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 ------------ -------------
153 Open vSwitch userspace should also work with the Linux kernel module
154 built into Linux 3.3 and later.
156 Open vSwitch userspace is not sensitive to the Linux kernel version.
157 It should build against almost any kernel, certainly against 2.6.32
160 Q: What Linux kernel versions does IPFIX flow monitoring work with?
162 A: IPFIX flow monitoring requires the Linux kernel module from Open
163 vSwitch version 1.10.90 or later.
165 Q: Should userspace or kernel be upgraded first to minimize downtime?
167 In general, the Open vSwitch userspace should be used with the
168 kernel version included in the same release or with the version
169 from upstream Linux. However, when upgrading between two releases
170 of Open vSwitch it is best to migrate userspace first to reduce
171 the possibility of incompatibilities.
173 Q: What features are not available in the Open vSwitch kernel datapath
174 that ships as part of the upstream Linux kernel?
176 A: The kernel module in upstream Linux does not include support for
177 LISP. Work is in progress to add support for LISP to the upstream
178 Linux version of the Open vSwitch kernel module. For now, if you
179 need this feature, use the kernel module from the Open vSwitch
180 distribution instead of the upstream Linux kernel module.
182 Certain features require kernel support to function or to have
183 reasonable performance. If the ovs-vswitchd log file indicates that
184 a feature is not supported, consider upgrading to a newer upstream
185 Linux release or using the kernel module paired with the userspace
188 Q: What features are not available when using the userspace datapath?
190 A: Tunnel virtual ports are not supported, as described in the
191 previous answer. It is also not possible to use queue-related
192 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
193 may not be transmitted.
195 Q: What happened to the bridge compatibility feature?
197 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
198 When it was enabled, Open vSwitch imitated the interface of the
199 Linux kernel "bridge" module. This allowed users to drop Open
200 vSwitch into environments designed to use the Linux kernel bridge
201 module without adapting the environment to use Open vSwitch.
203 Open vSwitch 1.10 and later do not support bridge compatibility.
204 The feature was dropped because version 1.10 adopted a new internal
205 architecture that made bridge compatibility difficult to maintain.
206 Now that many environments use OVS directly, it would be rarely
209 To use bridge compatibility, install OVS 1.9 or earlier, including
210 the accompanying kernel modules (both the main and bridge
211 compatibility modules), following the instructions that come with
212 the release. Be sure to start the ovs-brcompatd daemon.
218 Q: I thought Open vSwitch was a virtual Ethernet switch, but the
219 documentation keeps talking about bridges. What's a bridge?
221 A: In networking, the terms "bridge" and "switch" are synonyms. Open
222 vSwitch implements an Ethernet switch, which means that it is also
227 A: See the "VLAN" section below.
233 Q: How do I configure a port as an access port?
235 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
236 the following commands configure br0 with eth0 as a trunk port (the
237 default) and tap0 as an access port for VLAN 9:
240 ovs-vsctl add-port br0 eth0
241 ovs-vsctl add-port br0 tap0 tag=9
243 If you want to configure an already added port as an access port,
244 use "ovs-vsctl set", e.g.:
246 ovs-vsctl set port tap0 tag=9
248 Q: How do I configure a port as a SPAN port, that is, enable mirroring
249 of all traffic to that port?
251 A: The following commands configure br0 with eth0 and tap0 as trunk
252 ports. All traffic coming in or going out on eth0 or tap0 is also
253 mirrored to tap1; any traffic arriving on tap1 is dropped:
256 ovs-vsctl add-port br0 eth0
257 ovs-vsctl add-port br0 tap0
258 ovs-vsctl add-port br0 tap1 \
259 -- --id=@p get port tap1 \
260 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
261 -- set bridge br0 mirrors=@m
263 To later disable mirroring, run:
265 ovs-vsctl clear bridge br0 mirrors
267 Q: Does Open vSwitch support configuring a port in promiscuous mode?
269 A: Yes. How you configure it depends on what you mean by "promiscuous
272 - Conventionally, "promiscuous mode" is a feature of a network
273 interface card. Ordinarily, a NIC passes to the CPU only the
274 packets actually destined to its host machine. It discards
275 the rest to avoid wasting memory and CPU cycles. When
276 promiscuous mode is enabled, however, it passes every packet
277 to the CPU. On an old-style shared-media or hub-based
278 network, this allows the host to spy on all packets on the
279 network. But in the switched networks that are almost
280 everywhere these days, promiscuous mode doesn't have much
281 effect, because few packets not destined to a host are
282 delivered to the host's NIC.
284 This form of promiscuous mode is configured in the guest OS of
285 the VMs on your bridge, e.g. with "ifconfig".
287 - The VMware vSwitch uses a different definition of "promiscuous
288 mode". When you configure promiscuous mode on a VMware vNIC,
289 the vSwitch sends a copy of every packet received by the
290 vSwitch to that vNIC. That has a much bigger effect than just
291 enabling promiscuous mode in a guest OS. Rather than getting
292 a few stray packets for which the switch does not yet know the
293 correct destination, the vNIC gets every packet. The effect
294 is similar to replacing the vSwitch by a virtual hub.
296 This "promiscuous mode" is what switches normally call "port
297 mirroring" or "SPAN". For information on how to configure
298 SPAN, see "How do I configure a port as a SPAN port, that is,
299 enable mirroring of all traffic to that port?"
301 Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable
302 mirroring of all traffic to that VLAN?
304 A: The following commands configure br0 with eth0 as a trunk port and
305 tap0 as an access port for VLAN 10. All traffic coming in or going
306 out on tap0, as well as traffic coming in or going out on eth0 in
307 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
308 VLAN 10, in cases where one is present, is dropped as part of
312 ovs-vsctl add-port br0 eth0
313 ovs-vsctl add-port br0 tap0 tag=10
315 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
317 -- set bridge br0 mirrors=@m
319 To later disable mirroring, run:
321 ovs-vsctl clear bridge br0 mirrors
323 Mirroring to a VLAN can disrupt a network that contains unmanaged
324 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
325 GRE tunnel has fewer caveats than mirroring to a VLAN and should
326 generally be preferred.
328 Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
330 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
331 of the specified output-vlan. This loss of information may make
332 the mirrored traffic too hard to interpret.
334 To mirror multiple VLANs, use the commands above, but specify a
335 comma-separated list of VLANs as the value for select-vlan. To
336 mirror every VLAN, use the commands above, but omit select-vlan and
339 When a packet arrives on a VLAN that is used as a mirror output
340 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
341 floods the packet across all the ports for which the mirror output
342 VLAN is configured. (If an OpenFlow controller is in use, then it
343 can override this behavior through the flow table.) If OVS is used
344 as an intermediate switch, rather than an edge switch, this ensures
345 that the RSPAN traffic is distributed through the network.
347 Mirroring to a VLAN can disrupt a network that contains unmanaged
348 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
349 GRE tunnel has fewer caveats than mirroring to a VLAN and should
350 generally be preferred.
352 Q: How do I configure mirroring of all traffic to a GRE tunnel?
354 A: The following commands configure br0 with eth0 and tap0 as trunk
355 ports. All traffic coming in or going out on eth0 or tap0 is also
356 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
357 traffic arriving on gre0 is dropped:
360 ovs-vsctl add-port br0 eth0
361 ovs-vsctl add-port br0 tap0
362 ovs-vsctl add-port br0 gre0 \
363 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
364 -- --id=@p get port gre0 \
365 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
366 -- set bridge br0 mirrors=@m
368 To later disable mirroring and destroy the GRE tunnel:
370 ovs-vsctl clear bridge br0 mirrors
371 ovs-vcstl del-port br0 gre0
373 Q: Does Open vSwitch support ERSPAN?
375 A: No. ERSPAN is an undocumented proprietary protocol. As an
376 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
379 Q: How do I connect two bridges?
381 A: First, why do you want to do this? Two connected bridges are not
382 much different from a single bridge, so you might as well just have
383 a single bridge with all your ports on it.
385 If you still want to connect two bridges, you can use a pair of
386 patch ports. The following example creates bridges br0 and br1,
387 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
388 and br1 with a pair of patch ports.
391 ovs-vsctl add-port br0 eth0
392 ovs-vsctl add-port br0 tap0
394 ovs-vsctl add-port br1 tap1
396 -- add-port br0 patch0 \
397 -- set interface patch0 type=patch options:peer=patch1 \
398 -- add-port br1 patch1 \
399 -- set interface patch1 type=patch options:peer=patch0
401 Bridges connected with patch ports are much like a single bridge.
402 For instance, if the example above also added eth1 to br1, and both
403 eth0 and eth1 happened to be connected to the same next-hop switch,
404 then you could loop your network just as you would if you added
405 eth0 and eth1 to the same bridge (see the "Configuration Problems"
406 section below for more information).
408 If you are using Open vSwitch 1.9 or an earlier version, then you
409 need to be using the kernel module bundled with Open vSwitch rather
410 than the one that is integrated into Linux 3.3 and later, because
411 Open vSwitch 1.9 and earlier versions need kernel support for patch
412 ports. This also means that in Open vSwitch 1.9 and earlier, patch
413 ports will not work with the userspace datapath, only with the
417 Implementation Details
418 ----------------------
420 Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
422 A: Open vSwitch uses different kinds of flows for different purposes:
424 - OpenFlow flows are the most important kind of flow. OpenFlow
425 controllers use these flows to define a switch's policy.
426 OpenFlow flows support wildcards, priorities, and multiple
429 When in-band control is in use, Open vSwitch sets up a few
430 "hidden" flows, with priority higher than a controller or the
431 user can configure, that are not visible via OpenFlow. (See
432 the "Controller" section of the FAQ for more information
435 - The Open vSwitch software switch implementation uses a second
436 kind of flow internally. These flows, called "datapath" or
437 "kernel" flows, do not support priorities and comprise only a
438 single table, which makes them suitable for caching. (Like
439 OpenFlow flows, datapath flows do support wildcarding, in Open
440 vSwitch 1.11 and later.) OpenFlow flows and datapath flows
441 also support different actions and number ports differently.
443 Datapath flows are an implementation detail that is subject to
444 change in future versions of Open vSwitch. Even with the
445 current version of Open vSwitch, hardware switch
446 implementations do not necessarily use this architecture.
448 Users and controllers directly control only the OpenFlow flow
449 table. Open vSwitch manages the datapath flow table itself, so
450 users should not normally be concerned with it.
452 Q: Why are there so many different ways to dump flows?
454 A: Open vSwitch has two kinds of flows (see the previous question), so
455 it has commands with different purposes for dumping each kind of
458 - "ovs-ofctl dump-flows <br>" dumps OpenFlow flows, excluding
459 hidden flows. This is the most commonly useful form of flow
460 dump. (Unlike the other commands, this should work with any
461 OpenFlow switch, not just Open vSwitch.)
463 - "ovs-appctl bridge/dump-flows <br>" dumps OpenFlow flows,
464 including hidden flows. This is occasionally useful for
465 troubleshooting suspected issues with in-band control.
467 - "ovs-dpctl dump-flows [dp]" dumps the datapath flow table
468 entries for a Linux kernel-based datapath. In Open vSwitch
469 1.10 and later, ovs-vswitchd merges multiple switches into a
470 single datapath, so it will show all the flows on all your
471 kernel-based switches. This command can occasionally be
472 useful for debugging.
474 - "ovs-appctl dpif/dump-flows <br>", new in Open vSwitch 1.10,
475 dumps datapath flows for only the specified bridge, regardless
482 Q: I just upgraded and I see a performance drop. Why?
484 A: The OVS kernel datapath may have been updated to a newer version than
485 the OVS userspace components. Sometimes new versions of OVS kernel
486 module add functionality that is backwards compatible with older
487 userspace components but may cause a drop in performance with them.
488 Especially, if a kernel module from OVS 2.1 or newer is paired with
489 OVS userspace 1.10 or older, there will be a performance drop for
492 Updating the OVS userspace components to the latest released
493 version should fix the performance degradation.
495 To get the best possible performance and functionality, it is
496 recommended to pair the same versions of the kernel module and OVS
500 Configuration Problems
501 ----------------------
503 Q: I created a bridge and added my Ethernet port to it, using commands
507 ovs-vsctl add-port br0 eth0
509 and as soon as I ran the "add-port" command I lost all connectivity
512 A: A physical Ethernet device that is part of an Open vSwitch bridge
513 should not have an IP address. If one does, then that IP address
514 will not be fully functional.
516 You can restore functionality by moving the IP address to an Open
517 vSwitch "internal" device, such as the network device named after
518 the bridge itself. For example, assuming that eth0's IP address is
519 192.168.128.5, you could run the commands below to fix up the
522 ifconfig eth0 0.0.0.0
523 ifconfig br0 192.168.128.5
525 (If your only connection to the machine running OVS is through the
526 IP address in question, then you would want to run all of these
527 commands on a single command line, or put them into a script.) If
528 there were any additional routes assigned to eth0, then you would
529 also want to use commands to adjust these routes to go through br0.
531 If you use DHCP to obtain an IP address, then you should kill the
532 DHCP client that was listening on the physical Ethernet interface
533 (e.g. eth0) and start one listening on the internal interface
534 (e.g. br0). You might still need to manually clear the IP address
535 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
537 There is no compelling reason why Open vSwitch must work this way.
538 However, this is the way that the Linux kernel bridge module has
539 always worked, so it's a model that those accustomed to Linux
540 bridging are already used to. Also, the model that most people
541 expect is not implementable without kernel changes on all the
542 versions of Linux that Open vSwitch supports.
544 By the way, this issue is not specific to physical Ethernet
545 devices. It applies to all network devices except Open vswitch
548 Q: I created a bridge and added a couple of Ethernet ports to it,
549 using commands like these:
552 ovs-vsctl add-port br0 eth0
553 ovs-vsctl add-port br0 eth1
555 and now my network seems to have melted: connectivity is unreliable
556 (even connectivity that doesn't go through Open vSwitch), all the
557 LEDs on my physical switches are blinking, wireshark shows
558 duplicated packets, and CPU usage is very high.
560 A: More than likely, you've looped your network. Probably, eth0 and
561 eth1 are connected to the same physical Ethernet switch. This
562 yields a scenario where OVS receives a broadcast packet on eth0 and
563 sends it out on eth1, then the physical switch connected to eth1
564 sends the packet back on eth0, and so on forever. More complicated
565 scenarios, involving a loop through multiple switches, are possible
568 The solution depends on what you are trying to do:
570 - If you added eth0 and eth1 to get higher bandwidth or higher
571 reliability between OVS and your physical Ethernet switch,
572 use a bond. The following commands create br0 and then add
573 eth0 and eth1 as a bond:
576 ovs-vsctl add-bond br0 bond0 eth0 eth1
578 Bonds have tons of configuration options. Please read the
579 documentation on the Port table in ovs-vswitchd.conf.db(5)
582 - Perhaps you don't actually need eth0 and eth1 to be on the
583 same bridge. For example, if you simply want to be able to
584 connect each of them to virtual machines, then you can put
585 each of them on a bridge of its own:
588 ovs-vsctl add-port br0 eth0
591 ovs-vsctl add-port br1 eth1
593 and then connect VMs to br0 and br1. (A potential
594 disadvantage is that traffic cannot directly pass between br0
595 and br1. Instead, it will go out eth0 and come back in eth1,
598 - If you have a redundant or complex network topology and you
599 want to prevent loops, turn on spanning tree protocol (STP).
600 The following commands create br0, enable STP, and add eth0
601 and eth1 to the bridge. The order is important because you
602 don't want have to have a loop in your network even
606 ovs-vsctl set bridge br0 stp_enable=true
607 ovs-vsctl add-port br0 eth0
608 ovs-vsctl add-port br0 eth1
610 The Open vSwitch implementation of STP is not well tested.
611 Please report any bugs you observe, but if you'd rather avoid
612 acting as a beta tester then another option might be your
615 Q: I can't seem to use Open vSwitch in a wireless network.
617 A: Wireless base stations generally only allow packets with the source
618 MAC address of NIC that completed the initial handshake.
619 Therefore, without MAC rewriting, only a single device can
620 communicate over a single wireless link.
622 This isn't specific to Open vSwitch, it's enforced by the access
623 point, so the same problems will show up with the Linux bridge or
624 any other way to do bridging.
626 Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
628 A: PPP most commonly carries IP packets, but Open vSwitch works only
629 with Ethernet frames. The correct way to interface PPP to an
630 Ethernet network is usually to use routing instead of switching.
632 Q: Is there any documentation on the database tables and fields?
634 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
636 Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
637 I only see a datapath called "ovs-system". How can I see datapath
638 information about a particular bridge?
640 A: In version 1.9.0, OVS switched to using a single datapath that is
641 shared by all bridges of that type. The "ovs-appctl dpif/*"
642 commands provide similar functionality that is scoped by the bridge.
644 Q: I created a GRE port using ovs-vsctl so why can't I send traffic or
645 see the port in the datapath?
647 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
648 cannot be loaded at the same time. It is likely that on your system the
649 Linux GRE module is already loaded and blocking OVS (to confirm, check
650 dmesg for errors regarding GRE registration). To fix this, unload all
651 GRE modules that appear in lsmod as well as the OVS kernel module. You
652 can then reload the OVS module following the directions in INSTALL,
653 which will ensure that dependencies are satisfied.
656 Quality of Service (QoS)
657 ------------------------
659 Q: How do I configure Quality of Service (QoS)?
661 A: Suppose that you want to set up bridge br0 connected to physical
662 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
663 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
664 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
665 could configure the bridge this way:
669 add-port br0 eth0 -- \
670 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
671 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
672 set port eth0 qos=@newqos -- \
673 --id=@newqos create qos type=linux-htb \
674 other-config:max-rate=1000000000 \
675 queues:123=@vif10queue \
676 queues:234=@vif20queue -- \
677 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
678 --id=@vif20queue create queue other-config:max-rate=20000000
680 At this point, bridge br0 is configured with the ports and eth0 is
681 configured with the queues that you need for QoS, but nothing is
682 actually directing packets from vif1.0 or vif2.0 to the queues that
683 we have set up for them. That means that all of the packets to
684 eth0 are going to the "default queue", which is not what we want.
686 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
687 queues reserved for them:
689 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
690 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
692 Each of the above flows matches on the input port, sets up the
693 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
694 executes the "normal" action, which performs the same switching
695 that Open vSwitch would have done without any OpenFlow flows being
696 present. (We know that vif1.0 and vif2.0 have OpenFlow port
697 numbers 5 and 6, respectively, because we set their ofport_request
698 columns above. If we had not done that, then we would have needed
699 to find out their port numbers before setting up these flows.)
701 Now traffic going from vif1.0 or vif2.0 to eth0 should be
704 By the way, if you delete the bridge created by the above commands,
709 then that will leave one unreferenced QoS record and two
710 unreferenced Queue records in the Open vSwich database. One way to
711 clear them out, assuming you don't have other QoS or Queue records
712 that you want to keep, is:
714 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
716 If you do want to keep some QoS or Queue records, or the Open
717 vSwitch you are using is older than version 1.8 (which added the
718 --all option), then you will have to destroy QoS and Queue records
721 Q: I configured Quality of Service (QoS) in my OpenFlow network by
722 adding records to the QoS and Queue table, but the results aren't
725 A: Did you install OpenFlow flows that use your queues? This is the
726 primary way to tell Open vSwitch which queues you want to use. If
727 you don't do this, then the default queue will be used, which will
728 probably not have the effect you want.
730 Refer to the previous question for an example.
732 Q: I'd like to take advantage of some QoS feature that Open vSwitch
733 doesn't yet support. How do I do that?
735 A: Open vSwitch does not implement QoS itself. Instead, it can
736 configure some, but not all, of the QoS features built into the
737 Linux kernel. If you need some QoS feature that OVS cannot
738 configure itself, then the first step is to figure out whether
739 Linux QoS supports that feature. If it does, then you can submit a
740 patch to support Open vSwitch configuration for that feature, or
741 you can use "tc" directly to configure the feature in Linux. (If
742 Linux QoS doesn't support the feature you want, then first you have
743 to add that support to Linux.)
745 Q: I configured QoS, correctly, but my measurements show that it isn't
746 working as well as I expect.
748 A: With the Linux kernel, the Open vSwitch implementation of QoS has
751 - Open vSwitch configures a subset of Linux kernel QoS
752 features, according to what is in OVSDB. It is possible that
753 this code has bugs. If you believe that this is so, then you
754 can configure the Linux traffic control (QoS) stack directly
755 with the "tc" program. If you get better results that way,
756 you can send a detailed bug report to bugs@openvswitch.org.
758 It is certain that Open vSwitch cannot configure every Linux
759 kernel QoS feature. If you need some feature that OVS cannot
760 configure, then you can also use "tc" directly (or add that
763 - The Open vSwitch implementation of OpenFlow allows flows to
764 be directed to particular queues. This is pretty simple and
765 unlikely to have serious bugs at this point.
767 However, most problems with QoS on Linux are not bugs in Open
768 vSwitch at all. They tend to be either configuration errors
769 (please see the earlier questions in this section) or issues with
770 the traffic control (QoS) stack in Linux. The Open vSwitch
771 developers are not experts on Linux traffic control. We suggest
772 that, if you believe you are encountering a problem with Linux
773 traffic control, that you consult the tc manpages (e.g. tc(8),
774 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
775 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
777 Q: Does Open vSwitch support OpenFlow meters?
779 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
780 OpenFlow meters. There is no implementation of meters in the Open
781 vSwitch software switch (neither the kernel-based nor userspace
790 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
791 partition a single switch into multiple switches. Suppose, for
792 example, that you have two groups of machines, group A and group B.
793 You want the machines in group A to be able to talk to each other,
794 and you want the machine in group B to be able to talk to each
795 other, but you don't want the machines in group A to be able to
796 talk to the machines in group B. You can do this with two
797 switches, by plugging the machines in group A into one switch and
798 the machines in group B into the other switch.
800 If you only have one switch, then you can use VLANs to do the same
801 thing, by configuring the ports for machines in group A as VLAN
802 "access ports" for one VLAN and the ports for group B as "access
803 ports" for a different VLAN. The switch will only forward packets
804 between ports that are assigned to the same VLAN, so this
805 effectively subdivides your single switch into two independent
806 switches, one for each group of machines.
808 So far we haven't said anything about VLAN headers. With access
809 ports, like we've described so far, no VLAN header is present in
810 the Ethernet frame. This means that the machines (or switches)
811 connected to access ports need not be aware that VLANs are
812 involved, just like in the case where we use two different physical
815 Now suppose that you have a whole bunch of switches in your
816 network, instead of just one, and that some machines in group A are
817 connected directly to both switches 1 and 2. To allow these
818 machines to talk to each other, you could add an access port for
819 group A's VLAN to switch 1 and another to switch 2, and then
820 connect an Ethernet cable between those ports. That works fine,
821 but it doesn't scale well as the number of switches and the number
822 of VLANs increases, because you use up a lot of valuable switch
823 ports just connecting together your VLANs.
825 This is where VLAN headers come in. Instead of using one cable and
826 two ports per VLAN to connect a pair of switches, we configure a
827 port on each switch as a VLAN "trunk port". Packets sent and
828 received on a trunk port carry a VLAN header that says what VLAN
829 the packet belongs to, so that only two ports total are required to
830 connect the switches, regardless of the number of VLANs in use.
831 Normally, only switches (either physical or virtual) are connected
832 to a trunk port, not individual hosts, because individual hosts
833 don't expect to see a VLAN header in the traffic that they receive.
835 None of the above discussion says anything about particular VLAN
836 numbers. This is because VLAN numbers are completely arbitrary.
837 One must only ensure that a given VLAN is numbered consistently
838 throughout a network and that different VLANs are given different
839 numbers. (That said, VLAN 0 is usually synonymous with a packet
840 that has no VLAN header, and VLAN 4095 is reserved.)
844 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
845 bugs. If you are having problems with VLANs that you suspect to be
846 driver related, then you have several options:
848 - Upgrade to Linux 3.3 or later.
850 - Build and install a fixed version of the particular driver
851 that is causing trouble, if one is available.
853 - Use a NIC whose driver does not have VLAN problems.
855 - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later
856 that works around bugs in kernel drivers. To enable VLAN
857 splinters on interface eth0, use the command:
859 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
861 For VLAN splinters to be effective, Open vSwitch must know
862 which VLANs are in use. See the "VLAN splinters" section in
863 the Interface table in ovs-vswitchd.conf.db(5) for details on
864 how Open vSwitch infers in-use VLANs.
866 VLAN splinters increase memory use and reduce performance, so
867 use them only if needed.
869 - Apply the "vlan workaround" patch from the XenServer kernel
870 patch queue, build Open vSwitch against this patched kernel,
871 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
872 workaround for each interface whose driver is buggy.
874 (This is a nontrivial exercise, so this option is included
875 only for completeness.)
877 It is not always easy to tell whether a Linux kernel driver has
878 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
879 can help you test. See their manpages for details. Of the two
880 utilities, ovs-test(8) is newer and more thorough, but
881 ovs-vlan-test(8) may be easier to use.
883 Q: VLANs still don't work. I've tested the driver so I know that it's OK.
885 A: Do you have VLANs enabled on the physical switch that OVS is
886 attached to? Make sure that the port is configured to trunk the
887 VLAN or VLANs that you are using with OVS.
889 Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
890 and to its destination host, but OVS seems to drop incoming return
893 A: It's possible that you have the VLAN configured on your physical
894 switch as the "native" VLAN. In this mode, the switch treats
895 incoming packets either tagged with the native VLAN or untagged as
896 part of the native VLAN. It may also send outgoing packets in the
897 native VLAN without a VLAN tag.
899 If this is the case, you have two choices:
901 - Change the physical switch port configuration to tag packets
902 it forwards to OVS with the native VLAN instead of forwarding
905 - Change the OVS configuration for the physical port to a
906 native VLAN mode. For example, the following sets up a
907 bridge with port eth0 in "native-tagged" mode in VLAN 9:
910 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
912 In this situation, "native-untagged" mode will probably work
913 equally well. Refer to the documentation for the Port table
914 in ovs-vswitchd.conf.db(5) for more information.
916 Q: I added a pair of VMs on different VLANs, like this:
919 ovs-vsctl add-port br0 eth0
920 ovs-vsctl add-port br0 tap0 tag=9
921 ovs-vsctl add-port br0 tap1 tag=10
923 but the VMs can't access each other, the external network, or the
926 A: It is to be expected that the VMs can't access each other. VLANs
927 are a means to partition a network. When you configured tap0 and
928 tap1 as access ports for different VLANs, you indicated that they
929 should be isolated from each other.
931 As for the external network and the Internet, it seems likely that
932 the machines you are trying to access are not on VLAN 9 (or 10) and
933 that the Internet is not available on VLAN 9 (or 10).
935 Q: I added a pair of VMs on the same VLAN, like this:
938 ovs-vsctl add-port br0 eth0
939 ovs-vsctl add-port br0 tap0 tag=9
940 ovs-vsctl add-port br0 tap1 tag=9
942 The VMs can access each other, but not the external network or the
945 A: It seems likely that the machines you are trying to access in the
946 external network are not on VLAN 9 and that the Internet is not
947 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
948 trunk VLAN on the upstream switch port to which eth0 is connected.
950 Q: Can I configure an IP address on a VLAN?
952 A: Yes. Use an "internal port" configured as an access port. For
953 example, the following configures IP address 192.168.0.7 on VLAN 9.
954 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
955 they have an 802.1Q header with VLAN 9. Conversely, traffic
956 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
960 ovs-vsctl add-port br0 eth0
961 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
962 ifconfig vlan9 192.168.0.7
964 See also the following question.
966 Q: I configured one IP address on VLAN 0 and another on VLAN 9, like
970 ovs-vsctl add-port br0 eth0
971 ifconfig br0 192.168.0.5
972 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
973 ifconfig vlan9 192.168.0.9
975 but other hosts that are only on VLAN 0 can reach the IP address
976 configured on VLAN 9. What's going on?
978 A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two
979 approaches to IP address handling in Internet hosts:
981 - In the "Strong ES Model", where an ES is a host ("End
982 System"), an IP address is primarily associated with a
983 particular interface. The host discards packets that arrive
984 on interface A if they are destined for an IP address that is
985 configured on interface B. The host never sends packets from
986 interface A using a source address configured on interface B.
988 - In the "Weak ES Model", an IP address is primarily associated
989 with a host. The host accepts packets that arrive on any
990 interface if they are destined for any of the host's IP
991 addresses, even if the address is configured on some
992 interface other than the one on which it arrived. The host
993 does not restrict itself to sending packets from an IP
994 address associated with the originating interface.
996 Linux uses the weak ES model. That means that when packets
997 destined to the VLAN 9 IP address arrive on eth0 and are bridged to
998 br0, the kernel IP stack accepts them there for the VLAN 9 IP
999 address, even though they were not received on vlan9, the network
1002 To simulate the strong ES model on Linux, one may add iptables rule
1003 to filter packets based on source and destination address and
1004 adjust ARP configuration with sysctls.
1006 BSD uses the strong ES model.
1008 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1010 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1011 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1012 implementation of the OpenFlow "normal switching" action. By
1013 default, when Open vSwitch isn't connected to a controller and
1014 nothing has been manually configured in the flow table, all traffic
1015 goes through the "normal switching" action. But, if you set up
1016 OpenFlow flows on your own, through a controller or using ovs-ofctl
1017 or through other means, then you have to implement VLAN handling
1020 You can use "normal switching" as a component of your OpenFlow
1021 actions, e.g. by putting "normal" into the lists of actions on
1022 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
1023 controller. In situations where this is not suitable, you can
1024 implement VLAN handling yourself, e.g.:
1026 - If a packet comes in on an access port, and the flow table
1027 needs to send it out on a trunk port, then the flow can add
1028 the appropriate VLAN tag with the "mod_vlan_vid" action.
1030 - If a packet comes in on a trunk port, and the flow table
1031 needs to send it out on an access port, then the flow can
1032 strip the VLAN tag with the "strip_vlan" action.
1034 Q: I configured ports on a bridge as access ports with different VLAN
1037 ovs-vsctl add-br br0
1038 ovs-vsctl set-controller br0 tcp:192.168.0.10:6633
1039 ovs-vsctl add-port br0 eth0
1040 ovs-vsctl add-port br0 tap0 tag=9
1041 ovs-vsctl add-port br0 tap1 tag=10
1043 but the VMs running behind tap0 and tap1 can still communicate,
1044 that is, they are not isolated from each other even though they are
1047 A: Do you have a controller configured on br0 (as the commands above
1048 do)? If so, then this is a variant on the previous question, "My
1049 OpenFlow controller doesn't see the VLANs that I expect," and you
1050 can refer to the answer there for more information.
1058 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1059 to solve the scaling challenges of VLAN networks in a multi-tenant
1060 environment. VXLAN is an overlay network which transports an L2 network
1061 over an existing L3 network. For more information on VXLAN, please see
1062 the IETF draft available here:
1064 http://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-03
1066 Q: How much of the VXLAN protocol does Open vSwitch currently support?
1068 A: Open vSwitch currently supports the framing format for packets on the
1069 wire. There is currently no support for the multicast aspects of VXLAN.
1070 To get around the lack of multicast support, it is possible to
1071 pre-provision MAC to IP address mappings either manually or from a
1074 Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1077 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
1078 is 4789. However, it is possible to configure the destination UDP port
1079 manually on a per-VXLAN tunnel basis. An example of this configuration is
1082 ovs-vsctl add-br br0
1083 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
1084 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
1085 options:dst_port=8472
1088 Using OpenFlow (Manually or Via Controller)
1089 -------------------------------------------
1091 Q: What versions of OpenFlow does Open vSwitch support?
1093 A: The following table lists the versions of OpenFlow supported by
1094 each version of Open vSwitch:
1096 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4
1097 =============== ===== ===== ===== ===== =====
1098 1.9 and earlier yes --- --- --- ---
1099 1.10 yes --- [*] [*] ---
1100 1.11 yes --- [*] [*] ---
1101 2.0 yes [*] [*] [*] ---
1102 2.1 yes [*] [*] [*] ---
1103 2.2 yes [*] [*] [*] [%]
1105 [*] Supported, with one or more missing features.
1106 [%] Support is unsafe: ovs-vswitchd will abort when certain
1107 unimplemented features are tested.
1109 Because of missing features, OpenFlow 1.1, 1.2, and 1.3 must be
1110 enabled manually. The following command enables OpenFlow 1.0, 1.1,
1111 1.2, and 1.3 on bridge br0:
1113 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1115 Use the -O option to enable support for later versions of OpenFlow
1116 in ovs-ofctl. For example:
1118 ovs-ofctl -O OpenFlow13 dump-flows br0
1120 OpenFlow 1.4 is a special case, because it is not implemented
1121 safely: ovs-vswitchd will abort when certain unimplemented features
1122 are tested. Thus, for now it is suitable only for experimental
1123 use. ovs-vswitchd will only allow OpenFlow 1.4 to be enabled
1124 (which must be done in the same way described above) when it is
1125 invoked with a special --enable-of14 command line option.
1127 OPENFLOW-1.1+ in the Open vSwitch source tree tracks support for
1128 OpenFlow 1.1 and later features. When support for a given OpenFlow
1129 version is solidly implemented, Open vSwitch will enable that
1132 Q: Does Open vSwitch support MPLS?
1134 A: Before version 1.11, Open vSwitch did not support MPLS. That is,
1135 these versions can match on MPLS Ethernet types, but they cannot
1136 match, push, or pop MPLS labels, nor can they look past MPLS labels
1137 into the encapsulated packet.
1139 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support
1140 for MPLS. With the userspace datapath only, these versions can
1141 match, push, or pop a single MPLS label, but they still cannot look
1142 past MPLS labels (even after popping them) into the encapsulated
1143 packet. Kernel datapath support is unchanged from earlier
1146 Open vSwitch version 2.2 will be able to match, push, or pop up to
1147 3 MPLS labels. Looking past MPLS labels into the encapsulated
1148 packet will still be unsupported. Both userspace and kernel
1149 datapaths will be supported, but MPLS processing always happens in
1150 userspace either way, so kernel datapath performance will be
1153 Q: I'm getting "error type 45250 code 0". What's that?
1155 A: This is a Open vSwitch extension to OpenFlow error codes. Open
1156 vSwitch uses this extension when it must report an error to an
1157 OpenFlow controller but no standard OpenFlow error code is
1160 Open vSwitch logs the errors that it sends to controllers, so the
1161 easiest thing to do is probably to look at the ovs-vswitchd log to
1162 find out what the error was.
1164 If you want to dissect the extended error message yourself, the
1165 format is documented in include/openflow/nicira-ext.h in the Open
1166 vSwitch source distribution. The extended error codes are
1167 documented in lib/ofp-errors.h.
1169 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
1170 doesn't actually appear through the OpenFlow connection, even
1171 though I know that it's going through.
1172 Q2: Some of the OpenFlow flows that my controller sets up don't seem
1173 to apply to certain traffic, especially traffic between OVS and
1174 the controller itself.
1176 A: By default, Open vSwitch assumes that OpenFlow controllers are
1177 connected "in-band", that is, that the controllers are actually
1178 part of the network that is being controlled. In in-band mode,
1179 Open vSwitch sets up special "hidden" flows to make sure that
1180 traffic can make it back and forth between OVS and the controllers.
1181 These hidden flows are higher priority than any flows that can be
1182 set up through OpenFlow, and they are not visible through normal
1183 OpenFlow flow table dumps.
1185 Usually, the hidden flows are desirable and helpful, but
1186 occasionally they can cause unexpected behavior. You can view the
1187 full OpenFlow flow table, including hidden flows, on bridge br0
1190 ovs-appctl bridge/dump-flows br0
1192 to help you debug. The hidden flows are those with priorities
1193 greater than 65535 (the maximum priority that can be set with
1196 The DESIGN file at the top level of the Open vSwitch source
1197 distribution describes the in-band model in detail.
1199 If your controllers are not actually in-band (e.g. they are on
1200 localhost via 127.0.0.1, or on a separate network), then you should
1201 configure your controllers in "out-of-band" mode. If you have one
1202 controller on bridge br0, then you can configure out-of-band mode
1205 ovs-vsctl set controller br0 connection-mode=out-of-band
1207 Q: I configured all my controllers for out-of-band control mode but
1208 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1210 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1211 set-manager"). By default, Open vSwitch assumes that managers need
1212 in-band rules set up on every bridge. You can disable these rules
1215 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1217 This actually disables in-band control entirely for the bridge, as
1218 if all the bridge's controllers were configured for out-of-band
1221 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1223 A: See answer under "VLANs", above.
1225 Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1226 but I got a funny message like this:
1228 ofp_util|INFO|normalization changed ofp_match, details:
1229 ofp_util|INFO| pre: nw_dst=192.168.0.1
1232 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1233 match had disappeared, so that the flow ends up matching every
1236 A: The term "normalization" in the log message means that a flow
1237 cannot match on an L3 field without saying what L3 protocol is in
1238 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1239 so the L3 field match was dropped.
1241 In this case, the L3 protocol could be IP or ARP. A correct
1242 command for each possibility is, respectively:
1244 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1248 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1250 Similarly, a flow cannot match on an L4 field without saying what
1251 L4 protocol is in use. For example, the flow match "tp_src=1234"
1252 is, by itself, meaningless and will be ignored. Instead, to match
1253 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1254 source port 1234, write "udp,tp_src=1234".
1256 Q: How can I figure out the OpenFlow port number for a given port?
1258 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1259 respond with an OFPT_FEATURES_REPLY that, among other information,
1260 includes a mapping between OpenFlow port names and numbers. From a
1261 command prompt, "ovs-ofctl show br0" makes such a request and
1262 prints the response for switch br0.
1264 The Interface table in the Open vSwitch database also maps OpenFlow
1265 port names to numbers. To print the OpenFlow port number
1266 associated with interface eth0, run:
1268 ovs-vsctl get Interface eth0 ofport
1270 You can print the entire mapping with:
1272 ovs-vsctl -- --columns=name,ofport list Interface
1274 but the output mixes together interfaces from all bridges in the
1275 database, so it may be confusing if more than one bridge exists.
1277 In the Open vSwitch database, ofport value -1 means that the
1278 interface could not be created due to an error. (The Open vSwitch
1279 log should indicate the reason.) ofport value [] (the empty set)
1280 means that the interface hasn't been created yet. The latter is
1281 normally an intermittent condition (unless ovs-vswitchd is not
1284 Q: I added some flows with my controller or with ovs-ofctl, but when I
1285 run "ovs-dpctl dump-flows" I don't see them.
1287 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1288 won't display the information that you want. You want to use
1289 "ovs-ofctl dump-flows" instead.
1291 Q: It looks like each of the interfaces in my bonded port shows up
1292 as an individual OpenFlow port. Is that right?
1294 A: Yes, Open vSwitch makes individual bond interfaces visible as
1295 OpenFlow ports, rather than the bond as a whole. The interfaces
1296 are treated together as a bond for only a few purposes:
1298 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1299 controller is not configured, this happens implicitly to
1302 - Mirrors configured for output to a bonded port.
1304 It would make a lot of sense for Open vSwitch to present a bond as
1305 a single OpenFlow port. If you want to contribute an
1306 implementation of such a feature, please bring it up on the Open
1307 vSwitch development mailing list at dev@openvswitch.org.
1309 Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1310 multiple hosts, and other components. The behavior isn't what I
1313 A: To debug network behavior problems, trace the path of a packet,
1314 hop-by-hop, from its origin in one host to a remote host. If
1315 that's correct, then trace the path of the response packet back to
1318 Usually a simple ICMP echo request and reply ("ping") packet is
1319 good enough. Start by initiating an ongoing "ping" from the origin
1320 host to a remote host. If you are tracking down a connectivity
1321 problem, the "ping" will not display any successful output, but
1322 packets are still being sent. (In this case the packets being sent
1323 are likely ARP rather than ICMP.)
1325 Tools available for tracing include the following:
1327 - "tcpdump" and "wireshark" for observing hops across network
1328 devices, such as Open vSwitch internal devices and physical
1331 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1332 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1333 These tools allow one to observe the actions being taken on
1334 packets in ongoing flows.
1336 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1337 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1338 documentation, and "Why are there so many different ways to
1339 dump flows?" above for some background.
1341 - "ovs-appctl ofproto/trace" to observe the logic behind how
1342 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1343 documentation. You can out more details about a given flow
1344 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1345 a flow from the output into an "ovs-appctl ofproto/trace"
1348 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1349 observe what goes on at these physical hops.
1351 Starting at the origin of a given packet, observe the packet at
1352 each hop in turn. For example, in one plausible scenario, you
1355 1. "tcpdump" the "eth" interface through which an ARP egresses
1356 a VM, from inside the VM.
1358 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1359 ingresses the host machine.
1361 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1362 the host interface through which the ARP egresses the
1363 physical machine. You may need to use "ovs-dpctl show" to
1364 interpret the port numbers. If the output seems surprising,
1365 you can use "ovs-appctl ofproto/trace" to observe details of
1366 how ovs-vswitchd determined the actions in the "ovs-dpctl
1369 4. "tcpdump" the "eth" interface through which the ARP egresses
1370 the physical machine.
1372 5. "tcpdump" the "eth" interface through which the ARP
1373 ingresses the physical machine, at the remote host that
1376 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1377 remote host that receives the ARP and observe the VM "vif"
1378 or "tap" interface to which the flow is directed. Again,
1379 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1381 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1384 8. "tcpdump" the "eth" interface through which the ARP
1385 ingresses a VM, from inside the VM.
1387 It is likely that during one of these steps you will figure out the
1388 problem. If not, then follow the ARP reply back to the origin, in
1391 Q: How do I make a flow drop packets?
1393 A: To drop a packet is to receive it without forwarding it. OpenFlow
1394 explicitly specifies forwarding actions. Thus, a flow with an
1395 empty set of actions does not forward packets anywhere, causing
1396 them to be dropped. You can specify an empty set of actions with
1397 "actions=" on the ovs-ofctl command line. For example:
1399 ovs-ofctl add-flow br0 priority=65535,actions=
1401 would cause every packet entering switch br0 to be dropped.
1403 You can write "drop" explicitly if you like. The effect is the
1404 same. Thus, the following command also causes every packet
1405 entering switch br0 to be dropped:
1407 ovs-ofctl add-flow br0 priority=65535,actions=drop
1409 "drop" is not an action, either in OpenFlow or Open vSwitch.
1410 Rather, it is only a way to say that there are no actions.
1412 Q: I added a flow to send packets out the ingress port, like this:
1414 ovs-ofctl add-flow br0 in_port=2,actions=2
1416 but OVS drops the packets instead.
1418 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet
1419 out its ingress port. The rationale is that dropping these packets
1420 makes it harder to loop the network. Sometimes this behavior can
1421 even be convenient, e.g. it is often the desired behavior in a flow
1422 that forwards a packet to several ports ("floods" the packet).
1424 Sometimes one really needs to send a packet out its ingress port
1425 ("hairpin"). In this case, output to OFPP_IN_PORT, which in
1426 ovs-ofctl syntax is expressed as just "in_port", e.g.:
1428 ovs-ofctl add-flow br0 in_port=2,actions=in_port
1430 This also works in some circumstances where the flow doesn't match
1431 on the input port. For example, if you know that your switch has
1432 five ports numbered 2 through 6, then the following will send every
1433 received packet out every port, even its ingress port:
1435 ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1439 ovs-ofctl add-flow br0 actions=all,in_port
1441 Sometimes, in complicated flow tables with multiple levels of
1442 "resubmit" actions, a flow needs to output to a particular port
1443 that may or may not be the ingress port. It's difficult to take
1444 advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch
1445 provides, as an OpenFlow extension, the ability to modify the
1446 in_port field. Whatever value is currently in the in_port field is
1447 the port to which outputs will be dropped, as well as the
1448 destination for OFPP_IN_PORT. This means that the following will
1449 reliably output to port 2 or to ports 2 through 6, respectively:
1451 ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1452 ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1454 If the input port is important, then one may save and restore it on
1457 ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1458 load:0->NXM_OF_IN_PORT[],\
1460 pop:NXM_OF_IN_PORT[]
1462 Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2
1463 on port 2. I set up flows to forward only traffic destined to the
1464 other host and drop other traffic, like this:
1466 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1467 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1468 priority=0,actions=drop
1470 But it doesn't work--I don't get any connectivity when I do this.
1473 A: These flows drop the ARP packets that IP hosts use to establish IP
1474 connectivity over Ethernet. To solve the problem, add flows to
1475 allow ARP to pass between the hosts:
1477 priority=5,in_port=1,arp,actions=2
1478 priority=5,in_port=2,arp,actions=1
1480 This issue can manifest other ways, too. The following flows that
1481 match on Ethernet addresses instead of IP addresses will also drop
1482 ARP packets, because ARP requests are broadcast instead of being
1483 directed to a specific host:
1485 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1486 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1487 priority=0,actions=drop
1489 The solution already described above will also work in this case.
1490 It may be better to add flows to allow all multicast and broadcast
1493 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1494 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1500 Q: How do I implement a new OpenFlow message?
1502 A: Add your new message to "enum ofpraw" and "enum ofptype" in
1503 lib/ofp-msgs.h, following the existing pattern. Then recompile and
1504 fix all of the new warnings, implementing new functionality for the
1505 new message as needed. (If you configure with --enable-Werror, as
1506 described in INSTALL, then it is impossible to miss any warnings.)
1508 If you need to add an OpenFlow vendor extension message for a
1509 vendor that doesn't yet have any extension messages, then you will
1510 also need to edit build-aux/extract-ofp-msgs.
1516 bugs@openvswitch.org
1517 http://openvswitch.org/