 yes [*] [*] [*] [%] [*] ---
1535 2.3 yes yes yes yes [*] [*] ---
1536 2.4 yes yes yes yes [*] [*] ---
1537 2.5 yes yes yes yes [*] [*] [*]
1539 [*] Supported, with one or more missing features.
1540 [%] Experimental, unsafe implementation.
1542 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default
1543 in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1,
1544 1.2, and 1.3 must be enabled manually in ovs-vswitchd.
1546 Some versions of OpenFlow are supported with missing features and
1547 therefore not enabled by default: OpenFlow 1.4 and 1.5, in Open
1548 vSwitch 2.3 and later, as well as OpenFlow 1.6 in Open vSwitch 2.5
1549 and later. Also, the OpenFlow 1.6 specification is still under
1550 development and thus subject to change.
1552 In any case, the user may override the default:
1554 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0:
1556 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1558 - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0:
1560 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15
1562 - To enable only OpenFlow 1.0 on bridge br0:
1564 ovs-vsctl set bridge br0 protocols=OpenFlow10
1566 All current versions of ovs-ofctl enable only OpenFlow 1.0 by
1567 default. Use the -O option to enable support for later versions of
1568 OpenFlow in ovs-ofctl. For example:
1570 ovs-ofctl -O OpenFlow13 dump-flows br0
1572 (Open vSwitch 2.2 had an experimental implementation of OpenFlow
1573 1.4 that could cause crashes. We don't recommend enabling it.)
1575 [OPENFLOW-1.1+.md] in the Open vSwitch source tree tracks support for
1576 OpenFlow 1.1 and later features. When support for OpenFlow 1.4 and
1577 1.5 is solidly implemented, Open vSwitch will enable those version
1580 ### Q: Does Open vSwitch support MPLS?
1582 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1583 these versions can match on MPLS Ethernet types, but they cannot
1584 match, push, or pop MPLS labels, nor can they look past MPLS labels
1585 into the encapsulated packet.
1587 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1588 for MPLS. With the userspace datapath only, these versions can
1589 match, push, or pop a single MPLS label, but they still cannot look
1590 past MPLS labels (even after popping them) into the encapsulated
1591 packet. Kernel datapath support is unchanged from earlier
1594 Open vSwitch version 2.3 can match, push, or pop a single MPLS
1595 label and look past the MPLS label into the encapsulated packet.
1596 Both userspace and kernel datapaths will be supported, but MPLS
1597 processing always happens in userspace either way, so kernel
1598 datapath performance will be disappointing.
1600 Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS
1601 labels and look past the MPLS label into the encapsulated packet.
1602 It will have kernel support for MPLS, yielding improved
1605 ### Q: I'm getting "error type 45250 code 0". What's that?
1607 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1608 vSwitch uses this extension when it must report an error to an
1609 OpenFlow controller but no standard OpenFlow error code is
1612 Open vSwitch logs the errors that it sends to controllers, so the
1613 easiest thing to do is probably to look at the ovs-vswitchd log to
1614 find out what the error was.
1616 If you want to dissect the extended error message yourself, the
1617 format is documented in include/openflow/nicira-ext.h in the Open
1618 vSwitch source distribution. The extended error codes are
1619 documented in include/openvswitch/ofp-errors.h.
1621 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1622 doesn't actually appear through the OpenFlow connection, even
1623 though I know that it's going through.
1624 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1625 to apply to certain traffic, especially traffic between OVS and
1626 the controller itself.
1628 A: By default, Open vSwitch assumes that OpenFlow controllers are
1629 connected "in-band", that is, that the controllers are actually
1630 part of the network that is being controlled. In in-band mode,
1631 Open vSwitch sets up special "hidden" flows to make sure that
1632 traffic can make it back and forth between OVS and the controllers.
1633 These hidden flows are higher priority than any flows that can be
1634 set up through OpenFlow, and they are not visible through normal
1635 OpenFlow flow table dumps.
1637 Usually, the hidden flows are desirable and helpful, but
1638 occasionally they can cause unexpected behavior. You can view the
1639 full OpenFlow flow table, including hidden flows, on bridge br0
1642 ovs-appctl bridge/dump-flows br0
1644 to help you debug. The hidden flows are those with priorities
1645 greater than 65535 (the maximum priority that can be set with
1648 The DESIGN file at the top level of the Open vSwitch source
1649 distribution describes the in-band model in detail.
1651 If your controllers are not actually in-band (e.g. they are on
1652 localhost via 127.0.0.1, or on a separate network), then you should
1653 configure your controllers in "out-of-band" mode. If you have one
1654 controller on bridge br0, then you can configure out-of-band mode
1657 ovs-vsctl set controller br0 connection-mode=out-of-band
1659 ### Q: I configured all my controllers for out-of-band control mode but
1660 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1662 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1663 set-manager"). By default, Open vSwitch assumes that managers need
1664 in-band rules set up on every bridge. You can disable these rules
1667 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1669 This actually disables in-band control entirely for the bridge, as
1670 if all the bridge's controllers were configured for out-of-band
1673 ### Q: My OpenFlow controller doesn't see the VLANs that I expect.
1675 A: See answer under "VLANs", above.
1677 ### Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1678 but I got a funny message like this:
1680 ofp_util|INFO|normalization changed ofp_match, details:
1681 ofp_util|INFO| pre: nw_dst=192.168.0.1
1684 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1685 match had disappeared, so that the flow ends up matching every
1688 A: The term "normalization" in the log message means that a flow
1689 cannot match on an L3 field without saying what L3 protocol is in
1690 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1691 so the L3 field match was dropped.
1693 In this case, the L3 protocol could be IP or ARP. A correct
1694 command for each possibility is, respectively:
1696 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1700 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1702 Similarly, a flow cannot match on an L4 field without saying what
1703 L4 protocol is in use. For example, the flow match "tp_src=1234"
1704 is, by itself, meaningless and will be ignored. Instead, to match
1705 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1706 source port 1234, write "udp,tp_src=1234".
1708 ### Q: How can I figure out the OpenFlow port number for a given port?
1710 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1711 respond with an OFPT_FEATURES_REPLY that, among other information,
1712 includes a mapping between OpenFlow port names and numbers. From a
1713 command prompt, "ovs-ofctl show br0" makes such a request and
1714 prints the response for switch br0.
1716 The Interface table in the Open vSwitch database also maps OpenFlow
1717 port names to numbers. To print the OpenFlow port number
1718 associated with interface eth0, run:
1720 ovs-vsctl get Interface eth0 ofport
1722 You can print the entire mapping with:
1724 ovs-vsctl -- --columns=name,ofport list Interface
1726 but the output mixes together interfaces from all bridges in the
1727 database, so it may be confusing if more than one bridge exists.
1729 In the Open vSwitch database, ofport value -1 means that the
1730 interface could not be created due to an error. (The Open vSwitch
1731 log should indicate the reason.) ofport value [] (the empty set)
1732 means that the interface hasn't been created yet. The latter is
1733 normally an intermittent condition (unless ovs-vswitchd is not
1736 ### Q: I added some flows with my controller or with ovs-ofctl, but when I
1737 run "ovs-dpctl dump-flows" I don't see them.
1739 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1740 won't display the information that you want. You want to use
1741 "ovs-ofctl dump-flows" instead.
1743 ### Q: It looks like each of the interfaces in my bonded port shows up
1744 as an individual OpenFlow port. Is that right?
1746 A: Yes, Open vSwitch makes individual bond interfaces visible as
1747 OpenFlow ports, rather than the bond as a whole. The interfaces
1748 are treated together as a bond for only a few purposes:
1750 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1751 controller is not configured, this happens implicitly to
1754 - Mirrors configured for output to a bonded port.
1756 It would make a lot of sense for Open vSwitch to present a bond as
1757 a single OpenFlow port. If you want to contribute an
1758 implementation of such a feature, please bring it up on the Open
1759 vSwitch development mailing list at dev@openvswitch.org.
1761 ### Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1762 multiple hosts, and other components. The behavior isn't what I
1765 A: To debug network behavior problems, trace the path of a packet,
1766 hop-by-hop, from its origin in one host to a remote host. If
1767 that's correct, then trace the path of the response packet back to
1770 The open source tool called "plotnetcfg" can help to understand the
1771 relationship between the networking devices on a single host.
1773 Usually a simple ICMP echo request and reply ("ping") packet is
1774 good enough. Start by initiating an ongoing "ping" from the origin
1775 host to a remote host. If you are tracking down a connectivity
1776 problem, the "ping" will not display any successful output, but
1777 packets are still being sent. (In this case the packets being sent
1778 are likely ARP rather than ICMP.)
1780 Tools available for tracing include the following:
1782 - "tcpdump" and "wireshark" for observing hops across network
1783 devices, such as Open vSwitch internal devices and physical
1786 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1787 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1788 These tools allow one to observe the actions being taken on
1789 packets in ongoing flows.
1791 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1792 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1793 documentation, and "Why are there so many different ways to
1794 dump flows?" above for some background.
1796 - "ovs-appctl ofproto/trace" to observe the logic behind how
1797 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1798 documentation. You can out more details about a given flow
1799 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1800 a flow from the output into an "ovs-appctl ofproto/trace"
1803 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1804 observe what goes on at these physical hops.
1806 Starting at the origin of a given packet, observe the packet at
1807 each hop in turn. For example, in one plausible scenario, you
1810 1. "tcpdump" the "eth" interface through which an ARP egresses
1811 a VM, from inside the VM.
1813 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1814 ingresses the host machine.
1816 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1817 the host interface through which the ARP egresses the
1818 physical machine. You may need to use "ovs-dpctl show" to
1819 interpret the port numbers. If the output seems surprising,
1820 you can use "ovs-appctl ofproto/trace" to observe details of
1821 how ovs-vswitchd determined the actions in the "ovs-dpctl
1824 4. "tcpdump" the "eth" interface through which the ARP egresses
1825 the physical machine.
1827 5. "tcpdump" the "eth" interface through which the ARP
1828 ingresses the physical machine, at the remote host that
1831 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1832 remote host that receives the ARP and observe the VM "vif"
1833 or "tap" interface to which the flow is directed. Again,
1834 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1836 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1839 8. "tcpdump" the "eth" interface through which the ARP
1840 ingresses a VM, from inside the VM.
1842 It is likely that during one of these steps you will figure out the
1843 problem. If not, then follow the ARP reply back to the origin, in
1846 ### Q: How do I make a flow drop packets?
1848 A: To drop a packet is to receive it without forwarding it. OpenFlow
1849 explicitly specifies forwarding actions. Thus, a flow with an
1850 empty set of actions does not forward packets anywhere, causing
1851 them to be dropped. You can specify an empty set of actions with
1852 "actions=" on the ovs-ofctl command line. For example:
1854 ovs-ofctl add-flow br0 priority=65535,actions=
1856 would cause every packet entering switch br0 to be dropped.
1858 You can write "drop" explicitly if you like. The effect is the
1859 same. Thus, the following command also causes every packet
1860 entering switch br0 to be dropped:
1862 ovs-ofctl add-flow br0 priority=65535,actions=drop
1864 "drop" is not an action, either in OpenFlow or Open vSwitch.
1865 Rather, it is only a way to say that there are no actions.
1867 ### Q: I added a flow to send packets out the ingress port, like this:
1869 ovs-ofctl add-flow br0 in_port=2,actions=2
1871 but OVS drops the packets instead.
1873 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1874 out its ingress port. The rationale is that dropping these packets
1875 makes it harder to loop the network. Sometimes this behavior can
1876 even be convenient, e.g. it is often the desired behavior in a flow
1877 that forwards a packet to several ports ("floods" the packet).
1879 Sometimes one really needs to send a packet out its ingress port
1880 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1881 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1883 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1885 This also works in some circumstances where the flow doesn't match
1886 on the input port. For example, if you know that your switch has
1887 five ports numbered 2 through 6, then the following will send every
1888 received packet out every port, even its ingress port:
1890 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1894 ovs-ofctl add-flow br0 actions=all,in_port
1896 Sometimes, in complicated flow tables with multiple levels of
1897 "resubmit" actions, a flow needs to output to a particular port
1898 that may or may not be the ingress port. It's difficult to take
1899 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1900 provides, as an OpenFlow extension, the ability to modify the
1901 in_port field. Whatever value is currently in the in_port field is
1902 the port to which outputs will be dropped, as well as the
1903 destination for OFPP_IN_PORT. This means that the following will
1904 reliably output to port 2 or to ports 2 through 6, respectively:
1906 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1907 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1909 If the input port is important, then one may save and restore it on
1912 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1913 load:0->NXM_OF_IN_PORT[],\
1915 pop:NXM_OF_IN_PORT[]
1917 ### Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1918 on port 2. I set up flows to forward only traffic destined to the
1919 other host and drop other traffic, like this:
1921 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1922 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1923 priority=0,actions=drop
1925 But it doesn't work--I don't get any connectivity when I do this.
1928 A: These flows drop the ARP packets that IP hosts use to establish IP
1929 connectivity over Ethernet. To solve the problem, add flows to
1930 allow ARP to pass between the hosts:
1932 priority=5,in_port=1,arp,actions=2
1933 priority=5,in_port=2,arp,actions=1
1935 This issue can manifest other ways, too. The following flows that
1936 match on Ethernet addresses instead of IP addresses will also drop
1937 ARP packets, because ARP requests are broadcast instead of being
1938 directed to a specific host:
1940 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1941 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1942 priority=0,actions=drop
1944 The solution already described above will also work in this case.
1945 It may be better to add flows to allow all multicast and broadcast
1948 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1949 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1951 ### Q: My bridge disconnects from my controller on add-port/del-port.
1953 A: Reconfiguring your bridge can change your bridge's datapath-id because
1954 Open vSwitch generates datapath-id from the MAC address of one of its ports.
1955 In that case, Open vSwitch disconnects from controllers because there's
1956 no graceful way to notify controllers about the change of datapath-id.
1958 To avoid the behaviour, you can configure datapath-id manually.
1960 ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef
1962 ### Q: My controller is getting errors about "buffers". What's going on?
1964 A: When a switch sends a packet to an OpenFlow controller using a
1965 "packet-in" message, it can also keep a copy of that packet in a
1966 "buffer", identified by a 32-bit integer "buffer_id". There are
1967 two advantages to buffering. First, when the controller wants to
1968 tell the switch to do something with the buffered packet (with a
1969 "packet-out" OpenFlow request), it does not need to send another
1970 copy of the packet back across the OpenFlow connection, which
1971 reduces the bandwidth cost of the connection and improves latency.
1972 This enables the second advantage: the switch can optionally send
1973 only the first part of the packet to the controller (assuming that
1974 the switch only needs to look at the first few bytes of the
1975 packet), further reducing bandwidth and improving latency.
1977 However, buffering introduces some issues of its own. First, any
1978 switch has limited resources, so if the controller does not use a
1979 buffered packet, the switch has to decide how long to keep it
1980 buffered. When many packets are sent to a controller and buffered,
1981 Open vSwitch can discard buffered packets that the controller has
1982 not used after as little as 5 seconds. This means that
1983 controllers, if they make use of packet buffering, should use the
1984 buffered packets promptly. (This includes sending a "packet-out"
1985 with no actions if the controller does not want to do anything with
1986 a buffered packet, to clear the packet buffer and effectively
1989 Second, packet buffers are one-time-use, meaning that a controller
1990 cannot use a single packet buffer in two or more "packet-out"
1991 commands. Open vSwitch will respond with an error to the second
1992 and subsequent "packet-out"s in such a case.
1994 Finally, a common error early in controller development is to try
1995 to use buffer_id 0 in a "packet-out" message as if 0 represented
1996 "no buffered packet". This is incorrect usage: the buffer_id with
1997 this meaning is actually 0xffffffff.
1999 ovs-vswitchd(8) describes some details of Open vSwitch packet
2000 buffering that the OpenFlow specification requires implementations
2003 ### Q: How does OVS divide flows among buckets in an OpenFlow "select" group?
2005 A: In Open vSwitch 2.3 and earlier, Open vSwitch used the destination
2006 Ethernet address to choose a bucket in a select group.
2008 Open vSwitch 2.4 and later by default hashes the source and
2009 destination Ethernet address, VLAN ID, Ethernet type, IPv4/v6
2010 source and destination address and protocol, and for TCP and SCTP
2011 only, the source and destination ports. The hash is "symmetric",
2012 meaning that exchanging source and destination addresses does not
2013 change the bucket selection.
2015 Select groups in Open vSwitch 2.4 and later can be configured to
2016 use a different hash function, using a Netronome extension to the
2017 OpenFlow 1.5+ group_mod message. For more information, see
2018 Documentation/group-selection-method-property.txt in the Open
2019 vSwitch source tree. (OpenFlow 1.5 support in Open vSwitch is still
2026 ### Q: How do I implement a new OpenFlow message?
2028 A: Add your new message to "enum ofpraw" and "enum ofptype" in
2029 lib/ofp-msgs.h, following the existing pattern. Then recompile and
2030 fix all of the new warnings, implementing new functionality for the
2031 new message as needed. (If you configure with --enable-Werror, as
2032 described in [INSTALL.md], then it is impossible to miss any warnings.)
2034 If you need to add an OpenFlow vendor extension message for a
2035 vendor that doesn't yet have any extension messages, then you will
2036 also need to edit build-aux/extract-ofp-msgs.
2038 ### Q: How do I add support for a new field or header?
2040 A: Add new members for your field to "struct flow" in lib/flow.h, and
2041 add new enumerations for your new field to "enum mf_field_id" in
2042 lib/meta-flow.h, following the existing pattern. Also, add support
2043 to miniflow_extract() in lib/flow.c for extracting your new field
2044 from a packet into struct miniflow, and to nx_put_raw() in
2045 lib/nx-match.c to output your new field in OXM matches. Then
2046 recompile and fix all of the new warnings, implementing new
2047 functionality for the new field or header as needed. (If you
2048 configure with --enable-Werror, as described in [INSTALL.md], then
2049 it is impossible to miss any warnings.)
2051 If you want kernel datapath support for your new field, you also
2052 need to modify the kernel module for the operating systems you are
2053 interested in. This isn't mandatory, since fields understood only
2054 by userspace work too (with a performance penalty), so it's
2055 reasonable to start development without it. If you implement
2056 kernel module support for Linux, then the Linux kernel "netdev"
2057 mailing list is the place to submit that support first; please read
2058 up on the Linux kernel development process separately. The Windows
2059 datapath kernel module support, on the other hand, is maintained
2060 within the OVS tree, so patches for that can go directly to
2063 ### Q: How do I add support for a new OpenFlow action?
2065 A: Add your new action to "enum ofp_raw_action_type" in
2066 lib/ofp-actions.c, following the existing pattern. Then recompile
2067 and fix all of the new warnings, implementing new functionality for
2068 the new action as needed. (If you configure with --enable-Werror,
2069 as described in [INSTALL.md], then it is impossible to miss any
2072 If you need to add an OpenFlow vendor extension action for a vendor
2073 that doesn't yet have any extension actions, then you will also
2074 need to edit build-aux/extract-ofp-actions.
2080 bugs@openvswitch.org
2081 http://openvswitch.org/
2083 [PORTING.md]:PORTING.md
2084 [WHY-OVS.md]:WHY-OVS.md
2085 [INSTALL.md]:INSTALL.md
2086 [OPENFLOW-1.1+.md]:OPENFLOW-1.1+.md
2087 [INSTALL.DPDK.md]:INSTALL.DPDK.md
projects / ovs.git / blob