1 Open vSwitch <http://openvswitch.org>
3 Frequently Asked Questions
4 ==========================
9 Q: What is Open vSwitch?
11 A: Open vSwitch is a production quality open source software switch
12 designed to be used as a vswitch in virtualized server
13 environments. A vswitch forwards traffic between different VMs on
14 the same physical host and also forwards traffic between VMs and
15 the physical network. Open vSwitch supports standard management
16 interfaces (e.g. sFlow, NetFlow, IPFIX, RSPAN, CLI), and is open to
17 programmatic extension and control using OpenFlow and the OVSDB
20 Open vSwitch as designed to be compatible with modern switching
21 chipsets. This means that it can be ported to existing high-fanout
22 switches allowing the same flexible control of the physical
23 infrastructure as the virtual infrastructure. It also means that
24 Open vSwitch will be able to take advantage of on-NIC switching
25 chipsets as their functionality matures.
27 Q: What virtualization platforms can use Open vSwitch?
29 A: Open vSwitch can currently run on any Linux-based virtualization
30 platform (kernel 2.6.32 and newer), including: KVM, VirtualBox, Xen,
31 Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the
32 mainline kernel. The bulk of the code is written in platform-
33 independent C and is easily ported to other environments. We welcome
34 inquires about integrating Open vSwitch with other virtualization
37 Q: How can I try Open vSwitch?
39 A: The Open vSwitch source code can be built on a Linux system. You can
40 build and experiment with Open vSwitch on any Linux machine.
41 Packages for various Linux distributions are available on many
42 platforms, including: Debian, Ubuntu, Fedora.
44 You may also download and run a virtualization platform that already
45 has Open vSwitch integrated. For example, download a recent ISO for
46 XenServer or Xen Cloud Platform. Be aware that the version
47 integrated with a particular platform may not be the most recent Open
50 Q: Does Open vSwitch only work on Linux?
52 A: No, Open vSwitch has been ported to a number of different operating
53 systems and hardware platforms. Most of the development work occurs
54 on Linux, but the code should be portable to any POSIX system. We've
55 seen Open vSwitch ported to a number of different platforms,
56 including FreeBSD, Windows, and even non-POSIX embedded systems.
58 By definition, the Open vSwitch Linux kernel module only works on
59 Linux and will provide the highest performance. However, a userspace
60 datapath is available that should be very portable.
62 Q: What's involved with porting Open vSwitch to a new platform or
65 A: The PORTING document describes how one would go about porting Open
66 vSwitch to a new operating system or hardware platform.
68 Q: Why would I use Open vSwitch instead of the Linux bridge?
70 A: Open vSwitch is specially designed to make it easier to manage VM
71 network configuration and monitor state spread across many physical
72 hosts in dynamic virtualized environments. Please see WHY-OVS for a
73 more detailed description of how Open vSwitch relates to the Linux
76 Q: How is Open vSwitch related to distributed virtual switches like the
77 VMware vNetwork distributed switch or the Cisco Nexus 1000V?
79 A: Distributed vswitch applications (e.g., VMware vNetwork distributed
80 switch, Cisco Nexus 1000V) provide a centralized way to configure and
81 monitor the network state of VMs that are spread across many physical
82 hosts. Open vSwitch is not a distributed vswitch itself, rather it
83 runs on each physical host and supports remote management in a way
84 that makes it easier for developers of virtualization/cloud
85 management platforms to offer distributed vswitch capabilities.
87 To aid in distribution, Open vSwitch provides two open protocols that
88 are specially designed for remote management in virtualized network
89 environments: OpenFlow, which exposes flow-based forwarding state,
90 and the OVSDB management protocol, which exposes switch port state.
91 In addition to the switch implementation itself, Open vSwitch
92 includes tools (ovs-ofctl, ovs-vsctl) that developers can script and
93 extend to provide distributed vswitch capabilities that are closely
94 integrated with their virtualization management platform.
96 Q: Why doesn't Open vSwitch support distribution?
98 A: Open vSwitch is intended to be a useful component for building
99 flexible network infrastructure. There are many different approaches
100 to distribution which balance trade-offs between simplicity,
101 scalability, hardware compatibility, convergence times, logical
102 forwarding model, etc. The goal of Open vSwitch is to be able to
103 support all as a primitive building block rather than choose a
104 particular point in the distributed design space.
106 Q: How can I contribute to the Open vSwitch Community?
108 A: You can start by joining the mailing lists and helping to answer
109 questions. You can also suggest improvements to documentation. If
110 you have a feature or bug you would like to work on, send a mail to
111 one of the mailing lists:
113 http://openvswitch.org/mlists/
119 Q: What does it mean for an Open vSwitch release to be LTS (long-term
122 A: All official releases have been through a comprehensive testing
123 process and are suitable for production use. Planned releases will
124 occur several times a year. If a significant bug is identified in an
125 LTS release, we will provide an updated release that includes the
126 fix. Releases that are not LTS may not be fixed and may just be
127 supplanted by the next major release. The current LTS release is
130 Q: What Linux kernel versions does each Open vSwitch release work with?
132 A: The following table lists the Linux kernel versions against which the
133 given versions of the Open vSwitch kernel module will successfully
134 build. The Linux kernel versions are upstream kernel versions, so
135 Linux kernels modified from the upstream sources may not build in
136 some cases even if they are based on a supported version. This is
137 most notably true of Red Hat Enterprise Linux (RHEL) kernels, which
138 are extensively modified from upstream.
140 Open vSwitch Linux kernel
141 ------------ -------------
154 Open vSwitch userspace should also work with the Linux kernel module
155 built into Linux 3.3 and later.
157 Open vSwitch userspace is not sensitive to the Linux kernel version.
158 It should build against almost any kernel, certainly against 2.6.32
161 Q: I get an error like this when I configure Open vSwitch:
163 configure: error: Linux kernel in <dir> is version <x>, but
164 version newer than <y> is not supported (please refer to the
169 A: If there is a newer version of Open vSwitch, consider building that
170 one, because it may support the kernel that you are building
171 against. (To find out, consult the table in the previous answer.)
173 Otherwise, use the Linux kernel module supplied with the kernel
174 that you are using. All versions of Open vSwitch userspace are
175 compatible with all versions of the Open vSwitch kernel module, so
176 this will also work. See also the following question.
178 Q: What features are not available in the Open vSwitch kernel datapath
179 that ships as part of the upstream Linux kernel?
181 A: The kernel module in upstream Linux does not include support for
182 LISP. Work is in progress to add support for LISP to the upstream
183 Linux version of the Open vSwitch kernel module. For now, if you
184 need this feature, use the kernel module from the Open vSwitch
185 distribution instead of the upstream Linux kernel module.
187 Certain features require kernel support to function or to have
188 reasonable performance. If the ovs-vswitchd log file indicates that
189 a feature is not supported, consider upgrading to a newer upstream
190 Linux release or using the kernel module paired with the userspace
193 Q: What features are not available when using the userspace datapath?
195 A: Tunnel virtual ports are not supported, as described in the
196 previous answer. It is also not possible to use queue-related
197 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
198 may not be transmitted.
200 Q: What Linux kernel versions does IPFIX flow monitoring work with?
202 A: IPFIX flow monitoring requires the Linux kernel module from Open
203 vSwitch version 1.10.90 or later.
205 Q: Should userspace or kernel be upgraded first to minimize downtime?
207 In general, the Open vSwitch userspace should be used with the
208 kernel version included in the same release or with the version
209 from upstream Linux. However, when upgrading between two releases
210 of Open vSwitch it is best to migrate userspace first to reduce
211 the possibility of incompatibilities.
213 Q: What happened to the bridge compatibility feature?
215 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
216 When it was enabled, Open vSwitch imitated the interface of the
217 Linux kernel "bridge" module. This allowed users to drop Open
218 vSwitch into environments designed to use the Linux kernel bridge
219 module without adapting the environment to use Open vSwitch.
221 Open vSwitch 1.10 and later do not support bridge compatibility.
222 The feature was dropped because version 1.10 adopted a new internal
223 architecture that made bridge compatibility difficult to maintain.
224 Now that many environments use OVS directly, it would be rarely
227 To use bridge compatibility, install OVS 1.9 or earlier, including
228 the accompanying kernel modules (both the main and bridge
229 compatibility modules), following the instructions that come with
230 the release. Be sure to start the ovs-brcompatd daemon.
236 Q: I thought Open vSwitch was a virtual Ethernet switch, but the
237 documentation keeps talking about bridges. What's a bridge?
239 A: In networking, the terms "bridge" and "switch" are synonyms. Open
240 vSwitch implements an Ethernet switch, which means that it is also
245 A: See the "VLAN" section below.
251 Q: How do I configure a port as an access port?
253 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
254 the following commands configure br0 with eth0 as a trunk port (the
255 default) and tap0 as an access port for VLAN 9:
258 ovs-vsctl add-port br0 eth0
259 ovs-vsctl add-port br0 tap0 tag=9
261 If you want to configure an already added port as an access port,
262 use "ovs-vsctl set", e.g.:
264 ovs-vsctl set port tap0 tag=9
266 Q: How do I configure a port as a SPAN port, that is, enable mirroring
267 of all traffic to that port?
269 A: The following commands configure br0 with eth0 and tap0 as trunk
270 ports. All traffic coming in or going out on eth0 or tap0 is also
271 mirrored to tap1; any traffic arriving on tap1 is dropped:
274 ovs-vsctl add-port br0 eth0
275 ovs-vsctl add-port br0 tap0
276 ovs-vsctl add-port br0 tap1 \
277 -- --id=@p get port tap1 \
278 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
279 -- set bridge br0 mirrors=@m
281 To later disable mirroring, run:
283 ovs-vsctl clear bridge br0 mirrors
285 Q: Does Open vSwitch support configuring a port in promiscuous mode?
287 A: Yes. How you configure it depends on what you mean by "promiscuous
290 - Conventionally, "promiscuous mode" is a feature of a network
291 interface card. Ordinarily, a NIC passes to the CPU only the
292 packets actually destined to its host machine. It discards
293 the rest to avoid wasting memory and CPU cycles. When
294 promiscuous mode is enabled, however, it passes every packet
295 to the CPU. On an old-style shared-media or hub-based
296 network, this allows the host to spy on all packets on the
297 network. But in the switched networks that are almost
298 everywhere these days, promiscuous mode doesn't have much
299 effect, because few packets not destined to a host are
300 delivered to the host's NIC.
302 This form of promiscuous mode is configured in the guest OS of
303 the VMs on your bridge, e.g. with "ifconfig".
305 - The VMware vSwitch uses a different definition of "promiscuous
306 mode". When you configure promiscuous mode on a VMware vNIC,
307 the vSwitch sends a copy of every packet received by the
308 vSwitch to that vNIC. That has a much bigger effect than just
309 enabling promiscuous mode in a guest OS. Rather than getting
310 a few stray packets for which the switch does not yet know the
311 correct destination, the vNIC gets every packet. The effect
312 is similar to replacing the vSwitch by a virtual hub.
314 This "promiscuous mode" is what switches normally call "port
315 mirroring" or "SPAN". For information on how to configure
316 SPAN, see "How do I configure a port as a SPAN port, that is,
317 enable mirroring of all traffic to that port?"
319 Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable
320 mirroring of all traffic to that VLAN?
322 A: The following commands configure br0 with eth0 as a trunk port and
323 tap0 as an access port for VLAN 10. All traffic coming in or going
324 out on tap0, as well as traffic coming in or going out on eth0 in
325 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
326 VLAN 10, in cases where one is present, is dropped as part of
330 ovs-vsctl add-port br0 eth0
331 ovs-vsctl add-port br0 tap0 tag=10
333 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
335 -- set bridge br0 mirrors=@m
337 To later disable mirroring, run:
339 ovs-vsctl clear bridge br0 mirrors
341 Mirroring to a VLAN can disrupt a network that contains unmanaged
342 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
343 GRE tunnel has fewer caveats than mirroring to a VLAN and should
344 generally be preferred.
346 Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
348 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
349 of the specified output-vlan. This loss of information may make
350 the mirrored traffic too hard to interpret.
352 To mirror multiple VLANs, use the commands above, but specify a
353 comma-separated list of VLANs as the value for select-vlan. To
354 mirror every VLAN, use the commands above, but omit select-vlan and
357 When a packet arrives on a VLAN that is used as a mirror output
358 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
359 floods the packet across all the ports for which the mirror output
360 VLAN is configured. (If an OpenFlow controller is in use, then it
361 can override this behavior through the flow table.) If OVS is used
362 as an intermediate switch, rather than an edge switch, this ensures
363 that the RSPAN traffic is distributed through the network.
365 Mirroring to a VLAN can disrupt a network that contains unmanaged
366 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
367 GRE tunnel has fewer caveats than mirroring to a VLAN and should
368 generally be preferred.
370 Q: How do I configure mirroring of all traffic to a GRE tunnel?
372 A: The following commands configure br0 with eth0 and tap0 as trunk
373 ports. All traffic coming in or going out on eth0 or tap0 is also
374 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
375 traffic arriving on gre0 is dropped:
378 ovs-vsctl add-port br0 eth0
379 ovs-vsctl add-port br0 tap0
380 ovs-vsctl add-port br0 gre0 \
381 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
382 -- --id=@p get port gre0 \
383 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
384 -- set bridge br0 mirrors=@m
386 To later disable mirroring and destroy the GRE tunnel:
388 ovs-vsctl clear bridge br0 mirrors
389 ovs-vcstl del-port br0 gre0
391 Q: Does Open vSwitch support ERSPAN?
393 A: No. ERSPAN is an undocumented proprietary protocol. As an
394 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
397 Q: How do I connect two bridges?
399 A: First, why do you want to do this? Two connected bridges are not
400 much different from a single bridge, so you might as well just have
401 a single bridge with all your ports on it.
403 If you still want to connect two bridges, you can use a pair of
404 patch ports. The following example creates bridges br0 and br1,
405 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
406 and br1 with a pair of patch ports.
409 ovs-vsctl add-port br0 eth0
410 ovs-vsctl add-port br0 tap0
412 ovs-vsctl add-port br1 tap1
414 -- add-port br0 patch0 \
415 -- set interface patch0 type=patch options:peer=patch1 \
416 -- add-port br1 patch1 \
417 -- set interface patch1 type=patch options:peer=patch0
419 Bridges connected with patch ports are much like a single bridge.
420 For instance, if the example above also added eth1 to br1, and both
421 eth0 and eth1 happened to be connected to the same next-hop switch,
422 then you could loop your network just as you would if you added
423 eth0 and eth1 to the same bridge (see the "Configuration Problems"
424 section below for more information).
426 If you are using Open vSwitch 1.9 or an earlier version, then you
427 need to be using the kernel module bundled with Open vSwitch rather
428 than the one that is integrated into Linux 3.3 and later, because
429 Open vSwitch 1.9 and earlier versions need kernel support for patch
430 ports. This also means that in Open vSwitch 1.9 and earlier, patch
431 ports will not work with the userspace datapath, only with the
434 Q: How do I configure a bridge without an OpenFlow local port?
435 (Local port in the sense of OFPP_LOCAL)
437 A: Open vSwitch does not support such a configuration.
438 Bridges always have their local ports.
441 Implementation Details
442 ----------------------
444 Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
446 A: Open vSwitch uses different kinds of flows for different purposes:
448 - OpenFlow flows are the most important kind of flow. OpenFlow
449 controllers use these flows to define a switch's policy.
450 OpenFlow flows support wildcards, priorities, and multiple
453 When in-band control is in use, Open vSwitch sets up a few
454 "hidden" flows, with priority higher than a controller or the
455 user can configure, that are not visible via OpenFlow. (See
456 the "Controller" section of the FAQ for more information
459 - The Open vSwitch software switch implementation uses a second
460 kind of flow internally. These flows, called "datapath" or
461 "kernel" flows, do not support priorities and comprise only a
462 single table, which makes them suitable for caching. (Like
463 OpenFlow flows, datapath flows do support wildcarding, in Open
464 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
465 also support different actions and number ports differently.
467 Datapath flows are an implementation detail that is subject to
468 change in future versions of Open vSwitch. Even with the
469 current version of Open vSwitch, hardware switch
470 implementations do not necessarily use this architecture.
472 Users and controllers directly control only the OpenFlow flow
473 table. Open vSwitch manages the datapath flow table itself, so
474 users should not normally be concerned with it.
476 Q: Why are there so many different ways to dump flows?
478 A: Open vSwitch has two kinds of flows (see the previous question), so
479 it has commands with different purposes for dumping each kind of
482 - "ovs-ofctl dump-flows <br>" dumps OpenFlow flows, excluding
483 hidden flows. This is the most commonly useful form of flow
484 dump. (Unlike the other commands, this should work with any
485 OpenFlow switch, not just Open vSwitch.)
487 - "ovs-appctl bridge/dump-flows <br>" dumps OpenFlow flows,
488 including hidden flows. This is occasionally useful for
489 troubleshooting suspected issues with in-band control.
491 - "ovs-dpctl dump-flows [dp]" dumps the datapath flow table
492 entries for a Linux kernel-based datapath. In Open vSwitch
493 1.10 and later, ovs-vswitchd merges multiple switches into a
494 single datapath, so it will show all the flows on all your
495 kernel-based switches. This command can occasionally be
496 useful for debugging.
498 - "ovs-appctl dpif/dump-flows <br>", new in Open vSwitch 1.10,
499 dumps datapath flows for only the specified bridge, regardless
506 Q: I just upgraded and I see a performance drop. Why?
508 A: The OVS kernel datapath may have been updated to a newer version than
509 the OVS userspace components. Sometimes new versions of OVS kernel
510 module add functionality that is backwards compatible with older
511 userspace components but may cause a drop in performance with them.
512 Especially, if a kernel module from OVS 2.1 or newer is paired with
513 OVS userspace 1.10 or older, there will be a performance drop for
516 Updating the OVS userspace components to the latest released
517 version should fix the performance degradation.
519 To get the best possible performance and functionality, it is
520 recommended to pair the same versions of the kernel module and OVS
524 Configuration Problems
525 ----------------------
527 Q: I created a bridge and added my Ethernet port to it, using commands
531 ovs-vsctl add-port br0 eth0
533 and as soon as I ran the "add-port" command I lost all connectivity
536 A: A physical Ethernet device that is part of an Open vSwitch bridge
537 should not have an IP address. If one does, then that IP address
538 will not be fully functional.
540 You can restore functionality by moving the IP address to an Open
541 vSwitch "internal" device, such as the network device named after
542 the bridge itself. For example, assuming that eth0's IP address is
543 192.168.128.5, you could run the commands below to fix up the
546 ifconfig eth0 0.0.0.0
547 ifconfig br0 192.168.128.5
549 (If your only connection to the machine running OVS is through the
550 IP address in question, then you would want to run all of these
551 commands on a single command line, or put them into a script.) If
552 there were any additional routes assigned to eth0, then you would
553 also want to use commands to adjust these routes to go through br0.
555 If you use DHCP to obtain an IP address, then you should kill the
556 DHCP client that was listening on the physical Ethernet interface
557 (e.g. eth0) and start one listening on the internal interface
558 (e.g. br0). You might still need to manually clear the IP address
559 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
561 There is no compelling reason why Open vSwitch must work this way.
562 However, this is the way that the Linux kernel bridge module has
563 always worked, so it's a model that those accustomed to Linux
564 bridging are already used to. Also, the model that most people
565 expect is not implementable without kernel changes on all the
566 versions of Linux that Open vSwitch supports.
568 By the way, this issue is not specific to physical Ethernet
569 devices. It applies to all network devices except Open vswitch
572 Q: I created a bridge and added a couple of Ethernet ports to it,
573 using commands like these:
576 ovs-vsctl add-port br0 eth0
577 ovs-vsctl add-port br0 eth1
579 and now my network seems to have melted: connectivity is unreliable
580 (even connectivity that doesn't go through Open vSwitch), all the
581 LEDs on my physical switches are blinking, wireshark shows
582 duplicated packets, and CPU usage is very high.
584 A: More than likely, you've looped your network. Probably, eth0 and
585 eth1 are connected to the same physical Ethernet switch. This
586 yields a scenario where OVS receives a broadcast packet on eth0 and
587 sends it out on eth1, then the physical switch connected to eth1
588 sends the packet back on eth0, and so on forever. More complicated
589 scenarios, involving a loop through multiple switches, are possible
592 The solution depends on what you are trying to do:
594 - If you added eth0 and eth1 to get higher bandwidth or higher
595 reliability between OVS and your physical Ethernet switch,
596 use a bond. The following commands create br0 and then add
597 eth0 and eth1 as a bond:
600 ovs-vsctl add-bond br0 bond0 eth0 eth1
602 Bonds have tons of configuration options. Please read the
603 documentation on the Port table in ovs-vswitchd.conf.db(5)
606 - Perhaps you don't actually need eth0 and eth1 to be on the
607 same bridge. For example, if you simply want to be able to
608 connect each of them to virtual machines, then you can put
609 each of them on a bridge of its own:
612 ovs-vsctl add-port br0 eth0
615 ovs-vsctl add-port br1 eth1
617 and then connect VMs to br0 and br1. (A potential
618 disadvantage is that traffic cannot directly pass between br0
619 and br1. Instead, it will go out eth0 and come back in eth1,
622 - If you have a redundant or complex network topology and you
623 want to prevent loops, turn on spanning tree protocol (STP).
624 The following commands create br0, enable STP, and add eth0
625 and eth1 to the bridge. The order is important because you
626 don't want have to have a loop in your network even
630 ovs-vsctl set bridge br0 stp_enable=true
631 ovs-vsctl add-port br0 eth0
632 ovs-vsctl add-port br0 eth1
634 The Open vSwitch implementation of STP is not well tested.
635 Please report any bugs you observe, but if you'd rather avoid
636 acting as a beta tester then another option might be your
639 Q: I can't seem to use Open vSwitch in a wireless network.
641 A: Wireless base stations generally only allow packets with the source
642 MAC address of NIC that completed the initial handshake.
643 Therefore, without MAC rewriting, only a single device can
644 communicate over a single wireless link.
646 This isn't specific to Open vSwitch, it's enforced by the access
647 point, so the same problems will show up with the Linux bridge or
648 any other way to do bridging.
650 Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
652 A: PPP most commonly carries IP packets, but Open vSwitch works only
653 with Ethernet frames. The correct way to interface PPP to an
654 Ethernet network is usually to use routing instead of switching.
656 Q: Is there any documentation on the database tables and fields?
658 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
660 Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
661 I only see a datapath called "ovs-system". How can I see datapath
662 information about a particular bridge?
664 A: In version 1.9.0, OVS switched to using a single datapath that is
665 shared by all bridges of that type. The "ovs-appctl dpif/*"
666 commands provide similar functionality that is scoped by the bridge.
668 Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
669 see the port in the datapath?
671 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
672 cannot be loaded at the same time. It is likely that on your system the
673 Linux GRE module is already loaded and blocking OVS (to confirm, check
674 dmesg for errors regarding GRE registration). To fix this, unload all
675 GRE modules that appear in lsmod as well as the OVS kernel module. You
676 can then reload the OVS module following the directions in INSTALL,
677 which will ensure that dependencies are satisfied.
679 Q: Open vSwitch does not seem to obey my packet filter rules.
681 A: It depends on mechanisms and configurations you want to use.
683 You cannot usefully use typical packet filters, like iptables, on
684 physical Ethernet ports that you add to an Open vSwitch bridge.
685 This is because Open vSwitch captures packets from the interface at
686 a layer lower below where typical packet-filter implementations
687 install their hooks. (This actually applies to any interface of
688 type "system" that you might add to an Open vSwitch bridge.)
690 You can usefully use typical packet filters on Open vSwitch
691 internal ports as they are mostly ordinary interfaces from the point
692 of view of packet filters.
694 For example, suppose you create a bridge br0 and add Ethernet port
695 eth0 to it. Then you can usefully add iptables rules to affect the
696 internal interface br0, but not the physical interface eth0. (br0
697 is also where you would add an IP address, as discussed elsewhere
700 For simple filtering rules, it might be possible to achieve similar
701 results by installing appropriate OpenFlow flows instead.
703 If the use of a particular packet filter setup is essential, Open
704 vSwitch might not be the best choice for you. On Linux, you might
705 want to consider using the Linux Bridge. (This is the only choice if
706 you want to use ebtables rules.) On NetBSD, you might want to
707 consider using the bridge(4) with BRIDGE_IPF option.
710 Quality of Service (QoS)
711 ------------------------
713 Q: How do I configure Quality of Service (QoS)?
715 A: Suppose that you want to set up bridge br0 connected to physical
716 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
717 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
718 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
719 could configure the bridge this way:
723 add-port br0 eth0 -- \
724 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
725 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
726 set port eth0 qos=@newqos -- \
727 --id=@newqos create qos type=linux-htb \
728 other-config:max-rate=1000000000 \
729 queues:123=@vif10queue \
730 queues:234=@vif20queue -- \
731 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
732 --id=@vif20queue create queue other-config:max-rate=20000000
734 At this point, bridge br0 is configured with the ports and eth0 is
735 configured with the queues that you need for QoS, but nothing is
736 actually directing packets from vif1.0 or vif2.0 to the queues that
737 we have set up for them. That means that all of the packets to
738 eth0 are going to the "default queue", which is not what we want.
740 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
741 queues reserved for them:
743 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
744 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
746 Each of the above flows matches on the input port, sets up the
747 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
748 executes the "normal" action, which performs the same switching
749 that Open vSwitch would have done without any OpenFlow flows being
750 present. (We know that vif1.0 and vif2.0 have OpenFlow port
751 numbers 5 and 6, respectively, because we set their ofport_request
752 columns above. If we had not done that, then we would have needed
753 to find out their port numbers before setting up these flows.)
755 Now traffic going from vif1.0 or vif2.0 to eth0 should be
758 By the way, if you delete the bridge created by the above commands,
763 then that will leave one unreferenced QoS record and two
764 unreferenced Queue records in the Open vSwich database. One way to
765 clear them out, assuming you don't have other QoS or Queue records
766 that you want to keep, is:
768 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
770 If you do want to keep some QoS or Queue records, or the Open
771 vSwitch you are using is older than version 1.8 (which added the
772 --all option), then you will have to destroy QoS and Queue records
775 Q: I configured Quality of Service (QoS) in my OpenFlow network by
776 adding records to the QoS and Queue table, but the results aren't
779 A: Did you install OpenFlow flows that use your queues? This is the
780 primary way to tell Open vSwitch which queues you want to use. If
781 you don't do this, then the default queue will be used, which will
782 probably not have the effect you want.
784 Refer to the previous question for an example.
786 Q: I'd like to take advantage of some QoS feature that Open vSwitch
787 doesn't yet support. How do I do that?
789 A: Open vSwitch does not implement QoS itself. Instead, it can
790 configure some, but not all, of the QoS features built into the
791 Linux kernel. If you need some QoS feature that OVS cannot
792 configure itself, then the first step is to figure out whether
793 Linux QoS supports that feature. If it does, then you can submit a
794 patch to support Open vSwitch configuration for that feature, or
795 you can use "tc" directly to configure the feature in Linux. (If
796 Linux QoS doesn't support the feature you want, then first you have
797 to add that support to Linux.)
799 Q: I configured QoS, correctly, but my measurements show that it isn't
800 working as well as I expect.
802 A: With the Linux kernel, the Open vSwitch implementation of QoS has
805 - Open vSwitch configures a subset of Linux kernel QoS
806 features, according to what is in OVSDB. It is possible that
807 this code has bugs. If you believe that this is so, then you
808 can configure the Linux traffic control (QoS) stack directly
809 with the "tc" program. If you get better results that way,
810 you can send a detailed bug report to bugs@openvswitch.org.
812 It is certain that Open vSwitch cannot configure every Linux
813 kernel QoS feature. If you need some feature that OVS cannot
814 configure, then you can also use "tc" directly (or add that
817 - The Open vSwitch implementation of OpenFlow allows flows to
818 be directed to particular queues. This is pretty simple and
819 unlikely to have serious bugs at this point.
821 However, most problems with QoS on Linux are not bugs in Open
822 vSwitch at all. They tend to be either configuration errors
823 (please see the earlier questions in this section) or issues with
824 the traffic control (QoS) stack in Linux. The Open vSwitch
825 developers are not experts on Linux traffic control. We suggest
826 that, if you believe you are encountering a problem with Linux
827 traffic control, that you consult the tc manpages (e.g. tc(8),
828 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
829 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
831 Q: Does Open vSwitch support OpenFlow meters?
833 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
834 OpenFlow meters. There is no implementation of meters in the Open
835 vSwitch software switch (neither the kernel-based nor userspace
844 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
845 partition a single switch into multiple switches. Suppose, for
846 example, that you have two groups of machines, group A and group B.
847 You want the machines in group A to be able to talk to each other,
848 and you want the machine in group B to be able to talk to each
849 other, but you don't want the machines in group A to be able to
850 talk to the machines in group B. You can do this with two
851 switches, by plugging the machines in group A into one switch and
852 the machines in group B into the other switch.
854 If you only have one switch, then you can use VLANs to do the same
855 thing, by configuring the ports for machines in group A as VLAN
856 "access ports" for one VLAN and the ports for group B as "access
857 ports" for a different VLAN. The switch will only forward packets
858 between ports that are assigned to the same VLAN, so this
859 effectively subdivides your single switch into two independent
860 switches, one for each group of machines.
862 So far we haven't said anything about VLAN headers. With access
863 ports, like we've described so far, no VLAN header is present in
864 the Ethernet frame. This means that the machines (or switches)
865 connected to access ports need not be aware that VLANs are
866 involved, just like in the case where we use two different physical
869 Now suppose that you have a whole bunch of switches in your
870 network, instead of just one, and that some machines in group A are
871 connected directly to both switches 1 and 2. To allow these
872 machines to talk to each other, you could add an access port for
873 group A's VLAN to switch 1 and another to switch 2, and then
874 connect an Ethernet cable between those ports. That works fine,
875 but it doesn't scale well as the number of switches and the number
876 of VLANs increases, because you use up a lot of valuable switch
877 ports just connecting together your VLANs.
879 This is where VLAN headers come in. Instead of using one cable and
880 two ports per VLAN to connect a pair of switches, we configure a
881 port on each switch as a VLAN "trunk port". Packets sent and
882 received on a trunk port carry a VLAN header that says what VLAN
883 the packet belongs to, so that only two ports total are required to
884 connect the switches, regardless of the number of VLANs in use.
885 Normally, only switches (either physical or virtual) are connected
886 to a trunk port, not individual hosts, because individual hosts
887 don't expect to see a VLAN header in the traffic that they receive.
889 None of the above discussion says anything about particular VLAN
890 numbers. This is because VLAN numbers are completely arbitrary.
891 One must only ensure that a given VLAN is numbered consistently
892 throughout a network and that different VLANs are given different
893 numbers. (That said, VLAN 0 is usually synonymous with a packet
894 that has no VLAN header, and VLAN 4095 is reserved.)
898 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
899 bugs. If you are having problems with VLANs that you suspect to be
900 driver related, then you have several options:
902 - Upgrade to Linux 3.3 or later.
904 - Build and install a fixed version of the particular driver
905 that is causing trouble, if one is available.
907 - Use a NIC whose driver does not have VLAN problems.
909 - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later
910 that works around bugs in kernel drivers. To enable VLAN
911 splinters on interface eth0, use the command:
913 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
915 For VLAN splinters to be effective, Open vSwitch must know
916 which VLANs are in use. See the "VLAN splinters" section in
917 the Interface table in ovs-vswitchd.conf.db(5) for details on
918 how Open vSwitch infers in-use VLANs.
920 VLAN splinters increase memory use and reduce performance, so
921 use them only if needed.
923 - Apply the "vlan workaround" patch from the XenServer kernel
924 patch queue, build Open vSwitch against this patched kernel,
925 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
926 workaround for each interface whose driver is buggy.
928 (This is a nontrivial exercise, so this option is included
929 only for completeness.)
931 It is not always easy to tell whether a Linux kernel driver has
932 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
933 can help you test. See their manpages for details. Of the two
934 utilities, ovs-test(8) is newer and more thorough, but
935 ovs-vlan-test(8) may be easier to use.
937 Q: VLANs still don't work. I've tested the driver so I know that it's OK.
939 A: Do you have VLANs enabled on the physical switch that OVS is
940 attached to? Make sure that the port is configured to trunk the
941 VLAN or VLANs that you are using with OVS.
943 Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
944 and to its destination host, but OVS seems to drop incoming return
947 A: It's possible that you have the VLAN configured on your physical
948 switch as the "native" VLAN. In this mode, the switch treats
949 incoming packets either tagged with the native VLAN or untagged as
950 part of the native VLAN. It may also send outgoing packets in the
951 native VLAN without a VLAN tag.
953 If this is the case, you have two choices:
955 - Change the physical switch port configuration to tag packets
956 it forwards to OVS with the native VLAN instead of forwarding
959 - Change the OVS configuration for the physical port to a
960 native VLAN mode. For example, the following sets up a
961 bridge with port eth0 in "native-tagged" mode in VLAN 9:
964 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
966 In this situation, "native-untagged" mode will probably work
967 equally well. Refer to the documentation for the Port table
968 in ovs-vswitchd.conf.db(5) for more information.
970 Q: I added a pair of VMs on different VLANs, like this:
973 ovs-vsctl add-port br0 eth0
974 ovs-vsctl add-port br0 tap0 tag=9
975 ovs-vsctl add-port br0 tap1 tag=10
977 but the VMs can't access each other, the external network, or the
980 A: It is to be expected that the VMs can't access each other. VLANs
981 are a means to partition a network. When you configured tap0 and
982 tap1 as access ports for different VLANs, you indicated that they
983 should be isolated from each other.
985 As for the external network and the Internet, it seems likely that
986 the machines you are trying to access are not on VLAN 9 (or 10) and
987 that the Internet is not available on VLAN 9 (or 10).
989 Q: I added a pair of VMs on the same VLAN, like this:
992 ovs-vsctl add-port br0 eth0
993 ovs-vsctl add-port br0 tap0 tag=9
994 ovs-vsctl add-port br0 tap1 tag=9
996 The VMs can access each other, but not the external network or the
999 A: It seems likely that the machines you are trying to access in the
1000 external network are not on VLAN 9 and that the Internet is not
1001 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
1002 trunk VLAN on the upstream switch port to which eth0 is connected.
1004 Q: Can I configure an IP address on a VLAN?
1006 A: Yes. Use an "internal port" configured as an access port. For
1007 example, the following configures IP address 192.168.0.7 on VLAN 9.
1008 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
1009 they have an 802.1Q header with VLAN 9. Conversely, traffic
1010 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
1013 ovs-vsctl add-br br0
1014 ovs-vsctl add-port br0 eth0
1015 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1016 ifconfig vlan9 192.168.0.7
1018 See also the following question.
1020 Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
1023 ovs-vsctl add-br br0
1024 ovs-vsctl add-port br0 eth0
1025 ifconfig br0 192.168.0.5
1026 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1027 ifconfig vlan9 192.168.0.9
1029 but other hosts that are only on VLAN 0 can reach the IP address
1030 configured on VLAN 9. What's going on?
1032 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
1033 approaches to IP address handling in Internet hosts:
1035 - In the "Strong ES Model", where an ES is a host ("End
1036 System"), an IP address is primarily associated with a
1037 particular interface. The host discards packets that arrive
1038 on interface A if they are destined for an IP address that is
1039 configured on interface B. The host never sends packets from
1040 interface A using a source address configured on interface B.
1042 - In the "Weak ES Model", an IP address is primarily associated
1043 with a host. The host accepts packets that arrive on any
1044 interface if they are destined for any of the host's IP
1045 addresses, even if the address is configured on some
1046 interface other than the one on which it arrived. The host
1047 does not restrict itself to sending packets from an IP
1048 address associated with the originating interface.
1050 Linux uses the weak ES model. That means that when packets
1051 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
1052 br0, the kernel IP stack accepts them there for the VLAN 9 IP
1053 address, even though they were not received on vlan9, the network
1056 To simulate the strong ES model on Linux, one may add iptables rule
1057 to filter packets based on source and destination address and
1058 adjust ARP configuration with sysctls.
1060 BSD uses the strong ES model.
1062 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1064 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1065 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1066 implementation of the OpenFlow "normal switching" action. By
1067 default, when Open vSwitch isn't connected to a controller and
1068 nothing has been manually configured in the flow table, all traffic
1069 goes through the "normal switching" action. But, if you set up
1070 OpenFlow flows on your own, through a controller or using ovs-ofctl
1071 or through other means, then you have to implement VLAN handling
1074 You can use "normal switching" as a component of your OpenFlow
1075 actions, e.g. by putting "normal" into the lists of actions on
1076 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1077 controller. In situations where this is not suitable, you can
1078 implement VLAN handling yourself, e.g.:
1080 - If a packet comes in on an access port, and the flow table
1081 needs to send it out on a trunk port, then the flow can add
1082 the appropriate VLAN tag with the "mod_vlan_vid" action.
1084 - If a packet comes in on a trunk port, and the flow table
1085 needs to send it out on an access port, then the flow can
1086 strip the VLAN tag with the "strip_vlan" action.
1088 Q: I configured ports on a bridge as access ports with different VLAN
1091 ovs-vsctl add-br br0
1092 ovs-vsctl set-controller br0 tcp:192.168.0.10:6633
1093 ovs-vsctl add-port br0 eth0
1094 ovs-vsctl add-port br0 tap0 tag=9
1095 ovs-vsctl add-port br0 tap1 tag=10
1097 but the VMs running behind tap0 and tap1 can still communicate,
1098 that is, they are not isolated from each other even though they are
1101 A: Do you have a controller configured on br0 (as the commands above
1102 do)? If so, then this is a variant on the previous question, "My
1103 OpenFlow controller doesn't see the VLANs that I expect," and you
1104 can refer to the answer there for more information.
1112 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1113 to solve the scaling challenges of VLAN networks in a multi-tenant
1114 environment. VXLAN is an overlay network which transports an L2 network
1115 over an existing L3 network. For more information on VXLAN, please see
1116 the IETF draft available here:
1118 http://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-03
1120 Q: How much of the VXLAN protocol does Open vSwitch currently support?
1122 A: Open vSwitch currently supports the framing format for packets on the
1123 wire. There is currently no support for the multicast aspects of VXLAN.
1124 To get around the lack of multicast support, it is possible to
1125 pre-provision MAC to IP address mappings either manually or from a
1128 Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1131 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1132 is 4789. However, it is possible to configure the destination UDP port
1133 manually on a per-VXLAN tunnel basis. An example of this configuration is
1136 ovs-vsctl add-br br0
1137 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1138 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1139 options:dst_port=8472
1142 Using OpenFlow (Manually or Via Controller)
1143 -------------------------------------------
1145 Q: What versions of OpenFlow does Open vSwitch support?
1147 A: The following table lists the versions of OpenFlow supported by
1148 each version of Open vSwitch:
1150 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4
1151 =============== ===== ===== ===== ===== =====
1152 1.9 and earlier yes --- --- --- ---
1153 1.10 yes --- [*] [*] ---
1154 1.11 yes --- [*] [*] ---
1155 2.0 yes [*] [*] [*] ---
1156 2.1 yes [*] [*] [*] ---
1157 2.2 yes [*] [*] [*] [%]
1158 2.3 yes yes yes yes [%]
1160 [*] Supported, with one or more missing features.
1161 [%] Support is unsafe: ovs-vswitchd will abort when certain
1162 unimplemented features are tested.
1164 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default
1165 in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1,
1166 1.2, and 1.3 must be enabled manually in ovs-vswitchd. Either way,
1167 the user may override the default:
1169 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0:
1171 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1173 - To enable only OpenFlow 1.0 on bridge br0:
1175 ovs-vsctl set bridge br0 protocols=OpenFlow10
1177 All current versions of ovs-ofctl enable only OpenFlow 1.0 by
1178 default. Use the -O option to enable support for later versions of
1179 OpenFlow in ovs-ofctl. For example:
1181 ovs-ofctl -O OpenFlow13 dump-flows br0
1183 OpenFlow 1.4 is a special case, because it is not implemented
1184 safely: ovs-vswitchd will abort when certain unimplemented features
1185 are tested. Thus, for now it is suitable only for experimental
1186 use. ovs-vswitchd will only allow OpenFlow 1.4 to be enabled
1187 (which must be done in the same way described above) when it is
1188 invoked with a special --enable-of14 command line option.
1190 OPENFLOW-1.1+ in the Open vSwitch source tree tracks support for
1191 OpenFlow 1.1 and later features. When support for OpenFlow 1.4 is
1192 solidly implemented, Open vSwitch will enable that version by
1195 Q: Does Open vSwitch support MPLS?
1197 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1198 these versions can match on MPLS Ethernet types, but they cannot
1199 match, push, or pop MPLS labels, nor can they look past MPLS labels
1200 into the encapsulated packet.
1202 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1203 for MPLS. With the userspace datapath only, these versions can
1204 match, push, or pop a single MPLS label, but they still cannot look
1205 past MPLS labels (even after popping them) into the encapsulated
1206 packet. Kernel datapath support is unchanged from earlier
1209 Open vSwitch version 2.2 will be able to match, push, or pop up to
1210 3 MPLS labels. Looking past MPLS labels into the encapsulated
1211 packet will still be unsupported. Both userspace and kernel
1212 datapaths will be supported, but MPLS processing always happens in
1213 userspace either way, so kernel datapath performance will be
1216 Q: I'm getting "error type 45250 code 0". What's that?
1218 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1219 vSwitch uses this extension when it must report an error to an
1220 OpenFlow controller but no standard OpenFlow error code is
1223 Open vSwitch logs the errors that it sends to controllers, so the
1224 easiest thing to do is probably to look at the ovs-vswitchd log to
1225 find out what the error was.
1227 If you want to dissect the extended error message yourself, the
1228 format is documented in include/openflow/nicira-ext.h in the Open
1229 vSwitch source distribution. The extended error codes are
1230 documented in lib/ofp-errors.h.
1232 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1233 doesn't actually appear through the OpenFlow connection, even
1234 though I know that it's going through.
1235 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1236 to apply to certain traffic, especially traffic between OVS and
1237 the controller itself.
1239 A: By default, Open vSwitch assumes that OpenFlow controllers are
1240 connected "in-band", that is, that the controllers are actually
1241 part of the network that is being controlled. In in-band mode,
1242 Open vSwitch sets up special "hidden" flows to make sure that
1243 traffic can make it back and forth between OVS and the controllers.
1244 These hidden flows are higher priority than any flows that can be
1245 set up through OpenFlow, and they are not visible through normal
1246 OpenFlow flow table dumps.
1248 Usually, the hidden flows are desirable and helpful, but
1249 occasionally they can cause unexpected behavior. You can view the
1250 full OpenFlow flow table, including hidden flows, on bridge br0
1253 ovs-appctl bridge/dump-flows br0
1255 to help you debug. The hidden flows are those with priorities
1256 greater than 65535 (the maximum priority that can be set with
1259 The DESIGN file at the top level of the Open vSwitch source
1260 distribution describes the in-band model in detail.
1262 If your controllers are not actually in-band (e.g. they are on
1263 localhost via 127.0.0.1, or on a separate network), then you should
1264 configure your controllers in "out-of-band" mode. If you have one
1265 controller on bridge br0, then you can configure out-of-band mode
1268 ovs-vsctl set controller br0 connection-mode=out-of-band
1270 Q: I configured all my controllers for out-of-band control mode but
1271 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1273 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1274 set-manager"). By default, Open vSwitch assumes that managers need
1275 in-band rules set up on every bridge. You can disable these rules
1278 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1280 This actually disables in-band control entirely for the bridge, as
1281 if all the bridge's controllers were configured for out-of-band
1284 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1286 A: See answer under "VLANs", above.
1288 Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1289 but I got a funny message like this:
1291 ofp_util|INFO|normalization changed ofp_match, details:
1292 ofp_util|INFO| pre: nw_dst=192.168.0.1
1295 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1296 match had disappeared, so that the flow ends up matching every
1299 A: The term "normalization" in the log message means that a flow
1300 cannot match on an L3 field without saying what L3 protocol is in
1301 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1302 so the L3 field match was dropped.
1304 In this case, the L3 protocol could be IP or ARP. A correct
1305 command for each possibility is, respectively:
1307 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1311 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1313 Similarly, a flow cannot match on an L4 field without saying what
1314 L4 protocol is in use. For example, the flow match "tp_src=1234"
1315 is, by itself, meaningless and will be ignored. Instead, to match
1316 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1317 source port 1234, write "udp,tp_src=1234".
1319 Q: How can I figure out the OpenFlow port number for a given port?
1321 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1322 respond with an OFPT_FEATURES_REPLY that, among other information,
1323 includes a mapping between OpenFlow port names and numbers. From a
1324 command prompt, "ovs-ofctl show br0" makes such a request and
1325 prints the response for switch br0.
1327 The Interface table in the Open vSwitch database also maps OpenFlow
1328 port names to numbers. To print the OpenFlow port number
1329 associated with interface eth0, run:
1331 ovs-vsctl get Interface eth0 ofport
1333 You can print the entire mapping with:
1335 ovs-vsctl -- --columns=name,ofport list Interface
1337 but the output mixes together interfaces from all bridges in the
1338 database, so it may be confusing if more than one bridge exists.
1340 In the Open vSwitch database, ofport value -1 means that the
1341 interface could not be created due to an error. (The Open vSwitch
1342 log should indicate the reason.) ofport value [] (the empty set)
1343 means that the interface hasn't been created yet. The latter is
1344 normally an intermittent condition (unless ovs-vswitchd is not
1347 Q: I added some flows with my controller or with ovs-ofctl, but when I
1348 run "ovs-dpctl dump-flows" I don't see them.
1350 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1351 won't display the information that you want. You want to use
1352 "ovs-ofctl dump-flows" instead.
1354 Q: It looks like each of the interfaces in my bonded port shows up
1355 as an individual OpenFlow port. Is that right?
1357 A: Yes, Open vSwitch makes individual bond interfaces visible as
1358 OpenFlow ports, rather than the bond as a whole. The interfaces
1359 are treated together as a bond for only a few purposes:
1361 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1362 controller is not configured, this happens implicitly to
1365 - Mirrors configured for output to a bonded port.
1367 It would make a lot of sense for Open vSwitch to present a bond as
1368 a single OpenFlow port. If you want to contribute an
1369 implementation of such a feature, please bring it up on the Open
1370 vSwitch development mailing list at dev@openvswitch.org.
1372 Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1373 multiple hosts, and other components. The behavior isn't what I
1376 A: To debug network behavior problems, trace the path of a packet,
1377 hop-by-hop, from its origin in one host to a remote host. If
1378 that's correct, then trace the path of the response packet back to
1381 Usually a simple ICMP echo request and reply ("ping") packet is
1382 good enough. Start by initiating an ongoing "ping" from the origin
1383 host to a remote host. If you are tracking down a connectivity
1384 problem, the "ping" will not display any successful output, but
1385 packets are still being sent. (In this case the packets being sent
1386 are likely ARP rather than ICMP.)
1388 Tools available for tracing include the following:
1390 - "tcpdump" and "wireshark" for observing hops across network
1391 devices, such as Open vSwitch internal devices and physical
1394 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1395 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1396 These tools allow one to observe the actions being taken on
1397 packets in ongoing flows.
1399 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1400 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1401 documentation, and "Why are there so many different ways to
1402 dump flows?" above for some background.
1404 - "ovs-appctl ofproto/trace" to observe the logic behind how
1405 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1406 documentation. You can out more details about a given flow
1407 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1408 a flow from the output into an "ovs-appctl ofproto/trace"
1411 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1412 observe what goes on at these physical hops.
1414 Starting at the origin of a given packet, observe the packet at
1415 each hop in turn. For example, in one plausible scenario, you
1418 1. "tcpdump" the "eth" interface through which an ARP egresses
1419 a VM, from inside the VM.
1421 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1422 ingresses the host machine.
1424 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1425 the host interface through which the ARP egresses the
1426 physical machine. You may need to use "ovs-dpctl show" to
1427 interpret the port numbers. If the output seems surprising,
1428 you can use "ovs-appctl ofproto/trace" to observe details of
1429 how ovs-vswitchd determined the actions in the "ovs-dpctl
1432 4. "tcpdump" the "eth" interface through which the ARP egresses
1433 the physical machine.
1435 5. "tcpdump" the "eth" interface through which the ARP
1436 ingresses the physical machine, at the remote host that
1439 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1440 remote host that receives the ARP and observe the VM "vif"
1441 or "tap" interface to which the flow is directed. Again,
1442 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1444 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1447 8. "tcpdump" the "eth" interface through which the ARP
1448 ingresses a VM, from inside the VM.
1450 It is likely that during one of these steps you will figure out the
1451 problem. If not, then follow the ARP reply back to the origin, in
1454 Q: How do I make a flow drop packets?
1456 A: To drop a packet is to receive it without forwarding it. OpenFlow
1457 explicitly specifies forwarding actions. Thus, a flow with an
1458 empty set of actions does not forward packets anywhere, causing
1459 them to be dropped. You can specify an empty set of actions with
1460 "actions=" on the ovs-ofctl command line. For example:
1462 ovs-ofctl add-flow br0 priority=65535,actions=
1464 would cause every packet entering switch br0 to be dropped.
1466 You can write "drop" explicitly if you like. The effect is the
1467 same. Thus, the following command also causes every packet
1468 entering switch br0 to be dropped:
1470 ovs-ofctl add-flow br0 priority=65535,actions=drop
1472 "drop" is not an action, either in OpenFlow or Open vSwitch.
1473 Rather, it is only a way to say that there are no actions.
1475 Q: I added a flow to send packets out the ingress port, like this:
1477 ovs-ofctl add-flow br0 in_port=2,actions=2
1479 but OVS drops the packets instead.
1481 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1482 out its ingress port. The rationale is that dropping these packets
1483 makes it harder to loop the network. Sometimes this behavior can
1484 even be convenient, e.g. it is often the desired behavior in a flow
1485 that forwards a packet to several ports ("floods" the packet).
1487 Sometimes one really needs to send a packet out its ingress port
1488 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1489 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1491 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1493 This also works in some circumstances where the flow doesn't match
1494 on the input port. For example, if you know that your switch has
1495 five ports numbered 2 through 6, then the following will send every
1496 received packet out every port, even its ingress port:
1498 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1502 ovs-ofctl add-flow br0 actions=all,in_port
1504 Sometimes, in complicated flow tables with multiple levels of
1505 "resubmit" actions, a flow needs to output to a particular port
1506 that may or may not be the ingress port. It's difficult to take
1507 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1508 provides, as an OpenFlow extension, the ability to modify the
1509 in_port field. Whatever value is currently in the in_port field is
1510 the port to which outputs will be dropped, as well as the
1511 destination for OFPP_IN_PORT. This means that the following will
1512 reliably output to port 2 or to ports 2 through 6, respectively:
1514 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1515 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1517 If the input port is important, then one may save and restore it on
1520 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1521 load:0->NXM_OF_IN_PORT[],\
1523 pop:NXM_OF_IN_PORT[]
1525 Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1526 on port 2. I set up flows to forward only traffic destined to the
1527 other host and drop other traffic, like this:
1529 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1530 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1531 priority=0,actions=drop
1533 But it doesn't work--I don't get any connectivity when I do this.
1536 A: These flows drop the ARP packets that IP hosts use to establish IP
1537 connectivity over Ethernet. To solve the problem, add flows to
1538 allow ARP to pass between the hosts:
1540 priority=5,in_port=1,arp,actions=2
1541 priority=5,in_port=2,arp,actions=1
1543 This issue can manifest other ways, too. The following flows that
1544 match on Ethernet addresses instead of IP addresses will also drop
1545 ARP packets, because ARP requests are broadcast instead of being
1546 directed to a specific host:
1548 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1549 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1550 priority=0,actions=drop
1552 The solution already described above will also work in this case.
1553 It may be better to add flows to allow all multicast and broadcast
1556 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1557 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1563 Q: How do I implement a new OpenFlow message?
1565 A: Add your new message to "enum ofpraw" and "enum ofptype" in
1566 lib/ofp-msgs.h, following the existing pattern. Then recompile and
1567 fix all of the new warnings, implementing new functionality for the
1568 new message as needed. (If you configure with --enable-Werror, as
1569 described in INSTALL, then it is impossible to miss any warnings.)
1571 If you need to add an OpenFlow vendor extension message for a
1572 vendor that doesn't yet have any extension messages, then you will
1573 also need to edit build-aux/extract-ofp-msgs.
1579 bugs@openvswitch.org
1580 http://openvswitch.org/