2 Licensed under the Apache License, Version 2.0 (the "License"); you may
3 not use this file except in compliance with the License. You may obtain
4 a copy of the License at
6 http://www.apache.org/licenses/LICENSE-2.0
8 Unless required by applicable law or agreed to in writing, software
9 distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
10 WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
11 License for the specific language governing permissions and limitations
14 Convention for heading levels in Open vSwitch documentation:
16 ======= Heading 0 (reserved for the title in a document)
22 Avoid deeper levels because they do not render well.
24 ==========================
25 Frequently Asked Questions
26 ==========================
31 Q: What is Open vSwitch?
33 A: Open vSwitch is a production quality open source software switch
34 designed to be used as a vswitch in virtualized server environments. A
35 vswitch forwards traffic between different VMs on the same physical host
36 and also forwards traffic between VMs and the physical network. Open
37 vSwitch supports standard management interfaces (e.g. sFlow, NetFlow,
38 IPFIX, RSPAN, CLI), and is open to programmatic extension and control using
39 OpenFlow and the OVSDB management protocol.
41 Open vSwitch as designed to be compatible with modern switching chipsets.
42 This means that it can be ported to existing high-fanout switches allowing
43 the same flexible control of the physical infrastructure as the virtual
44 infrastructure. It also means that Open vSwitch will be able to take
45 advantage of on-NIC switching chipsets as their functionality matures.
47 Q: What virtualization platforms can use Open vSwitch?
49 A: Open vSwitch can currently run on any Linux-based virtualization
50 platform (kernel 3.10 and newer), including: KVM, VirtualBox, Xen, Xen
51 Cloud Platform, XenServer. As of Linux 3.3 it is part of the mainline
52 kernel. The bulk of the code is written in platform- independent C and is
53 easily ported to other environments. We welcome inquires about integrating
54 Open vSwitch with other virtualization platforms.
56 Q: How can I try Open vSwitch?
58 A: The Open vSwitch source code can be built on a Linux system. You can
59 build and experiment with Open vSwitch on any Linux machine. Packages for
60 various Linux distributions are available on many platforms, including:
61 Debian, Ubuntu, Fedora.
63 You may also download and run a virtualization platform that already has
64 Open vSwitch integrated. For example, download a recent ISO for XenServer
65 or Xen Cloud Platform. Be aware that the version integrated with a
66 particular platform may not be the most recent Open vSwitch release.
68 Q: Does Open vSwitch only work on Linux?
70 A: No, Open vSwitch has been ported to a number of different operating
71 systems and hardware platforms. Most of the development work occurs on
72 Linux, but the code should be portable to any POSIX system. We've seen
73 Open vSwitch ported to a number of different platforms, including FreeBSD,
74 Windows, and even non-POSIX embedded systems.
76 By definition, the Open vSwitch Linux kernel module only works on Linux and
77 will provide the highest performance. However, a userspace datapath is
78 available that should be very portable.
80 Q: What's involved with porting Open vSwitch to a new platform or switching ASIC?
82 A: The `porting document <Documentation/development-guide/porting.rst>`__
83 describes how one would go about porting Open vSwitch to a new operating
84 system or hardware platform.
86 Q: Why would I use Open vSwitch instead of the Linux bridge?
88 A: Open vSwitch is specially designed to make it easier to manage VM
89 network configuration and monitor state spread across many physical hosts
90 in dynamic virtualized environments. Refer to `WHY-OVS
91 <Documentation/intro/why-ovs.rst>`__ for a more detailed description of how
92 Open vSwitch relates to the Linux Bridge.
94 Q: How is Open vSwitch related to distributed virtual switches like the VMware
95 vNetwork distributed switch or the Cisco Nexus 1000V?
97 A: Distributed vswitch applications (e.g., VMware vNetwork distributed
98 switch, Cisco Nexus 1000V) provide a centralized way to configure and
99 monitor the network state of VMs that are spread across many physical
100 hosts. Open vSwitch is not a distributed vswitch itself, rather it runs on
101 each physical host and supports remote management in a way that makes it
102 easier for developers of virtualization/cloud management platforms to offer
103 distributed vswitch capabilities.
105 To aid in distribution, Open vSwitch provides two open protocols that are
106 specially designed for remote management in virtualized network
107 environments: OpenFlow, which exposes flow-based forwarding state, and the
108 OVSDB management protocol, which exposes switch port state. In addition to
109 the switch implementation itself, Open vSwitch includes tools (ovs-ofctl,
110 ovs-vsctl) that developers can script and extend to provide distributed
111 vswitch capabilities that are closely integrated with their virtualization
114 Q: Why doesn't Open vSwitch support distribution?
116 A: Open vSwitch is intended to be a useful component for building flexible
117 network infrastructure. There are many different approaches to distribution
118 which balance trade-offs between simplicity, scalability, hardware
119 compatibility, convergence times, logical forwarding model, etc. The goal
120 of Open vSwitch is to be able to support all as a primitive building block
121 rather than choose a particular point in the distributed design space.
123 Q: How can I contribute to the Open vSwitch Community?
125 A: You can start by joining the mailing lists and helping to answer
126 questions. You can also suggest improvements to documentation. If you
127 have a feature or bug you would like to work on, send a mail to `one of the
128 mailing lists <http://openvswitch.org/mlists/>`__.
130 Q: Why can I no longer connect to my OpenFlow controller or OVSDB manager?
132 A: Starting in OVS 2.4, we switched the default ports to the IANA-specified
133 port numbers for OpenFlow (6633->6653) and OVSDB (6632->6640). We
134 recommend using these port numbers, but if you cannot, all the programs
135 allow overriding the default port. See the appropriate man page.
140 Q: What does it mean for an Open vSwitch release to be LTS (long-term support)?
142 A: All official releases have been through a comprehensive testing process
143 and are suitable for production use. Planned releases occur twice a year.
144 If a significant bug is identified in an LTS release, we will provide an
145 updated release that includes the fix. Releases that are not LTS may not
146 be fixed and may just be supplanted by the next major release. The current
147 LTS release is 2.5.x.
149 For more information on the Open vSwitch release process, refer to `release
150 process <Documentation/release-process.rst>`__.
152 Q: What Linux kernel versions does each Open vSwitch release work with?
154 A: The following table lists the Linux kernel versions against which the
155 given versions of the Open vSwitch kernel module will successfully build.
156 The Linux kernel versions are upstream kernel versions, so Linux kernels
157 modified from the upstream sources may not build in some cases even if they
158 are based on a supported version. This is most notably true of Red Hat
159 Enterprise Linux (RHEL) kernels, which are extensively modified from
162 ============ ==============
163 Open vSwitch Linux kernel
164 ============ ==============
179 ============ ==============
181 Open vSwitch userspace should also work with the Linux kernel module built
182 into Linux 3.3 and later.
184 Open vSwitch userspace is not sensitive to the Linux kernel version. It
185 should build against almost any kernel, certainly against 2.6.32 and later.
187 Q: Are all features available with all datapaths?
189 A: Open vSwitch supports different datapaths on different platforms. Each
190 datapath has a different feature set: the following tables try to summarize
196 The datapath implemented by the kernel module shipped with Linux
197 upstream. Since features have been gradually introduced into the kernel,
198 the table mentions the first Linux release whose OVS module supports the
202 The datapath implemented by the Linux kernel module distributed with the
206 Also known as DPDK, dpif-netdev or dummy datapath. It is the only
207 datapath that works on NetBSD, FreeBSD and Mac OSX.
210 Also known as the Windows datapath.
212 The following table lists the datapath supported features from an Open
213 vSwitch user's perspective.
215 ===================== ============== ============== ========= =======
216 Feature Linux upstream Linux OVS tree Userspace Hyper-V
217 ===================== ============== ============== ========= =======
219 Connection tracking 4.3 YES PARTIAL PARTIAL
220 Tunnel - LISP NO YES NO NO
221 Tunnel - STT NO YES NO YES
222 Tunnel - GRE 3.11 YES YES YES
223 Tunnel - VXLAN 3.12 YES YES YES
224 Tunnel - Geneve 3.18 YES YES YES
225 Tunnel - GRE-IPv6 NO NO YES NO
226 Tunnel - VXLAN-IPv6 4.3 YES YES NO
227 Tunnel - Geneve-IPv6 4.4 YES YES NO
228 QoS - Policing YES YES YES NO
229 QoS - Shaping YES YES NO NO
231 IPFIX 3.10 YES YES NO
232 Set action YES YES YES PARTIAL
233 NIC Bonding YES YES YES NO
234 Multiple VTEPs YES YES YES NO
235 ===================== ============== ============== ========= =======
239 * Only a limited set of flow fields is modifiable via the set action by the
241 * The Hyper-V datapath only supports one physical NIC per datapath. This is
242 why bonding is not supported.
243 * The Hyper-V datapath can have at most one IP address configured as a
246 The following table lists features that do not *directly* impact an Open
247 vSwitch user, e.g. because their absence can be hidden by the ofproto layer
248 (usually this comes with a performance penalty).
250 ===================== ============== ============== ========= =======
251 Feature Linux upstream Linux OVS tree Userspace Hyper-V
252 ===================== ============== ============== ========= =======
253 SCTP flows 3.12 YES YES YES
254 MPLS 3.19 YES YES YES
256 Megaflows 3.12 YES YES NO
257 Masked set action 4.0 YES YES NO
258 Recirculation 3.19 YES YES YES
259 TCP flags matching 3.13 YES YES NO
260 Validate flow actions YES YES N/A NO
261 Multiple datapaths YES YES YES NO
262 Tunnel TSO - STT N/A YES NO YES
263 ===================== ============== ============== ========= =======
265 Q: What DPDK version does each Open vSwitch release work with?
267 A: The following table lists the DPDK version against which the given
268 versions of Open vSwitch will successfully build.
280 Q: I get an error like this when I configure Open vSwitch::
282 configure: error: Linux kernel in <dir> is version <x>, but
283 version newer than <y> is not supported (please refer to the
288 A: You have the following options:
290 - Use the Linux kernel module supplied with the kernel that you are using.
291 (See also the following FAQ.)
293 - If there is a newer released version of Open vSwitch, consider building
294 that one, because it may support the kernel that you are building
295 against. (To find out, consult the table in the previous FAQ.)
297 - The Open vSwitch "master" branch may support the kernel that you are
298 using, so consider building the kernel module from "master".
300 All versions of Open vSwitch userspace are compatible with all versions of
301 the Open vSwitch kernel module, so you do not have to use the kernel module
302 from one source along with the userspace programs from the same source.
304 Q: What features are not available in the Open vSwitch kernel datapath that
305 ships as part of the upstream Linux kernel?
307 A: The kernel module in upstream Linux does not include support for LISP.
308 Work is in progress to add support for LISP to the upstream Linux version
309 of the Open vSwitch kernel module. For now, if you need this feature, use
310 the kernel module from the Open vSwitch distribution instead of the
311 upstream Linux kernel module.
313 Certain features require kernel support to function or to have reasonable
314 performance. If the ovs-vswitchd log file indicates that a feature is not
315 supported, consider upgrading to a newer upstream Linux release or using
316 the kernel module paired with the userspace distribution.
318 Q: Why do tunnels not work when using a kernel module other than the one
319 packaged with Open vSwitch?
321 A: Support for tunnels was added to the upstream Linux kernel module after
322 the rest of Open vSwitch. As a result, some kernels may contain support for
323 Open vSwitch but not tunnels. The minimum kernel version that supports each
326 ======== ============
327 Protocol Linux Kernel
328 ======== ============
334 ======== ============
336 If you are using a version of the kernel that is older than the one listed
337 above, it is still possible to use that tunnel protocol. However, you must
338 compile and install the kernel module included with the Open vSwitch
339 distribution rather than the one on your machine. If problems persist after
340 doing this, check to make sure that the module that is loaded is the one
343 Q: Why are UDP tunnel checksums not computed for VXLAN or Geneve?
345 A: Generating outer UDP checksums requires kernel support that was not part
346 of the initial implementation of these protocols. If using the upstream
347 Linux Open vSwitch module, you must use kernel 4.0 or newer. The
348 out-of-tree modules from Open vSwitch release 2.4 and later support UDP
351 Q: What features are not available when using the userspace datapath?
353 A: Tunnel virtual ports are not supported, as described in the previous
354 answer. It is also not possible to use queue-related actions. On Linux
355 kernels before 2.6.39, maximum-sized VLAN packets may not be transmitted.
357 Q: Should userspace or kernel be upgraded first to minimize downtime?
359 A. In general, the Open vSwitch userspace should be used with the kernel
360 version included in the same release or with the version from upstream
361 Linux. However, when upgrading between two releases of Open vSwitch it is
362 best to migrate userspace first to reduce the possibility of
365 Q: What happened to the bridge compatibility feature?
367 A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier.
368 When it was enabled, Open vSwitch imitated the interface of the Linux
369 kernel "bridge" module. This allowed users to drop Open vSwitch into
370 environments designed to use the Linux kernel bridge module without
371 adapting the environment to use Open vSwitch.
373 Open vSwitch 1.10 and later do not support bridge compatibility. The
374 feature was dropped because version 1.10 adopted a new internal
375 architecture that made bridge compatibility difficult to maintain. Now
376 that many environments use OVS directly, it would be rarely useful in any
379 To use bridge compatibility, install OVS 1.9 or earlier, including the
380 accompanying kernel modules (both the main and bridge compatibility
381 modules), following the instructions that come with the release. Be sure
382 to start the ovs-brcompatd daemon.
387 Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation
388 keeps talking about bridges. What's a bridge?
390 A: In networking, the terms "bridge" and "switch" are synonyms. Open
391 vSwitch implements an Ethernet switch, which means that it is also an
396 A: See the "VLAN" section below.
401 Q: How do I configure a port as an access port?
403 A. Add ``tag=VLAN`` to your ``ovs-vsctl add-port`` command. For example,
404 the following commands configure br0 with eth0 as a trunk port (the
405 default) and tap0 as an access port for VLAN 9:
409 $ ovs-vsctl add-br br0
410 $ ovs-vsctl add-port br0 eth0
411 $ ovs-vsctl add-port br0 tap0 tag=9
413 If you want to configure an already added port as an access port, use
414 ``ovs-vsctl set``, e.g.:
418 $ ovs-vsctl set port tap0 tag=9
420 Q: How do I configure a port as a SPAN port, that is, enable mirroring of all
421 traffic to that port?
423 A. The following commands configure br0 with eth0 and tap0 as trunk ports.
424 All traffic coming in or going out on eth0 or tap0 is also mirrored to
425 tap1; any traffic arriving on tap1 is dropped:
429 $ ovs-vsctl add-br br0
430 $ ovs-vsctl add-port br0 eth0
431 $ ovs-vsctl add-port br0 tap0
432 $ ovs-vsctl add-port br0 tap1 \
433 -- --id=@p get port tap1 \
434 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
435 -- set bridge br0 mirrors=@m
437 To later disable mirroring, run:
441 $ ovs-vsctl clear bridge br0 mirrors
443 Q: Does Open vSwitch support configuring a port in promiscuous mode?
445 A: Yes. How you configure it depends on what you mean by "promiscuous
448 - Conventionally, "promiscuous mode" is a feature of a network interface
449 card. Ordinarily, a NIC passes to the CPU only the packets actually
450 destined to its host machine. It discards the rest to avoid wasting
451 memory and CPU cycles. When promiscuous mode is enabled, however, it
452 passes every packet to the CPU. On an old-style shared-media or
453 hub-based network, this allows the host to spy on all packets on the
454 network. But in the switched networks that are almost everywhere these
455 days, promiscuous mode doesn't have much effect, because few packets not
456 destined to a host are delivered to the host's NIC.
458 This form of promiscuous mode is configured in the guest OS of the VMs on
459 your bridge, e.g. with "ifconfig".
461 - The VMware vSwitch uses a different definition of "promiscuous mode".
462 When you configure promiscuous mode on a VMware vNIC, the vSwitch sends a
463 copy of every packet received by the vSwitch to that vNIC. That has a
464 much bigger effect than just enabling promiscuous mode in a guest OS.
465 Rather than getting a few stray packets for which the switch does not yet
466 know the correct destination, the vNIC gets every packet. The effect is
467 similar to replacing the vSwitch by a virtual hub.
469 This "promiscuous mode" is what switches normally call "port mirroring"
470 or "SPAN". For information on how to configure SPAN, see "How do I
471 configure a port as a SPAN port, that is, enable mirroring of all traffic
474 Q: How do I configure a DPDK port as an access port?
476 A: Firstly, you must have a DPDK-enabled version of Open vSwitch.
478 If your version is DPDK-enabled it will support the other-config:dpdk-init
479 configuration in the database and will display lines with "EAL:..." during
480 startup when other_config:dpdk-init is set to 'true'.
482 Secondly, when adding a DPDK port, unlike a system port, the type for the
483 interface must be specified. For example::
485 $ ovs-vsctl add-br br0
486 $ ovs-vsctl add-port br0 dpdk0 -- set Interface dpdk0 type=dpdk
488 Finally, it is required that DPDK port names begin with ``dpdk``.
490 Refer to the `DPDK installation guide`_ for more information on enabling
491 and using DPDK with Open vSwitch.
493 Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all
494 traffic to that VLAN?
496 A: The following commands configure br0 with eth0 as a trunk port and tap0
497 as an access port for VLAN 10. All traffic coming in or going out on tap0,
498 as well as traffic coming in or going out on eth0 in VLAN 10, is also
499 mirrored to VLAN 15 on eth0. The original tag for VLAN 10, in cases where
500 one is present, is dropped as part of mirroring:
504 $ ovs-vsctl add-br br0
505 $ ovs-vsctl add-port br0 eth0
506 $ ovs-vsctl add-port br0 tap0 tag=10
508 -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \
510 -- set bridge br0 mirrors=@m
512 To later disable mirroring, run:
516 $ ovs-vsctl clear bridge br0 mirrors
518 Mirroring to a VLAN can disrupt a network that contains unmanaged switches.
519 See ovs-vswitchd.conf.db(5) for details. Mirroring to a GRE tunnel has
520 fewer caveats than mirroring to a VLAN and should generally be preferred.
522 Q: Can I mirror more than one input VLAN to an RSPAN VLAN?
524 A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor of
525 the specified output-vlan. This loss of information may make the mirrored
526 traffic too hard to interpret.
528 To mirror multiple VLANs, use the commands above, but specify a
529 comma-separated list of VLANs as the value for select-vlan. To mirror
530 every VLAN, use the commands above, but omit select-vlan and its value
533 When a packet arrives on a VLAN that is used as a mirror output VLAN, the
534 mirror is disregarded. Instead, in standalone mode, OVS floods the packet
535 across all the ports for which the mirror output VLAN is configured. (If
536 an OpenFlow controller is in use, then it can override this behavior
537 through the flow table.) If OVS is used as an intermediate switch, rather
538 than an edge switch, this ensures that the RSPAN traffic is distributed
541 Mirroring to a VLAN can disrupt a network that contains unmanaged switches.
542 See ovs-vswitchd.conf.db(5) for details. Mirroring to a GRE tunnel has
543 fewer caveats than mirroring to a VLAN and should generally be preferred.
545 Q: How do I configure mirroring of all traffic to a GRE tunnel?
547 A: The following commands configure br0 with eth0 and tap0 as trunk ports.
548 All traffic coming in or going out on eth0 or tap0 is also mirrored to
549 gre0, a GRE tunnel to the remote host 192.168.1.10; any traffic arriving on
552 $ ovs-vsctl add-br br0
553 $ ovs-vsctl add-port br0 eth0
554 $ ovs-vsctl add-port br0 tap0
555 $ ovs-vsctl add-port br0 gre0 \
556 -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \
557 -- --id=@p get port gre0 \
558 -- --id=@m create mirror name=m0 select-all=true output-port=@p \
559 -- set bridge br0 mirrors=@m
561 To later disable mirroring and destroy the GRE tunnel::
563 $ ovs-vsctl clear bridge br0 mirrors
564 $ ovs-vsctl del-port br0 gre0
566 Q: Does Open vSwitch support ERSPAN?
568 A: No. As an alternative, Open vSwitch supports mirroring to a GRE tunnel
571 Q: How do I connect two bridges?
573 A: First, why do you want to do this? Two connected bridges are not much
574 different from a single bridge, so you might as well just have a single
575 bridge with all your ports on it.
577 If you still want to connect two bridges, you can use a pair of patch
578 ports. The following example creates bridges br0 and br1, adds eth0 and
579 tap0 to br0, adds tap1 to br1, and then connects br0 and br1 with a pair of
584 $ ovs-vsctl add-br br0
585 $ ovs-vsctl add-port br0 eth0
586 $ ovs-vsctl add-port br0 tap0
587 $ ovs-vsctl add-br br1
588 $ ovs-vsctl add-port br1 tap1
590 -- add-port br0 patch0 \
591 -- set interface patch0 type=patch options:peer=patch1 \
592 -- add-port br1 patch1 \
593 -- set interface patch1 type=patch options:peer=patch0
595 Bridges connected with patch ports are much like a single bridge. For
596 instance, if the example above also added eth1 to br1, and both eth0 and
597 eth1 happened to be connected to the same next-hop switch, then you could
598 loop your network just as you would if you added eth0 and eth1 to the same
599 bridge (see the "Configuration Problems" section below for more
602 If you are using Open vSwitch 1.9 or an earlier version, then you need to
603 be using the kernel module bundled with Open vSwitch rather than the one
604 that is integrated into Linux 3.3 and later, because Open vSwitch 1.9 and
605 earlier versions need kernel support for patch ports. This also means that
606 in Open vSwitch 1.9 and earlier, patch ports will not work with the
607 userspace datapath, only with the kernel module.
609 Q: How do I configure a bridge without an OpenFlow local port? (Local port in
610 the sense of OFPP_LOCAL)
612 A: Open vSwitch does not support such a configuration. Bridges always have
615 Implementation Details
616 ----------------------
618 Q: I hear OVS has a couple of kinds of flows. Can you tell me about them?
620 A: Open vSwitch uses different kinds of flows for different purposes:
622 - OpenFlow flows are the most important kind of flow. OpenFlow controllers
623 use these flows to define a switch's policy. OpenFlow flows support
624 wildcards, priorities, and multiple tables.
626 When in-band control is in use, Open vSwitch sets up a few "hidden"
627 flows, with priority higher than a controller or the user can configure,
628 that are not visible via OpenFlow. (See the "Controller" section of the
629 FAQ for more information about hidden flows.)
631 - The Open vSwitch software switch implementation uses a second kind of
632 flow internally. These flows, called "datapath" or "kernel" flows, do
633 not support priorities and comprise only a single table, which makes them
634 suitable for caching. (Like OpenFlow flows, datapath flows do support
635 wildcarding, in Open vSwitch 1.11 and later.) OpenFlow flows and
636 datapath flows also support different actions and number ports
639 Datapath flows are an implementation detail that is subject to change in
640 future versions of Open vSwitch. Even with the current version of Open
641 vSwitch, hardware switch implementations do not necessarily use this
644 Users and controllers directly control only the OpenFlow flow table. Open
645 vSwitch manages the datapath flow table itself, so users should not normally be
648 Q: Why are there so many different ways to dump flows?
650 A: Open vSwitch has two kinds of flows (see the previous question), so it
651 has commands with different purposes for dumping each kind of flow:
653 - ``ovs-ofctl dump-flows <br>`` dumps OpenFlow flows, excluding hidden
654 flows. This is the most commonly useful form of flow dump. (Unlike the
655 other commands, this should work with any OpenFlow switch, not just Open
658 - ``ovs-appctl bridge/dump-flows <br>`` dumps OpenFlow flows, including
659 hidden flows. This is occasionally useful for troubleshooting suspected
660 issues with in-band control.
662 - ``ovs-dpctl dump-flows [dp]`` dumps the datapath flow table entries for a
663 Linux kernel-based datapath. In Open vSwitch 1.10 and later,
664 ovs-vswitchd merges multiple switches into a single datapath, so it will
665 show all the flows on all your kernel-based switches. This command can
666 occasionally be useful for debugging.
668 - ``ovs-appctl dpif/dump-flows <br>``, new in Open vSwitch 1.10, dumps
669 datapath flows for only the specified bridge, regardless of the type.
671 Q: How does multicast snooping works with VLANs?
673 A: Open vSwitch maintains snooping tables for each VLAN.
675 Q: Can OVS populate the kernel flow table in advance instead of in reaction to
678 A: No. There are several reasons:
680 - Kernel flows are not as sophisticated as OpenFlow flows, which means that
681 some OpenFlow policies could require a large number of kernel flows. The
682 "conjunctive match" feature is an extreme example: the number of kernel
683 flows it requires is the product of the number of flows in each
686 - With multiple OpenFlow flow tables and simple sets of actions, the number
687 of kernel flows required can be as large as the product of the number of
688 flows in each dimension. With more sophisticated actions, the number of
689 kernel flows could be even larger.
691 - Open vSwitch is designed so that any version of OVS userspace
692 interoperates with any version of the OVS kernel module. This forward
693 and backward compatibility requires that userspace observe how the kernel
694 module parses received packets. This is only possible in a
695 straightforward way when userspace adds kernel flows in reaction to
698 For more relevant information on the architecture of Open vSwitch, please
699 read "The Design and Implementation of Open vSwitch", published in USENIX
705 Q: I just upgraded and I see a performance drop. Why?
707 A: The OVS kernel datapath may have been updated to a newer version than
708 the OVS userspace components. Sometimes new versions of OVS kernel module
709 add functionality that is backwards compatible with older userspace
710 components but may cause a drop in performance with them. Especially, if a
711 kernel module from OVS 2.1 or newer is paired with OVS userspace 1.10 or
712 older, there will be a performance drop for TCP traffic.
714 Updating the OVS userspace components to the latest released version should
715 fix the performance degradation.
717 To get the best possible performance and functionality, it is recommended to
718 pair the same versions of the kernel module and OVS userspace.
720 Configuration Problems
721 ----------------------
723 Q: I created a bridge and added my Ethernet port to it, using commands like
727 ovs-vsctl add-port br0 eth0
729 and as soon as I ran the "add-port" command I lost all connectivity through
732 A: A physical Ethernet device that is part of an Open vSwitch bridge should
733 not have an IP address. If one does, then that IP address will not be
736 You can restore functionality by moving the IP address to an Open vSwitch
737 "internal" device, such as the network device named after the bridge
738 itself. For example, assuming that eth0's IP address is 192.168.128.5, you
739 could run the commands below to fix up the situation::
741 $ ifconfig eth0 0.0.0.0
742 $ ifconfig br0 192.168.128.5
744 (If your only connection to the machine running OVS is through the IP
745 address in question, then you would want to run all of these commands on a
746 single command line, or put them into a script.) If there were any
747 additional routes assigned to eth0, then you would also want to use
748 commands to adjust these routes to go through br0.
750 If you use DHCP to obtain an IP address, then you should kill the DHCP
751 client that was listening on the physical Ethernet interface (e.g. eth0)
752 and start one listening on the internal interface (e.g. br0). You might
753 still need to manually clear the IP address from the physical interface
754 (e.g. with "ifconfig eth0 0.0.0.0").
756 There is no compelling reason why Open vSwitch must work this way.
757 However, this is the way that the Linux kernel bridge module has always
758 worked, so it's a model that those accustomed to Linux bridging are already
759 used to. Also, the model that most people expect is not implementable
760 without kernel changes on all the versions of Linux that Open vSwitch
763 By the way, this issue is not specific to physical Ethernet devices. It
764 applies to all network devices except Open vSwitch "internal" devices.
766 Q: I created a bridge and added a couple of Ethernet ports to it, using
767 commands like these::
769 $ ovs-vsctl add-br br0
770 $ ovs-vsctl add-port br0 eth0
771 $ ovs-vsctl add-port br0 eth1
773 and now my network seems to have melted: connectivity is unreliable (even
774 connectivity that doesn't go through Open vSwitch), all the LEDs on my physical
775 switches are blinking, wireshark shows duplicated packets, and CPU usage is
778 A: More than likely, you've looped your network. Probably, eth0 and eth1
779 are connected to the same physical Ethernet switch. This yields a scenario
780 where OVS receives a broadcast packet on eth0 and sends it out on eth1,
781 then the physical switch connected to eth1 sends the packet back on eth0,
782 and so on forever. More complicated scenarios, involving a loop through
783 multiple switches, are possible too.
785 The solution depends on what you are trying to do:
787 - If you added eth0 and eth1 to get higher bandwidth or higher reliability
788 between OVS and your physical Ethernet switch, use a bond. The following
789 commands create br0 and then add eth0 and eth1 as a bond::
791 $ ovs-vsctl add-br br0
792 $ ovs-vsctl add-bond br0 bond0 eth0 eth1
794 Bonds have tons of configuration options. Please read the documentation
795 on the Port table in ovs-vswitchd.conf.db(5) for all the details.
797 Configuration for DPDK-enabled interfaces is slightly less
798 straightforward. Refer to the `DPDK installation guide`_ for more
801 - Perhaps you don't actually need eth0 and eth1 to be on the same bridge.
802 For example, if you simply want to be able to connect each of them to
803 virtual machines, then you can put each of them on a bridge of its own:
805 $ ovs-vsctl add-br br0
806 $ ovs-vsctl add-port br0 eth0
808 $ ovs-vsctl add-br br1
809 $ ovs-vsctl add-port br1 eth1
811 and then connect VMs to br0 and br1. (A potential disadvantage is that
812 traffic cannot directly pass between br0 and br1. Instead, it will go
813 out eth0 and come back in eth1, or vice versa.)
815 - If you have a redundant or complex network topology and you want to
816 prevent loops, turn on spanning tree protocol (STP). The following
817 commands create br0, enable STP, and add eth0 and eth1 to the bridge.
818 The order is important because you don't want have to have a loop in your
819 network even transiently::
821 $ ovs-vsctl add-br br0
822 $ ovs-vsctl set bridge br0 stp_enable=true
823 $ ovs-vsctl add-port br0 eth0
824 $ ovs-vsctl add-port br0 eth1
826 The Open vSwitch implementation of STP is not well tested. Report any
827 bugs you observe, but if you'd rather avoid acting as a beta tester then
828 another option might be your best shot.
830 Q: I can't seem to use Open vSwitch in a wireless network.
832 A: Wireless base stations generally only allow packets with the source MAC
833 address of NIC that completed the initial handshake. Therefore, without
834 MAC rewriting, only a single device can communicate over a single wireless
837 This isn't specific to Open vSwitch, it's enforced by the access point, so
838 the same problems will show up with the Linux bridge or any other way to do
841 Q: I can't seem to add my PPP interface to an Open vSwitch bridge.
843 A: PPP most commonly carries IP packets, but Open vSwitch works only with
844 Ethernet frames. The correct way to interface PPP to an Ethernet network
845 is usually to use routing instead of switching.
847 Q: Is there any documentation on the database tables and fields?
849 A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference.
851 Q: When I run ovs-dpctl I no longer see the bridges I created. Instead, I only
852 see a datapath called "ovs-system". How can I see datapath information about a
855 A: In version 1.9.0, OVS switched to using a single datapath that is shared
856 by all bridges of that type. The ``ovs-appctl dpif/*`` commands provide
857 similar functionality that is scoped by the bridge.
859 Q: I created a GRE port using ovs-vsctl so why can't I send traffic or see the
860 port in the datapath?
862 A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module
863 cannot be loaded at the same time. It is likely that on your system the
864 Linux GRE module is already loaded and blocking OVS (to confirm, check
865 dmesg for errors regarding GRE registration). To fix this, unload all GRE
866 modules that appear in lsmod as well as the OVS kernel module. You can then
867 reload the OVS module following the directions in the `general installation
868 guide`_, which will ensure that dependencies are satisfied.
870 Q: Open vSwitch does not seem to obey my packet filter rules.
872 A: It depends on mechanisms and configurations you want to use.
874 You cannot usefully use typical packet filters, like iptables, on physical
875 Ethernet ports that you add to an Open vSwitch bridge. This is because
876 Open vSwitch captures packets from the interface at a layer lower below
877 where typical packet-filter implementations install their hooks. (This
878 actually applies to any interface of type "system" that you might add to an
879 Open vSwitch bridge.)
881 You can usefully use typical packet filters on Open vSwitch internal ports
882 as they are mostly ordinary interfaces from the point of view of packet
885 For example, suppose you create a bridge br0 and add Ethernet port eth0 to
886 it. Then you can usefully add iptables rules to affect the internal
887 interface br0, but not the physical interface eth0. (br0 is also where you
888 would add an IP address, as discussed elsewhere in the FAQ.)
890 For simple filtering rules, it might be possible to achieve similar results
891 by installing appropriate OpenFlow flows instead. The OVS conntrack
892 feature (see the "ct" action in ovs-ofctl(8)) can implement a stateful
895 If the use of a particular packet filter setup is essential, Open vSwitch
896 might not be the best choice for you. On Linux, you might want to consider
897 using the Linux Bridge. (This is the only choice if you want to use
898 ebtables rules.) On NetBSD, you might want to consider using the bridge(4)
899 with BRIDGE_IPF option.
901 Q: It seems that Open vSwitch does nothing when I removed a port and then
902 immediately put it back. For example, consider that p1 is a port of
905 $ ovs-vsctl del-port br0 p1 -- \
907 set interface p1 type=internal
909 A: It's an expected behaviour.
911 If del-port and add-port happen in a single OVSDB transaction as your
912 example, Open vSwitch always "skips" the intermediate steps. Even if they
913 are done in multiple transactions, it's still allowed for Open vSwitch to
914 skip the intermediate steps and just implement the overall effect. In both
915 cases, your example would be turned into a no-op.
917 If you want to make Open vSwitch actually destroy and then re-create the
918 port for some side effects like resetting kernel setting for the
919 corresponding interface, you need to separate operations into multiple
920 OVSDB transactions and ensure that at least the first one does not have
921 ``--no-wait``. In the following example, the first ovs-vsctl will block
922 until Open vSwitch reloads the new configuration and removes the port::
924 $ ovs-vsctl del-port br0 p1
925 $ ovs-vsctl add-port br0 p1 -- \
926 set interface p1 type=internal
928 Q: I want to add thousands of ports to an Open vSwitch bridge, but it takes too
929 long (minutes or hours) to do it with ovs-vsctl. How can I do it faster?
931 A: If you add them one at a time with ovs-vsctl, it can take a long time to
932 add thousands of ports to an Open vSwitch bridge. This is because every
933 invocation of ovs-vsctl first reads the current configuration from OVSDB.
934 As the number of ports grows, this starts to take an appreciable amount of
935 time, and when it is repeated thousands of times the total time becomes
938 The solution is to add the ports in one invocation of ovs-vsctl (or a small
939 number of them). For example, using bash::
941 $ ovs-vsctl add-br br0
942 $ cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done
945 takes seconds, not minutes or hours, in the OVS sandbox environment.
947 Q: I created a bridge named br0. My bridge shows up in "ovs-vsctl show", but
948 "ovs-ofctl show br0" just prints "br0 is not a bridge or a socket".
950 A: Open vSwitch wasn't able to create the bridge. Check the ovs-vswitchd
951 log for details (Debian and Red Hat packaging for Open vSwitch put it in
952 /var/log/openvswitch/ovs-vswitchd.log).
954 In general, the Open vSwitch database reflects the desired configuration
955 state. ovs-vswitchd monitors the database and, when it changes,
956 reconfigures the system to reflect the new desired state. This normally
957 happens very quickly. Thus, a discrepancy between the database and the
958 actual state indicates that ovs-vswitchd could not implement the
959 configuration, and so one should check the log to find out why. (Another
960 possible cause is that ovs-vswitchd is not running. This will make
961 ovs-vsctl commands hang, if they change the configuration, unless one
962 specifies ``--no-wait``.)
964 Q: I have a bridge br0. I added a new port vif1.0, and it shows up in
965 "ovs-vsctl show", but "ovs-vsctl list port" says that it has OpenFlow port
966 ("ofport") -1, and "ovs-ofctl show br0" doesn't show vif1.0 at all.
968 A: Open vSwitch wasn't able to create the port. Check the ovs-vswitchd log
969 for details (Debian and Red Hat packaging for Open vSwitch put it in
970 /var/log/openvswitch/ovs-vswitchd.log). Please see the previous question
971 for more information.
973 You may want to upgrade to Open vSwitch 2.3 (or later), in which ovs-vsctl
974 will immediately report when there is an issue creating a port.
976 Q: I created a tap device tap0, configured an IP address on it, and added it to
977 a bridge, like this::
980 $ ifconfig tap0 192.168.0.123
981 $ ovs-vsctl add-br br0
982 $ ovs-vsctl add-port br0 tap0
984 I expected that I could then use this IP address to contact other hosts on the
985 network, but it doesn't work. Why not?
987 A: The short answer is that this is a misuse of a "tap" device. Use an
988 "internal" device implemented by Open vSwitch, which works differently and
989 is designed for this use. To solve this problem with an internal device,
992 $ ovs-vsctl add-br br0
993 $ ovs-vsctl add-port br0 int0 -- set Interface int0 type=internal
994 $ ifconfig int0 192.168.0.123
996 Even more simply, you can take advantage of the internal port that every
997 bridge has under the name of the bridge::
999 $ ovs-vsctl add-br br0
1000 $ ifconfig br0 192.168.0.123
1002 In more detail, a "tap" device is an interface between the Linux (or BSD)
1003 network stack and a user program that opens it as a socket. When the "tap"
1004 device transmits a packet, it appears in the socket opened by the userspace
1005 program. Conversely, when the userspace program writes to the "tap"
1006 socket, the kernel TCP/IP stack processes the packet as if it had been
1007 received by the "tap" device.
1009 Consider the configuration above. Given this configuration, if you "ping"
1010 an IP address in the 192.168.0.x subnet, the Linux kernel routing stack
1011 will transmit an ARP on the tap0 device. Open vSwitch userspace treats
1012 "tap" devices just like any other network device; that is, it doesn't open
1013 them as "tap" sockets. That means that the ARP packet will simply get
1016 You might wonder why the Open vSwitch kernel module doesn't intercept the
1017 ARP packet and bridge it. After all, Open vSwitch intercepts packets on
1018 other devices. The answer is that Open vSwitch only intercepts *received*
1019 packets, but this is a packet being transmitted. The same thing happens
1020 for all other types of network devices, except for Open vSwitch "internal"
1021 ports. If you, for example, add a physical Ethernet port to an OVS bridge,
1022 configure an IP address on a physical Ethernet port, and then issue a
1023 "ping" to an address in that subnet, the same thing happens: an ARP gets
1024 transmitted on the physical Ethernet port and Open vSwitch never sees it.
1025 (You should not do that, as documented at the beginning of this section.)
1027 It can make sense to add a "tap" device to an Open vSwitch bridge, if some
1028 userspace program (other than Open vSwitch) has opened the tap socket.
1029 This is the case, for example, if the "tap" device was created by KVM (or
1030 QEMU) to simulate a virtual NIC. In such a case, when OVS bridges a packet
1031 to the "tap" device, the kernel forwards that packet to KVM in userspace,
1032 which passes it along to the VM, and in the other direction, when the VM
1033 sends a packet, KVM writes it to the "tap" socket, which causes OVS to
1034 receive it and bridge it to the other OVS ports. Please note that in such
1035 a case no IP address is configured on the "tap" device (there is normally
1036 an IP address configured in the virtual NIC inside the VM, but this is not
1037 visible to the host Linux kernel or to Open vSwitch).
1039 There is one special case in which Open vSwitch does directly read and
1040 write "tap" sockets. This is an implementation detail of the Open vSwitch
1041 userspace switch, which implements its "internal" ports as Linux (or BSD)
1042 "tap" sockets. In such a userspace switch, OVS receives packets sent on
1043 the "tap" device used to implement an "internal" port by reading the
1044 associated "tap" socket, and bridges them to the rest of the switch. In
1045 the other direction, OVS transmits packets bridged to the "internal" port
1046 by writing them to the "tap" socket, causing them to be processed by the
1047 kernel TCP/IP stack as if they had been received on the "tap" device.
1048 Users should not need to be concerned with this implementation detail.
1050 Open vSwitch has a network device type called "tap". This is intended only
1051 for implementing "internal" ports in the OVS userspace switch and should
1052 not be used otherwise. In particular, users should not configure KVM "tap"
1053 devices as type "tap" (use type "system", the default, instead).
1055 Q: I observe packet loss at the beginning of RFC2544 tests on a server running
1056 few hundred container apps bridged to OVS with traffic generated by HW traffic
1057 generator. How can I fix this?
1059 A: This is expected behavior on virtual switches. RFC2544 tests were
1060 designed for hardware switches, which don't have caches on the fastpath
1061 that need to be heated. Traffic generators in order to prime the switch
1062 use learning phase to heat the caches before sending the actual traffic in
1063 test phase. In case of OVS the cache is flushed quickly and to accommodate
1064 the traffic generator's delay between learning and test phase, the max-idle
1065 timeout settings should be changed to 50000 ms.::
1067 $ ovs-vsctl --no-wait set Open_vSwitch . other_config:max-idle=50000
1069 Q: How can I configure the bridge internal interface MTU? Why does Open vSwitch
1070 keep changing internal ports MTU?
1072 A: By default Open vSwitch overrides the internal interfaces (e.g. br0)
1073 MTU. If you have just an internal interface (e.g. br0) and a physical
1074 interface (e.g. eth0), then every change in MTU to eth0 will be reflected
1075 to br0. Any manual MTU configuration using `ip` or `ifconfig` on internal
1076 interfaces is going to be overridden by Open vSwitch to match the current
1079 Sometimes this behavior is not desirable, for example with tunnels. The
1080 MTU of an internal interface can be explicitly set using the following
1083 $ ovs-vsctl set int br0 mtu_request=1450
1085 After this, Open vSwitch will configure br0 MTU to 1450. Since this
1086 setting is in the database it will be persistent (compared to what happens
1087 with `ip` or `ifconfig`).
1089 The MTU configuration can be removed to restore the default behavior
1092 $ ovs-vsctl set int br0 mtu_request=[]
1094 The mtu_request column can be used to configure MTU even for physical
1095 interfaces (e.g. eth0).
1100 Q: Does OVS support Quality of Service (QoS)?
1102 A: Yes. For traffic that egresses from a switch, OVS supports traffic
1103 shaping; for traffic that ingresses into a switch, OVS support policing.
1104 Policing is a simple form of quality-of-service that simply drops packets
1105 received in excess of the configured rate. Due to its simplicity, policing
1106 is usually less accurate and less effective than egress traffic shaping,
1107 which queues packets.
1109 Keep in mind that ingress and egress are from the perspective of the
1110 switch. That means that egress shaping limits the rate at which traffic is
1111 allowed to transmit from a physical interface, but not the rate at which
1112 traffic will be received on a virtual machine's VIF. For ingress policing,
1113 the behavior is the opposite.
1115 Q: How do I configure egress traffic shaping?
1117 A: Suppose that you want to set up bridge br0 connected to physical
1118 Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces vif1.0
1119 and vif2.0, and that you want to limit traffic from vif1.0 to eth0 to 10
1120 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you could configure the
1125 add-port br0 eth0 -- \
1126 add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \
1127 add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \
1128 set port eth0 qos=@newqos -- \
1129 --id=@newqos create qos type=linux-htb \
1130 other-config:max-rate=1000000000 \
1131 queues:123=@vif10queue \
1132 queues:234=@vif20queue -- \
1133 --id=@vif10queue create queue other-config:max-rate=10000000 -- \
1134 --id=@vif20queue create queue other-config:max-rate=20000000
1136 At this point, bridge br0 is configured with the ports and eth0 is
1137 configured with the queues that you need for QoS, but nothing is actually
1138 directing packets from vif1.0 or vif2.0 to the queues that we have set up
1139 for them. That means that all of the packets to eth0 are going to the
1140 "default queue", which is not what we want.
1142 We use OpenFlow to direct packets from vif1.0 and vif2.0 to the queues
1145 $ ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal
1146 $ ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal
1148 Each of the above flows matches on the input port, sets up the appropriate
1149 queue (123 for vif1.0, 234 for vif2.0), and then executes the "normal"
1150 action, which performs the same switching that Open vSwitch would have done
1151 without any OpenFlow flows being present. (We know that vif1.0 and vif2.0
1152 have OpenFlow port numbers 5 and 6, respectively, because we set their
1153 ofport_request columns above. If we had not done that, then we would have
1154 needed to find out their port numbers before setting up these flows.)
1156 Now traffic going from vif1.0 or vif2.0 to eth0 should be rate-limited.
1158 By the way, if you delete the bridge created by the above commands, with::
1160 $ ovs-vsctl del-br br0
1162 then that will leave one unreferenced QoS record and two unreferenced Queue
1163 records in the Open vSwich database. One way to clear them out, assuming
1164 you don't have other QoS or Queue records that you want to keep, is::
1166 $ ovs-vsctl -- --all destroy QoS -- --all destroy Queue
1168 If you do want to keep some QoS or Queue records, or the Open vSwitch you
1169 are using is older than version 1.8 (which added the ``--all`` option),
1170 then you will have to destroy QoS and Queue records individually.
1172 Q: How do I configure ingress policing?
1174 A: A policing policy can be configured on an interface to drop packets that
1175 arrive at a higher rate than the configured value. For example, the
1176 following commands will rate-limit traffic that vif1.0 may generate to
1179 $ ovs-vsctl set interface vif1.0 ingress_policing_rate=10000
1180 $ ovs-vsctl set interface vif1.0 ingress_policing_burst=8000
1182 Traffic policing can interact poorly with some network protocols and can
1183 have surprising results. The "Ingress Policing" section of
1184 ovs-vswitchd.conf.db(5) discusses the issues in greater detail.
1186 Q: I configured Quality of Service (QoS) in my OpenFlow network by adding
1187 records to the QoS and Queue table, but the results aren't what I expect.
1189 A: Did you install OpenFlow flows that use your queues? This is the
1190 primary way to tell Open vSwitch which queues you want to use. If you
1191 don't do this, then the default queue will be used, which will probably not
1192 have the effect you want.
1194 Refer to the previous question for an example.
1196 Q: I'd like to take advantage of some QoS feature that Open vSwitch doesn't yet
1197 support. How do I do that?
1199 A: Open vSwitch does not implement QoS itself. Instead, it can configure
1200 some, but not all, of the QoS features built into the Linux kernel. If you
1201 need some QoS feature that OVS cannot configure itself, then the first step
1202 is to figure out whether Linux QoS supports that feature. If it does, then
1203 you can submit a patch to support Open vSwitch configuration for that
1204 feature, or you can use "tc" directly to configure the feature in Linux.
1205 (If Linux QoS doesn't support the feature you want, then first you have to
1206 add that support to Linux.)
1208 Q: I configured QoS, correctly, but my measurements show that it isn't working
1209 as well as I expect.
1211 A: With the Linux kernel, the Open vSwitch implementation of QoS has two
1214 - Open vSwitch configures a subset of Linux kernel QoS features, according
1215 to what is in OVSDB. It is possible that this code has bugs. If you
1216 believe that this is so, then you can configure the Linux traffic control
1217 (QoS) stack directly with the "tc" program. If you get better results
1218 that way, you can send a detailed bug report to bugs@openvswitch.org.
1220 It is certain that Open vSwitch cannot configure every Linux kernel QoS
1221 feature. If you need some feature that OVS cannot configure, then you
1222 can also use "tc" directly (or add that feature to OVS).
1224 - The Open vSwitch implementation of OpenFlow allows flows to be directed
1225 to particular queues. This is pretty simple and unlikely to have serious
1228 However, most problems with QoS on Linux are not bugs in Open vSwitch at
1229 all. They tend to be either configuration errors (please see the earlier
1230 questions in this section) or issues with the traffic control (QoS) stack
1231 in Linux. The Open vSwitch developers are not experts on Linux traffic
1232 control. We suggest that, if you believe you are encountering a problem
1233 with Linux traffic control, that you consult the tc manpages (e.g. tc(8),
1234 tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or mailing
1235 lists (e.g. http://vger.kernel.org/vger-lists.html#netdev).
1237 Q: Does Open vSwitch support OpenFlow meters?
1239 A: Since version 2.0, Open vSwitch has OpenFlow protocol support for
1240 OpenFlow meters. There is no implementation of meters in the Open vSwitch
1241 software switch (neither the kernel-based nor userspace switches).
1248 A: At the simplest level, a VLAN (short for "virtual LAN") is a way to
1249 partition a single switch into multiple switches. Suppose, for example,
1250 that you have two groups of machines, group A and group B. You want the
1251 machines in group A to be able to talk to each other, and you want the
1252 machine in group B to be able to talk to each other, but you don't want the
1253 machines in group A to be able to talk to the machines in group B. You can
1254 do this with two switches, by plugging the machines in group A into one
1255 switch and the machines in group B into the other switch.
1257 If you only have one switch, then you can use VLANs to do the same thing,
1258 by configuring the ports for machines in group A as VLAN "access ports" for
1259 one VLAN and the ports for group B as "access ports" for a different VLAN.
1260 The switch will only forward packets between ports that are assigned to the
1261 same VLAN, so this effectively subdivides your single switch into two
1262 independent switches, one for each group of machines.
1264 So far we haven't said anything about VLAN headers. With access ports,
1265 like we've described so far, no VLAN header is present in the Ethernet
1266 frame. This means that the machines (or switches) connected to access
1267 ports need not be aware that VLANs are involved, just like in the case
1268 where we use two different physical switches.
1270 Now suppose that you have a whole bunch of switches in your network,
1271 instead of just one, and that some machines in group A are connected
1272 directly to both switches 1 and 2. To allow these machines to talk to each
1273 other, you could add an access port for group A's VLAN to switch 1 and
1274 another to switch 2, and then connect an Ethernet cable between those
1275 ports. That works fine, but it doesn't scale well as the number of
1276 switches and the number of VLANs increases, because you use up a lot of
1277 valuable switch ports just connecting together your VLANs.
1279 This is where VLAN headers come in. Instead of using one cable and two
1280 ports per VLAN to connect a pair of switches, we configure a port on each
1281 switch as a VLAN "trunk port". Packets sent and received on a trunk port
1282 carry a VLAN header that says what VLAN the packet belongs to, so that only
1283 two ports total are required to connect the switches, regardless of the
1284 number of VLANs in use. Normally, only switches (either physical or
1285 virtual) are connected to a trunk port, not individual hosts, because
1286 individual hosts don't expect to see a VLAN header in the traffic that they
1289 None of the above discussion says anything about particular VLAN numbers.
1290 This is because VLAN numbers are completely arbitrary. One must only
1291 ensure that a given VLAN is numbered consistently throughout a network and
1292 that different VLANs are given different numbers. (That said, VLAN 0 is
1293 usually synonymous with a packet that has no VLAN header, and VLAN 4095 is
1296 Q: VLANs don't work.
1298 A: Many drivers in Linux kernels before version 3.3 had VLAN-related bugs.
1299 If you are having problems with VLANs that you suspect to be driver
1300 related, then you have several options:
1302 - Upgrade to Linux 3.3 or later.
1304 - Build and install a fixed version of the particular driver that is
1305 causing trouble, if one is available.
1307 - Use a NIC whose driver does not have VLAN problems.
1309 - Use "VLAN splinters", a feature in Open vSwitch 1.4 upto 2.5 that works
1310 around bugs in kernel drivers. To enable VLAN splinters on interface
1311 eth0, use the command::
1313 $ ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true
1315 For VLAN splinters to be effective, Open vSwitch must know which VLANs
1316 are in use. See the "VLAN splinters" section in the Interface table in
1317 ovs-vswitchd.conf.db(5) for details on how Open vSwitch infers in-use
1320 VLAN splinters increase memory use and reduce performance, so use them
1323 - Apply the "vlan workaround" patch from the XenServer kernel patch queue,
1324 build Open vSwitch against this patched kernel, and then use
1325 ovs-vlan-bug-workaround(8) to enable the VLAN workaround for each
1326 interface whose driver is buggy.
1328 (This is a nontrivial exercise, so this option is included only for
1331 It is not always easy to tell whether a Linux kernel driver has buggy VLAN
1332 support. The ovs-vlan-test(8) and ovs-test(8) utilities can help you test.
1333 See their manpages for details. Of the two utilities, ovs-test(8) is newer
1334 and more thorough, but ovs-vlan-test(8) may be easier to use.
1336 Q: VLANs still don't work. I've tested the driver so I know that it's OK.
1338 A: Do you have VLANs enabled on the physical switch that OVS is attached
1339 to? Make sure that the port is configured to trunk the VLAN or VLANs that
1340 you are using with OVS.
1342 Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch
1343 and to its destination host, but OVS seems to drop incoming return
1346 A: It's possible that you have the VLAN configured on your physical switch
1347 as the "native" VLAN. In this mode, the switch treats incoming packets
1348 either tagged with the native VLAN or untagged as part of the native VLAN.
1349 It may also send outgoing packets in the native VLAN without a VLAN tag.
1351 If this is the case, you have two choices:
1353 - Change the physical switch port configuration to tag packets it forwards
1354 to OVS with the native VLAN instead of forwarding them untagged.
1356 - Change the OVS configuration for the physical port to a native VLAN mode.
1357 For example, the following sets up a bridge with port eth0 in
1358 "native-tagged" mode in VLAN 9::
1360 $ ovs-vsctl add-br br0 $ ovs-vsctl add-port br0 eth0 tag=9
1361 vlan_mode=native-tagged
1363 In this situation, "native-untagged" mode will probably work equally
1364 well. Refer to the documentation for the Port table in
1365 ovs-vswitchd.conf.db(5) for more information.
1367 Q: I added a pair of VMs on different VLANs, like this::
1369 $ ovs-vsctl add-br br0
1370 $ ovs-vsctl add-port br0 eth0
1371 $ ovs-vsctl add-port br0 tap0 tag=9
1372 $ ovs-vsctl add-port br0 tap1 tag=10
1374 but the VMs can't access each other, the external network, or the Internet.
1376 A: It is to be expected that the VMs can't access each other. VLANs are a
1377 means to partition a network. When you configured tap0 and tap1 as access
1378 ports for different VLANs, you indicated that they should be isolated from
1381 As for the external network and the Internet, it seems likely that the
1382 machines you are trying to access are not on VLAN 9 (or 10) and that the
1383 Internet is not available on VLAN 9 (or 10).
1385 Q: I added a pair of VMs on the same VLAN, like this::
1387 $ ovs-vsctl add-br br0
1388 $ ovs-vsctl add-port br0 eth0
1389 $ ovs-vsctl add-port br0 tap0 tag=9
1390 $ ovs-vsctl add-port br0 tap1 tag=9
1392 The VMs can access each other, but not the external network or the Internet.
1394 A: It seems likely that the machines you are trying to access in the
1395 external network are not on VLAN 9 and that the Internet is not available
1396 on VLAN 9. Also, ensure VLAN 9 is set up as an allowed trunk VLAN on the
1397 upstream switch port to which eth0 is connected.
1399 Q: Can I configure an IP address on a VLAN?
1401 A: Yes. Use an "internal port" configured as an access port. For example,
1402 the following configures IP address 192.168.0.7 on VLAN 9. That is, OVS
1403 will forward packets from eth0 to 192.168.0.7 only if they have an 802.1Q
1404 header with VLAN 9. Conversely, traffic forwarded from 192.168.0.7 to eth0
1405 will be tagged with an 802.1Q header with VLAN 9::
1407 $ ovs-vsctl add-br br0
1408 $ ovs-vsctl add-port br0 eth0
1409 $ ovs-vsctl add-port br0 vlan9 tag=9 \
1410 -- set interface vlan9 type=internal
1411 $ ifconfig vlan9 192.168.0.7
1413 See also the following question.
1415 Q: I configured one IP address on VLAN 0 and another on VLAN 9, like this::
1417 $ ovs-vsctl add-br br0
1418 $ ovs-vsctl add-port br0 eth0
1419 $ ifconfig br0 192.168.0.5
1420 $ ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal
1421 $ ifconfig vlan9 192.168.0.9
1423 but other hosts that are only on VLAN 0 can reach the IP address configured on
1424 VLAN 9. What's going on?
1426 A: `RFC 1122 section 3.3.4.2 "Multihoming Requirements"
1427 <https://tools.ietf.org/html/rfc1122#section-3.3.4.2>`__ describes two
1428 approaches to IP address handling in Internet hosts:
1430 - In the "Strong ES Model", where an ES is a host ("End System"), an IP
1431 address is primarily associated with a particular interface. The host
1432 discards packets that arrive on interface A if they are destined for an
1433 IP address that is configured on interface B. The host never sends
1434 packets from interface A using a source address configured on interface
1437 - In the "Weak ES Model", an IP address is primarily associated with a
1438 host. The host accepts packets that arrive on any interface if they are
1439 destined for any of the host's IP addresses, even if the address is
1440 configured on some interface other than the one on which it arrived. The
1441 host does not restrict itself to sending packets from an IP address
1442 associated with the originating interface.
1444 Linux uses the weak ES model. That means that when packets destined to the
1445 VLAN 9 IP address arrive on eth0 and are bridged to br0, the kernel IP
1446 stack accepts them there for the VLAN 9 IP address, even though they were
1447 not received on vlan9, the network device for vlan9.
1449 To simulate the strong ES model on Linux, one may add iptables rule to
1450 filter packets based on source and destination address and adjust ARP
1451 configuration with sysctls.
1453 BSD uses the strong ES model.
1455 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1457 A: The configuration for VLANs in the Open vSwitch database (e.g. via
1458 ovs-vsctl) only affects traffic that goes through Open vSwitch's
1459 implementation of the OpenFlow "normal switching" action. By default, when
1460 Open vSwitch isn't connected to a controller and nothing has been manually
1461 configured in the flow table, all traffic goes through the "normal
1462 switching" action. But, if you set up OpenFlow flows on your own, through
1463 a controller or using ovs-ofctl or through other means, then you have to
1464 implement VLAN handling yourself.
1466 You can use "normal switching" as a component of your OpenFlow actions,
1467 e.g. by putting "normal" into the lists of actions on ovs-ofctl or by
1468 outputting to OFPP_NORMAL from an OpenFlow controller. In situations where
1469 this is not suitable, you can implement VLAN handling yourself, e.g.:
1471 - If a packet comes in on an access port, and the flow table needs to send
1472 it out on a trunk port, then the flow can add the appropriate VLAN tag
1473 with the "mod_vlan_vid" action.
1475 - If a packet comes in on a trunk port, and the flow table needs to send it
1476 out on an access port, then the flow can strip the VLAN tag with the
1477 "strip_vlan" action.
1479 Q: I configured ports on a bridge as access ports with different VLAN tags,
1482 $ ovs-vsctl add-br br0
1483 $ ovs-vsctl set-controller br0 tcp:192.168.0.10:6653
1484 $ ovs-vsctl add-port br0 eth0
1485 $ ovs-vsctl add-port br0 tap0 tag=9
1486 $ ovs-vsctl add-port br0 tap1 tag=10
1488 but the VMs running behind tap0 and tap1 can still communicate, that is, they
1489 are not isolated from each other even though they are on different VLANs.
1491 A: Do you have a controller configured on br0 (as the commands above do)?
1492 If so, then this is a variant on the previous question, "My OpenFlow
1493 controller doesn't see the VLANs that I expect," and you can refer to the
1494 answer there for more information.
1496 Q: How MAC learning works with VLANs?
1498 A: Open vSwitch implements Independent VLAN Learning (IVL) for
1499 ``OFPP_NORMAL`` action, e.g. it logically has separate learning tables for
1507 A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means
1508 to solve the scaling challenges of VLAN networks in a multi-tenant
1509 environment. VXLAN is an overlay network which transports an L2 network
1510 over an existing L3 network. For more information on VXLAN, please see `RFC
1511 7348 <http://tools.ietf.org/html/rfc7348>`__.
1513 Q: How much of the VXLAN protocol does Open vSwitch currently support?
1515 A: Open vSwitch currently supports the framing format for packets on the
1516 wire. There is currently no support for the multicast aspects of VXLAN. To
1517 get around the lack of multicast support, it is possible to pre-provision
1518 MAC to IP address mappings either manually or from a controller.
1520 Q: What destination UDP port does the VXLAN implementation in Open vSwitch
1523 A: By default, Open vSwitch will use the assigned IANA port for VXLAN,
1524 which is 4789. However, it is possible to configure the destination UDP
1525 port manually on a per-VXLAN tunnel basis. An example of this configuration
1526 is provided below.::
1528 $ ovs-vsctl add-br br0
1529 $ ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1 type=vxlan \
1530 options:remote_ip=192.168.1.2 options:key=flow options:dst_port=8472
1535 Q: What versions of OpenFlow does Open vSwitch support?
1537 A: The following table lists the versions of OpenFlow supported by each
1538 version of Open vSwitch:
1540 =============== ===== ===== ===== ===== ===== ===== =====
1541 Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5 OF1.6
1542 =============== ===== ===== ===== ===== ===== ===== =====
1543 1.9 and earlier yes --- --- --- --- --- ---
1544 1.10 yes --- (*) (*) --- --- ---
1545 1.11 yes --- (*) (*) --- --- ---
1546 2.0 yes (*) (*) (*) --- --- ---
1547 2.1 yes (*) (*) (*) --- --- ---
1548 2.2 yes (*) (*) (*) (%) (*) ---
1549 2.3 yes yes yes yes (*) (*) ---
1550 2.4 yes yes yes yes (*) (*) ---
1551 2.5 yes yes yes yes (*) (*) (*)
1552 =============== ===== ===== ===== ===== ===== ===== =====
1554 (*) Supported, with one or more missing features.
1555 (%) Experimental, unsafe implementation.
1557 Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default in
1558 ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1, 1.2, and 1.3
1559 must be enabled manually in ovs-vswitchd.
1561 Some versions of OpenFlow are supported with missing features and therefore
1562 not enabled by default: OpenFlow 1.4 and 1.5, in Open vSwitch 2.3 and
1563 later, as well as OpenFlow 1.6 in Open vSwitch 2.5 and later. Also, the
1564 OpenFlow 1.6 specification is still under development and thus subject to
1567 In any case, the user may override the default:
1569 - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0::
1571 $ ovs-vsctl set bridge br0 \
1572 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13
1574 - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0::
1576 $ ovs-vsctl set bridge br0 \
1577 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15
1579 - To enable only OpenFlow 1.0 on bridge br0::
1581 $ ovs-vsctl set bridge br0 protocols=OpenFlow10
1583 All current versions of ovs-ofctl enable only OpenFlow 1.0 by default. Use
1584 the -O option to enable support for later versions of OpenFlow in
1585 ovs-ofctl. For example::
1587 $ ovs-ofctl -O OpenFlow13 dump-flows br0
1589 (Open vSwitch 2.2 had an experimental implementation of OpenFlow 1.4 that
1590 could cause crashes. We don't recommend enabling it.)
1592 The `OpenFlow guide <Documentation/development-guide/openflow.rst>`__
1593 tracks support for OpenFlow 1.1 and later features. When support for
1594 OpenFlow 1.4 and 1.5 is solidly implemented, Open vSwitch will enable those
1597 Q: Does Open vSwitch support MPLS?
1599 A: Before version 1.11, Open vSwitch did not support MPLS. That is, these
1600 versions can match on MPLS Ethernet types, but they cannot match, push, or
1601 pop MPLS labels, nor can they look past MPLS labels into the encapsulated
1604 Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support for
1605 MPLS. With the userspace datapath only, these versions can match, push, or
1606 pop a single MPLS label, but they still cannot look past MPLS labels (even
1607 after popping them) into the encapsulated packet. Kernel datapath support
1608 is unchanged from earlier versions.
1610 Open vSwitch version 2.3 can match, push, or pop a single MPLS label and
1611 look past the MPLS label into the encapsulated packet. Both userspace and
1612 kernel datapaths will be supported, but MPLS processing always happens in
1613 userspace either way, so kernel datapath performance will be disappointing.
1615 Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS labels and
1616 look past the MPLS label into the encapsulated packet. It will have kernel
1617 support for MPLS, yielding improved performance.
1619 Q: I'm getting "error type 45250 code 0". What's that?
1621 A: This is a Open vSwitch extension to OpenFlow error codes. Open vSwitch
1622 uses this extension when it must report an error to an OpenFlow controller
1623 but no standard OpenFlow error code is suitable.
1625 Open vSwitch logs the errors that it sends to controllers, so the easiest
1626 thing to do is probably to look at the ovs-vswitchd log to find out what
1629 If you want to dissect the extended error message yourself, the format is
1630 documented in include/openflow/nicira-ext.h in the Open vSwitch source
1631 distribution. The extended error codes are documented in
1632 include/openvswitch/ofp-errors.h.
1634 Q: Some of the traffic that I'd expect my OpenFlow controller to see doesn't
1635 actually appear through the OpenFlow connection, even though I know that it's
1638 A: By default, Open vSwitch assumes that OpenFlow controllers are connected
1639 "in-band", that is, that the controllers are actually part of the network
1640 that is being controlled. In in-band mode, Open vSwitch sets up special
1641 "hidden" flows to make sure that traffic can make it back and forth between
1642 OVS and the controllers. These hidden flows are higher priority than any
1643 flows that can be set up through OpenFlow, and they are not visible through
1644 normal OpenFlow flow table dumps.
1646 Usually, the hidden flows are desirable and helpful, but occasionally they
1647 can cause unexpected behavior. You can view the full OpenFlow flow table,
1648 including hidden flows, on bridge br0 with the command::
1650 $ ovs-appctl bridge/dump-flows br0
1652 to help you debug. The hidden flows are those with priorities
1653 greater than 65535 (the maximum priority that can be set with
1656 The ``Documentation/topics/design`` doc describes the in-band model in
1659 If your controllers are not actually in-band (e.g. they are on
1660 localhost via 127.0.0.1, or on a separate network), then you should
1661 configure your controllers in "out-of-band" mode. If you have one
1662 controller on bridge br0, then you can configure out-of-band mode
1665 $ ovs-vsctl set controller br0 connection-mode=out-of-band
1667 Q: Some of the OpenFlow flows that my controller sets up don't seem to apply to
1668 certain traffic, especially traffic between OVS and the controller itself.
1672 Q: I configured all my controllers for out-of-band control mode but "ovs-appctl
1673 bridge/dump-flows" still shows some hidden flows.
1675 A: You probably have a remote manager configured (e.g. with "ovs-vsctl
1676 set-manager"). By default, Open vSwitch assumes that managers need in-band
1677 rules set up on every bridge. You can disable these rules on bridge br0
1680 $ ovs-vsctl set bridge br0 other-config:disable-in-band=true
1682 This actually disables in-band control entirely for the bridge, as if all
1683 the bridge's controllers were configured for out-of-band control.
1685 Q: My OpenFlow controller doesn't see the VLANs that I expect.
1687 A: See answer under "VLANs", above.
1689 Q: I ran ``ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop`` but I got a
1690 funny message like this::
1692 ofp_util|INFO|normalization changed ofp_match, details:
1693 ofp_util|INFO| pre: nw_dst=192.168.0.1
1696 and when I ran ``ovs-ofctl dump-flows br0`` I saw that my nw_dst match had
1697 disappeared, so that the flow ends up matching every packet.
1699 A: The term "normalization" in the log message means that a flow cannot
1700 match on an L3 field without saying what L3 protocol is in use. The
1701 "ovs-ofctl" command above didn't specify an L3 protocol, so the L3 field
1704 In this case, the L3 protocol could be IP or ARP. A correct command for
1705 each possibility is, respectively::
1707 $ ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop
1711 $ ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop
1713 Similarly, a flow cannot match on an L4 field without saying what L4
1714 protocol is in use. For example, the flow match ``tp_src=1234`` is, by
1715 itself, meaningless and will be ignored. Instead, to match TCP source port
1716 1234, write ``tcp,tp_src=1234``, or to match UDP source port 1234, write
1717 ``udp,tp_src=1234``.
1719 Q: How can I figure out the OpenFlow port number for a given port?
1721 A: The ``OFPT_FEATURES_REQUEST`` message requests an OpenFlow switch to
1722 respond with an ``OFPT_FEATURES_REPLY`` that, among other information,
1723 includes a mapping between OpenFlow port names and numbers. From a command
1724 prompt, ``ovs-ofctl show br0`` makes such a request and prints the response
1727 The Interface table in the Open vSwitch database also maps OpenFlow port
1728 names to numbers. To print the OpenFlow port number associated with
1729 interface eth0, run::
1731 $ ovs-vsctl get Interface eth0 ofport
1733 You can print the entire mapping with::
1735 $ ovs-vsctl -- --columns=name,ofport list Interface
1737 but the output mixes together interfaces from all bridges in the database,
1738 so it may be confusing if more than one bridge exists.
1740 In the Open vSwitch database, ofport value ``-1`` means that the interface
1741 could not be created due to an error. (The Open vSwitch log should
1742 indicate the reason.) ofport value ``[]`` (the empty set) means that the
1743 interface hasn't been created yet. The latter is normally an intermittent
1744 condition (unless ovs-vswitchd is not running).
1746 Q: I added some flows with my controller or with ovs-ofctl, but when I run
1747 "ovs-dpctl dump-flows" I don't see them.
1749 A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It won't
1750 display the information that you want. You want to use ``ovs-ofctl
1751 dump-flows`` instead.
1753 Q: It looks like each of the interfaces in my bonded port shows up as an
1754 individual OpenFlow port. Is that right?
1756 A: Yes, Open vSwitch makes individual bond interfaces visible as OpenFlow
1757 ports, rather than the bond as a whole. The interfaces are treated
1758 together as a bond for only a few purposes:
1760 - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow controller
1761 is not configured, this happens implicitly to every packet.)
1763 - Mirrors configured for output to a bonded port.
1765 It would make a lot of sense for Open vSwitch to present a bond as a single
1766 OpenFlow port. If you want to contribute an implementation of such a
1767 feature, please bring it up on the Open vSwitch development mailing list at
1768 dev@openvswitch.org.
1770 Q: I have a sophisticated network setup involving Open vSwitch, VMs or multiple
1771 hosts, and other components. The behavior isn't what I expect. Help!
1773 A: To debug network behavior problems, trace the path of a packet,
1774 hop-by-hop, from its origin in one host to a remote host. If that's
1775 correct, then trace the path of the response packet back to the origin.
1777 The open source tool called ``plotnetcfg`` can help to understand the
1778 relationship between the networking devices on a single host.
1780 Usually a simple ICMP echo request and reply (``ping``) packet is good
1781 enough. Start by initiating an ongoing ``ping`` from the origin host to a
1782 remote host. If you are tracking down a connectivity problem, the "ping"
1783 will not display any successful output, but packets are still being sent.
1784 (In this case the packets being sent are likely ARP rather than ICMP.)
1786 Tools available for tracing include the following:
1788 - ``tcpdump`` and ``wireshark`` for observing hops across network devices,
1789 such as Open vSwitch internal devices and physical wires.
1791 - ``ovs-appctl dpif/dump-flows <br>`` in Open vSwitch 1.10 and later or
1792 ``ovs-dpctl dump-flows <br>`` in earlier versions. These tools allow one
1793 to observe the actions being taken on packets in ongoing flows.
1795 See ovs-vswitchd(8) for ``ovs-appctl dpif/dump-flows`` documentation,
1796 ovs-dpctl(8) for ``ovs-dpctl dump-flows`` documentation, and "Why are
1797 there so many different ways to dump flows?" above for some background.
1799 - ``ovs-appctl ofproto/trace`` to observe the logic behind how ovs-vswitchd
1800 treats packets. See ovs-vswitchd(8) for documentation. You can out more
1801 details about a given flow that ``ovs-dpctl dump-flows`` displays, by
1802 cutting and pasting a flow from the output into an ``ovs-appctl
1803 ofproto/trace`` command.
1805 - SPAN, RSPAN, and ERSPAN features of physical switches, to observe what
1806 goes on at these physical hops.
1808 Starting at the origin of a given packet, observe the packet at each hop in
1809 turn. For example, in one plausible scenario, you might:
1811 1. ``tcpdump`` the ``eth`` interface through which an ARP egresses a VM,
1814 2. ``tcpdump`` the ``vif`` or ``tap`` interface through which the ARP
1815 ingresses the host machine.
1817 3. Use ``ovs-dpctl dump-flows`` to spot the ARP flow and observe the host
1818 interface through which the ARP egresses the physical machine. You may
1819 need to use ``ovs-dpctl show`` to interpret the port numbers. If the
1820 output seems surprising, you can use ``ovs-appctl ofproto/trace`` to
1821 observe details of how ovs-vswitchd determined the actions in the
1822 ``ovs-dpctl dump-flows`` output.
1824 4. ``tcpdump`` the ``eth`` interface through which the ARP egresses the
1827 5. ``tcpdump`` the ``eth`` interface through which the ARP ingresses the
1828 physical machine, at the remote host that receives the ARP.
1830 6. Use ``ovs-dpctl dump-flows`` to spot the ARP flow on the remote host
1831 remote host that receives the ARP and observe the VM ``vif`` or ``tap``
1832 interface to which the flow is directed. Again, ``ovs-dpctl show`` and
1833 ``ovs-appctl ofproto/trace`` might help.
1835 7. ``tcpdump`` the ``vif`` or ``tap`` interface to which the ARP is
1838 8. ``tcpdump`` the ``eth`` interface through which the ARP ingresses a VM,
1841 It is likely that during one of these steps you will figure out the
1842 problem. If not, then follow the ARP reply back to the origin, in reverse.
1844 Q: How do I make a flow drop packets?
1846 A: To drop a packet is to receive it without forwarding it. OpenFlow
1847 explicitly specifies forwarding actions. Thus, a flow with an empty set of
1848 actions does not forward packets anywhere, causing them to be dropped. You
1849 can specify an empty set of actions with ``actions=`` on the ovs-ofctl
1850 command line. For example::
1852 $ ovs-ofctl add-flow br0 priority=65535,actions=
1854 would cause every packet entering switch br0 to be dropped.
1856 You can write "drop" explicitly if you like. The effect is the same.
1857 Thus, the following command also causes every packet entering switch br0 to
1860 $ ovs-ofctl add-flow br0 priority=65535,actions=drop
1862 ``drop`` is not an action, either in OpenFlow or Open vSwitch. Rather, it
1863 is only a way to say that there are no actions.
1865 Q: I added a flow to send packets out the ingress port, like this::
1867 $ ovs-ofctl add-flow br0 in_port=2,actions=2
1869 but OVS drops the packets instead.
1871 A: Yes, OpenFlow requires a switch to ignore attempts to send a packet out
1872 its ingress port. The rationale is that dropping these packets makes it
1873 harder to loop the network. Sometimes this behavior can even be
1874 convenient, e.g. it is often the desired behavior in a flow that forwards a
1875 packet to several ports ("floods" the packet).
1877 Sometimes one really needs to send a packet out its ingress port
1878 ("hairpin"). In this case, output to ``OFPP_IN_PORT``, which in ovs-ofctl
1879 syntax is expressed as just ``in_port``, e.g.::
1881 $ ovs-ofctl add-flow br0 in_port=2,actions=in_port
1883 This also works in some circumstances where the flow doesn't match on the
1884 input port. For example, if you know that your switch has five ports
1885 numbered 2 through 6, then the following will send every received packet
1886 out every port, even its ingress port::
1888 $ ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port
1892 $ ovs-ofctl add-flow br0 actions=all,in_port
1894 Sometimes, in complicated flow tables with multiple levels of ``resubmit``
1895 actions, a flow needs to output to a particular port that may or may not be
1896 the ingress port. It's difficult to take advantage of ``OFPP_IN_PORT`` in
1897 this situation. To help, Open vSwitch provides, as an OpenFlow extension,
1898 the ability to modify the in_port field. Whatever value is currently in
1899 the in_port field is the port to which outputs will be dropped, as well as
1900 the destination for ``OFPP_IN_PORT``. This means that the following will
1901 reliably output to port 2 or to ports 2 through 6, respectively::
1903 $ ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2
1904 $ ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6
1906 If the input port is important, then one may save and restore it on the
1909 $ ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\
1910 load:0->NXM_OF_IN_PORT[],\
1912 pop:NXM_OF_IN_PORT[]
1914 Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2 on port 2.
1915 I set up flows to forward only traffic destined to the other host and drop
1916 other traffic, like this::
1918 priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2
1919 priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1
1920 priority=0,actions=drop
1922 But it doesn't work--I don't get any connectivity when I do this. Why?
1924 A: These flows drop the ARP packets that IP hosts use to establish IP
1925 connectivity over Ethernet. To solve the problem, add flows to allow ARP
1926 to pass between the hosts::
1928 priority=5,in_port=1,arp,actions=2
1929 priority=5,in_port=2,arp,actions=1
1931 This issue can manifest other ways, too. The following flows that match on
1932 Ethernet addresses instead of IP addresses will also drop ARP packets,
1933 because ARP requests are broadcast instead of being directed to a specific
1936 priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2
1937 priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1
1938 priority=0,actions=drop
1940 The solution already described above will also work in this case. It may
1941 be better to add flows to allow all multicast and broadcast traffic::
1943 priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2
1944 priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1
1946 Q: My bridge disconnects from my controller on add-port/del-port.
1948 A: Reconfiguring your bridge can change your bridge's datapath-id because
1949 Open vSwitch generates datapath-id from the MAC address of one of its
1950 ports. In that case, Open vSwitch disconnects from controllers because
1951 there's no graceful way to notify controllers about the change of
1954 To avoid the behaviour, you can configure datapath-id manually.::
1956 $ ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef
1958 Q: My controller complains that OVS is not buffering packets.
1961 A: "Packet buffering" is an optional OpenFlow feature, and controllers
1962 should detect how many "buffers" an OpenFlow switch implements. It was
1963 recently noticed that OVS implementation of the buffering feature was not
1964 compliant to OpenFlow specifications. Rather than fix it and risk
1965 controller incompatibility, the buffering feature is removed as of OVS 2.7.
1966 Controllers are already expected to work properly in cases where the switch
1967 can not buffer packets, but sends full packets in "packet-in" messages
1968 instead, so this change should not affect existing users. After the change
1969 OVS always sends the ``buffer_id`` as ``0xffffffff`` in "packet-in"
1970 messages and will send an error response if any other value of this field
1971 is included in a "packet-out" or a "flow mod" sent by a controller.
1973 Q: How does OVS divide flows among buckets in an OpenFlow "select" group?
1975 A: In Open vSwitch 2.3 and earlier, Open vSwitch used the destination
1976 Ethernet address to choose a bucket in a select group.
1978 Open vSwitch 2.4 and later by default hashes the source and destination
1979 Ethernet address, VLAN ID, Ethernet type, IPv4/v6 source and destination
1980 address and protocol, and for TCP and SCTP only, the source and destination
1981 ports. The hash is "symmetric", meaning that exchanging source and
1982 destination addresses does not change the bucket selection.
1984 Select groups in Open vSwitch 2.4 and later can be configured to use a
1985 different hash function, using a Netronome extension to the OpenFlow 1.5+
1986 group_mod message. For more information, see
1987 Documentation/group-selection-method-property.txt in the Open vSwitch
1988 source tree. (OpenFlow 1.5 support in Open vSwitch is still experimental.)
1990 Q: I added a flow to accept packets on VLAN 123 and output them on VLAN 456,
1993 $ ovs-ofctl add-flow br0 dl_vlan=123,actions=output:1,mod_vlan_vid:456
1995 but the packets are actually being output in VLAN 123. Why?
1997 A: OpenFlow actions are executed in the order specified. Thus, the actions
1998 above first output the packet, then change its VLAN. Since the output
1999 occurs before changing the VLAN, the change in VLAN will have no visible
2002 To solve this and similar problems, order actions so that changes to
2003 headers happen before output, e.g.::
2005 $ ovs-ofctl add-flow br0 dl_vlan=123,actions=mod_vlan_vid:456,output:1
2007 Q: The "learn" action can't learn the action I want, can you improve it?
2009 A: By itself, the "learn" action can only put two kinds of actions into the
2010 flows that it creates: "load" and "output" actions. If "learn" is used in
2011 isolation, these are severe limits.
2013 However, "learn" is not meant to be used in isolation. It is a primitive
2014 meant to be used together with other Open vSwitch features to accomplish a
2015 task. Its existing features are enough to accomplish most tasks.
2017 Here is an outline of a typical pipeline structure that allows for
2018 versatile behavior using "learn":
2020 - Flows in table A contain a "learn" action, that populates flows in table
2021 L, that use a "load" action to populate register R with information about
2024 - Flows in table B contain two sequential resubmit actions: one to table L
2025 and another one to table B+1.
2027 - Flows in table B+1 match on register R and act differently depending on
2028 what the flows in table L loaded into it.
2030 This approach can be used to implement many "learn"-based features. For
2033 - Resubmit to a table selected based on learned information, e.g. see:
2034 http://openvswitch.org/pipermail/discuss/2016-June/021694.html
2036 - MAC learning in the middle of a pipeline, as described in `the tutorial
2037 <Documentation/tutorials/ovs-advanced.rst>`__.
2039 - TCP state based firewalling, by learning outgoing connections based on
2040 SYN packets and matching them up with incoming packets.
2042 - At least some of the features described in T. A. Hoff, "Extending Open
2043 vSwitch to Facilitate Creation of Stateful SDN Applications".
2048 Q: How do I implement a new OpenFlow message?
2050 A: Add your new message to ``enum ofpraw`` and ``enum ofptype`` in
2051 ``lib/ofp-msgs.h``, following the existing pattern. Then recompile and fix
2052 all of the new warnings, implementing new functionality for the new message
2053 as needed. (If you configure with ``--enable-Werror``, as described in the
2054 `general installation guide`_, then it is impossible to miss any warnings.)
2056 If you need to add an OpenFlow vendor extension message for a vendor that
2057 doesn't yet have any extension messages, then you will also need to edit
2058 ``build-aux/extract-ofp-msgs``.
2060 Q: How do I add support for a new field or header?
2062 A: Add new members for your field to ``struct flow`` in ``lib/flow.h``, and
2063 add new enumerations for your new field to ``enum mf_field_id`` in
2064 ``lib/meta-flow.h``, following the existing pattern. Also, add support to
2065 ``miniflow_extract()`` in ``lib/flow.c`` for extracting your new field from
2066 a packet into struct miniflow, and to ``nx_put_raw()`` in
2067 ``lib/nx-match.c`` to output your new field in OXM matches. Then recompile
2068 and fix all of the new warnings, implementing new functionality for the new
2069 field or header as needed. (If you configure with ``--enable-Werror``, as
2070 described in the `general installation guide`_, then it is impossible to
2073 If you want kernel datapath support for your new field, you also need to
2074 modify the kernel module for the operating systems you are interested in.
2075 This isn't mandatory, since fields understood only by userspace work too
2076 (with a performance penalty), so it's reasonable to start development
2077 without it. If you implement kernel module support for Linux, then the
2078 Linux kernel "netdev" mailing list is the place to submit that support
2079 first; please read up on the Linux kernel development process separately.
2080 The Windows datapath kernel module support, on the other hand, is
2081 maintained within the OVS tree, so patches for that can go directly to
2084 Q: How do I add support for a new OpenFlow action?
2086 A: Add your new action to ``enum ofp_raw_action_type`` in
2087 ``lib/ofp-actions.c``, following the existing pattern. Then recompile and
2088 fix all of the new warnings, implementing new functionality for the new
2089 action as needed. (If you configure with ``--enable-Werror``, as described
2090 in the `general installation guide`_, then it is impossible to miss any
2093 If you need to add an OpenFlow vendor extension action for a vendor that
2094 doesn't yet have any extension actions, then you will also need to edit
2095 ``build-aux/extract-ofp-actions``.
2097 .. _general installation guide: Documentation/intro/install/general.rst
2098 .. _DPDK installation guide: Documentation/intro/install/dpdk.rst
projects / mirror_ovs.git / blob