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 environments. A
13 vswitch forwards traffic between different VMs on the same physical host
14 and also forwards traffic between VMs and the physical network. Open
15 vSwitch supports standard management interfaces (e.g. sFlow, NetFlow,
16 RSPAN, CLI), and is open to programmatic extension and control using
17 OpenFlow and the OVSDB management protocol.
19 Open vSwitch as designed to be compatible with modern switching
20 chipsets. This means that it can be ported to existing high-fanout
21 switches allowing the same flexible control of the physical
22 infrastructure as the virtual infrastructure. It also means that
23 Open vSwitch will be able to take advantage of on-NIC switching
24 chipsets as their functionality matures.
26 Q: What virtualization platforms can use Open vSwitch?
28 A: Open vSwitch can currently run on any Linux-based virtualization
29 platform (kernel 2.6.18 and newer), including: KVM, VirtualBox, Xen,
30 Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the
31 mainline kernel. The bulk of the code is written in platform-
32 independent C and is easily ported to other environments. We welcome
33 inquires about integrating Open vSwitch with other virtualization
36 Q: How can I try Open vSwitch?
38 A: The Open vSwitch source code can be built on a Linux system. You can
39 build and experiment with Open vSwitch on any Linux machine.
40 Packages for various Linux distributions are available on many
41 platforms, including: Debian, Ubuntu, Fedora.
43 You may also download and run a virtualization platform that already
44 has Open vSwitch integrated. For example, download a recent ISO for
45 XenServer or Xen Cloud Platform. Be aware that the version
46 integrated with a particular platform may not be the most recent Open
49 Q: Does Open vSwitch only work on Linux?
51 A: No, Open vSwitch has been ported to a number of different operating
52 systems and hardware platforms. Most of the development work occurs
53 on Linux, but the code should be portable to any POSIX system. We've
54 seen Open vSwitch ported to a number of different platforms,
55 including FreeBSD, Windows, and even non-POSIX embedded systems.
57 By definition, the Open vSwitch Linux kernel module only works on
58 Linux and will provide the highest performance. However, a userspace
59 datapath is available that should be very portable.
61 Q: What's involved with porting Open vSwitch to a new platform or
64 A: The PORTING document describes how one would go about porting Open
65 vSwitch to a new operating system or hardware platform.
67 Q: Why would I use Open vSwitch instead of the Linux bridge?
69 A: Open vSwitch is specially designed to make it easier to manage VM
70 network configuration and monitor state spread across many physical
71 hosts in dynamic virtualized environments. Please see WHY-OVS for a
72 more detailed description of how Open vSwitch relates to the Linux
75 Q: How is Open vSwitch related to distributed virtual switches like the
76 VMware vNetwork distributed switch or the Cisco Nexus 1000V?
78 A: Distributed vswitch applications (e.g., VMware vNetwork distributed
79 switch, Cisco Nexus 1000V) provide a centralized way to configure and
80 monitor the network state of VMs that are spread across many physical
81 hosts. Open vSwitch is not a distributed vswitch itself, rather it
82 runs on each physical host and supports remote management in a way
83 that makes it easier for developers of virtualization/cloud
84 management platforms to offer distributed vswitch capabilities.
86 To aid in distribution, Open vSwitch provides two open protocols that
87 are specially designed for remote management in virtualized network
88 environments: OpenFlow, which exposes flow-based forwarding state,
89 and the OVSDB management protocol, which exposes switch port state.
90 In addition to the switch implementation itself, Open vSwitch
91 includes tools (ovs-controller, ovs-ofctl, ovs-vsctl) that developers
92 can script and extend to provide distributed vswitch capabilities
93 that are closely integrated with their virtualization management
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/
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: Should userspace or kernel be upgraded first to minimize downtime?
159 In general, the Open vSwitch userspace should be used with the
160 kernel version included in the same release or with the version
161 from upstream Linux. However, when upgrading between two releases
162 of Open vSwitch it is best to migrate userspace first to reduce
163 the possbility of incompatibilities.
165 Q: What features are not available in the Open vSwitch kernel datapath
166 that ships as part of the upstream Linux kernel?
168 A: The kernel module in upstream Linux 3.3 and later does not include
169 tunnel virtual ports, that is, interfaces with type "gre",
170 "ipsec_gre", "gre64", "ipsec_gre64", or "vxlan". It is
171 possible to create tunnels in Linux and attach them to Open vSwitch
172 as system devices. However, they cannot be dynamically created
173 through the OVSDB protocol or set the tunnel ids as a flow action.
175 Work is in progress in adding tunnel virtual ports to the upstream
176 Linux version of the Open vSwitch kernel module. For now, if you
177 need these features, use the kernel module from the Open vSwitch
178 distribution instead of the upstream Linux kernel module.
180 The upstream kernel module does not include patch ports, but this
181 only matters for Open vSwitch 1.9 and earlier, because Open vSwitch
182 1.10 and later implement patch ports without using this kernel
185 Q: What features are not available when using the userspace datapath?
187 A: Tunnel virtual ports are not supported, as described in the
188 previous answer. It is also not possible to use queue-related
189 actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets
190 may not be transmitted.
196 Q: I thought Open vSwitch was a virtual Ethernet switch, but the
197 documentation keeps talking about bridges. What's a bridge?
199 A: In networking, the terms "bridge" and "switch" are synonyms. Open
200 vSwitch implements an Ethernet switch, which means that it is also
205 A: See the "VLAN" section below.
211 Q: How do I configure a port as an access port?
213 A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example,
214 the following commands configure br0 with eth0 as a trunk port (the
215 default) and tap0 as an access port for VLAN 9:
218 ovs-vsctl add-port br0 eth0
219 ovs-vsctl add-port br0 tap0 tag=9
221 If you want to configure an already added port as an access port,
222 use "ovs-vsctl set", e.g.:
224 ovs-vsctl set port tap0 tag=9
226 Q: How do I configure a port as a SPAN port, that is, enable mirroring
227 of all traffic to that port?
229 A: The following commands configure br0 with eth0 and tap0 as trunk
230 ports. All traffic coming in or going out on eth0 or tap0 is also
231 mirrored to tap1; any traffic arriving on tap1 is dropped:
234 ovs-vsctl add-port br0 eth0
235 ovs-vsctl add-port br0 tap0
236 ovs-vsctl add-port br0 tap1 \
237 -- --id=@p get port tap1 \
238 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
239 -- set bridge br0 mirrors=@m
241 To later disable mirroring, run:
243 ovs-vsctl clear bridge br0 mirrors
245 Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable
246 mirroring of all traffic to that VLAN?
248 A: The following commands configure br0 with eth0 as a trunk port and
249 tap0 as an access port for VLAN 10. All traffic coming in or going
250 out on tap0, as well as traffic coming in or going out on eth0 in
251 VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for
252 VLAN 10, in cases where one is present, is dropped as part of
256 ovs-vsctl add-port br0 eth0
257 ovs-vsctl add-port br0 tap0 tag=10
259 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
261 -- set bridge br0 mirrors=@m
263 To later disable mirroring, run:
265 ovs-vsctl clear bridge br0 mirrors
267 Mirroring to a VLAN can disrupt a network that contains unmanaged
268 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
269 GRE tunnel has fewer caveats than mirroring to a VLAN and should
270 generally be preferred.
272 Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
274 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor
275 of the specified output-vlan. This loss of information may make
276 the mirrored traffic too hard to interpret.
278 To mirror multiple VLANs, use the commands above, but specify a
279 comma-separated list of VLANs as the value for select-vlan. To
280 mirror every VLAN, use the commands above, but omit select-vlan and
283 When a packet arrives on a VLAN that is used as a mirror output
284 VLAN, the mirror is disregarded. Instead, in standalone mode, OVS
285 floods the packet across all the ports for which the mirror output
286 VLAN is configured. (If an OpenFlow controller is in use, then it
287 can override this behavior through the flow table.) If OVS is used
288 as an intermediate switch, rather than an edge switch, this ensures
289 that the RSPAN traffic is distributed through the network.
291 Mirroring to a VLAN can disrupt a network that contains unmanaged
292 switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a
293 GRE tunnel has fewer caveats than mirroring to a VLAN and should
294 generally be preferred.
296 Q: How do I configure mirroring of all traffic to a GRE tunnel?
298 A: The following commands configure br0 with eth0 and tap0 as trunk
299 ports. All traffic coming in or going out on eth0 or tap0 is also
300 mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any
301 traffic arriving on gre0 is dropped:
304 ovs-vsctl add-port br0 eth0
305 ovs-vsctl add-port br0 tap0
306 ovs-vsctl add-port br0 gre0 \
307 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
308 -- --id=@p get port gre0 \
309 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
310 -- set bridge br0 mirrors=@m
312 To later disable mirroring and destroy the GRE tunnel:
314 ovs-vsctl clear bridge br0 mirrors
315 ovs-vcstl del-port br0 gre0
317 Q: Does Open vSwitch support ERSPAN?
319 A: No. ERSPAN is an undocumented proprietary protocol. As an
320 alternative, Open vSwitch supports mirroring to a GRE tunnel (see
323 Q: Why are there so many different ways to dump flows?
325 A: Open vSwitch uses different kinds of flows for different purposes:
327 - OpenFlow flows are the most important kind of flow. OpenFlow
328 controllers use these flows to define a switch's policy.
329 OpenFlow flows support wildcards, priorities, and multiple
332 When in-band control is in use, Open vSwitch sets up a few
333 "hidden" flows, with priority higher than a controller or the
334 user can configure, that are not visible via OpenFlow. (See
335 the "Controller" section of the FAQ for more information
338 - The Open vSwitch software switch implementation uses a second
339 kind of flow internally. These flows, called "exact-match"
340 or "datapath" or "kernel" flows, do not support wildcards or
341 priorities and comprise only a single table, which makes them
342 suitable for caching. OpenFlow flows and exact-match flows
343 also support different actions and number ports differently.
345 Exact-match flows are an implementation detail that is
346 subject to change in future versions of Open vSwitch. Even
347 with the current version of Open vSwitch, hardware switch
348 implementations do not necessarily use exact-match flows.
350 Each of the commands for dumping flows has a different purpose:
352 - "ovs-ofctl dump-flows <br>" dumps OpenFlow flows, excluding
353 hidden flows. This is the most commonly useful form of flow
354 dump. (Unlike the other commands, this should work with any
355 OpenFlow switch, not just Open vSwitch.)
357 - "ovs-appctl bridge/dump-flows <br>" dumps OpenFlow flows,
358 including hidden flows. This is occasionally useful for
359 troubleshooting suspected issues with in-band control.
361 - "ovs-dpctl dump-flows [dp]" dumps the exact-match flow table
362 entries for a Linux kernel-based datapath. In Open vSwitch
363 1.10 and later, ovs-vswitchd merges multiple switches into a
364 single datapath, so it will show all the flows on all your
365 kernel-based switches. This command can occasionally be
366 useful for debugging.
368 - "ovs-appctl dpif/dump-flows <br>", new in Open vSwitch 1.10,
369 dumps exact-match flows for only the specified bridge,
370 regardless of the type.
373 Configuration Problems
374 ----------------------
376 Q: I created a bridge and added my Ethernet port to it, using commands
380 ovs-vsctl add-port br0 eth0
382 and as soon as I ran the "add-port" command I lost all connectivity
385 A: A physical Ethernet device that is part of an Open vSwitch bridge
386 should not have an IP address. If one does, then that IP address
387 will not be fully functional.
389 You can restore functionality by moving the IP address to an Open
390 vSwitch "internal" device, such as the network device named after
391 the bridge itself. For example, assuming that eth0's IP address is
392 192.168.128.5, you could run the commands below to fix up the
395 ifconfig eth0 0.0.0.0
396 ifconfig br0 192.168.128.5
398 (If your only connection to the machine running OVS is through the
399 IP address in question, then you would want to run all of these
400 commands on a single command line, or put them into a script.) If
401 there were any additional routes assigned to eth0, then you would
402 also want to use commands to adjust these routes to go through br0.
404 If you use DHCP to obtain an IP address, then you should kill the
405 DHCP client that was listening on the physical Ethernet interface
406 (e.g. eth0) and start one listening on the internal interface
407 (e.g. br0). You might still need to manually clear the IP address
408 from the physical interface (e.g. with "ifconfig eth0 0.0.0.0").
410 There is no compelling reason why Open vSwitch must work this way.
411 However, this is the way that the Linux kernel bridge module has
412 always worked, so it's a model that those accustomed to Linux
413 bridging are already used to. Also, the model that most people
414 expect is not implementable without kernel changes on all the
415 versions of Linux that Open vSwitch supports.
417 By the way, this issue is not specific to physical Ethernet
418 devices. It applies to all network devices except Open vswitch
421 Q: I created a bridge and added a couple of Ethernet ports to it,
422 using commands like these:
425 ovs-vsctl add-port br0 eth0
426 ovs-vsctl add-port br0 eth1
428 and now my network seems to have melted: connectivity is unreliable
429 (even connectivity that doesn't go through Open vSwitch), all the
430 LEDs on my physical switches are blinking, wireshark shows
431 duplicated packets, and CPU usage is very high.
433 A: More than likely, you've looped your network. Probably, eth0 and
434 eth1 are connected to the same physical Ethernet switch. This
435 yields a scenario where OVS receives a broadcast packet on eth0 and
436 sends it out on eth1, then the physical switch connected to eth1
437 sends the packet back on eth0, and so on forever. More complicated
438 scenarios, involving a loop through multiple switches, are possible
441 The solution depends on what you are trying to do:
443 - If you added eth0 and eth1 to get higher bandwidth or higher
444 reliability between OVS and your physical Ethernet switch,
445 use a bond. The following commands create br0 and then add
446 eth0 and eth1 as a bond:
449 ovs-vsctl add-bond br0 bond0 eth0 eth1
451 Bonds have tons of configuration options. Please read the
452 documentation on the Port table in ovs-vswitchd.conf.db(5)
455 - Perhaps you don't actually need eth0 and eth1 to be on the
456 same bridge. For example, if you simply want to be able to
457 connect each of them to virtual machines, then you can put
458 each of them on a bridge of its own:
461 ovs-vsctl add-port br0 eth0
464 ovs-vsctl add-port br1 eth1
466 and then connect VMs to br0 and br1. (A potential
467 disadvantage is that traffic cannot directly pass between br0
468 and br1. Instead, it will go out eth0 and come back in eth1,
471 - If you have a redundant or complex network topology and you
472 want to prevent loops, turn on spanning tree protocol (STP).
473 The following commands create br0, enable STP, and add eth0
474 and eth1 to the bridge. The order is important because you
475 don't want have to have a loop in your network even
479 ovs-vsctl set bridge br0 stp_enable=true
480 ovs-vsctl add-port br0 eth0
481 ovs-vsctl add-port br0 eth1
483 The Open vSwitch implementation of STP is not well tested.
484 Please report any bugs you observe, but if you'd rather avoid
485 acting as a beta tester then another option might be your
488 Q: I can't seem to use Open vSwitch in a wireless network.
490 A: Wireless base stations generally only allow packets with the source
491 MAC address of NIC that completed the initial handshake.
492 Therefore, without MAC rewriting, only a single device can
493 communicate over a single wireless link.
495 This isn't specific to Open vSwitch, it's enforced by the access
496 point, so the same problems will show up with the Linux bridge or
497 any other way to do bridging.
499 Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
501 A: PPP most commonly carries IP packets, but Open vSwitch works only
502 with Ethernet frames. The correct way to interface PPP to an
503 Ethernet network is usually to use routing instead of switching.
505 Q: Is there any documentation on the database tables and fields?
507 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
509 Q: When I run ovs-dpctl I no longer see the bridges I created. Instead,
510 I only see a datapath called "ovs-system". How can I see datapath
511 information about a particular bridge?
513 A: In version 1.9.0, OVS switched to using a single datapath that is
514 shared by all bridges of that type. The "ovs-appctl dpif/*"
515 commands provide similar functionality that is scoped by the bridge.
518 Quality of Service (QoS)
519 ------------------------
521 Q: How do I configure Quality of Service (QoS)?
523 A: Suppose that you want to set up bridge br0 connected to physical
524 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces
525 vif1.0 and vif2.0, and that you want to limit traffic from vif1.0
526 to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you
527 could configure the bridge this way:
531 add-port br0 eth0 -- \
532 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
533 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
534 set port eth0 qos=@newqos -- \
535 --id=@newqos create qos type=linux-htb \
536 other-config:max-rate=1000000000 \
537 queues:123=@vif10queue \
538 queues:234=@vif20queue -- \
539 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
540 --id=@vif20queue create queue other-config:max-rate=20000000
542 At this point, bridge br0 is configured with the ports and eth0 is
543 configured with the queues that you need for QoS, but nothing is
544 actually directing packets from vif1.0 or vif2.0 to the queues that
545 we have set up for them. That means that all of the packets to
546 eth0 are going to the "default queue", which is not what we want.
548 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the
549 queues reserved for them:
551 ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
552 ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
554 Each of the above flows matches on the input port, sets up the
555 appropriate queue (123 for vif1.0, 234 for vif2.0), and then
556 executes the "normal" action, which performs the same switching
557 that Open vSwitch would have done without any OpenFlow flows being
558 present. (We know that vif1.0 and vif2.0 have OpenFlow port
559 numbers 5 and 6, respectively, because we set their ofport_request
560 columns above. If we had not done that, then we would have needed
561 to find out their port numbers before setting up these flows.)
563 Now traffic going from vif1.0 or vif2.0 to eth0 should be
566 By the way, if you delete the bridge created by the above commands,
571 then that will leave one unreferenced QoS record and two
572 unreferenced Queue records in the Open vSwich database. One way to
573 clear them out, assuming you don't have other QoS or Queue records
574 that you want to keep, is:
576 ovs-vsctl -- --all destroy QoS -- --all destroy Queue
578 Q: I configured Quality of Service (QoS) in my OpenFlow network by
579 adding records to the QoS and Queue table, but the results aren't
582 A: Did you install OpenFlow flows that use your queues? This is the
583 primary way to tell Open vSwitch which queues you want to use. If
584 you don't do this, then the default queue will be used, which will
585 probably not have the effect you want.
587 Refer to the previous question for an example.
589 Q: I configured QoS, correctly, but my measurements show that it isn't
590 working as well as I expect.
592 A: With the Linux kernel, the Open vSwitch implementation of QoS has
595 - Open vSwitch configures a subset of Linux kernel QoS
596 features, according to what is in OVSDB. It is possible that
597 this code has bugs. If you believe that this is so, then you
598 can configure the Linux traffic control (QoS) stack directly
599 with the "tc" program. If you get better results that way,
600 you can send a detailed bug report to bugs@openvswitch.org.
602 It is certain that Open vSwitch cannot configure every Linux
603 kernel QoS feature. If you need some feature that OVS cannot
604 configure, then you can also use "tc" directly (or add that
607 - The Open vSwitch implementation of OpenFlow allows flows to
608 be directed to particular queues. This is pretty simple and
609 unlikely to have serious bugs at this point.
611 However, most problems with QoS on Linux are not bugs in Open
612 vSwitch at all. They tend to be either configuration errors
613 (please see the earlier questions in this section) or issues with
614 the traffic control (QoS) stack in Linux. The Open vSwitch
615 developers are not experts on Linux traffic control. We suggest
616 that, if you believe you are encountering a problem with Linux
617 traffic control, that you consult the tc manpages (e.g. tc(8),
618 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or
619 mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
627 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
628 partition a single switch into multiple switches. Suppose, for
629 example, that you have two groups of machines, group A and group B.
630 You want the machines in group A to be able to talk to each other,
631 and you want the machine in group B to be able to talk to each
632 other, but you don't want the machines in group A to be able to
633 talk to the machines in group B. You can do this with two
634 switches, by plugging the machines in group A into one switch and
635 the machines in group B into the other switch.
637 If you only have one switch, then you can use VLANs to do the same
638 thing, by configuring the ports for machines in group A as VLAN
639 "access ports" for one VLAN and the ports for group B as "access
640 ports" for a different VLAN. The switch will only forward packets
641 between ports that are assigned to the same VLAN, so this
642 effectively subdivides your single switch into two independent
643 switches, one for each group of machines.
645 So far we haven't said anything about VLAN headers. With access
646 ports, like we've described so far, no VLAN header is present in
647 the Ethernet frame. This means that the machines (or switches)
648 connected to access ports need not be aware that VLANs are
649 involved, just like in the case where we use two different physical
652 Now suppose that you have a whole bunch of switches in your
653 network, instead of just one, and that some machines in group A are
654 connected directly to both switches 1 and 2. To allow these
655 machines to talk to each other, you could add an access port for
656 group A's VLAN to switch 1 and another to switch 2, and then
657 connect an Ethernet cable between those ports. That works fine,
658 but it doesn't scale well as the number of switches and the number
659 of VLANs increases, because you use up a lot of valuable switch
660 ports just connecting together your VLANs.
662 This is where VLAN headers come in. Instead of using one cable and
663 two ports per VLAN to connect a pair of switches, we configure a
664 port on each switch as a VLAN "trunk port". Packets sent and
665 received on a trunk port carry a VLAN header that says what VLAN
666 the packet belongs to, so that only two ports total are required to
667 connect the switches, regardless of the number of VLANs in use.
668 Normally, only switches (either physical or virtual) are connected
669 to a trunk port, not individual hosts, because individual hosts
670 don't expect to see a VLAN header in the traffic that they receive.
672 None of the above discussion says anything about particular VLAN
673 numbers. This is because VLAN numbers are completely arbitrary.
674 One must only ensure that a given VLAN is numbered consistently
675 throughout a network and that different VLANs are given different
676 numbers. (That said, VLAN 0 is usually synonymous with a packet
677 that has no VLAN header, and VLAN 4095 is reserved.)
681 A: Many drivers in Linux kernels before version 3.3 had VLAN-related
682 bugs. If you are having problems with VLANs that you suspect to be
683 driver related, then you have several options:
685 - Upgrade to Linux 3.3 or later.
687 - Build and install a fixed version of the particular driver
688 that is causing trouble, if one is available.
690 - Use a NIC whose driver does not have VLAN problems.
692 - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later
693 that works around bugs in kernel drivers. To enable VLAN
694 splinters on interface eth0, use the command:
696 ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
698 For VLAN splinters to be effective, Open vSwitch must know
699 which VLANs are in use. See the "VLAN splinters" section in
700 the Interface table in ovs-vswitchd.conf.db(5) for details on
701 how Open vSwitch infers in-use VLANs.
703 VLAN splinters increase memory use and reduce performance, so
704 use them only if needed.
706 - Apply the "vlan workaround" patch from the XenServer kernel
707 patch queue, build Open vSwitch against this patched kernel,
708 and then use ovs-vlan-bug-workaround(8) to enable the VLAN
709 workaround for each interface whose driver is buggy.
711 (This is a nontrivial exercise, so this option is included
712 only for completeness.)
714 It is not always easy to tell whether a Linux kernel driver has
715 buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities
716 can help you test. See their manpages for details. Of the two
717 utilities, ovs-test(8) is newer and more thorough, but
718 ovs-vlan-test(8) may be easier to use.
720 Q: VLANs still don't work. I've tested the driver so I know that it's OK.
722 A: Do you have VLANs enabled on the physical switch that OVS is
723 attached to? Make sure that the port is configured to trunk the
724 VLAN or VLANs that you are using with OVS.
726 Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
727 and to its destination host, but OVS seems to drop incoming return
730 A: It's possible that you have the VLAN configured on your physical
731 switch as the "native" VLAN. In this mode, the switch treats
732 incoming packets either tagged with the native VLAN or untagged as
733 part of the native VLAN. It may also send outgoing packets in the
734 native VLAN without a VLAN tag.
736 If this is the case, you have two choices:
738 - Change the physical switch port configuration to tag packets
739 it forwards to OVS with the native VLAN instead of forwarding
742 - Change the OVS configuration for the physical port to a
743 native VLAN mode. For example, the following sets up a
744 bridge with port eth0 in "native-tagged" mode in VLAN 9:
747 ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged
749 In this situation, "native-untagged" mode will probably work
750 equally well. Refer to the documentation for the Port table
751 in ovs-vswitchd.conf.db(5) for more information.
753 Q: I added a pair of VMs on different VLANs, like this:
756 ovs-vsctl add-port br0 eth0
757 ovs-vsctl add-port br0 tap0 tag=9
758 ovs-vsctl add-port br0 tap1 tag=10
760 but the VMs can't access each other, the external network, or the
763 A: It is to be expected that the VMs can't access each other. VLANs
764 are a means to partition a network. When you configured tap0 and
765 tap1 as access ports for different VLANs, you indicated that they
766 should be isolated from each other.
768 As for the external network and the Internet, it seems likely that
769 the machines you are trying to access are not on VLAN 9 (or 10) and
770 that the Internet is not available on VLAN 9 (or 10).
772 Q: Can I configure an IP address on a VLAN?
774 A: Yes. Use an "internal port" configured as an access port. For
775 example, the following configures IP address 192.168.0.7 on VLAN 9.
776 That is, OVS will forward packets from eth0 to 192.168.0.7 only if
777 they have an 802.1Q header with VLAN 9. Conversely, traffic
778 forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q
782 ovs-vsctl add-port br0 eth0
783 ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
784 ifconfig vlan9 192.168.0.7
786 Q: My OpenFlow controller doesn't see the VLANs that I expect.
788 A: The configuration for VLANs in the Open vSwitch database (e.g. via
789 ovs-vsctl) only affects traffic that goes through Open vSwitch's
790 implementation of the OpenFlow "normal switching" action. By
791 default, when Open vSwitch isn't connected to a controller and
792 nothing has been manually configured in the flow table, all traffic
793 goes through the "normal switching" action. But, if you set up
794 OpenFlow flows on your own, through a controller or using ovs-ofctl
795 or through other means, then you have to implement VLAN handling
798 You can use "normal switching" as a component of your OpenFlow
799 actions, e.g. by putting "normal" into the lists of actions on
800 ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow
801 controller. In situations where this is not suitable, you can
802 implement VLAN handling yourself, e.g.:
804 - If a packet comes in on an access port, and the flow table
805 needs to send it out on a trunk port, then the flow can add
806 the appropriate VLAN tag with the "mod_vlan_vid" action.
808 - If a packet comes in on a trunk port, and the flow table
809 needs to send it out on an access port, then the flow can
810 strip the VLAN tag with the "strip_vlan" action.
812 Q: I configured ports on a bridge as access ports with different VLAN
816 ovs-vsctl set-controller br0 tcp:192.168.0.10:6633
817 ovs-vsctl add-port br0 eth0
818 ovs-vsctl add-port br0 tap0 tag=9
819 ovs-vsctl add-port br0 tap1 tag=10
821 but the VMs running behind tap0 and tap1 can still communicate,
822 that is, they are not isolated from each other even though they are
825 A: Do you have a controller configured on br0 (as the commands above
826 do)? If so, then this is a variant on the previous question, "My
827 OpenFlow controller doesn't see the VLANs that I expect," and you
828 can refer to the answer there for more information.
834 Q: What versions of OpenFlow does Open vSwitch support?
836 A: Open vSwitch 1.9 and earlier support only OpenFlow 1.0 (plus
837 extensions that bring in many of the features from later versions
840 Open vSwitch versions 1.10 and later will have experimental support
841 for OpenFlow 1.2 and 1.3. On these versions of Open vSwitch, the
842 following command enables OpenFlow 1.0, 1.2, and 1.3 on bridge br0:
844 ovs-vsctl set bridge br0 protocols=openflow10,openflow12,openflow13
846 Support for OpenFlow 1.1 is incomplete enough that it cannot yet be
847 enabled, even experimentally.
849 Support for OpenFlow 1.2 and 1.3 is still incomplete. Work to be
850 done is tracked in OPENFLOW-1.1+ in the Open vSwitch source tree
851 (also via http://openvswitch.org/development/openflow-1-x-plan/).
852 When support for a given OpenFlow version is solidly implemented,
853 Open vSwitch will enable that version by default.
855 Q: I'm getting "error type 45250 code 0". What's that?
857 A: This is a Open vSwitch extension to OpenFlow error codes. Open
858 vSwitch uses this extension when it must report an error to an
859 OpenFlow controller but no standard OpenFlow error code is
862 Open vSwitch logs the errors that it sends to controllers, so the
863 easiest thing to do is probably to look at the ovs-vswitchd log to
864 find out what the error was.
866 If you want to dissect the extended error message yourself, the
867 format is documented in include/openflow/nicira-ext.h in the Open
868 vSwitch source distribution. The extended error codes are
869 documented in lib/ofp-errors.h.
871 Q1: Some of the traffic that I'd expect my OpenFlow controller to see
872 doesn't actually appear through the OpenFlow connection, even
873 though I know that it's going through.
874 Q2: Some of the OpenFlow flows that my controller sets up don't seem
875 to apply to certain traffic, especially traffic between OVS and
876 the controller itself.
878 A: By default, Open vSwitch assumes that OpenFlow controllers are
879 connected "in-band", that is, that the controllers are actually
880 part of the network that is being controlled. In in-band mode,
881 Open vSwitch sets up special "hidden" flows to make sure that
882 traffic can make it back and forth between OVS and the controllers.
883 These hidden flows are higher priority than any flows that can be
884 set up through OpenFlow, and they are not visible through normal
885 OpenFlow flow table dumps.
887 Usually, the hidden flows are desirable and helpful, but
888 occasionally they can cause unexpected behavior. You can view the
889 full OpenFlow flow table, including hidden flows, on bridge br0
892 ovs-appctl bridge/dump-flows br0
894 to help you debug. The hidden flows are those with priorities
895 greater than 65535 (the maximum priority that can be set with
898 The DESIGN file at the top level of the Open vSwitch source
899 distribution describes the in-band model in detail.
901 If your controllers are not actually in-band (e.g. they are on
902 localhost via 127.0.0.1, or on a separate network), then you should
903 configure your controllers in "out-of-band" mode. If you have one
904 controller on bridge br0, then you can configure out-of-band mode
907 ovs-vsctl set controller br0 connection-mode=out-of-band
909 Q: I configured all my controllers for out-of-band control mode but
910 "ovs-appctl bridge/dump-flows" still shows some hidden flows.
912 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
913 set-manager"). By default, Open vSwitch assumes that managers need
914 in-band rules set up on every bridge. You can disable these rules
917 ovs-vsctl set bridge br0 other-config:disable-in-band=true
919 This actually disables in-band control entirely for the bridge, as
920 if all the bridge's controllers were configured for out-of-band
923 Q: My OpenFlow controller doesn't see the VLANs that I expect.
925 A: See answer under "VLANs", above.
927 Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop"
928 but I got a funny message like this:
930 ofp_util|INFO|normalization changed ofp_match, details:
931 ofp_util|INFO| pre: nw_dst=192.168.0.1
934 and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst
935 match had disappeared, so that the flow ends up matching every
938 A: The term "normalization" in the log message means that a flow
939 cannot match on an L3 field without saying what L3 protocol is in
940 use. The "ovs-ofctl" command above didn't specify an L3 protocol,
941 so the L3 field match was dropped.
943 In this case, the L3 protocol could be IP or ARP. A correct
944 command for each possibility is, respectively:
946 ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
950 ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
952 Similarly, a flow cannot match on an L4 field without saying what
953 L4 protocol is in use. For example, the flow match "tp_src=1234"
954 is, by itself, meaningless and will be ignored. Instead, to match
955 TCP source port 1234, write "tcp,tp_src=1234", or to match UDP
956 source port 1234, write "udp,tp_src=1234".
958 Q: How can I figure out the OpenFlow port number for a given port?
960 A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to
961 respond with an OFPT_FEATURES_REPLY that, among other information,
962 includes a mapping between OpenFlow port names and numbers. From a
963 command prompt, "ovs-ofctl show br0" makes such a request and
964 prints the response for switch br0.
966 The Interface table in the Open vSwitch database also maps OpenFlow
967 port names to numbers. To print the OpenFlow port number
968 associated with interface eth0, run:
970 ovs-vsctl get Interface eth0 ofport
972 You can print the entire mapping with:
974 ovs-vsctl -- --columns=name,ofport list Interface
976 but the output mixes together interfaces from all bridges in the
977 database, so it may be confusing if more than one bridge exists.
979 In the Open vSwitch database, ofport value -1 means that the
980 interface could not be created due to an error. (The Open vSwitch
981 log should indicate the reason.) ofport value [] (the empty set)
982 means that the interface hasn't been created yet. The latter is
983 normally an intermittent condition (unless ovs-vswitchd is not
986 Q: I added some flows with my controller or with ovs-ofctl, but when I
987 run "ovs-dpctl dump-flows" I don't see them.
989 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It
990 won't display the information that you want. You want to use
991 "ovs-ofctl dump-flows" instead.
993 Q: It looks like each of the interfaces in my bonded port shows up
994 as an individual OpenFlow port. Is that right?
996 A: Yes, Open vSwitch makes individual bond interfaces visible as
997 OpenFlow ports, rather than the bond as a whole. The interfaces
998 are treated together as a bond for only a few purposes:
1000 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow
1001 controller is not configured, this happens implicitly to
1004 - Mirrors configured for output to a bonded port.
1006 It would make a lot of sense for Open vSwitch to present a bond as
1007 a single OpenFlow port. If you want to contribute an
1008 implementation of such a feature, please bring it up on the Open
1009 vSwitch development mailing list at dev@openvswitch.org.
1011 Q: I have a sophisticated network setup involving Open vSwitch, VMs or
1012 multiple hosts, and other components. The behavior isn't what I
1015 A: To debug network behavior problems, trace the path of a packet,
1016 hop-by-hop, from its origin in one host to a remote host. If
1017 that's correct, then trace the path of the response packet back to
1020 Usually a simple ICMP echo request and reply ("ping") packet is
1021 good enough. Start by initiating an ongoing "ping" from the origin
1022 host to a remote host. If you are tracking down a connectivity
1023 problem, the "ping" will not display any successful output, but
1024 packets are still being sent. (In this case the packets being sent
1025 are likely ARP rather than ICMP.)
1027 Tools available for tracing include the following:
1029 - "tcpdump" and "wireshark" for observing hops across network
1030 devices, such as Open vSwitch internal devices and physical
1033 - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and
1034 later or "ovs-dpctl dump-flows <br>" in earlier versions.
1035 These tools allow one to observe the actions being taken on
1036 packets in ongoing flows.
1038 See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows"
1039 documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows"
1040 documentation, and "Why are there so many different ways to
1041 dump flows?" above for some background.
1043 - "ovs-appctl ofproto/trace" to observe the logic behind how
1044 ovs-vswitchd treats packets. See ovs-vswitchd(8) for
1045 documentation. You can out more details about a given flow
1046 that "ovs-dpctl dump-flows" displays, by cutting and pasting
1047 a flow from the output into an "ovs-appctl ofproto/trace"
1050 - SPAN, RSPAN, and ERSPAN features of physical switches, to
1051 observe what goes on at these physical hops.
1053 Starting at the origin of a given packet, observe the packet at
1054 each hop in turn. For example, in one plausible scenario, you
1057 1. "tcpdump" the "eth" interface through which an ARP egresses
1058 a VM, from inside the VM.
1060 2. "tcpdump" the "vif" or "tap" interface through which the ARP
1061 ingresses the host machine.
1063 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe
1064 the host interface through which the ARP egresses the
1065 physical machine. You may need to use "ovs-dpctl show" to
1066 interpret the port numbers. If the output seems surprising,
1067 you can use "ovs-appctl ofproto/trace" to observe details of
1068 how ovs-vswitchd determined the actions in the "ovs-dpctl
1071 4. "tcpdump" the "eth" interface through which the ARP egresses
1072 the physical machine.
1074 5. "tcpdump" the "eth" interface through which the ARP
1075 ingresses the physical machine, at the remote host that
1078 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the
1079 remote host that receives the ARP and observe the VM "vif"
1080 or "tap" interface to which the flow is directed. Again,
1081 "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help.
1083 7. "tcpdump" the "vif" or "tap" interface to which the ARP is
1086 8. "tcpdump" the "eth" interface through which the ARP
1087 ingresses a VM, from inside the VM.
1089 It is likely that during one of these steps you will figure out the
1090 problem. If not, then follow the ARP reply back to the origin, in
1096 bugs@openvswitch.org
1097 http://openvswitch.org/
git://git.onelab.eu / sliver-openvswitch.git / blob