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 2.6.18 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-controller, ovs-ofctl, ovs-vsctl) that developers
93 can script and extend to provide distributed vswitch capabilities
94 that are closely integrated with their virtualization management
97 Q: Why doesn't Open vSwitch support distribution?
99 A: Open vSwitch is intended to be a useful component for building
100 flexible network infrastructure. There are many different approaches
101 to distribution which balance trade-offs between simplicity,
102 scalability, hardware compatibility, convergence times, logical
103 forwarding model, etc. The goal of Open vSwitch is to be able to
104 support all as a primitive building block rather than choose a
105 particular point in the distributed design space.
107 Q: How can I contribute to the Open vSwitch Community?
109 A: You can start by joining the mailing lists and helping to answer
110 questions. You can also suggest improvements to documentation. If
111 you have a feature or bug you would like to work on, send a mail to
112 one of the mailing lists:
114 http://openvswitch.org/mlists/
120 Q: What does it mean for an Open vSwitch release to be LTS (long-term
123 A: All official releases have been through a comprehensive testing
124 process and are suitable for production use. Planned releases will
125 occur several times a year. If a significant bug is identified in an
126 LTS release, we will provide an updated release that includes the
127 fix. Releases that are not LTS may not be fixed and may just be
128 supplanted by the next major release. The current LTS release is
131 Q: What Linux kernel versions does each Open vSwitch release work with?
133 A: The following table lists the Linux kernel versions against which the
134 given versions of the Open vSwitch kernel module will successfully
135 build. The Linux kernel versions are upstream kernel versions, so
136 Linux kernels modified from the upstream sources may not build in
137 some cases even if they are based on a supported version. This is
138 most notably true of Red Hat Enterprise Linux (RHEL) kernels, which
139 are extensively modified from upstream.
141 Open vSwitch Linux kernel
142 ------------ -------------
150 Open vSwitch userspace should also work with the Linux kernel module
151 built into Linux 3.3 and later.
153 Open vSwitch userspace is not sensitive to the Linux kernel version.
154 It should build against almost any kernel, certainly against 2.6.18
157 Q: What Linux kernel versions does IPFIX flow monitoring work with?
159 A: IPFIX flow monitoring requires the Linux kernel module from Open
160 vSwitch version 1.10.90 or later.
162 Q: Should userspace or kernel be upgraded first to minimize downtime?
164 In general, the Open vSwitch userspace should be used with the
165 kernel version included in the same release or with the version
166 from upstream Linux. However, when upgrading between two releases
167 of Open vSwitch it is best to migrate userspace first to reduce
168 the possbility of incompatibilities.
170 Q: What features are not available in the Open vSwitch kernel datapath
171 that ships as part of the upstream Linux kernel?
173 A: The kernel module in upstream Linux 3.3 and later does not include
174 tunnel virtual ports, that is, interfaces with type "gre",
175 "ipsec_gre", "gre64", "ipsec_gre64", "vxlan", or "lisp". It is
176 possible to create tunnels in Linux and attach them to Open vSwitch
177 as system devices. However, they cannot be dynamically created
178 through the OVSDB protocol or set the tunnel ids as a flow action.
180 Work is in progress in adding tunnel virtual ports to the upstream
181 Linux version of the Open vSwitch kernel module. For now, if you
182 need these features, use the kernel module from the Open vSwitch
183 distribution instead of the upstream Linux kernel module.
185 The upstream kernel module does not include patch ports, but this
186 only matters for Open vSwitch 1.9 and earlier, because Open vSwitch
187 1.10 and later implement patch ports without using this kernel
190 Q: What features are not available when using the userspace datapath?
192 A: Tunnel virtual ports are not supported, as described in the
193 previous answer. It is also not possible to use queue-related
194 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
195 may not be transmitted.
201 Q: I thought Open vSwitch was a virtual Ethernet switch, but the
202 documentation keeps talking about bridges. What's a bridge?
204 A: In networking, the terms "bridge" and "switch" are synonyms. Open
205 vSwitch implements an Ethernet switch, which means that it is also
210 A: See the "VLAN" section below.
216 Q: How do I configure a port as an access port?
218 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
219 the following commands configure br0 with eth0 as a trunk port (the
220 default) and tap0 as an access port for VLAN 9:
223 ovs-vsctl add-port br0 eth0
224 ovs-vsctl add-port br0 tap0 tag=9
226 If you want to configure an already added port as an access port,
227 use "ovs-vsctl set", e.g.:
229 ovs-vsctl set port tap0 tag=9
231 Q: How do I configure a port as a SPAN port, that is, enable mirroring
232 of all traffic to that port?
234 A: The following commands configure br0 with eth0 and tap0 as trunk
235 ports. All traffic coming in or going out on eth0 or tap0 is also
236 mirrored to tap1; any traffic arriving on tap1 is dropped:
239 ovs-vsctl add-port br0 eth0
240 ovs-vsctl add-port br0 tap0
241 ovs-vsctl add-port br0 tap1 \
242 -- --id=@p get port tap1 \
243 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
244 -- set bridge br0 mirrors=@m
246 To later disable mirroring, run:
248 ovs-vsctl clear bridge br0 mirrors
250 Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable
251 mirroring of all traffic to that VLAN?
253 A: The following commands configure br0 with eth0 as a trunk port and
254 tap0 as an access port for VLAN 10. All traffic coming in or going
255 out on tap0, as well as traffic coming in or going out on eth0 in
256 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
257 VLAN 10, in cases where one is present, is dropped as part of
261 ovs-vsctl add-port br0 eth0
262 ovs-vsctl add-port br0 tap0 tag=10
264 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
266 -- set bridge br0 mirrors=@m
268 To later disable mirroring, run:
270 ovs-vsctl clear bridge br0 mirrors
272 Mirroring to a VLAN can disrupt a network that contains unmanaged
273 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
274 GRE tunnel has fewer caveats than mirroring to a VLAN and should
275 generally be preferred.
277 Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
279 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
280 of the specified output-vlan. This loss of information may make
281 the mirrored traffic too hard to interpret.
283 To mirror multiple VLANs, use the commands above, but specify a
284 comma-separated list of VLANs as the value for select-vlan. To
285 mirror every VLAN, use the commands above, but omit select-vlan and
288 When a packet arrives on a VLAN that is used as a mirror output
289 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
290 floods the packet across all the ports for which the mirror output
291 VLAN is configured. (If an OpenFlow controller is in use, then it
292 can override this behavior through the flow table.) If OVS is used
293 as an intermediate switch, rather than an edge switch, this ensures
294 that the RSPAN traffic is distributed through the network.
296 Mirroring to a VLAN can disrupt a network that contains unmanaged
297 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
298 GRE tunnel has fewer caveats than mirroring to a VLAN and should
299 generally be preferred.
301 Q: How do I configure mirroring of all traffic to a GRE tunnel?
303 A: The following commands configure br0 with eth0 and tap0 as trunk
304 ports. All traffic coming in or going out on eth0 or tap0 is also
305 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
306 traffic arriving on gre0 is dropped:
309 ovs-vsctl add-port br0 eth0
310 ovs-vsctl add-port br0 tap0
311 ovs-vsctl add-port br0 gre0 \
312 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
313 -- --id=@p get port gre0 \
314 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
315 -- set bridge br0 mirrors=@m
317 To later disable mirroring and destroy the GRE tunnel:
319 ovs-vsctl clear bridge br0 mirrors
320 ovs-vcstl del-port br0 gre0
322 Q: Does Open vSwitch support ERSPAN?
324 A: No. ERSPAN is an undocumented proprietary protocol. As an
325 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
328 Q: How do I connect two bridges?
330 A: First, why do you want to do this? Two connected bridges are not
331 much different from a single bridge, so you might as well just have
332 a single bridge with all your ports on it.
334 If you still want to connect two bridges, you can use a pair of
335 patch ports. The following example creates bridges br0 and br1,
336 adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0
337 and br1 with a pair of patch ports.
340 ovs-vsctl add-port br0 eth0
341 ovs-vsctl add-port br0 tap0
343 ovs-vsctl add-port br1 tap1
345 -- add-port br0 patch0 \
346 -- set interface patch0 type=patch options:peer=patch1 \
347 -- add-port br1 patch1 \
348 -- set interface patch1 type=patch options:peer=patch0
350 Bridges connected with patch ports are much like a single bridge.
351 For instance, if the example above also added eth1 to br1, and both
352 eth0 and eth1 happened to be connected to the same next-hop switch,
353 then you could loop your network just as you would if you added
354 eth0 and eth1 to the same bridge (see the "Configuration Problems"
355 section below for more information).
357 If you are using Open vSwitch 1.9 or an earlier version, then you
358 need to be using the kernel module bundled with Open vSwitch rather
359 than the one that is integrated into Linux 3.3 and later, because
360 Open vSwitch 1.9 and earlier versions need kernel support for patch
361 ports. This also means that in Open vSwitch 1.9 and earlier, patch
362 ports will not work with the userspace datapath, only with the
365 Q: Why are there so many different ways to dump flows?
367 A: Open vSwitch uses different kinds of flows for different purposes:
369 - OpenFlow flows are the most important kind of flow. OpenFlow
370 controllers use these flows to define a switch's policy.
371 OpenFlow flows support wildcards, priorities, and multiple
374 When in-band control is in use, Open vSwitch sets up a few
375 "hidden" flows, with priority higher than a controller or the
376 user can configure, that are not visible via OpenFlow. (See
377 the "Controller" section of the FAQ for more information
380 - The Open vSwitch software switch implementation uses a second
381 kind of flow internally. These flows, called "exact-match"
382 or "datapath" or "kernel" flows, do not support wildcards or
383 priorities and comprise only a single table, which makes them
384 suitable for caching. OpenFlow flows and exact-match flows
385 also support different actions and number ports differently.
387 Exact-match flows are an implementation detail that is
388 subject to change in future versions of Open vSwitch. Even
389 with the current version of Open vSwitch, hardware switch
390 implementations do not necessarily use exact-match flows.
392 Each of the commands for dumping flows has a different purpose:
394 - "ovs-ofctl dump-flows <br>" dumps OpenFlow flows, excluding
395 hidden flows. This is the most commonly useful form of flow
396 dump. (Unlike the other commands, this should work with any
397 OpenFlow switch, not just Open vSwitch.)
399 - "ovs-appctl bridge/dump-flows <br>" dumps OpenFlow flows,
400 including hidden flows. This is occasionally useful for
401 troubleshooting suspected issues with in-band control.
403 - "ovs-dpctl dump-flows [dp]" dumps the exact-match flow table
404 entries for a Linux kernel-based datapath. In Open vSwitch
405 1.10 and later, ovs-vswitchd merges multiple switches into a
406 single datapath, so it will show all the flows on all your
407 kernel-based switches. This command can occasionally be
408 useful for debugging.
410 - "ovs-appctl dpif/dump-flows <br>", new in Open vSwitch 1.10,
411 dumps exact-match flows for only the specified bridge,
412 regardless of the type.
415 Configuration Problems
416 ----------------------
418 Q: I created a bridge and added my Ethernet port to it, using commands
422 ovs-vsctl add-port br0 eth0
424 and as soon as I ran the "add-port" command I lost all connectivity
427 A: A physical Ethernet device that is part of an Open vSwitch bridge
428 should not have an IP address. If one does, then that IP address
429 will not be fully functional.
431 You can restore functionality by moving the IP address to an Open
432 vSwitch "internal" device, such as the network device named after
433 the bridge itself. For example, assuming that eth0's IP address is
434 192.168.128.5, you could run the commands below to fix up the
437 ifconfig eth0 0.0.0.0
438 ifconfig br0 192.168.128.5
440 (If your only connection to the machine running OVS is through the
441 IP address in question, then you would want to run all of these
442 commands on a single command line, or put them into a script.) If
443 there were any additional routes assigned to eth0, then you would
444 also want to use commands to adjust these routes to go through br0.
446 If you use DHCP to obtain an IP address, then you should kill the
447 DHCP client that was listening on the physical Ethernet interface
448 (e.g. eth0) and start one listening on the internal interface
449 (e.g. br0). You might still need to manually clear the IP address
450 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
452 There is no compelling reason why Open vSwitch must work this way.
453 However, this is the way that the Linux kernel bridge module has
454 always worked, so it's a model that those accustomed to Linux
455 bridging are already used to. Also, the model that most people
456 expect is not implementable without kernel changes on all the
457 versions of Linux that Open vSwitch supports.
459 By the way, this issue is not specific to physical Ethernet
460 devices. It applies to all network devices except Open vswitch
463 Q: I created a bridge and added a couple of Ethernet ports to it,
464 using commands like these:
467 ovs-vsctl add-port br0 eth0
468 ovs-vsctl add-port br0 eth1
470 and now my network seems to have melted: connectivity is unreliable
471 (even connectivity that doesn't go through Open vSwitch), all the
472 LEDs on my physical switches are blinking, wireshark shows
473 duplicated packets, and CPU usage is very high.
475 A: More than likely, you've looped your network. Probably, eth0 and
476 eth1 are connected to the same physical Ethernet switch. This
477 yields a scenario where OVS receives a broadcast packet on eth0 and
478 sends it out on eth1, then the physical switch connected to eth1
479 sends the packet back on eth0, and so on forever. More complicated
480 scenarios, involving a loop through multiple switches, are possible
483 The solution depends on what you are trying to do:
485 - If you added eth0 and eth1 to get higher bandwidth or higher
486 reliability between OVS and your physical Ethernet switch,
487 use a bond. The following commands create br0 and then add
488 eth0 and eth1 as a bond:
491 ovs-vsctl add-bond br0 bond0 eth0 eth1
493 Bonds have tons of configuration options. Please read the
494 documentation on the Port table in ovs-vswitchd.conf.db(5)
497 - Perhaps you don't actually need eth0 and eth1 to be on the
498 same bridge. For example, if you simply want to be able to
499 connect each of them to virtual machines, then you can put
500 each of them on a bridge of its own:
503 ovs-vsctl add-port br0 eth0
506 ovs-vsctl add-port br1 eth1
508 and then connect VMs to br0 and br1. (A potential
509 disadvantage is that traffic cannot directly pass between br0
510 and br1. Instead, it will go out eth0 and come back in eth1,
513 - If you have a redundant or complex network topology and you
514 want to prevent loops, turn on spanning tree protocol (STP).
515 The following commands create br0, enable STP, and add eth0
516 and eth1 to the bridge. The order is important because you
517 don't want have to have a loop in your network even
521 ovs-vsctl set bridge br0 stp_enable=true
522 ovs-vsctl add-port br0 eth0
523 ovs-vsctl add-port br0 eth1
525 The Open vSwitch implementation of STP is not well tested.
526 Please report any bugs you observe, but if you'd rather avoid
527 acting as a beta tester then another option might be your
530 Q: I can't seem to use Open vSwitch in a wireless network.
532 A: Wireless base stations generally only allow packets with the source
533 MAC address of NIC that completed the initial handshake.
534 Therefore, without MAC rewriting, only a single device can
535 communicate over a single wireless link.
537 This isn't specific to Open vSwitch, it's enforced by the access
538 point, so the same problems will show up with the Linux bridge or
539 any other way to do bridging.
541 Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
543 A: PPP most commonly carries IP packets, but Open vSwitch works only
544 with Ethernet frames. The correct way to interface PPP to an
545 Ethernet network is usually to use routing instead of switching.
547 Q: Is there any documentation on the database tables and fields?
549 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
551 Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
552 I only see a datapath called "ovs-system". How can I see datapath
553 information about a particular bridge?
555 A: In version 1.9.0, OVS switched to using a single datapath that is
556 shared by all bridges of that type. The "ovs-appctl dpif/*"
557 commands provide similar functionality that is scoped by the bridge.
560 Quality of Service (QoS)
561 ------------------------
563 Q: How do I configure Quality of Service (QoS)?
565 A: Suppose that you want to set up bridge br0 connected to physical
566 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
567 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
568 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
569 could configure the bridge this way:
573 add-port br0 eth0 -- \
574 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
575 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
576 set port eth0 qos=@newqos -- \
577 --id=@newqos create qos type=linux-htb \
578 other-config:max-rate=1000000000 \
579 queues:123=@vif10queue \
580 queues:234=@vif20queue -- \
581 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
582 --id=@vif20queue create queue other-config:max-rate=20000000
584 At this point, bridge br0 is configured with the ports and eth0 is
585 configured with the queues that you need for QoS, but nothing is
586 actually directing packets from vif1.0 or vif2.0 to the queues that
587 we have set up for them. That means that all of the packets to
588 eth0 are going to the "default queue", which is not what we want.
590 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
591 queues reserved for them:
593 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
594 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
596 Each of the above flows matches on the input port, sets up the
597 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
598 executes the "normal" action, which performs the same switching
599 that Open vSwitch would have done without any OpenFlow flows being
600 present. (We know that vif1.0 and vif2.0 have OpenFlow port
601 numbers 5 and 6, respectively, because we set their ofport_request
602 columns above. If we had not done that, then we would have needed
603 to find out their port numbers before setting up these flows.)
605 Now traffic going from vif1.0 or vif2.0 to eth0 should be
608 By the way, if you delete the bridge created by the above commands,
613 then that will leave one unreferenced QoS record and two
614 unreferenced Queue records in the Open vSwich database. One way to
615 clear them out, assuming you don't have other QoS or Queue records
616 that you want to keep, is:
618 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
620 If you do want to keep some QoS or Queue records, or the Open
621 vSwitch you are using is older than version 1.8 (which added the
622 --all option), then you will have to destroy QoS and Queue records
625 Q: I configured Quality of Service (QoS) in my OpenFlow network by
626 adding records to the QoS and Queue table, but the results aren't
629 A: Did you install OpenFlow flows that use your queues? This is the
630 primary way to tell Open vSwitch which queues you want to use. If
631 you don't do this, then the default queue will be used, which will
632 probably not have the effect you want.
634 Refer to the previous question for an example.
636 Q: I configured QoS, correctly, but my measurements show that it isn't
637 working as well as I expect.
639 A: With the Linux kernel, the Open vSwitch implementation of QoS has
642 - Open vSwitch configures a subset of Linux kernel QoS
643 features, according to what is in OVSDB. It is possible that
644 this code has bugs. If you believe that this is so, then you
645 can configure the Linux traffic control (QoS) stack directly
646 with the "tc" program. If you get better results that way,
647 you can send a detailed bug report to bugs@openvswitch.org.
649 It is certain that Open vSwitch cannot configure every Linux
650 kernel QoS feature. If you need some feature that OVS cannot
651 configure, then you can also use "tc" directly (or add that
654 - The Open vSwitch implementation of OpenFlow allows flows to
655 be directed to particular queues. This is pretty simple and
656 unlikely to have serious bugs at this point.
658 However, most problems with QoS on Linux are not bugs in Open
659 vSwitch at all. They tend to be either configuration errors
660 (please see the earlier questions in this section) or issues with
661 the traffic control (QoS) stack in Linux. The Open vSwitch
662 developers are not experts on Linux traffic control. We suggest
663 that, if you believe you are encountering a problem with Linux
664 traffic control, that you consult the tc manpages (e.g. tc(8),
665 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
666 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
674 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
675 partition a single switch into multiple switches. Suppose, for
676 example, that you have two groups of machines, group A and group B.
677 You want the machines in group A to be able to talk to each other,
678 and you want the machine in group B to be able to talk to each
679 other, but you don't want the machines in group A to be able to
680 talk to the machines in group B. You can do this with two
681 switches, by plugging the machines in group A into one switch and
682 the machines in group B into the other switch.
684 If you only have one switch, then you can use VLANs to do the same
685 thing, by configuring the ports for machines in group A as VLAN
686 "access ports" for one VLAN and the ports for group B as "access
687 ports" for a different VLAN. The switch will only forward packets
688 between ports that are assigned to the same VLAN, so this
689 effectively subdivides your single switch into two independent
690 switches, one for each group of machines.
692 So far we haven't said anything about VLAN headers. With access
693 ports, like we've described so far, no VLAN header is present in
694 the Ethernet frame. This means that the machines (or switches)
695 connected to access ports need not be aware that VLANs are
696 involved, just like in the case where we use two different physical
699 Now suppose that you have a whole bunch of switches in your
700 network, instead of just one, and that some machines in group A are
701 connected directly to both switches 1 and 2. To allow these
702 machines to talk to each other, you could add an access port for
703 group A's VLAN to switch 1 and another to switch 2, and then
704 connect an Ethernet cable between those ports. That works fine,
705 but it doesn't scale well as the number of switches and the number
706 of VLANs increases, because you use up a lot of valuable switch
707 ports just connecting together your VLANs.
709 This is where VLAN headers come in. Instead of using one cable and
710 two ports per VLAN to connect a pair of switches, we configure a
711 port on each switch as a VLAN "trunk port". Packets sent and
712 received on a trunk port carry a VLAN header that says what VLAN
713 the packet belongs to, so that only two ports total are required to
714 connect the switches, regardless of the number of VLANs in use.
715 Normally, only switches (either physical or virtual) are connected
716 to a trunk port, not individual hosts, because individual hosts
717 don't expect to see a VLAN header in the traffic that they receive.
719 None of the above discussion says anything about particular VLAN
720 numbers. This is because VLAN numbers are completely arbitrary.
721 One must only ensure that a given VLAN is numbered consistently
722 throughout a network and that different VLANs are given different
723 numbers. (That said, VLAN 0 is usually synonymous with a packet
724 that has no VLAN header, and VLAN 4095 is reserved.)
728 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
729 bugs. If you are having problems with VLANs that you suspect to be
730 driver related, then you have several options:
732 - Upgrade to Linux 3.3 or later.
734 - Build and install a fixed version of the particular driver
735 that is causing trouble, if one is available.
737 - Use a NIC whose driver does not have VLAN problems.
739 - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later
740 that works around bugs in kernel drivers. To enable VLAN
741 splinters on interface eth0, use the command:
743 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
745 For VLAN splinters to be effective, Open vSwitch must know
746 which VLANs are in use. See the "VLAN splinters" section in
747 the Interface table in ovs-vswitchd.conf.db(5) for details on
748 how Open vSwitch infers in-use VLANs.
750 VLAN splinters increase memory use and reduce performance, so
751 use them only if needed.
753 - Apply the "vlan workaround" patch from the XenServer kernel
754 patch queue, build Open vSwitch against this patched kernel,
755 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
756 workaround for each interface whose driver is buggy.
758 (This is a nontrivial exercise, so this option is included
759 only for completeness.)
761 It is not always easy to tell whether a Linux kernel driver has
762 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
763 can help you test. See their manpages for details. Of the two
764 utilities, ovs-test(8) is newer and more thorough, but
765 ovs-vlan-test(8) may be easier to use.
767 Q: VLANs still don't work. I've tested the driver so I know that it's OK.
769 A: Do you have VLANs enabled on the physical switch that OVS is
770 attached to? Make sure that the port is configured to trunk the
771 VLAN or VLANs that you are using with OVS.
773 Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
774 and to its destination host, but OVS seems to drop incoming return
777 A: It's possible that you have the VLAN configured on your physical
778 switch as the "native" VLAN. In this mode, the switch treats
779 incoming packets either tagged with the native VLAN or untagged as
780 part of the native VLAN. It may also send outgoing packets in the
781 native VLAN without a VLAN tag.
783 If this is the case, you have two choices:
785 - Change the physical switch port configuration to tag packets
786 it forwards to OVS with the native VLAN instead of forwarding
789 - Change the OVS configuration for the physical port to a
790 native VLAN mode. For example, the following sets up a
791 bridge with port eth0 in "native-tagged" mode in VLAN 9:
794 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
796 In this situation, "native-untagged" mode will probably work
797 equally well. Refer to the documentation for the Port table
798 in ovs-vswitchd.conf.db(5) for more information.
800 Q: I added a pair of VMs on different VLANs, like this:
803 ovs-vsctl add-port br0 eth0
804 ovs-vsctl add-port br0 tap0 tag=9
805 ovs-vsctl add-port br0 tap1 tag=10
807 but the VMs can't access each other, the external network, or the
810 A: It is to be expected that the VMs can't access each other. VLANs
811 are a means to partition a network. When you configured tap0 and
812 tap1 as access ports for different VLANs, you indicated that they
813 should be isolated from each other.
815 As for the external network and the Internet, it seems likely that
816 the machines you are trying to access are not on VLAN 9 (or 10) and
817 that the Internet is not available on VLAN 9 (or 10).
819 Q: I added a pair of VMs on the same VLAN, like this:
822 ovs-vsctl add-port br0 eth0
823 ovs-vsctl add-port br0 tap0 tag=9
824 ovs-vsctl add-port br0 tap1 tag=9
826 The VMs can access each other, but not the external network or the
829 A: It seems likely that the machines you are trying to access in the
830 external network are not on VLAN 9 and that the Internet is not
831 available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed
832 trunk VLAN on the upstream switch port to which eth0 is connected.
834 Q: Can I configure an IP address on a VLAN?
836 A: Yes. Use an "internal port" configured as an access port. For
837 example, the following configures IP address 192.168.0.7 on VLAN 9.
838 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
839 they have an 802.1Q header with VLAN 9. Conversely, traffic
840 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
844 ovs-vsctl add-port br0 eth0
845 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
846 ifconfig vlan9 192.168.0.7
848 Q: My OpenFlow controller doesn't see the VLANs that I expect.
850 A: The configuration for VLANs in the Open vSwitch database (e.g. via
851 ovs-vsctl) only affects traffic that goes through Open vSwitch's
852 implementation of the OpenFlow "normal switching" action. By
853 default, when Open vSwitch isn't connected to a controller and
854 nothing has been manually configured in the flow table, all traffic
855 goes through the "normal switching" action. But, if you set up
856 OpenFlow flows on your own, through a controller or using ovs-ofctl
857 or through other means, then you have to implement VLAN handling
860 You can use "normal switching" as a component of your OpenFlow
861 actions, e.g. by putting "normal" into the lists of actions on
862 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
863 controller. In situations where this is not suitable, you can
864 implement VLAN handling yourself, e.g.:
866 - If a packet comes in on an access port, and the flow table
867 needs to send it out on a trunk port, then the flow can add
868 the appropriate VLAN tag with the "mod_vlan_vid" action.
870 - If a packet comes in on a trunk port, and the flow table
871 needs to send it out on an access port, then the flow can
872 strip the VLAN tag with the "strip_vlan" action.
874 Q: I configured ports on a bridge as access ports with different VLAN
878 ovs-vsctl set-controller br0 tcp:192.168.0.10:6633
879 ovs-vsctl add-port br0 eth0
880 ovs-vsctl add-port br0 tap0 tag=9
881 ovs-vsctl add-port br0 tap1 tag=10
883 but the VMs running behind tap0 and tap1 can still communicate,
884 that is, they are not isolated from each other even though they are
887 A: Do you have a controller configured on br0 (as the commands above
888 do)? If so, then this is a variant on the previous question, "My
889 OpenFlow controller doesn't see the VLANs that I expect," and you
890 can refer to the answer there for more information.
898 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
899 to solve the scaling challenges of VLAN networks in a multi-tenant
900 environment. VXLAN is an overlay network which transports an L2 network
901 over an existing L3 network. For more information on VXLAN, please see
902 the IETF draft available here:
904 http://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-03
906 Q: How much of the VXLAN protocol does Open vSwitch currently support?
908 A: Open vSwitch currently supports the framing format for packets on the
909 wire. There is currently no support for the multicast aspects of VXLAN.
910 To get around the lack of multicast support, it is possible to
911 pre-provision MAC to IP address mappings either manually or from a
914 Q: What destination UDP port does the VXLAN implementation in Open vSwitch
917 A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which
918 is 4789. However, it is possible to configure the destination UDP port
919 manually on a per-VXLAN tunnel basis. An example of this configuration is
923 ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1
924 type=vxlan options:remote_ip=192.168.1.2 options:key=flow
925 options:dst_port=8472
928 Using OpenFlow (Manually or Via Controller)
929 -------------------------------------------
931 Q: What versions of OpenFlow does Open vSwitch support?
933 A: Open vSwitch 1.9 and earlier support only OpenFlow 1.0 (plus
934 extensions that bring in many of the features from later versions
937 Open vSwitch versions 1.10 and later will have experimental support
938 for OpenFlow 1.2 and 1.3. On these versions of Open vSwitch, the
939 following command enables OpenFlow 1.0, 1.2, and 1.3 on bridge br0:
941 ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow12,OpenFlow13
943 Use the -O option to enable support for later versions of OpenFlow
944 in ovs-ofctl. For example:
946 ovs-ofctl -O OpenFlow13 dump-flows br0
948 Support for OpenFlow 1.1 is incomplete enough that it cannot yet be
949 enabled, even experimentally.
951 Support for OpenFlow 1.2 and 1.3 is still incomplete. Work to be
952 done is tracked in OPENFLOW-1.1+ in the Open vSwitch source tree
953 (also via http://openvswitch.org/development/openflow-1-x-plan/).
954 When support for a given OpenFlow version is solidly implemented,
955 Open vSwitch will enable that version by default.
957 Q: I'm getting "error type 45250 code 0". What's that?
959 A: This is a Open vSwitch extension to OpenFlow error codes. Open
960 vSwitch uses this extension when it must report an error to an
961 OpenFlow controller but no standard OpenFlow error code is
964 Open vSwitch logs the errors that it sends to controllers, so the
965 easiest thing to do is probably to look at the ovs-vswitchd log to
966 find out what the error was.
968 If you want to dissect the extended error message yourself, the
969 format is documented in include/openflow/nicira-ext.h in the Open
970 vSwitch source distribution. The extended error codes are
971 documented in lib/ofp-errors.h.
973 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
974 doesn't actually appear through the OpenFlow connection, even
975 though I know that it's going through.
976 Q2: Some of the OpenFlow flows that my controller sets up don't seem
977 to apply to certain traffic, especially traffic between OVS and
978 the controller itself.
980 A: By default, Open vSwitch assumes that OpenFlow controllers are
981 connected "in-band", that is, that the controllers are actually
982 part of the network that is being controlled. In in-band mode,
983 Open vSwitch sets up special "hidden" flows to make sure that
984 traffic can make it back and forth between OVS and the controllers.
985 These hidden flows are higher priority than any flows that can be
986 set up through OpenFlow, and they are not visible through normal
987 OpenFlow flow table dumps.
989 Usually, the hidden flows are desirable and helpful, but
990 occasionally they can cause unexpected behavior. You can view the
991 full OpenFlow flow table, including hidden flows, on bridge br0
994 ovs-appctl bridge/dump-flows br0
996 to help you debug. The hidden flows are those with priorities
997 greater than 65535 (the maximum priority that can be set with
1000 The DESIGN file at the top level of the Open vSwitch source
1001 distribution describes the in-band model in detail.
1003 If your controllers are not actually in-band (e.g. they are on
1004 localhost via 127.0.0.1, or on a separate network), then you should
1005 configure your controllers in "out-of-band" mode. If you have one
1006 controller on bridge br0, then you can configure out-of-band mode
1009 ovs-vsctl set controller br0 connection-mode=out-of-band
1011 Q: I configured all my controllers for out-of-band control mode but
1012 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
1014 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1015 set-manager"). By default, Open vSwitch assumes that managers need
1016 in-band rules set up on every bridge. You can disable these rules
1019 ovs-vsctl set bridge br0 other-config:disable-in-band=true
1021 This actually disables in-band control entirely for the bridge, as
1022 if all the bridge's controllers were configured for out-of-band
1025 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1027 A: See answer under "VLANs", above.
1029 Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
1030 but I got a funny message like this:
1032 ofp_util|INFO|normalization changed ofp_match, details:
1033 ofp_util|INFO| pre: nw_dst=192.168.0.1
1036 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
1037 match had disappeared, so that the flow ends up matching every
1040 A: The term "normalization" in the log message means that a flow
1041 cannot match on an L3 field without saying what L3 protocol is in
1042 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
1043 so the L3 field match was dropped.
1045 In this case, the L3 protocol could be IP or ARP. A correct
1046 command for each possibility is, respectively:
1048 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1052 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1054 Similarly, a flow cannot match on an L4 field without saying what
1055 L4 protocol is in use. For example, the flow match "tp_src=1234"
1056 is, by itself, meaningless and will be ignored. Instead, to match
1057 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
1058 source port 1234, write "udp,tp_src=1234".
1060 Q: How can I figure out the OpenFlow port number for a given port?
1062 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
1063 respond with an OFPT_FEATURES_REPLY that, among other information,
1064 includes a mapping between OpenFlow port names and numbers. From a
1065 command prompt, "ovs-ofctl show br0" makes such a request and
1066 prints the response for switch br0.
1068 The Interface table in the Open vSwitch database also maps OpenFlow
1069 port names to numbers. To print the OpenFlow port number
1070 associated with interface eth0, run:
1072 ovs-vsctl get Interface eth0 ofport
1074 You can print the entire mapping with:
1076 ovs-vsctl -- --columns=name,ofport list Interface
1078 but the output mixes together interfaces from all bridges in the
1079 database, so it may be confusing if more than one bridge exists.
1081 In the Open vSwitch database, ofport value -1 means that the
1082 interface could not be created due to an error. (The Open vSwitch
1083 log should indicate the reason.) ofport value [] (the empty set)
1084 means that the interface hasn't been created yet. The latter is
1085 normally an intermittent condition (unless ovs-vswitchd is not
1088 Q: I added some flows with my controller or with ovs-ofctl, but when I
1089 run "ovs-dpctl dump-flows" I don't see them.
1091 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
1092 won't display the information that you want. You want to use
1093 "ovs-ofctl dump-flows" instead.
1095 Q: It looks like each of the interfaces in my bonded port shows up
1096 as an individual OpenFlow port. Is that right?
1098 A: Yes, Open vSwitch makes individual bond interfaces visible as
1099 OpenFlow ports, rather than the bond as a whole. The interfaces
1100 are treated together as a bond for only a few purposes:
1102 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1103 controller is not configured, this happens implicitly to
1106 - Mirrors configured for output to a bonded port.
1108 It would make a lot of sense for Open vSwitch to present a bond as
1109 a single OpenFlow port. If you want to contribute an
1110 implementation of such a feature, please bring it up on the Open
1111 vSwitch development mailing list at dev@openvswitch.org.
1113 Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1114 multiple hosts, and other components. The behavior isn't what I
1117 A: To debug network behavior problems, trace the path of a packet,
1118 hop-by-hop, from its origin in one host to a remote host. If
1119 that's correct, then trace the path of the response packet back to
1122 Usually a simple ICMP echo request and reply ("ping") packet is
1123 good enough. Start by initiating an ongoing "ping" from the origin
1124 host to a remote host. If you are tracking down a connectivity
1125 problem, the "ping" will not display any successful output, but
1126 packets are still being sent. (In this case the packets being sent
1127 are likely ARP rather than ICMP.)
1129 Tools available for tracing include the following:
1131 - "tcpdump" and "wireshark" for observing hops across network
1132 devices, such as Open vSwitch internal devices and physical
1135 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1136 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1137 These tools allow one to observe the actions being taken on
1138 packets in ongoing flows.
1140 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1141 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1142 documentation, and "Why are there so many different ways to
1143 dump flows?" above for some background.
1145 - "ovs-appctl ofproto/trace" to observe the logic behind how
1146 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1147 documentation. You can out more details about a given flow
1148 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1149 a flow from the output into an "ovs-appctl ofproto/trace"
1152 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1153 observe what goes on at these physical hops.
1155 Starting at the origin of a given packet, observe the packet at
1156 each hop in turn. For example, in one plausible scenario, you
1159 1. "tcpdump" the "eth" interface through which an ARP egresses
1160 a VM, from inside the VM.
1162 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1163 ingresses the host machine.
1165 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1166 the host interface through which the ARP egresses the
1167 physical machine. You may need to use "ovs-dpctl show" to
1168 interpret the port numbers. If the output seems surprising,
1169 you can use "ovs-appctl ofproto/trace" to observe details of
1170 how ovs-vswitchd determined the actions in the "ovs-dpctl
1173 4. "tcpdump" the "eth" interface through which the ARP egresses
1174 the physical machine.
1176 5. "tcpdump" the "eth" interface through which the ARP
1177 ingresses the physical machine, at the remote host that
1180 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1181 remote host that receives the ARP and observe the VM "vif"
1182 or "tap" interface to which the flow is directed. Again,
1183 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1185 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1188 8. "tcpdump" the "eth" interface through which the ARP
1189 ingresses a VM, from inside the VM.
1191 It is likely that during one of these steps you will figure out the
1192 problem. If not, then follow the ARP reply back to the origin, in
1195 Q: How do I make a flow drop packets?
1197 A: An empty set of actions causes a packet to be dropped. You can
1198 specify an empty set of actions with "actions=" on the ovs-ofctl
1199 command line. For example:
1201 ovs-ofctl add-flow br0 priority=65535,actions=
1203 would cause every packet entering switch br0 to be dropped.
1205 You can write "drop" explicitly if you like. The effect is the
1206 same. Thus, the following command also causes every packet
1207 entering switch br0 to be dropped:
1209 ovs-ofctl add-flow br0 priority=65535,actions=drop
1215 bugs@openvswitch.org
1216 http://openvswitch.org/
git://git.onelab.eu / sliver-openvswitch.git / blob