]>
Commit | Line | Data |
---|---|---|
c483d489 BP |
1 | Frequently Asked Questions |
2 | ========================== | |
3 | ||
542cc9bb TG |
4 | Open vSwitch <http://openvswitch.org> |
5 | ||
3fc7dc18 JP |
6 | General |
7 | ------- | |
8 | ||
542cc9bb | 9 | ### Q: What is Open vSwitch? |
3fc7dc18 JP |
10 | |
11 | A: Open vSwitch is a production quality open source software switch | |
29089a54 RL |
12 | designed to be used as a vswitch in virtualized server |
13 | environments. A vswitch forwards traffic between different VMs on | |
14 | the same physical host and also forwards traffic between VMs and | |
15 | the physical network. Open vSwitch supports standard management | |
16 | interfaces (e.g. sFlow, NetFlow, IPFIX, RSPAN, CLI), and is open to | |
17 | programmatic extension and control using OpenFlow and the OVSDB | |
18 | management protocol. | |
3fc7dc18 JP |
19 | |
20 | Open vSwitch as designed to be compatible with modern switching | |
21 | chipsets. This means that it can be ported to existing high-fanout | |
22 | switches allowing the same flexible control of the physical | |
23 | infrastructure as the virtual infrastructure. It also means that | |
24 | Open vSwitch will be able to take advantage of on-NIC switching | |
25 | chipsets as their functionality matures. | |
26 | ||
542cc9bb | 27 | ### Q: What virtualization platforms can use Open vSwitch? |
3fc7dc18 JP |
28 | |
29 | A: Open vSwitch can currently run on any Linux-based virtualization | |
37418c86 | 30 | platform (kernel 2.6.32 and newer), including: KVM, VirtualBox, Xen, |
3fc7dc18 JP |
31 | Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the |
32 | mainline kernel. The bulk of the code is written in platform- | |
33 | independent C and is easily ported to other environments. We welcome | |
34 | inquires about integrating Open vSwitch with other virtualization | |
35 | platforms. | |
36 | ||
542cc9bb | 37 | ### Q: How can I try Open vSwitch? |
3fc7dc18 | 38 | |
7b287e99 JP |
39 | A: The Open vSwitch source code can be built on a Linux system. You can |
40 | build and experiment with Open vSwitch on any Linux machine. | |
41 | Packages for various Linux distributions are available on many | |
42 | platforms, including: Debian, Ubuntu, Fedora. | |
3fc7dc18 JP |
43 | |
44 | You may also download and run a virtualization platform that already | |
7b287e99 JP |
45 | has Open vSwitch integrated. For example, download a recent ISO for |
46 | XenServer or Xen Cloud Platform. Be aware that the version | |
47 | integrated with a particular platform may not be the most recent Open | |
48 | vSwitch release. | |
49 | ||
542cc9bb | 50 | ### Q: Does Open vSwitch only work on Linux? |
7b287e99 JP |
51 | |
52 | A: No, Open vSwitch has been ported to a number of different operating | |
53 | systems and hardware platforms. Most of the development work occurs | |
54 | on Linux, but the code should be portable to any POSIX system. We've | |
55 | seen Open vSwitch ported to a number of different platforms, | |
56 | including FreeBSD, Windows, and even non-POSIX embedded systems. | |
57 | ||
58 | By definition, the Open vSwitch Linux kernel module only works on | |
59 | Linux and will provide the highest performance. However, a userspace | |
60 | datapath is available that should be very portable. | |
61 | ||
542cc9bb | 62 | ### Q: What's involved with porting Open vSwitch to a new platform or switching ASIC? |
7b287e99 | 63 | |
9feb1017 | 64 | A: The [PORTING.md] document describes how one would go about |
542cc9bb | 65 | porting Open vSwitch to a new operating system or hardware platform. |
3fc7dc18 | 66 | |
542cc9bb | 67 | ### Q: Why would I use Open vSwitch instead of the Linux bridge? |
3fc7dc18 JP |
68 | |
69 | A: Open vSwitch is specially designed to make it easier to manage VM | |
7b287e99 | 70 | network configuration and monitor state spread across many physical |
542cc9bb | 71 | hosts in dynamic virtualized environments. Please see |
9feb1017 TG |
72 | [WHY-OVS.md] for a more detailed description of how Open vSwitch |
73 | relates to the Linux Bridge. | |
3fc7dc18 | 74 | |
542cc9bb | 75 | ### Q: How is Open vSwitch related to distributed virtual switches like the VMware vNetwork distributed switch or the Cisco Nexus 1000V? |
3fc7dc18 JP |
76 | |
77 | A: Distributed vswitch applications (e.g., VMware vNetwork distributed | |
78 | switch, Cisco Nexus 1000V) provide a centralized way to configure and | |
79 | monitor the network state of VMs that are spread across many physical | |
80 | hosts. Open vSwitch is not a distributed vswitch itself, rather it | |
81 | runs on each physical host and supports remote management in a way | |
82 | that makes it easier for developers of virtualization/cloud | |
83 | management platforms to offer distributed vswitch capabilities. | |
84 | ||
85 | To aid in distribution, Open vSwitch provides two open protocols that | |
86 | are specially designed for remote management in virtualized network | |
87 | environments: OpenFlow, which exposes flow-based forwarding state, | |
88 | and the OVSDB management protocol, which exposes switch port state. | |
89 | In addition to the switch implementation itself, Open vSwitch | |
1d5aaa61 BP |
90 | includes tools (ovs-ofctl, ovs-vsctl) that developers can script and |
91 | extend to provide distributed vswitch capabilities that are closely | |
92 | integrated with their virtualization management platform. | |
3fc7dc18 | 93 | |
542cc9bb | 94 | ### Q: Why doesn't Open vSwitch support distribution? |
3fc7dc18 JP |
95 | |
96 | A: Open vSwitch is intended to be a useful component for building | |
97 | flexible network infrastructure. There are many different approaches | |
98 | to distribution which balance trade-offs between simplicity, | |
99 | scalability, hardware compatibility, convergence times, logical | |
100 | forwarding model, etc. The goal of Open vSwitch is to be able to | |
101 | support all as a primitive building block rather than choose a | |
102 | particular point in the distributed design space. | |
103 | ||
542cc9bb | 104 | ### Q: How can I contribute to the Open vSwitch Community? |
3fc7dc18 JP |
105 | |
106 | A: You can start by joining the mailing lists and helping to answer | |
7b287e99 JP |
107 | questions. You can also suggest improvements to documentation. If |
108 | you have a feature or bug you would like to work on, send a mail to | |
109 | one of the mailing lists: | |
110 | ||
542cc9bb | 111 | http://openvswitch.org/mlists/ |
7b287e99 JP |
112 | |
113 | ||
7b287e99 JP |
114 | Releases |
115 | -------- | |
116 | ||
542cc9bb | 117 | ### Q: What does it mean for an Open vSwitch release to be LTS (long-term support)? |
3fc7dc18 | 118 | |
7b287e99 JP |
119 | A: All official releases have been through a comprehensive testing |
120 | process and are suitable for production use. Planned releases will | |
121 | occur several times a year. If a significant bug is identified in an | |
122 | LTS release, we will provide an updated release that includes the | |
123 | fix. Releases that are not LTS may not be fixed and may just be | |
124 | supplanted by the next major release. The current LTS release is | |
49b507d9 | 125 | 2.3.x. |
7b287e99 | 126 | |
542cc9bb | 127 | ### Q: What Linux kernel versions does each Open vSwitch release work with? |
314e60e1 BP |
128 | |
129 | A: The following table lists the Linux kernel versions against which the | |
130 | given versions of the Open vSwitch kernel module will successfully | |
131 | build. The Linux kernel versions are upstream kernel versions, so | |
a01b5c51 BP |
132 | Linux kernels modified from the upstream sources may not build in |
133 | some cases even if they are based on a supported version. This is | |
134 | most notably true of Red Hat Enterprise Linux (RHEL) kernels, which | |
135 | are extensively modified from upstream. | |
314e60e1 | 136 | |
542cc9bb TG |
137 | | Open vSwitch | Linux kernel |
138 | |:------------:|:-------------: | |
139 | | 1.4.x | 2.6.18 to 3.2 | |
140 | | 1.5.x | 2.6.18 to 3.2 | |
141 | | 1.6.x | 2.6.18 to 3.2 | |
142 | | 1.7.x | 2.6.18 to 3.3 | |
143 | | 1.8.x | 2.6.18 to 3.4 | |
144 | | 1.9.x | 2.6.18 to 3.8 | |
145 | | 1.10.x | 2.6.18 to 3.8 | |
146 | | 1.11.x | 2.6.18 to 3.8 | |
147 | | 2.0.x | 2.6.32 to 3.10 | |
148 | | 2.1.x | 2.6.32 to 3.11 | |
149 | | 2.3.x | 2.6.32 to 3.14 | |
adfaaeac | 150 | | 2.4.x | 2.6.32 to 3.19 |
314e60e1 BP |
151 | |
152 | Open vSwitch userspace should also work with the Linux kernel module | |
153 | built into Linux 3.3 and later. | |
154 | ||
155 | Open vSwitch userspace is not sensitive to the Linux kernel version. | |
37418c86 | 156 | It should build against almost any kernel, certainly against 2.6.32 |
314e60e1 BP |
157 | and later. |
158 | ||
542cc9bb | 159 | ### Q: I get an error like this when I configure Open vSwitch: |
29089a54 | 160 | |
a7ae9380 BP |
161 | configure: error: Linux kernel in <dir> is version <x>, but |
162 | version newer than <y> is not supported (please refer to the | |
163 | FAQ for advice) | |
29089a54 | 164 | |
a7ae9380 | 165 | What should I do? |
33cec590 | 166 | |
a7ae9380 BP |
167 | A: If there is a newer version of Open vSwitch, consider building that |
168 | one, because it may support the kernel that you are building | |
169 | against. (To find out, consult the table in the previous answer.) | |
170 | ||
171 | Otherwise, use the Linux kernel module supplied with the kernel | |
172 | that you are using. All versions of Open vSwitch userspace are | |
173 | compatible with all versions of the Open vSwitch kernel module, so | |
174 | this will also work. See also the following question. | |
33cec590 | 175 | |
542cc9bb | 176 | ### Q: What features are not available in the Open vSwitch kernel datapath that ships as part of the upstream Linux kernel? |
7b287e99 | 177 | |
9c333bff JG |
178 | A: The kernel module in upstream Linux does not include support for |
179 | LISP. Work is in progress to add support for LISP to the upstream | |
180 | Linux version of the Open vSwitch kernel module. For now, if you | |
181 | need this feature, use the kernel module from the Open vSwitch | |
0a740f48 EJ |
182 | distribution instead of the upstream Linux kernel module. |
183 | ||
9c333bff JG |
184 | Certain features require kernel support to function or to have |
185 | reasonable performance. If the ovs-vswitchd log file indicates that | |
186 | a feature is not supported, consider upgrading to a newer upstream | |
187 | Linux release or using the kernel module paired with the userspace | |
188 | distribution. | |
6302c641 | 189 | |
542cc9bb | 190 | ### Q: Why do tunnels not work when using a kernel module other than the one packaged with Open vSwitch? |
6814c630 JG |
191 | |
192 | A: Support for tunnels was added to the upstream Linux kernel module | |
193 | after the rest of Open vSwitch. As a result, some kernels may contain | |
194 | support for Open vSwitch but not tunnels. The minimum kernel version | |
195 | that supports each tunnel protocol is: | |
196 | ||
542cc9bb TG |
197 | | Protocol | Linux Kernel |
198 | |:--------:|:-------------: | |
199 | | GRE | 3.11 | |
200 | | VXLAN | 3.12 | |
201 | | LISP | <not upstream> | |
6814c630 JG |
202 | |
203 | If you are using a version of the kernel that is older than the one | |
204 | listed above, it is still possible to use that tunnel protocol. However, | |
205 | you must compile and install the kernel module included with the Open | |
206 | vSwitch distribution rather than the one on your machine. If problems | |
207 | persist after doing this, check to make sure that the module that is | |
208 | loaded is the one you expect. | |
209 | ||
542cc9bb | 210 | ### Q: What features are not available when using the userspace datapath? |
7b287e99 | 211 | |
0a740f48 | 212 | A: Tunnel virtual ports are not supported, as described in the |
7b287e99 JP |
213 | previous answer. It is also not possible to use queue-related |
214 | actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets | |
215 | may not be transmitted. | |
3fc7dc18 | 216 | |
542cc9bb | 217 | ### Q: What Linux kernel versions does IPFIX flow monitoring work with? |
a7ae9380 BP |
218 | |
219 | A: IPFIX flow monitoring requires the Linux kernel module from Open | |
220 | vSwitch version 1.10.90 or later. | |
221 | ||
542cc9bb | 222 | ### Q: Should userspace or kernel be upgraded first to minimize downtime? |
a7ae9380 BP |
223 | |
224 | In general, the Open vSwitch userspace should be used with the | |
225 | kernel version included in the same release or with the version | |
226 | from upstream Linux. However, when upgrading between two releases | |
227 | of Open vSwitch it is best to migrate userspace first to reduce | |
228 | the possibility of incompatibilities. | |
229 | ||
542cc9bb | 230 | ### Q: What happened to the bridge compatibility feature? |
900dc97c BP |
231 | |
232 | A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier. | |
233 | When it was enabled, Open vSwitch imitated the interface of the | |
234 | Linux kernel "bridge" module. This allowed users to drop Open | |
235 | vSwitch into environments designed to use the Linux kernel bridge | |
236 | module without adapting the environment to use Open vSwitch. | |
237 | ||
238 | Open vSwitch 1.10 and later do not support bridge compatibility. | |
239 | The feature was dropped because version 1.10 adopted a new internal | |
240 | architecture that made bridge compatibility difficult to maintain. | |
241 | Now that many environments use OVS directly, it would be rarely | |
242 | useful in any case. | |
243 | ||
244 | To use bridge compatibility, install OVS 1.9 or earlier, including | |
245 | the accompanying kernel modules (both the main and bridge | |
246 | compatibility modules), following the instructions that come with | |
247 | the release. Be sure to start the ovs-brcompatd daemon. | |
248 | ||
3fc7dc18 | 249 | |
79aa9fd0 BP |
250 | Terminology |
251 | ----------- | |
252 | ||
542cc9bb | 253 | ### Q: I thought Open vSwitch was a virtual Ethernet switch, but the documentation keeps talking about bridges. What's a bridge? |
79aa9fd0 BP |
254 | |
255 | A: In networking, the terms "bridge" and "switch" are synonyms. Open | |
256 | vSwitch implements an Ethernet switch, which means that it is also | |
257 | an Ethernet bridge. | |
258 | ||
542cc9bb | 259 | ### Q: What's a VLAN? |
79aa9fd0 BP |
260 | |
261 | A: See the "VLAN" section below. | |
262 | ||
263 | ||
717e7c8d BP |
264 | Basic Configuration |
265 | ------------------- | |
266 | ||
542cc9bb | 267 | ### Q: How do I configure a port as an access port? |
717e7c8d BP |
268 | |
269 | A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example, | |
270 | the following commands configure br0 with eth0 as a trunk port (the | |
271 | default) and tap0 as an access port for VLAN 9: | |
272 | ||
273 | ovs-vsctl add-br br0 | |
274 | ovs-vsctl add-port br0 eth0 | |
275 | ovs-vsctl add-port br0 tap0 tag=9 | |
276 | ||
277 | If you want to configure an already added port as an access port, | |
278 | use "ovs-vsctl set", e.g.: | |
279 | ||
280 | ovs-vsctl set port tap0 tag=9 | |
281 | ||
542cc9bb | 282 | ### Q: How do I configure a port as a SPAN port, that is, enable mirroring of all traffic to that port? |
717e7c8d BP |
283 | |
284 | A: The following commands configure br0 with eth0 and tap0 as trunk | |
285 | ports. All traffic coming in or going out on eth0 or tap0 is also | |
286 | mirrored to tap1; any traffic arriving on tap1 is dropped: | |
287 | ||
288 | ovs-vsctl add-br br0 | |
289 | ovs-vsctl add-port br0 eth0 | |
290 | ovs-vsctl add-port br0 tap0 | |
291 | ovs-vsctl add-port br0 tap1 \ | |
292 | -- --id=@p get port tap1 \ | |
0dc8b8c2 YT |
293 | -- --id=@m create mirror name=m0 select-all=true output-port=@p \ |
294 | -- set bridge br0 mirrors=@m | |
717e7c8d BP |
295 | |
296 | To later disable mirroring, run: | |
297 | ||
298 | ovs-vsctl clear bridge br0 mirrors | |
299 | ||
542cc9bb | 300 | ### Q: Does Open vSwitch support configuring a port in promiscuous mode? |
e253f732 BP |
301 | |
302 | A: Yes. How you configure it depends on what you mean by "promiscuous | |
303 | mode": | |
304 | ||
542cc9bb TG |
305 | - Conventionally, "promiscuous mode" is a feature of a network |
306 | interface card. Ordinarily, a NIC passes to the CPU only the | |
307 | packets actually destined to its host machine. It discards | |
308 | the rest to avoid wasting memory and CPU cycles. When | |
309 | promiscuous mode is enabled, however, it passes every packet | |
310 | to the CPU. On an old-style shared-media or hub-based | |
311 | network, this allows the host to spy on all packets on the | |
312 | network. But in the switched networks that are almost | |
313 | everywhere these days, promiscuous mode doesn't have much | |
314 | effect, because few packets not destined to a host are | |
315 | delivered to the host's NIC. | |
316 | ||
317 | This form of promiscuous mode is configured in the guest OS of | |
318 | the VMs on your bridge, e.g. with "ifconfig". | |
319 | ||
320 | - The VMware vSwitch uses a different definition of "promiscuous | |
321 | mode". When you configure promiscuous mode on a VMware vNIC, | |
322 | the vSwitch sends a copy of every packet received by the | |
323 | vSwitch to that vNIC. That has a much bigger effect than just | |
324 | enabling promiscuous mode in a guest OS. Rather than getting | |
325 | a few stray packets for which the switch does not yet know the | |
326 | correct destination, the vNIC gets every packet. The effect | |
327 | is similar to replacing the vSwitch by a virtual hub. | |
328 | ||
329 | This "promiscuous mode" is what switches normally call "port | |
330 | mirroring" or "SPAN". For information on how to configure | |
331 | SPAN, see "How do I configure a port as a SPAN port, that is, | |
332 | enable mirroring of all traffic to that port?" | |
333 | ||
334 | ### Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable mirroring of all traffic to that VLAN? | |
717e7c8d BP |
335 | |
336 | A: The following commands configure br0 with eth0 as a trunk port and | |
337 | tap0 as an access port for VLAN 10. All traffic coming in or going | |
338 | out on tap0, as well as traffic coming in or going out on eth0 in | |
339 | VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for | |
340 | VLAN 10, in cases where one is present, is dropped as part of | |
341 | mirroring: | |
342 | ||
343 | ovs-vsctl add-br br0 | |
344 | ovs-vsctl add-port br0 eth0 | |
345 | ovs-vsctl add-port br0 tap0 tag=10 | |
346 | ovs-vsctl \ | |
0dc8b8c2 | 347 | -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \ |
717e7c8d | 348 | output-vlan=15 \ |
0dc8b8c2 | 349 | -- set bridge br0 mirrors=@m |
717e7c8d BP |
350 | |
351 | To later disable mirroring, run: | |
352 | ||
353 | ovs-vsctl clear bridge br0 mirrors | |
354 | ||
355 | Mirroring to a VLAN can disrupt a network that contains unmanaged | |
356 | switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a | |
357 | GRE tunnel has fewer caveats than mirroring to a VLAN and should | |
358 | generally be preferred. | |
359 | ||
542cc9bb | 360 | ### Q: Can I mirror more than one input VLAN to an RSPAN VLAN? |
717e7c8d BP |
361 | |
362 | A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor | |
363 | of the specified output-vlan. This loss of information may make | |
364 | the mirrored traffic too hard to interpret. | |
365 | ||
366 | To mirror multiple VLANs, use the commands above, but specify a | |
367 | comma-separated list of VLANs as the value for select-vlan. To | |
368 | mirror every VLAN, use the commands above, but omit select-vlan and | |
369 | its value entirely. | |
370 | ||
371 | When a packet arrives on a VLAN that is used as a mirror output | |
372 | VLAN, the mirror is disregarded. Instead, in standalone mode, OVS | |
373 | floods the packet across all the ports for which the mirror output | |
374 | VLAN is configured. (If an OpenFlow controller is in use, then it | |
375 | can override this behavior through the flow table.) If OVS is used | |
376 | as an intermediate switch, rather than an edge switch, this ensures | |
377 | that the RSPAN traffic is distributed through the network. | |
378 | ||
379 | Mirroring to a VLAN can disrupt a network that contains unmanaged | |
380 | switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a | |
381 | GRE tunnel has fewer caveats than mirroring to a VLAN and should | |
382 | generally be preferred. | |
383 | ||
542cc9bb | 384 | ### Q: How do I configure mirroring of all traffic to a GRE tunnel? |
717e7c8d BP |
385 | |
386 | A: The following commands configure br0 with eth0 and tap0 as trunk | |
387 | ports. All traffic coming in or going out on eth0 or tap0 is also | |
388 | mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any | |
389 | traffic arriving on gre0 is dropped: | |
390 | ||
391 | ovs-vsctl add-br br0 | |
392 | ovs-vsctl add-port br0 eth0 | |
393 | ovs-vsctl add-port br0 tap0 | |
394 | ovs-vsctl add-port br0 gre0 \ | |
395 | -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \ | |
396 | -- --id=@p get port gre0 \ | |
0dc8b8c2 YT |
397 | -- --id=@m create mirror name=m0 select-all=true output-port=@p \ |
398 | -- set bridge br0 mirrors=@m | |
717e7c8d BP |
399 | |
400 | To later disable mirroring and destroy the GRE tunnel: | |
401 | ||
402 | ovs-vsctl clear bridge br0 mirrors | |
403 | ovs-vcstl del-port br0 gre0 | |
404 | ||
542cc9bb | 405 | ### Q: Does Open vSwitch support ERSPAN? |
717e7c8d BP |
406 | |
407 | A: No. ERSPAN is an undocumented proprietary protocol. As an | |
408 | alternative, Open vSwitch supports mirroring to a GRE tunnel (see | |
409 | above). | |
410 | ||
542cc9bb | 411 | ### Q: How do I connect two bridges? |
1ab9712b BP |
412 | |
413 | A: First, why do you want to do this? Two connected bridges are not | |
414 | much different from a single bridge, so you might as well just have | |
415 | a single bridge with all your ports on it. | |
416 | ||
417 | If you still want to connect two bridges, you can use a pair of | |
418 | patch ports. The following example creates bridges br0 and br1, | |
419 | adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0 | |
420 | and br1 with a pair of patch ports. | |
421 | ||
422 | ovs-vsctl add-br br0 | |
423 | ovs-vsctl add-port br0 eth0 | |
424 | ovs-vsctl add-port br0 tap0 | |
425 | ovs-vsctl add-br br1 | |
426 | ovs-vsctl add-port br1 tap1 | |
427 | ovs-vsctl \ | |
428 | -- add-port br0 patch0 \ | |
429 | -- set interface patch0 type=patch options:peer=patch1 \ | |
430 | -- add-port br1 patch1 \ | |
431 | -- set interface patch1 type=patch options:peer=patch0 | |
432 | ||
433 | Bridges connected with patch ports are much like a single bridge. | |
434 | For instance, if the example above also added eth1 to br1, and both | |
435 | eth0 and eth1 happened to be connected to the same next-hop switch, | |
436 | then you could loop your network just as you would if you added | |
437 | eth0 and eth1 to the same bridge (see the "Configuration Problems" | |
438 | section below for more information). | |
439 | ||
440 | If you are using Open vSwitch 1.9 or an earlier version, then you | |
441 | need to be using the kernel module bundled with Open vSwitch rather | |
442 | than the one that is integrated into Linux 3.3 and later, because | |
443 | Open vSwitch 1.9 and earlier versions need kernel support for patch | |
444 | ports. This also means that in Open vSwitch 1.9 and earlier, patch | |
445 | ports will not work with the userspace datapath, only with the | |
446 | kernel module. | |
447 | ||
542cc9bb | 448 | ### Q: How do I configure a bridge without an OpenFlow local port? (Local port in the sense of OFPP_LOCAL) |
5c9c1105 YT |
449 | |
450 | A: Open vSwitch does not support such a configuration. | |
451 | Bridges always have their local ports. | |
452 | ||
1a274bfe BP |
453 | |
454 | Implementation Details | |
455 | ---------------------- | |
456 | ||
542cc9bb | 457 | ### Q: I hear OVS has a couple of kinds of flows. Can you tell me about them? |
a70fc0cf JP |
458 | |
459 | A: Open vSwitch uses different kinds of flows for different purposes: | |
460 | ||
542cc9bb TG |
461 | - OpenFlow flows are the most important kind of flow. OpenFlow |
462 | controllers use these flows to define a switch's policy. | |
463 | OpenFlow flows support wildcards, priorities, and multiple | |
464 | tables. | |
465 | ||
466 | When in-band control is in use, Open vSwitch sets up a few | |
467 | "hidden" flows, with priority higher than a controller or the | |
468 | user can configure, that are not visible via OpenFlow. (See | |
469 | the "Controller" section of the FAQ for more information | |
470 | about hidden flows.) | |
471 | ||
472 | - The Open vSwitch software switch implementation uses a second | |
473 | kind of flow internally. These flows, called "datapath" or | |
474 | "kernel" flows, do not support priorities and comprise only a | |
475 | single table, which makes them suitable for caching. (Like | |
476 | OpenFlow flows, datapath flows do support wildcarding, in Open | |
477 | vSwitch 1.11 and later.) OpenFlow flows and datapath flows | |
478 | also support different actions and number ports differently. | |
479 | ||
480 | Datapath flows are an implementation detail that is subject to | |
481 | change in future versions of Open vSwitch. Even with the | |
482 | current version of Open vSwitch, hardware switch | |
483 | implementations do not necessarily use this architecture. | |
a70fc0cf | 484 | |
1a274bfe BP |
485 | Users and controllers directly control only the OpenFlow flow |
486 | table. Open vSwitch manages the datapath flow table itself, so | |
487 | users should not normally be concerned with it. | |
488 | ||
542cc9bb | 489 | ### Q: Why are there so many different ways to dump flows? |
1a274bfe BP |
490 | |
491 | A: Open vSwitch has two kinds of flows (see the previous question), so | |
492 | it has commands with different purposes for dumping each kind of | |
493 | flow: | |
a70fc0cf | 494 | |
542cc9bb TG |
495 | - `ovs-ofctl dump-flows <br>` dumps OpenFlow flows, excluding |
496 | hidden flows. This is the most commonly useful form of flow | |
497 | dump. (Unlike the other commands, this should work with any | |
498 | OpenFlow switch, not just Open vSwitch.) | |
a70fc0cf | 499 | |
542cc9bb TG |
500 | - `ovs-appctl bridge/dump-flows <br>` dumps OpenFlow flows, |
501 | including hidden flows. This is occasionally useful for | |
502 | troubleshooting suspected issues with in-band control. | |
a70fc0cf | 503 | |
542cc9bb TG |
504 | - `ovs-dpctl dump-flows [dp]` dumps the datapath flow table |
505 | entries for a Linux kernel-based datapath. In Open vSwitch | |
506 | 1.10 and later, ovs-vswitchd merges multiple switches into a | |
507 | single datapath, so it will show all the flows on all your | |
508 | kernel-based switches. This command can occasionally be | |
509 | useful for debugging. | |
a70fc0cf | 510 | |
542cc9bb TG |
511 | - `ovs-appctl dpif/dump-flows <br>`, new in Open vSwitch 1.10, |
512 | dumps datapath flows for only the specified bridge, regardless | |
513 | of the type. | |
a70fc0cf | 514 | |
542cc9bb | 515 | ### Q: How does multicast snooping works with VLANs? |
c81f359b FL |
516 | |
517 | A: Open vSwitch maintains snooping tables for each VLAN. | |
518 | ||
717e7c8d | 519 | |
bcb8bde4 JR |
520 | Performance |
521 | ----------- | |
522 | ||
542cc9bb | 523 | ### Q: I just upgraded and I see a performance drop. Why? |
bcb8bde4 JR |
524 | |
525 | A: The OVS kernel datapath may have been updated to a newer version than | |
526 | the OVS userspace components. Sometimes new versions of OVS kernel | |
527 | module add functionality that is backwards compatible with older | |
528 | userspace components but may cause a drop in performance with them. | |
529 | Especially, if a kernel module from OVS 2.1 or newer is paired with | |
530 | OVS userspace 1.10 or older, there will be a performance drop for | |
531 | TCP traffic. | |
532 | ||
533 | Updating the OVS userspace components to the latest released | |
534 | version should fix the performance degradation. | |
535 | ||
536 | To get the best possible performance and functionality, it is | |
537 | recommended to pair the same versions of the kernel module and OVS | |
538 | userspace. | |
539 | ||
540 | ||
c483d489 BP |
541 | Configuration Problems |
542 | ---------------------- | |
543 | ||
542cc9bb | 544 | ### Q: I created a bridge and added my Ethernet port to it, using commands |
c483d489 BP |
545 | like these: |
546 | ||
547 | ovs-vsctl add-br br0 | |
548 | ovs-vsctl add-port br0 eth0 | |
549 | ||
550 | and as soon as I ran the "add-port" command I lost all connectivity | |
551 | through eth0. Help! | |
552 | ||
553 | A: A physical Ethernet device that is part of an Open vSwitch bridge | |
554 | should not have an IP address. If one does, then that IP address | |
555 | will not be fully functional. | |
556 | ||
557 | You can restore functionality by moving the IP address to an Open | |
558 | vSwitch "internal" device, such as the network device named after | |
559 | the bridge itself. For example, assuming that eth0's IP address is | |
560 | 192.168.128.5, you could run the commands below to fix up the | |
561 | situation: | |
562 | ||
563 | ifconfig eth0 0.0.0.0 | |
564 | ifconfig br0 192.168.128.5 | |
565 | ||
566 | (If your only connection to the machine running OVS is through the | |
567 | IP address in question, then you would want to run all of these | |
568 | commands on a single command line, or put them into a script.) If | |
569 | there were any additional routes assigned to eth0, then you would | |
570 | also want to use commands to adjust these routes to go through br0. | |
571 | ||
572 | If you use DHCP to obtain an IP address, then you should kill the | |
573 | DHCP client that was listening on the physical Ethernet interface | |
574 | (e.g. eth0) and start one listening on the internal interface | |
575 | (e.g. br0). You might still need to manually clear the IP address | |
576 | from the physical interface (e.g. with "ifconfig eth0 0.0.0.0"). | |
577 | ||
578 | There is no compelling reason why Open vSwitch must work this way. | |
579 | However, this is the way that the Linux kernel bridge module has | |
580 | always worked, so it's a model that those accustomed to Linux | |
581 | bridging are already used to. Also, the model that most people | |
582 | expect is not implementable without kernel changes on all the | |
583 | versions of Linux that Open vSwitch supports. | |
584 | ||
585 | By the way, this issue is not specific to physical Ethernet | |
c7b0cfd3 | 586 | devices. It applies to all network devices except Open vSwitch |
c483d489 BP |
587 | "internal" devices. |
588 | ||
542cc9bb TG |
589 | ### Q: I created a bridge and added a couple of Ethernet ports to it, |
590 | ### using commands like these: | |
c483d489 BP |
591 | |
592 | ovs-vsctl add-br br0 | |
593 | ovs-vsctl add-port br0 eth0 | |
594 | ovs-vsctl add-port br0 eth1 | |
595 | ||
596 | and now my network seems to have melted: connectivity is unreliable | |
597 | (even connectivity that doesn't go through Open vSwitch), all the | |
629a6b48 BP |
598 | LEDs on my physical switches are blinking, wireshark shows |
599 | duplicated packets, and CPU usage is very high. | |
c483d489 BP |
600 | |
601 | A: More than likely, you've looped your network. Probably, eth0 and | |
602 | eth1 are connected to the same physical Ethernet switch. This | |
603 | yields a scenario where OVS receives a broadcast packet on eth0 and | |
604 | sends it out on eth1, then the physical switch connected to eth1 | |
605 | sends the packet back on eth0, and so on forever. More complicated | |
606 | scenarios, involving a loop through multiple switches, are possible | |
607 | too. | |
608 | ||
609 | The solution depends on what you are trying to do: | |
610 | ||
542cc9bb TG |
611 | - If you added eth0 and eth1 to get higher bandwidth or higher |
612 | reliability between OVS and your physical Ethernet switch, | |
613 | use a bond. The following commands create br0 and then add | |
614 | eth0 and eth1 as a bond: | |
c483d489 | 615 | |
542cc9bb TG |
616 | ovs-vsctl add-br br0 |
617 | ovs-vsctl add-bond br0 bond0 eth0 eth1 | |
c483d489 | 618 | |
542cc9bb TG |
619 | Bonds have tons of configuration options. Please read the |
620 | documentation on the Port table in ovs-vswitchd.conf.db(5) | |
621 | for all the details. | |
c483d489 | 622 | |
542cc9bb TG |
623 | - Perhaps you don't actually need eth0 and eth1 to be on the |
624 | same bridge. For example, if you simply want to be able to | |
625 | connect each of them to virtual machines, then you can put | |
626 | each of them on a bridge of its own: | |
c483d489 | 627 | |
542cc9bb TG |
628 | ovs-vsctl add-br br0 |
629 | ovs-vsctl add-port br0 eth0 | |
c483d489 | 630 | |
542cc9bb TG |
631 | ovs-vsctl add-br br1 |
632 | ovs-vsctl add-port br1 eth1 | |
c483d489 | 633 | |
542cc9bb TG |
634 | and then connect VMs to br0 and br1. (A potential |
635 | disadvantage is that traffic cannot directly pass between br0 | |
636 | and br1. Instead, it will go out eth0 and come back in eth1, | |
637 | or vice versa.) | |
c483d489 | 638 | |
542cc9bb TG |
639 | - If you have a redundant or complex network topology and you |
640 | want to prevent loops, turn on spanning tree protocol (STP). | |
641 | The following commands create br0, enable STP, and add eth0 | |
642 | and eth1 to the bridge. The order is important because you | |
643 | don't want have to have a loop in your network even | |
644 | transiently: | |
c483d489 | 645 | |
542cc9bb TG |
646 | ovs-vsctl add-br br0 |
647 | ovs-vsctl set bridge br0 stp_enable=true | |
648 | ovs-vsctl add-port br0 eth0 | |
649 | ovs-vsctl add-port br0 eth1 | |
c483d489 | 650 | |
542cc9bb TG |
651 | The Open vSwitch implementation of STP is not well tested. |
652 | Please report any bugs you observe, but if you'd rather avoid | |
653 | acting as a beta tester then another option might be your | |
654 | best shot. | |
c483d489 | 655 | |
542cc9bb | 656 | ### Q: I can't seem to use Open vSwitch in a wireless network. |
c483d489 BP |
657 | |
658 | A: Wireless base stations generally only allow packets with the source | |
659 | MAC address of NIC that completed the initial handshake. | |
660 | Therefore, without MAC rewriting, only a single device can | |
661 | communicate over a single wireless link. | |
662 | ||
663 | This isn't specific to Open vSwitch, it's enforced by the access | |
664 | point, so the same problems will show up with the Linux bridge or | |
665 | any other way to do bridging. | |
666 | ||
542cc9bb | 667 | ### Q: I can't seem to add my PPP interface to an Open vSwitch bridge. |
8748ec7b BP |
668 | |
669 | A: PPP most commonly carries IP packets, but Open vSwitch works only | |
670 | with Ethernet frames. The correct way to interface PPP to an | |
671 | Ethernet network is usually to use routing instead of switching. | |
672 | ||
542cc9bb | 673 | ### Q: Is there any documentation on the database tables and fields? |
5aa75474 BP |
674 | |
675 | A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference. | |
676 | ||
542cc9bb | 677 | ### Q: When I run ovs-dpctl I no longer see the bridges I created. Instead, |
acf60855 JP |
678 | I only see a datapath called "ovs-system". How can I see datapath |
679 | information about a particular bridge? | |
680 | ||
681 | A: In version 1.9.0, OVS switched to using a single datapath that is | |
682 | shared by all bridges of that type. The "ovs-appctl dpif/*" | |
683 | commands provide similar functionality that is scoped by the bridge. | |
684 | ||
542cc9bb | 685 | ### Q: I created a GRE port using ovs-vsctl so why can't I send traffic or |
004a6249 JG |
686 | see the port in the datapath? |
687 | ||
688 | A: On Linux kernels before 3.11, the OVS GRE module and Linux GRE module | |
689 | cannot be loaded at the same time. It is likely that on your system the | |
690 | Linux GRE module is already loaded and blocking OVS (to confirm, check | |
691 | dmesg for errors regarding GRE registration). To fix this, unload all | |
692 | GRE modules that appear in lsmod as well as the OVS kernel module. You | |
542cc9bb | 693 | can then reload the OVS module following the directions in |
9feb1017 | 694 | [INSTALL.md], which will ensure that dependencies are satisfied. |
004a6249 | 695 | |
542cc9bb | 696 | ### Q: Open vSwitch does not seem to obey my packet filter rules. |
c6bbc394 YT |
697 | |
698 | A: It depends on mechanisms and configurations you want to use. | |
699 | ||
700 | You cannot usefully use typical packet filters, like iptables, on | |
701 | physical Ethernet ports that you add to an Open vSwitch bridge. | |
702 | This is because Open vSwitch captures packets from the interface at | |
703 | a layer lower below where typical packet-filter implementations | |
704 | install their hooks. (This actually applies to any interface of | |
705 | type "system" that you might add to an Open vSwitch bridge.) | |
706 | ||
707 | You can usefully use typical packet filters on Open vSwitch | |
708 | internal ports as they are mostly ordinary interfaces from the point | |
709 | of view of packet filters. | |
710 | ||
711 | For example, suppose you create a bridge br0 and add Ethernet port | |
712 | eth0 to it. Then you can usefully add iptables rules to affect the | |
713 | internal interface br0, but not the physical interface eth0. (br0 | |
714 | is also where you would add an IP address, as discussed elsewhere | |
715 | in the FAQ.) | |
716 | ||
717 | For simple filtering rules, it might be possible to achieve similar | |
718 | results by installing appropriate OpenFlow flows instead. | |
719 | ||
720 | If the use of a particular packet filter setup is essential, Open | |
721 | vSwitch might not be the best choice for you. On Linux, you might | |
722 | want to consider using the Linux Bridge. (This is the only choice if | |
723 | you want to use ebtables rules.) On NetBSD, you might want to | |
724 | consider using the bridge(4) with BRIDGE_IPF option. | |
725 | ||
542cc9bb | 726 | ### Q: It seems that Open vSwitch does nothing when I removed a port and |
dd63a57e YT |
727 | then immediately put it back. For example, consider that p1 is |
728 | a port of type=internal: | |
729 | ||
730 | ovs-vsctl del-port br0 p1 -- \ | |
731 | add-port br0 p1 -- \ | |
732 | set interface p1 type=internal | |
733 | ||
734 | A: It's an expected behaviour. | |
735 | ||
7494dca8 YT |
736 | If del-port and add-port happen in a single OVSDB transaction as |
737 | your example, Open vSwitch always "skips" the intermediate steps. | |
dd63a57e YT |
738 | Even if they are done in multiple transactions, it's still allowed |
739 | for Open vSwitch to skip the intermediate steps and just implement | |
740 | the overall effect. In both cases, your example would be turned | |
741 | into a no-op. | |
742 | ||
743 | If you want to make Open vSwitch actually destroy and then re-create | |
744 | the port for some side effects like resetting kernel setting for the | |
745 | corresponding interface, you need to separate operations into multiple | |
746 | OVSDB transactions and ensure that at least the first one does not have | |
747 | --no-wait. In the following example, the first ovs-vsctl will block | |
748 | until Open vSwitch reloads the new configuration and removes the port: | |
749 | ||
750 | ovs-vsctl del-port br0 p1 | |
751 | ovs-vsctl add-port br0 p1 -- \ | |
752 | set interface p1 type=internal | |
c483d489 | 753 | |
ed5c5110 BP |
754 | ### Q: I want to add thousands of ports to an Open vSwitch bridge, but |
755 | it takes too long (minutes or hours) to do it with ovs-vsctl. How | |
756 | can I do it faster? | |
757 | ||
758 | A: If you add them one at a time with ovs-vsctl, it can take a long | |
759 | time to add thousands of ports to an Open vSwitch bridge. This is | |
760 | because every invocation of ovs-vsctl first reads the current | |
761 | configuration from OVSDB. As the number of ports grows, this | |
762 | starts to take an appreciable amount of time, and when it is | |
763 | repeated thousands of times the total time becomes significant. | |
764 | ||
765 | The solution is to add the ports in one invocation of ovs-vsctl (or | |
766 | a small number of them). For example, using bash: | |
767 | ||
768 | ovs-vsctl add-br br0 | |
769 | cmds=; for i in {1..5000}; do cmds+=" -- add-port br0 p$i"; done | |
770 | ovs-vsctl $cmds | |
771 | ||
772 | takes seconds, not minutes or hours, in the OVS sandbox environment. | |
773 | ||
bceafb63 BP |
774 | Quality of Service (QoS) |
775 | ------------------------ | |
776 | ||
542cc9bb | 777 | ### Q: How do I configure Quality of Service (QoS)? |
bceafb63 BP |
778 | |
779 | A: Suppose that you want to set up bridge br0 connected to physical | |
780 | Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces | |
781 | vif1.0 and vif2.0, and that you want to limit traffic from vif1.0 | |
782 | to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you | |
783 | could configure the bridge this way: | |
784 | ||
785 | ovs-vsctl -- \ | |
786 | add-br br0 -- \ | |
0dc8b8c2 YT |
787 | add-port br0 eth0 -- \ |
788 | add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \ | |
789 | add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \ | |
790 | set port eth0 qos=@newqos -- \ | |
791 | --id=@newqos create qos type=linux-htb \ | |
bceafb63 | 792 | other-config:max-rate=1000000000 \ |
0dc8b8c2 YT |
793 | queues:123=@vif10queue \ |
794 | queues:234=@vif20queue -- \ | |
bceafb63 BP |
795 | --id=@vif10queue create queue other-config:max-rate=10000000 -- \ |
796 | --id=@vif20queue create queue other-config:max-rate=20000000 | |
797 | ||
798 | At this point, bridge br0 is configured with the ports and eth0 is | |
799 | configured with the queues that you need for QoS, but nothing is | |
800 | actually directing packets from vif1.0 or vif2.0 to the queues that | |
801 | we have set up for them. That means that all of the packets to | |
802 | eth0 are going to the "default queue", which is not what we want. | |
803 | ||
804 | We use OpenFlow to direct packets from vif1.0 and vif2.0 to the | |
805 | queues reserved for them: | |
806 | ||
807 | ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal | |
808 | ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal | |
809 | ||
810 | Each of the above flows matches on the input port, sets up the | |
811 | appropriate queue (123 for vif1.0, 234 for vif2.0), and then | |
812 | executes the "normal" action, which performs the same switching | |
813 | that Open vSwitch would have done without any OpenFlow flows being | |
814 | present. (We know that vif1.0 and vif2.0 have OpenFlow port | |
815 | numbers 5 and 6, respectively, because we set their ofport_request | |
816 | columns above. If we had not done that, then we would have needed | |
817 | to find out their port numbers before setting up these flows.) | |
818 | ||
819 | Now traffic going from vif1.0 or vif2.0 to eth0 should be | |
820 | rate-limited. | |
821 | ||
822 | By the way, if you delete the bridge created by the above commands, | |
823 | with: | |
824 | ||
825 | ovs-vsctl del-br br0 | |
826 | ||
827 | then that will leave one unreferenced QoS record and two | |
828 | unreferenced Queue records in the Open vSwich database. One way to | |
829 | clear them out, assuming you don't have other QoS or Queue records | |
830 | that you want to keep, is: | |
831 | ||
832 | ovs-vsctl -- --all destroy QoS -- --all destroy Queue | |
833 | ||
7839bb41 BP |
834 | If you do want to keep some QoS or Queue records, or the Open |
835 | vSwitch you are using is older than version 1.8 (which added the | |
836 | --all option), then you will have to destroy QoS and Queue records | |
837 | individually. | |
838 | ||
542cc9bb | 839 | ### Q: I configured Quality of Service (QoS) in my OpenFlow network by |
bceafb63 BP |
840 | adding records to the QoS and Queue table, but the results aren't |
841 | what I expect. | |
842 | ||
843 | A: Did you install OpenFlow flows that use your queues? This is the | |
844 | primary way to tell Open vSwitch which queues you want to use. If | |
845 | you don't do this, then the default queue will be used, which will | |
846 | probably not have the effect you want. | |
847 | ||
848 | Refer to the previous question for an example. | |
849 | ||
542cc9bb | 850 | ### Q: I'd like to take advantage of some QoS feature that Open vSwitch |
e6d29aa7 BP |
851 | doesn't yet support. How do I do that? |
852 | ||
853 | A: Open vSwitch does not implement QoS itself. Instead, it can | |
854 | configure some, but not all, of the QoS features built into the | |
855 | Linux kernel. If you need some QoS feature that OVS cannot | |
856 | configure itself, then the first step is to figure out whether | |
857 | Linux QoS supports that feature. If it does, then you can submit a | |
858 | patch to support Open vSwitch configuration for that feature, or | |
859 | you can use "tc" directly to configure the feature in Linux. (If | |
860 | Linux QoS doesn't support the feature you want, then first you have | |
861 | to add that support to Linux.) | |
862 | ||
542cc9bb | 863 | ### Q: I configured QoS, correctly, but my measurements show that it isn't |
bceafb63 BP |
864 | working as well as I expect. |
865 | ||
866 | A: With the Linux kernel, the Open vSwitch implementation of QoS has | |
867 | two aspects: | |
868 | ||
542cc9bb TG |
869 | - Open vSwitch configures a subset of Linux kernel QoS |
870 | features, according to what is in OVSDB. It is possible that | |
871 | this code has bugs. If you believe that this is so, then you | |
872 | can configure the Linux traffic control (QoS) stack directly | |
873 | with the "tc" program. If you get better results that way, | |
874 | you can send a detailed bug report to bugs@openvswitch.org. | |
bceafb63 | 875 | |
542cc9bb TG |
876 | It is certain that Open vSwitch cannot configure every Linux |
877 | kernel QoS feature. If you need some feature that OVS cannot | |
878 | configure, then you can also use "tc" directly (or add that | |
879 | feature to OVS). | |
bceafb63 | 880 | |
542cc9bb TG |
881 | - The Open vSwitch implementation of OpenFlow allows flows to |
882 | be directed to particular queues. This is pretty simple and | |
883 | unlikely to have serious bugs at this point. | |
bceafb63 BP |
884 | |
885 | However, most problems with QoS on Linux are not bugs in Open | |
886 | vSwitch at all. They tend to be either configuration errors | |
887 | (please see the earlier questions in this section) or issues with | |
888 | the traffic control (QoS) stack in Linux. The Open vSwitch | |
889 | developers are not experts on Linux traffic control. We suggest | |
890 | that, if you believe you are encountering a problem with Linux | |
891 | traffic control, that you consult the tc manpages (e.g. tc(8), | |
892 | tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or | |
893 | mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev). | |
894 | ||
542cc9bb | 895 | ### Q: Does Open vSwitch support OpenFlow meters? |
733fd270 BP |
896 | |
897 | A: Since version 2.0, Open vSwitch has OpenFlow protocol support for | |
898 | OpenFlow meters. There is no implementation of meters in the Open | |
899 | vSwitch software switch (neither the kernel-based nor userspace | |
900 | switches). | |
901 | ||
bceafb63 | 902 | |
c483d489 BP |
903 | VLANs |
904 | ----- | |
905 | ||
542cc9bb | 906 | ### Q: What's a VLAN? |
14481051 BP |
907 | |
908 | A: At the simplest level, a VLAN (short for "virtual LAN") is a way to | |
909 | partition a single switch into multiple switches. Suppose, for | |
910 | example, that you have two groups of machines, group A and group B. | |
911 | You want the machines in group A to be able to talk to each other, | |
912 | and you want the machine in group B to be able to talk to each | |
913 | other, but you don't want the machines in group A to be able to | |
914 | talk to the machines in group B. You can do this with two | |
915 | switches, by plugging the machines in group A into one switch and | |
916 | the machines in group B into the other switch. | |
917 | ||
918 | If you only have one switch, then you can use VLANs to do the same | |
919 | thing, by configuring the ports for machines in group A as VLAN | |
920 | "access ports" for one VLAN and the ports for group B as "access | |
921 | ports" for a different VLAN. The switch will only forward packets | |
922 | between ports that are assigned to the same VLAN, so this | |
923 | effectively subdivides your single switch into two independent | |
924 | switches, one for each group of machines. | |
925 | ||
926 | So far we haven't said anything about VLAN headers. With access | |
927 | ports, like we've described so far, no VLAN header is present in | |
928 | the Ethernet frame. This means that the machines (or switches) | |
929 | connected to access ports need not be aware that VLANs are | |
930 | involved, just like in the case where we use two different physical | |
931 | switches. | |
932 | ||
933 | Now suppose that you have a whole bunch of switches in your | |
934 | network, instead of just one, and that some machines in group A are | |
935 | connected directly to both switches 1 and 2. To allow these | |
936 | machines to talk to each other, you could add an access port for | |
937 | group A's VLAN to switch 1 and another to switch 2, and then | |
938 | connect an Ethernet cable between those ports. That works fine, | |
939 | but it doesn't scale well as the number of switches and the number | |
940 | of VLANs increases, because you use up a lot of valuable switch | |
941 | ports just connecting together your VLANs. | |
942 | ||
943 | This is where VLAN headers come in. Instead of using one cable and | |
944 | two ports per VLAN to connect a pair of switches, we configure a | |
945 | port on each switch as a VLAN "trunk port". Packets sent and | |
946 | received on a trunk port carry a VLAN header that says what VLAN | |
947 | the packet belongs to, so that only two ports total are required to | |
948 | connect the switches, regardless of the number of VLANs in use. | |
949 | Normally, only switches (either physical or virtual) are connected | |
950 | to a trunk port, not individual hosts, because individual hosts | |
951 | don't expect to see a VLAN header in the traffic that they receive. | |
952 | ||
953 | None of the above discussion says anything about particular VLAN | |
954 | numbers. This is because VLAN numbers are completely arbitrary. | |
955 | One must only ensure that a given VLAN is numbered consistently | |
956 | throughout a network and that different VLANs are given different | |
957 | numbers. (That said, VLAN 0 is usually synonymous with a packet | |
958 | that has no VLAN header, and VLAN 4095 is reserved.) | |
959 | ||
542cc9bb | 960 | ### Q: VLANs don't work. |
c483d489 BP |
961 | |
962 | A: Many drivers in Linux kernels before version 3.3 had VLAN-related | |
963 | bugs. If you are having problems with VLANs that you suspect to be | |
964 | driver related, then you have several options: | |
965 | ||
542cc9bb | 966 | - Upgrade to Linux 3.3 or later. |
c483d489 | 967 | |
542cc9bb TG |
968 | - Build and install a fixed version of the particular driver |
969 | that is causing trouble, if one is available. | |
c483d489 | 970 | |
542cc9bb | 971 | - Use a NIC whose driver does not have VLAN problems. |
c483d489 | 972 | |
542cc9bb TG |
973 | - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later |
974 | that works around bugs in kernel drivers. To enable VLAN | |
975 | splinters on interface eth0, use the command: | |
c483d489 | 976 | |
542cc9bb | 977 | ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true |
c483d489 | 978 | |
542cc9bb TG |
979 | For VLAN splinters to be effective, Open vSwitch must know |
980 | which VLANs are in use. See the "VLAN splinters" section in | |
981 | the Interface table in ovs-vswitchd.conf.db(5) for details on | |
982 | how Open vSwitch infers in-use VLANs. | |
c483d489 | 983 | |
542cc9bb TG |
984 | VLAN splinters increase memory use and reduce performance, so |
985 | use them only if needed. | |
c483d489 | 986 | |
542cc9bb TG |
987 | - Apply the "vlan workaround" patch from the XenServer kernel |
988 | patch queue, build Open vSwitch against this patched kernel, | |
989 | and then use ovs-vlan-bug-workaround(8) to enable the VLAN | |
990 | workaround for each interface whose driver is buggy. | |
c483d489 | 991 | |
542cc9bb TG |
992 | (This is a nontrivial exercise, so this option is included |
993 | only for completeness.) | |
c483d489 BP |
994 | |
995 | It is not always easy to tell whether a Linux kernel driver has | |
996 | buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities | |
997 | can help you test. See their manpages for details. Of the two | |
998 | utilities, ovs-test(8) is newer and more thorough, but | |
999 | ovs-vlan-test(8) may be easier to use. | |
1000 | ||
542cc9bb | 1001 | ### Q: VLANs still don't work. I've tested the driver so I know that it's OK. |
c483d489 BP |
1002 | |
1003 | A: Do you have VLANs enabled on the physical switch that OVS is | |
1004 | attached to? Make sure that the port is configured to trunk the | |
1005 | VLAN or VLANs that you are using with OVS. | |
1006 | ||
542cc9bb | 1007 | ### Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch |
c483d489 BP |
1008 | and to its destination host, but OVS seems to drop incoming return |
1009 | traffic. | |
1010 | ||
1011 | A: It's possible that you have the VLAN configured on your physical | |
1012 | switch as the "native" VLAN. In this mode, the switch treats | |
1013 | incoming packets either tagged with the native VLAN or untagged as | |
1014 | part of the native VLAN. It may also send outgoing packets in the | |
1015 | native VLAN without a VLAN tag. | |
1016 | ||
1017 | If this is the case, you have two choices: | |
1018 | ||
542cc9bb TG |
1019 | - Change the physical switch port configuration to tag packets |
1020 | it forwards to OVS with the native VLAN instead of forwarding | |
1021 | them untagged. | |
c483d489 | 1022 | |
542cc9bb TG |
1023 | - Change the OVS configuration for the physical port to a |
1024 | native VLAN mode. For example, the following sets up a | |
1025 | bridge with port eth0 in "native-tagged" mode in VLAN 9: | |
c483d489 | 1026 | |
542cc9bb TG |
1027 | ovs-vsctl add-br br0 |
1028 | ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged | |
c483d489 | 1029 | |
542cc9bb TG |
1030 | In this situation, "native-untagged" mode will probably work |
1031 | equally well. Refer to the documentation for the Port table | |
1032 | in ovs-vswitchd.conf.db(5) for more information. | |
c483d489 | 1033 | |
542cc9bb | 1034 | ### Q: I added a pair of VMs on different VLANs, like this: |
8d45e938 BP |
1035 | |
1036 | ovs-vsctl add-br br0 | |
1037 | ovs-vsctl add-port br0 eth0 | |
1038 | ovs-vsctl add-port br0 tap0 tag=9 | |
1039 | ovs-vsctl add-port br0 tap1 tag=10 | |
1040 | ||
1041 | but the VMs can't access each other, the external network, or the | |
1042 | Internet. | |
1043 | ||
1044 | A: It is to be expected that the VMs can't access each other. VLANs | |
1045 | are a means to partition a network. When you configured tap0 and | |
1046 | tap1 as access ports for different VLANs, you indicated that they | |
1047 | should be isolated from each other. | |
1048 | ||
1049 | As for the external network and the Internet, it seems likely that | |
1050 | the machines you are trying to access are not on VLAN 9 (or 10) and | |
1051 | that the Internet is not available on VLAN 9 (or 10). | |
1052 | ||
542cc9bb | 1053 | ### Q: I added a pair of VMs on the same VLAN, like this: |
3c8399a2 BP |
1054 | |
1055 | ovs-vsctl add-br br0 | |
1056 | ovs-vsctl add-port br0 eth0 | |
1057 | ovs-vsctl add-port br0 tap0 tag=9 | |
1058 | ovs-vsctl add-port br0 tap1 tag=9 | |
1059 | ||
1060 | The VMs can access each other, but not the external network or the | |
1061 | Internet. | |
1062 | ||
1063 | A: It seems likely that the machines you are trying to access in the | |
1064 | external network are not on VLAN 9 and that the Internet is not | |
1065 | available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed | |
1066 | trunk VLAN on the upstream switch port to which eth0 is connected. | |
1067 | ||
542cc9bb | 1068 | ### Q: Can I configure an IP address on a VLAN? |
c483d489 BP |
1069 | |
1070 | A: Yes. Use an "internal port" configured as an access port. For | |
1071 | example, the following configures IP address 192.168.0.7 on VLAN 9. | |
1072 | That is, OVS will forward packets from eth0 to 192.168.0.7 only if | |
1073 | they have an 802.1Q header with VLAN 9. Conversely, traffic | |
1074 | forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q | |
1075 | header with VLAN 9: | |
1076 | ||
1077 | ovs-vsctl add-br br0 | |
1078 | ovs-vsctl add-port br0 eth0 | |
1079 | ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal | |
1080 | ifconfig vlan9 192.168.0.7 | |
1081 | ||
8dc54921 BP |
1082 | See also the following question. |
1083 | ||
542cc9bb | 1084 | ### Q: I configured one IP address on VLAN 0 and another on VLAN 9, like |
8dc54921 BP |
1085 | this: |
1086 | ||
1087 | ovs-vsctl add-br br0 | |
1088 | ovs-vsctl add-port br0 eth0 | |
1089 | ifconfig br0 192.168.0.5 | |
1090 | ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal | |
1091 | ifconfig vlan9 192.168.0.9 | |
1092 | ||
1093 | but other hosts that are only on VLAN 0 can reach the IP address | |
1094 | configured on VLAN 9. What's going on? | |
1095 | ||
1096 | A: RFC 1122 section 3.3.4.2 "Multihoming Requirements" describes two | |
1097 | approaches to IP address handling in Internet hosts: | |
1098 | ||
542cc9bb TG |
1099 | - In the "Strong ES Model", where an ES is a host ("End |
1100 | System"), an IP address is primarily associated with a | |
1101 | particular interface. The host discards packets that arrive | |
1102 | on interface A if they are destined for an IP address that is | |
1103 | configured on interface B. The host never sends packets from | |
1104 | interface A using a source address configured on interface B. | |
1105 | ||
1106 | - In the "Weak ES Model", an IP address is primarily associated | |
1107 | with a host. The host accepts packets that arrive on any | |
1108 | interface if they are destined for any of the host's IP | |
1109 | addresses, even if the address is configured on some | |
1110 | interface other than the one on which it arrived. The host | |
1111 | does not restrict itself to sending packets from an IP | |
1112 | address associated with the originating interface. | |
8dc54921 BP |
1113 | |
1114 | Linux uses the weak ES model. That means that when packets | |
1115 | destined to the VLAN 9 IP address arrive on eth0 and are bridged to | |
1116 | br0, the kernel IP stack accepts them there for the VLAN 9 IP | |
1117 | address, even though they were not received on vlan9, the network | |
1118 | device for vlan9. | |
1119 | ||
1120 | To simulate the strong ES model on Linux, one may add iptables rule | |
1121 | to filter packets based on source and destination address and | |
1122 | adjust ARP configuration with sysctls. | |
1123 | ||
1124 | BSD uses the strong ES model. | |
1125 | ||
542cc9bb | 1126 | ### Q: My OpenFlow controller doesn't see the VLANs that I expect. |
c483d489 BP |
1127 | |
1128 | A: The configuration for VLANs in the Open vSwitch database (e.g. via | |
1129 | ovs-vsctl) only affects traffic that goes through Open vSwitch's | |
1130 | implementation of the OpenFlow "normal switching" action. By | |
1131 | default, when Open vSwitch isn't connected to a controller and | |
1132 | nothing has been manually configured in the flow table, all traffic | |
1133 | goes through the "normal switching" action. But, if you set up | |
1134 | OpenFlow flows on your own, through a controller or using ovs-ofctl | |
1135 | or through other means, then you have to implement VLAN handling | |
1136 | yourself. | |
1137 | ||
1138 | You can use "normal switching" as a component of your OpenFlow | |
1139 | actions, e.g. by putting "normal" into the lists of actions on | |
1140 | ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow | |
241241f5 BP |
1141 | controller. In situations where this is not suitable, you can |
1142 | implement VLAN handling yourself, e.g.: | |
1143 | ||
542cc9bb TG |
1144 | - If a packet comes in on an access port, and the flow table |
1145 | needs to send it out on a trunk port, then the flow can add | |
1146 | the appropriate VLAN tag with the "mod_vlan_vid" action. | |
241241f5 | 1147 | |
542cc9bb TG |
1148 | - If a packet comes in on a trunk port, and the flow table |
1149 | needs to send it out on an access port, then the flow can | |
1150 | strip the VLAN tag with the "strip_vlan" action. | |
c483d489 | 1151 | |
542cc9bb | 1152 | ### Q: I configured ports on a bridge as access ports with different VLAN |
f0a0c1a6 BP |
1153 | tags, like this: |
1154 | ||
1155 | ovs-vsctl add-br br0 | |
1156 | ovs-vsctl set-controller br0 tcp:192.168.0.10:6633 | |
1157 | ovs-vsctl add-port br0 eth0 | |
1158 | ovs-vsctl add-port br0 tap0 tag=9 | |
1159 | ovs-vsctl add-port br0 tap1 tag=10 | |
1160 | ||
1161 | but the VMs running behind tap0 and tap1 can still communicate, | |
1162 | that is, they are not isolated from each other even though they are | |
1163 | on different VLANs. | |
1164 | ||
1165 | A: Do you have a controller configured on br0 (as the commands above | |
1166 | do)? If so, then this is a variant on the previous question, "My | |
1167 | OpenFlow controller doesn't see the VLANs that I expect," and you | |
1168 | can refer to the answer there for more information. | |
1169 | ||
542cc9bb | 1170 | ### Q: How MAC learning works with VLANs? |
f8003b53 YT |
1171 | |
1172 | A: Open vSwitch implements Independent VLAN Learning (IVL) for | |
1173 | OFPP_NORMAL action. I.e. it logically has separate learning tables | |
1174 | for each VLANs. | |
1175 | ||
c483d489 | 1176 | |
0edbe3fb KM |
1177 | VXLANs |
1178 | ----- | |
1179 | ||
542cc9bb | 1180 | ### Q: What's a VXLAN? |
0edbe3fb KM |
1181 | |
1182 | A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means | |
1183 | to solve the scaling challenges of VLAN networks in a multi-tenant | |
1184 | environment. VXLAN is an overlay network which transports an L2 network | |
1185 | over an existing L3 network. For more information on VXLAN, please see | |
bedd6854 | 1186 | RFC 7348: |
0edbe3fb | 1187 | |
bedd6854 | 1188 | http://tools.ietf.org/html/rfc7348 |
0edbe3fb | 1189 | |
542cc9bb | 1190 | ### Q: How much of the VXLAN protocol does Open vSwitch currently support? |
0edbe3fb KM |
1191 | |
1192 | A: Open vSwitch currently supports the framing format for packets on the | |
1193 | wire. There is currently no support for the multicast aspects of VXLAN. | |
1194 | To get around the lack of multicast support, it is possible to | |
1195 | pre-provision MAC to IP address mappings either manually or from a | |
1196 | controller. | |
1197 | ||
542cc9bb | 1198 | ### Q: What destination UDP port does the VXLAN implementation in Open vSwitch |
0edbe3fb KM |
1199 | use? |
1200 | ||
1201 | A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which | |
1202 | is 4789. However, it is possible to configure the destination UDP port | |
1203 | manually on a per-VXLAN tunnel basis. An example of this configuration is | |
1204 | provided below. | |
1205 | ||
b770275d RÃ…A |
1206 | ovs-vsctl add-br br0 |
1207 | ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1 | |
0edbe3fb KM |
1208 | type=vxlan options:remote_ip=192.168.1.2 options:key=flow |
1209 | options:dst_port=8472 | |
1210 | ||
1211 | ||
0f5edef0 BP |
1212 | Using OpenFlow (Manually or Via Controller) |
1213 | ------------------------------------------- | |
c483d489 | 1214 | |
542cc9bb | 1215 | ### Q: What versions of OpenFlow does Open vSwitch support? |
7b287e99 | 1216 | |
c37c0382 AC |
1217 | A: The following table lists the versions of OpenFlow supported by |
1218 | each version of Open vSwitch: | |
7b287e99 | 1219 | |
42dccab5 | 1220 | Open vSwitch OF1.0 OF1.1 OF1.2 OF1.3 OF1.4 OF1.5 |
542cc9bb | 1221 | ###============ ===== ===== ===== ===== ===== ===== |
42dccab5 BP |
1222 | 1.9 and earlier yes --- --- --- --- --- |
1223 | 1.10 yes --- [*] [*] --- --- | |
1224 | 1.11 yes --- [*] [*] --- --- | |
1225 | 2.0 yes [*] [*] [*] --- --- | |
1226 | 2.1 yes [*] [*] [*] --- --- | |
1227 | 2.2 yes [*] [*] [*] [%] [*] | |
1228 | 2.3 yes yes yes yes [*] [*] | |
8e70e196 | 1229 | |
c37c0382 | 1230 | [*] Supported, with one or more missing features. |
aa233d57 | 1231 | [%] Experimental, unsafe implementation. |
8e70e196 | 1232 | |
6dc53744 BP |
1233 | Open vSwitch 2.3 enables OpenFlow 1.0, 1.1, 1.2, and 1.3 by default |
1234 | in ovs-vswitchd. In Open vSwitch 1.10 through 2.2, OpenFlow 1.1, | |
aa233d57 | 1235 | 1.2, and 1.3 must be enabled manually in ovs-vswitchd. OpenFlow |
42dccab5 BP |
1236 | 1.4 and 1.5 are also supported, with missing features, in Open |
1237 | vSwitch 2.3 and later, but not enabled by default. In any case, | |
1238 | the user may override the default: | |
75fa58f8 | 1239 | |
542cc9bb | 1240 | - To enable OpenFlow 1.0, 1.1, 1.2, and 1.3 on bridge br0: |
75fa58f8 | 1241 | |
542cc9bb | 1242 | ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13 |
6dc53744 | 1243 | |
542cc9bb | 1244 | - To enable OpenFlow 1.0, 1.1, 1.2, 1.3, 1.4, and 1.5 on bridge br0: |
aa233d57 | 1245 | |
542cc9bb | 1246 | ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13,OpenFlow14,OpenFlow15 |
aa233d57 | 1247 | |
542cc9bb | 1248 | - To enable only OpenFlow 1.0 on bridge br0: |
6dc53744 | 1249 | |
542cc9bb | 1250 | ovs-vsctl set bridge br0 protocols=OpenFlow10 |
6dc53744 BP |
1251 | |
1252 | All current versions of ovs-ofctl enable only OpenFlow 1.0 by | |
1253 | default. Use the -O option to enable support for later versions of | |
1254 | OpenFlow in ovs-ofctl. For example: | |
ac12b4cb BP |
1255 | |
1256 | ovs-ofctl -O OpenFlow13 dump-flows br0 | |
1257 | ||
aa233d57 BP |
1258 | (Open vSwitch 2.2 had an experimental implementation of OpenFlow |
1259 | 1.4 that could cause crashes. We don't recommend enabling it.) | |
ecb229be | 1260 | |
9feb1017 | 1261 | [OPENFLOW-1.1+.md] in the Open vSwitch source tree tracks support for |
42dccab5 BP |
1262 | OpenFlow 1.1 and later features. When support for OpenFlow 1.4 and |
1263 | 1.5 is solidly implemented, Open vSwitch will enable those version | |
1264 | by default. Also, the OpenFlow 1.5 specification is still under | |
1265 | development and thus subject to change. | |
7b287e99 | 1266 | |
542cc9bb | 1267 | ### Q: Does Open vSwitch support MPLS? |
c78a9ead BP |
1268 | |
1269 | A: Before version 1.11, Open vSwitch did not support MPLS. That is, | |
1270 | these versions can match on MPLS Ethernet types, but they cannot | |
1271 | match, push, or pop MPLS labels, nor can they look past MPLS labels | |
1272 | into the encapsulated packet. | |
1273 | ||
1274 | Open vSwitch versions 1.11, 2.0, and 2.1 have very minimal support | |
1275 | for MPLS. With the userspace datapath only, these versions can | |
1276 | match, push, or pop a single MPLS label, but they still cannot look | |
1277 | past MPLS labels (even after popping them) into the encapsulated | |
7e6410d2 BP |
1278 | packet. Kernel datapath support is unchanged from earlier |
1279 | versions. | |
c78a9ead | 1280 | |
b6fe204d BP |
1281 | Open vSwitch version 2.3 can match, push, or pop a single MPLS |
1282 | label and look past the MPLS label into the encapsulated packet. | |
1283 | Both userspace and kernel datapaths will be supported, but MPLS | |
1284 | processing always happens in userspace either way, so kernel | |
1285 | datapath performance will be disappointing. | |
1286 | ||
1287 | Open vSwitch version 2.4 can match, push, or pop up to 3 MPLS | |
1288 | labels and look past the MPLS label into the encapsulated packet. | |
1289 | It will have kernel support for MPLS, yielding improved | |
1290 | performance. | |
c78a9ead | 1291 | |
542cc9bb | 1292 | ### Q: I'm getting "error type 45250 code 0". What's that? |
c483d489 BP |
1293 | |
1294 | A: This is a Open vSwitch extension to OpenFlow error codes. Open | |
1295 | vSwitch uses this extension when it must report an error to an | |
1296 | OpenFlow controller but no standard OpenFlow error code is | |
1297 | suitable. | |
1298 | ||
1299 | Open vSwitch logs the errors that it sends to controllers, so the | |
1300 | easiest thing to do is probably to look at the ovs-vswitchd log to | |
1301 | find out what the error was. | |
1302 | ||
1303 | If you want to dissect the extended error message yourself, the | |
1304 | format is documented in include/openflow/nicira-ext.h in the Open | |
1305 | vSwitch source distribution. The extended error codes are | |
1306 | documented in lib/ofp-errors.h. | |
1307 | ||
1308 | Q1: Some of the traffic that I'd expect my OpenFlow controller to see | |
1309 | doesn't actually appear through the OpenFlow connection, even | |
1310 | though I know that it's going through. | |
1311 | Q2: Some of the OpenFlow flows that my controller sets up don't seem | |
1312 | to apply to certain traffic, especially traffic between OVS and | |
1313 | the controller itself. | |
1314 | ||
1315 | A: By default, Open vSwitch assumes that OpenFlow controllers are | |
1316 | connected "in-band", that is, that the controllers are actually | |
1317 | part of the network that is being controlled. In in-band mode, | |
1318 | Open vSwitch sets up special "hidden" flows to make sure that | |
1319 | traffic can make it back and forth between OVS and the controllers. | |
1320 | These hidden flows are higher priority than any flows that can be | |
1321 | set up through OpenFlow, and they are not visible through normal | |
1322 | OpenFlow flow table dumps. | |
1323 | ||
1324 | Usually, the hidden flows are desirable and helpful, but | |
1325 | occasionally they can cause unexpected behavior. You can view the | |
1326 | full OpenFlow flow table, including hidden flows, on bridge br0 | |
1327 | with the command: | |
1328 | ||
1329 | ovs-appctl bridge/dump-flows br0 | |
1330 | ||
1331 | to help you debug. The hidden flows are those with priorities | |
1332 | greater than 65535 (the maximum priority that can be set with | |
1333 | OpenFlow). | |
1334 | ||
1335 | The DESIGN file at the top level of the Open vSwitch source | |
1336 | distribution describes the in-band model in detail. | |
1337 | ||
1338 | If your controllers are not actually in-band (e.g. they are on | |
1339 | localhost via 127.0.0.1, or on a separate network), then you should | |
1340 | configure your controllers in "out-of-band" mode. If you have one | |
1341 | controller on bridge br0, then you can configure out-of-band mode | |
1342 | on it with: | |
1343 | ||
1344 | ovs-vsctl set controller br0 connection-mode=out-of-band | |
1345 | ||
542cc9bb | 1346 | ### Q: I configured all my controllers for out-of-band control mode but |
c483d489 BP |
1347 | "ovs-appctl bridge/dump-flows" still shows some hidden flows. |
1348 | ||
1349 | A: You probably have a remote manager configured (e.g. with "ovs-vsctl | |
1350 | set-manager"). By default, Open vSwitch assumes that managers need | |
1351 | in-band rules set up on every bridge. You can disable these rules | |
1352 | on bridge br0 with: | |
1353 | ||
1354 | ovs-vsctl set bridge br0 other-config:disable-in-band=true | |
1355 | ||
1356 | This actually disables in-band control entirely for the bridge, as | |
1357 | if all the bridge's controllers were configured for out-of-band | |
1358 | control. | |
1359 | ||
542cc9bb | 1360 | ### Q: My OpenFlow controller doesn't see the VLANs that I expect. |
c483d489 BP |
1361 | |
1362 | A: See answer under "VLANs", above. | |
1363 | ||
542cc9bb | 1364 | ### Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop" |
5cb2356b BP |
1365 | but I got a funny message like this: |
1366 | ||
1367 | ofp_util|INFO|normalization changed ofp_match, details: | |
1368 | ofp_util|INFO| pre: nw_dst=192.168.0.1 | |
1369 | ofp_util|INFO|post: | |
1370 | ||
1371 | and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst | |
1372 | match had disappeared, so that the flow ends up matching every | |
1373 | packet. | |
1374 | ||
1375 | A: The term "normalization" in the log message means that a flow | |
1376 | cannot match on an L3 field without saying what L3 protocol is in | |
1377 | use. The "ovs-ofctl" command above didn't specify an L3 protocol, | |
1378 | so the L3 field match was dropped. | |
1379 | ||
1380 | In this case, the L3 protocol could be IP or ARP. A correct | |
1381 | command for each possibility is, respectively: | |
1382 | ||
1383 | ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop | |
1384 | ||
1385 | and | |
1386 | ||
1387 | ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop | |
1388 | ||
1389 | Similarly, a flow cannot match on an L4 field without saying what | |
1390 | L4 protocol is in use. For example, the flow match "tp_src=1234" | |
1391 | is, by itself, meaningless and will be ignored. Instead, to match | |
1392 | TCP source port 1234, write "tcp,tp_src=1234", or to match UDP | |
1393 | source port 1234, write "udp,tp_src=1234". | |
1394 | ||
542cc9bb | 1395 | ### Q: How can I figure out the OpenFlow port number for a given port? |
c5b25863 BP |
1396 | |
1397 | A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to | |
1398 | respond with an OFPT_FEATURES_REPLY that, among other information, | |
1399 | includes a mapping between OpenFlow port names and numbers. From a | |
1400 | command prompt, "ovs-ofctl show br0" makes such a request and | |
1401 | prints the response for switch br0. | |
1402 | ||
1403 | The Interface table in the Open vSwitch database also maps OpenFlow | |
1404 | port names to numbers. To print the OpenFlow port number | |
1405 | associated with interface eth0, run: | |
1406 | ||
1407 | ovs-vsctl get Interface eth0 ofport | |
1408 | ||
1409 | You can print the entire mapping with: | |
1410 | ||
1411 | ovs-vsctl -- --columns=name,ofport list Interface | |
1412 | ||
1413 | but the output mixes together interfaces from all bridges in the | |
1414 | database, so it may be confusing if more than one bridge exists. | |
1415 | ||
1416 | In the Open vSwitch database, ofport value -1 means that the | |
1417 | interface could not be created due to an error. (The Open vSwitch | |
1418 | log should indicate the reason.) ofport value [] (the empty set) | |
1419 | means that the interface hasn't been created yet. The latter is | |
1420 | normally an intermittent condition (unless ovs-vswitchd is not | |
1421 | running). | |
7b287e99 | 1422 | |
542cc9bb | 1423 | ### Q: I added some flows with my controller or with ovs-ofctl, but when I |
af1ac4b9 BP |
1424 | run "ovs-dpctl dump-flows" I don't see them. |
1425 | ||
1426 | A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It | |
1427 | won't display the information that you want. You want to use | |
1428 | "ovs-ofctl dump-flows" instead. | |
1429 | ||
542cc9bb | 1430 | ### Q: It looks like each of the interfaces in my bonded port shows up |
15d63ed3 BP |
1431 | as an individual OpenFlow port. Is that right? |
1432 | ||
1433 | A: Yes, Open vSwitch makes individual bond interfaces visible as | |
1434 | OpenFlow ports, rather than the bond as a whole. The interfaces | |
1435 | are treated together as a bond for only a few purposes: | |
1436 | ||
542cc9bb TG |
1437 | - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow |
1438 | controller is not configured, this happens implicitly to | |
1439 | every packet.) | |
15d63ed3 | 1440 | |
542cc9bb | 1441 | - Mirrors configured for output to a bonded port. |
15d63ed3 BP |
1442 | |
1443 | It would make a lot of sense for Open vSwitch to present a bond as | |
1444 | a single OpenFlow port. If you want to contribute an | |
1445 | implementation of such a feature, please bring it up on the Open | |
1446 | vSwitch development mailing list at dev@openvswitch.org. | |
1447 | ||
542cc9bb | 1448 | ### Q: I have a sophisticated network setup involving Open vSwitch, VMs or |
bb955418 BP |
1449 | multiple hosts, and other components. The behavior isn't what I |
1450 | expect. Help! | |
1451 | ||
1452 | A: To debug network behavior problems, trace the path of a packet, | |
1453 | hop-by-hop, from its origin in one host to a remote host. If | |
1454 | that's correct, then trace the path of the response packet back to | |
1455 | the origin. | |
1456 | ||
1457 | Usually a simple ICMP echo request and reply ("ping") packet is | |
1458 | good enough. Start by initiating an ongoing "ping" from the origin | |
1459 | host to a remote host. If you are tracking down a connectivity | |
1460 | problem, the "ping" will not display any successful output, but | |
1461 | packets are still being sent. (In this case the packets being sent | |
1462 | are likely ARP rather than ICMP.) | |
1463 | ||
1464 | Tools available for tracing include the following: | |
1465 | ||
542cc9bb TG |
1466 | - "tcpdump" and "wireshark" for observing hops across network |
1467 | devices, such as Open vSwitch internal devices and physical | |
1468 | wires. | |
bb955418 | 1469 | |
542cc9bb TG |
1470 | - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and |
1471 | later or "ovs-dpctl dump-flows <br>" in earlier versions. | |
1472 | These tools allow one to observe the actions being taken on | |
1473 | packets in ongoing flows. | |
bb955418 | 1474 | |
542cc9bb TG |
1475 | See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows" |
1476 | documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows" | |
1477 | documentation, and "Why are there so many different ways to | |
1478 | dump flows?" above for some background. | |
bb955418 | 1479 | |
542cc9bb TG |
1480 | - "ovs-appctl ofproto/trace" to observe the logic behind how |
1481 | ovs-vswitchd treats packets. See ovs-vswitchd(8) for | |
1482 | documentation. You can out more details about a given flow | |
1483 | that "ovs-dpctl dump-flows" displays, by cutting and pasting | |
1484 | a flow from the output into an "ovs-appctl ofproto/trace" | |
1485 | command. | |
bb955418 | 1486 | |
542cc9bb TG |
1487 | - SPAN, RSPAN, and ERSPAN features of physical switches, to |
1488 | observe what goes on at these physical hops. | |
bb955418 BP |
1489 | |
1490 | Starting at the origin of a given packet, observe the packet at | |
1491 | each hop in turn. For example, in one plausible scenario, you | |
1492 | might: | |
1493 | ||
542cc9bb TG |
1494 | 1. "tcpdump" the "eth" interface through which an ARP egresses |
1495 | a VM, from inside the VM. | |
bb955418 | 1496 | |
542cc9bb TG |
1497 | 2. "tcpdump" the "vif" or "tap" interface through which the ARP |
1498 | ingresses the host machine. | |
bb955418 | 1499 | |
542cc9bb TG |
1500 | 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe |
1501 | the host interface through which the ARP egresses the | |
1502 | physical machine. You may need to use "ovs-dpctl show" to | |
1503 | interpret the port numbers. If the output seems surprising, | |
1504 | you can use "ovs-appctl ofproto/trace" to observe details of | |
1505 | how ovs-vswitchd determined the actions in the "ovs-dpctl | |
1506 | dump-flows" output. | |
bb955418 | 1507 | |
542cc9bb TG |
1508 | 4. "tcpdump" the "eth" interface through which the ARP egresses |
1509 | the physical machine. | |
bb955418 | 1510 | |
542cc9bb TG |
1511 | 5. "tcpdump" the "eth" interface through which the ARP |
1512 | ingresses the physical machine, at the remote host that | |
1513 | receives the ARP. | |
bb955418 | 1514 | |
542cc9bb TG |
1515 | 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the |
1516 | remote host that receives the ARP and observe the VM "vif" | |
1517 | or "tap" interface to which the flow is directed. Again, | |
1518 | "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help. | |
bb955418 | 1519 | |
542cc9bb TG |
1520 | 7. "tcpdump" the "vif" or "tap" interface to which the ARP is |
1521 | directed. | |
bb955418 | 1522 | |
542cc9bb TG |
1523 | 8. "tcpdump" the "eth" interface through which the ARP |
1524 | ingresses a VM, from inside the VM. | |
bb955418 BP |
1525 | |
1526 | It is likely that during one of these steps you will figure out the | |
1527 | problem. If not, then follow the ARP reply back to the origin, in | |
1528 | reverse. | |
1529 | ||
542cc9bb | 1530 | ### Q: How do I make a flow drop packets? |
0f5edef0 | 1531 | |
e5f1da19 BP |
1532 | A: To drop a packet is to receive it without forwarding it. OpenFlow |
1533 | explicitly specifies forwarding actions. Thus, a flow with an | |
1534 | empty set of actions does not forward packets anywhere, causing | |
1535 | them to be dropped. You can specify an empty set of actions with | |
1536 | "actions=" on the ovs-ofctl command line. For example: | |
0f5edef0 BP |
1537 | |
1538 | ovs-ofctl add-flow br0 priority=65535,actions= | |
1539 | ||
1540 | would cause every packet entering switch br0 to be dropped. | |
1541 | ||
1542 | You can write "drop" explicitly if you like. The effect is the | |
1543 | same. Thus, the following command also causes every packet | |
1544 | entering switch br0 to be dropped: | |
1545 | ||
1546 | ovs-ofctl add-flow br0 priority=65535,actions=drop | |
1547 | ||
e5f1da19 BP |
1548 | "drop" is not an action, either in OpenFlow or Open vSwitch. |
1549 | Rather, it is only a way to say that there are no actions. | |
1550 | ||
542cc9bb | 1551 | ### Q: I added a flow to send packets out the ingress port, like this: |
2fafc091 BP |
1552 | |
1553 | ovs-ofctl add-flow br0 in_port=2,actions=2 | |
1554 | ||
1555 | but OVS drops the packets instead. | |
1556 | ||
1557 | A: Yes, OpenFlow requires a switch to ignore attempts to send a packet | |
1558 | out its ingress port. The rationale is that dropping these packets | |
1559 | makes it harder to loop the network. Sometimes this behavior can | |
1560 | even be convenient, e.g. it is often the desired behavior in a flow | |
1561 | that forwards a packet to several ports ("floods" the packet). | |
1562 | ||
6620f928 JS |
1563 | Sometimes one really needs to send a packet out its ingress port |
1564 | ("hairpin"). In this case, output to OFPP_IN_PORT, which in | |
1565 | ovs-ofctl syntax is expressed as just "in_port", e.g.: | |
2fafc091 BP |
1566 | |
1567 | ovs-ofctl add-flow br0 in_port=2,actions=in_port | |
1568 | ||
1569 | This also works in some circumstances where the flow doesn't match | |
1570 | on the input port. For example, if you know that your switch has | |
1571 | five ports numbered 2 through 6, then the following will send every | |
1572 | received packet out every port, even its ingress port: | |
1573 | ||
1574 | ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port | |
1575 | ||
1576 | or, equivalently: | |
1577 | ||
1578 | ovs-ofctl add-flow br0 actions=all,in_port | |
1579 | ||
1580 | Sometimes, in complicated flow tables with multiple levels of | |
1581 | "resubmit" actions, a flow needs to output to a particular port | |
1582 | that may or may not be the ingress port. It's difficult to take | |
1583 | advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch | |
1584 | provides, as an OpenFlow extension, the ability to modify the | |
1585 | in_port field. Whatever value is currently in the in_port field is | |
1586 | the port to which outputs will be dropped, as well as the | |
1587 | destination for OFPP_IN_PORT. This means that the following will | |
1588 | reliably output to port 2 or to ports 2 through 6, respectively: | |
1589 | ||
1590 | ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2 | |
1591 | ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6 | |
1592 | ||
1593 | If the input port is important, then one may save and restore it on | |
1594 | the stack: | |
1595 | ||
1596 | ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\ | |
1597 | load:0->NXM_OF_IN_PORT[],\ | |
1598 | 2,3,4,5,6,\ | |
1599 | pop:NXM_OF_IN_PORT[] | |
1600 | ||
542cc9bb | 1601 | ### Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2 |
d4ee72b4 BP |
1602 | on port 2. I set up flows to forward only traffic destined to the |
1603 | other host and drop other traffic, like this: | |
1604 | ||
1605 | priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2 | |
1606 | priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1 | |
1607 | priority=0,actions=drop | |
1608 | ||
1609 | But it doesn't work--I don't get any connectivity when I do this. | |
1610 | Why? | |
1611 | ||
1612 | A: These flows drop the ARP packets that IP hosts use to establish IP | |
1613 | connectivity over Ethernet. To solve the problem, add flows to | |
1614 | allow ARP to pass between the hosts: | |
1615 | ||
1616 | priority=5,in_port=1,arp,actions=2 | |
1617 | priority=5,in_port=2,arp,actions=1 | |
1618 | ||
1619 | This issue can manifest other ways, too. The following flows that | |
1620 | match on Ethernet addresses instead of IP addresses will also drop | |
1621 | ARP packets, because ARP requests are broadcast instead of being | |
1622 | directed to a specific host: | |
1623 | ||
1624 | priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2 | |
1625 | priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1 | |
1626 | priority=0,actions=drop | |
1627 | ||
1628 | The solution already described above will also work in this case. | |
1629 | It may be better to add flows to allow all multicast and broadcast | |
1630 | traffic: | |
1631 | ||
1632 | priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2 | |
1633 | priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1 | |
0f5edef0 | 1634 | |
542cc9bb | 1635 | ### Q: My bridge disconnects from my controller on add-port/del-port. |
9a7a9154 YT |
1636 | |
1637 | A: Reconfiguring your bridge can change your bridge's datapath-id because | |
1638 | Open vSwitch generates datapath-id from the MAC address of one of its ports. | |
1639 | In that case, Open vSwitch disconnects from controllers because there's | |
1640 | no graceful way to notify controllers about the change of datapath-id. | |
1641 | ||
1642 | To avoid the behaviour, you can configure datapath-id manually. | |
1643 | ||
1644 | ovs-vsctl set bridge br0 other-config:datapath-id=0123456789abcdef | |
1645 | ||
ae99ee45 BP |
1646 | ### Q: My controller is getting errors about "buffers". What's going on? |
1647 | ||
1648 | A: When a switch sends a packet to an OpenFlow controller using a | |
1649 | "packet-in" message, it can also keep a copy of that packet in a | |
1650 | "buffer", identified by a 32-bit integer "buffer_id". There are | |
1651 | two advantages to buffering. First, when the controller wants to | |
1652 | tell the switch to do something with the buffered packet (with a | |
1653 | "packet-out" OpenFlow request), it does not need to send another | |
1654 | copy of the packet back across the OpenFlow connection, which | |
1655 | reduces the bandwidth cost of the connection and improves latency. | |
1656 | This enables the second advantage: the switch can optionally send | |
1657 | only the first part of the packet to the controller (assuming that | |
1658 | the switch only needs to look at the first few bytes of the | |
1659 | packet), further reducing bandwidth and improving latency. | |
1660 | ||
1661 | However, buffering introduces some issues of its own. First, any | |
1662 | switch has limited resources, so if the controller does not use a | |
1663 | buffered packet, the switch has to decide how long to keep it | |
1664 | buffered. When many packets are sent to a controller and buffered, | |
1665 | Open vSwitch can discard buffered packets that the controller has | |
1666 | not used after as little as 5 seconds. This means that | |
1667 | controllers, if they make use of packet buffering, should use the | |
1668 | buffered packets promptly. (This includes sending a "packet-out" | |
1669 | with no actions if the controller does not want to do anything with | |
1670 | a buffered packet, to clear the packet buffer and effectively | |
1671 | "drop" its packet.) | |
1672 | ||
1673 | Second, packet buffers are one-time-use, meaning that a controller | |
1674 | cannot use a single packet buffer in two or more "packet-out" | |
1675 | commands. Open vSwitch will respond with an error to the second | |
1676 | and subsequent "packet-out"s in such a case. | |
1677 | ||
1678 | Finally, a common error early in controller development is to try | |
1679 | to use buffer_id 0 in a "packet-out" message as if 0 represented | |
1680 | "no buffered packet". This is incorrect usage: the buffer_id with | |
1681 | this meaning is actually 0xffffffff. | |
1682 | ||
1683 | ovs-vswitchd(8) describes some details of Open vSwitch packet | |
1684 | buffering that the OpenFlow specification requires implementations | |
1685 | to document. | |
1686 | ||
66679738 BP |
1687 | |
1688 | Development | |
1689 | ----------- | |
1690 | ||
542cc9bb | 1691 | ### Q: How do I implement a new OpenFlow message? |
66679738 BP |
1692 | |
1693 | A: Add your new message to "enum ofpraw" and "enum ofptype" in | |
1694 | lib/ofp-msgs.h, following the existing pattern. Then recompile and | |
1695 | fix all of the new warnings, implementing new functionality for the | |
1696 | new message as needed. (If you configure with --enable-Werror, as | |
9feb1017 | 1697 | described in [INSTALL.md], then it is impossible to miss any warnings.) |
66679738 BP |
1698 | |
1699 | If you need to add an OpenFlow vendor extension message for a | |
1700 | vendor that doesn't yet have any extension messages, then you will | |
1701 | also need to edit build-aux/extract-ofp-msgs. | |
1702 | ||
fe540ca9 BP |
1703 | ### Q: How do I add support for a new field or header? |
1704 | ||
1705 | A: Add new members for your field to "struct flow" in lib/flow.h, and | |
1706 | add new enumerations for your new field to "enum mf_field_id" in | |
1707 | lib/meta-flow.h, following the existing pattern. Then recompile | |
1708 | and fix all of the new warnings, implementing new functionality for | |
1709 | the new field or header as needed. (If you configure with | |
1710 | --enable-Werror, as described in [INSTALL.md], then it is | |
1711 | impossible to miss any warnings.) | |
1712 | ||
1713 | ### Q: How do I add support for a new OpenFlow action? | |
1714 | ||
1715 | A: Add your new action to "enum ofp_raw_action_type" in | |
1716 | lib/ofp-actions.c, following the existing pattern. Then recompile | |
1717 | and fix all of the new warnings, implementing new functionality for | |
1718 | the new action as needed. (If you configure with --enable-Werror, | |
1719 | as described in [INSTALL.md], then it is impossible to miss any | |
1720 | warnings.) | |
1721 | ||
1722 | If you need to add an OpenFlow vendor extension action for a vendor | |
1723 | that doesn't yet have any extension actions, then you will also | |
1724 | need to edit build-aux/extract-ofp-actions. | |
1725 | ||
66679738 | 1726 | |
c483d489 BP |
1727 | Contact |
1728 | ------- | |
1729 | ||
1730 | bugs@openvswitch.org | |
1731 | http://openvswitch.org/ | |
9feb1017 TG |
1732 | |
1733 | [PORTING.md]:PORTING.md | |
1734 | [WHY-OVS.md]:WHY-OVS.md | |
1735 | [INSTALL.md]:INSTALL.md | |
1736 | [OPENFLOW-1.1+.md]:OPENFLOW-1.1+.md |