]>
Commit | Line | Data |
---|---|---|
1 | Open vSwitch <http://openvswitch.org> | |
2 | ||
3 | Frequently Asked Questions | |
4 | ========================== | |
5 | ||
6 | General | |
7 | ------- | |
8 | ||
9 | Q: What is Open vSwitch? | |
10 | ||
11 | A: Open vSwitch is a production quality open source software switch | |
12 | designed to be used as a vswitch in virtualized server | |
13 | environments. A vswitch forwards traffic between different VMs on | |
14 | the same physical host and also forwards traffic between VMs and | |
15 | the physical network. Open vSwitch supports standard management | |
16 | interfaces (e.g. sFlow, NetFlow, IPFIX, RSPAN, CLI), and is open to | |
17 | programmatic extension and control using OpenFlow and the OVSDB | |
18 | management protocol. | |
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 | ||
27 | Q: What virtualization platforms can use Open vSwitch? | |
28 | ||
29 | A: Open vSwitch can currently run on any Linux-based virtualization | |
30 | platform (kernel 2.6.32 and newer), including: KVM, VirtualBox, Xen, | |
31 | Xen Cloud Platform, XenServer. As of Linux 3.3 it is part of the | |
32 | mainline kernel. The bulk of the code is written in platform- | |
33 | independent C and is easily ported to other environments. We welcome | |
34 | inquires about integrating Open vSwitch with other virtualization | |
35 | platforms. | |
36 | ||
37 | Q: How can I try Open vSwitch? | |
38 | ||
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. | |
43 | ||
44 | You may also download and run a virtualization platform that already | |
45 | has Open vSwitch integrated. For example, download a recent ISO for | |
46 | XenServer or Xen Cloud Platform. Be aware that the version | |
47 | integrated with a particular platform may not be the most recent Open | |
48 | vSwitch release. | |
49 | ||
50 | Q: Does Open vSwitch only work on Linux? | |
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 | ||
62 | Q: What's involved with porting Open vSwitch to a new platform or | |
63 | switching ASIC? | |
64 | ||
65 | A: The PORTING document describes how one would go about porting Open | |
66 | vSwitch to a new operating system or hardware platform. | |
67 | ||
68 | Q: Why would I use Open vSwitch instead of the Linux bridge? | |
69 | ||
70 | A: Open vSwitch is specially designed to make it easier to manage VM | |
71 | network configuration and monitor state spread across many physical | |
72 | hosts in dynamic virtualized environments. Please see WHY-OVS for a | |
73 | more detailed description of how Open vSwitch relates to the Linux | |
74 | Bridge. | |
75 | ||
76 | Q: How is Open vSwitch related to distributed virtual switches like the | |
77 | VMware vNetwork distributed switch or the Cisco Nexus 1000V? | |
78 | ||
79 | A: Distributed vswitch applications (e.g., VMware vNetwork distributed | |
80 | switch, Cisco Nexus 1000V) provide a centralized way to configure and | |
81 | monitor the network state of VMs that are spread across many physical | |
82 | hosts. Open vSwitch is not a distributed vswitch itself, rather it | |
83 | runs on each physical host and supports remote management in a way | |
84 | that makes it easier for developers of virtualization/cloud | |
85 | management platforms to offer distributed vswitch capabilities. | |
86 | ||
87 | To aid in distribution, Open vSwitch provides two open protocols that | |
88 | are specially designed for remote management in virtualized network | |
89 | environments: OpenFlow, which exposes flow-based forwarding state, | |
90 | and the OVSDB management protocol, which exposes switch port state. | |
91 | In addition to the switch implementation itself, Open vSwitch | |
92 | includes tools (ovs-ofctl, ovs-vsctl) that developers can script and | |
93 | extend to provide distributed vswitch capabilities that are closely | |
94 | integrated with their virtualization management platform. | |
95 | ||
96 | Q: Why doesn't Open vSwitch support distribution? | |
97 | ||
98 | A: Open vSwitch is intended to be a useful component for building | |
99 | flexible network infrastructure. There are many different approaches | |
100 | to distribution which balance trade-offs between simplicity, | |
101 | scalability, hardware compatibility, convergence times, logical | |
102 | forwarding model, etc. The goal of Open vSwitch is to be able to | |
103 | support all as a primitive building block rather than choose a | |
104 | particular point in the distributed design space. | |
105 | ||
106 | Q: How can I contribute to the Open vSwitch Community? | |
107 | ||
108 | A: You can start by joining the mailing lists and helping to answer | |
109 | questions. You can also suggest improvements to documentation. If | |
110 | you have a feature or bug you would like to work on, send a mail to | |
111 | one of the mailing lists: | |
112 | ||
113 | http://openvswitch.org/mlists/ | |
114 | ||
115 | ||
116 | Releases | |
117 | -------- | |
118 | ||
119 | Q: What does it mean for an Open vSwitch release to be LTS (long-term | |
120 | support)? | |
121 | ||
122 | A: All official releases have been through a comprehensive testing | |
123 | process and are suitable for production use. Planned releases will | |
124 | occur several times a year. If a significant bug is identified in an | |
125 | LTS release, we will provide an updated release that includes the | |
126 | fix. Releases that are not LTS may not be fixed and may just be | |
127 | supplanted by the next major release. The current LTS release is | |
128 | 1.9.x. | |
129 | ||
130 | Q: What Linux kernel versions does each Open vSwitch release work with? | |
131 | ||
132 | A: The following table lists the Linux kernel versions against which the | |
133 | given versions of the Open vSwitch kernel module will successfully | |
134 | build. The Linux kernel versions are upstream kernel versions, so | |
135 | Linux kernels modified from the upstream sources may not build in | |
136 | some cases even if they are based on a supported version. This is | |
137 | most notably true of Red Hat Enterprise Linux (RHEL) kernels, which | |
138 | are extensively modified from upstream. | |
139 | ||
140 | Open vSwitch Linux kernel | |
141 | ------------ ------------- | |
142 | 1.4.x 2.6.18 to 3.2 | |
143 | 1.5.x 2.6.18 to 3.2 | |
144 | 1.6.x 2.6.18 to 3.2 | |
145 | 1.7.x 2.6.18 to 3.3 | |
146 | 1.8.x 2.6.18 to 3.4 | |
147 | 1.9.x 2.6.18 to 3.8 | |
148 | 1.10.x 2.6.18 to 3.8 | |
149 | 1.11.x 2.6.18 to 3.8 | |
150 | 2.0.x 2.6.32 to 3.10 | |
151 | 2.1.x 2.6.32 to 3.11 | |
152 | ||
153 | Open vSwitch userspace should also work with the Linux kernel module | |
154 | built into Linux 3.3 and later. | |
155 | ||
156 | Open vSwitch userspace is not sensitive to the Linux kernel version. | |
157 | It should build against almost any kernel, certainly against 2.6.32 | |
158 | and later. | |
159 | ||
160 | Q: What Linux kernel versions does IPFIX flow monitoring work with? | |
161 | ||
162 | A: IPFIX flow monitoring requires the Linux kernel module from Open | |
163 | vSwitch version 1.10.90 or later. | |
164 | ||
165 | Q: Should userspace or kernel be upgraded first to minimize downtime? | |
166 | ||
167 | In general, the Open vSwitch userspace should be used with the | |
168 | kernel version included in the same release or with the version | |
169 | from upstream Linux. However, when upgrading between two releases | |
170 | of Open vSwitch it is best to migrate userspace first to reduce | |
171 | the possibility of incompatibilities. | |
172 | ||
173 | Q: What features are not available in the Open vSwitch kernel datapath | |
174 | that ships as part of the upstream Linux kernel? | |
175 | ||
176 | A: The kernel module in upstream Linux 3.3 and later does not include | |
177 | tunnel virtual ports, that is, interfaces with type "gre", | |
178 | "ipsec_gre", "gre64", "ipsec_gre64", "vxlan", or "lisp". It is | |
179 | possible to create tunnels in Linux and attach them to Open vSwitch | |
180 | as system devices. However, they cannot be dynamically created | |
181 | through the OVSDB protocol or set the tunnel ids as a flow action. | |
182 | ||
183 | Work is in progress in adding tunnel virtual ports to the upstream | |
184 | Linux version of the Open vSwitch kernel module. For now, if you | |
185 | need these features, use the kernel module from the Open vSwitch | |
186 | distribution instead of the upstream Linux kernel module. | |
187 | ||
188 | The upstream kernel module does not include patch ports, but this | |
189 | only matters for Open vSwitch 1.9 and earlier, because Open vSwitch | |
190 | 1.10 and later implement patch ports without using this kernel | |
191 | feature. | |
192 | ||
193 | Q: What features are not available when using the userspace datapath? | |
194 | ||
195 | A: Tunnel virtual ports are not supported, as described in the | |
196 | previous answer. It is also not possible to use queue-related | |
197 | actions. On Linux kernels before 2.6.39, maximum-sized VLAN packets | |
198 | may not be transmitted. | |
199 | ||
200 | Q: What happened to the bridge compatibility feature? | |
201 | ||
202 | A: Bridge compatibility was a feature of Open vSwitch 1.9 and earlier. | |
203 | When it was enabled, Open vSwitch imitated the interface of the | |
204 | Linux kernel "bridge" module. This allowed users to drop Open | |
205 | vSwitch into environments designed to use the Linux kernel bridge | |
206 | module without adapting the environment to use Open vSwitch. | |
207 | ||
208 | Open vSwitch 1.10 and later do not support bridge compatibility. | |
209 | The feature was dropped because version 1.10 adopted a new internal | |
210 | architecture that made bridge compatibility difficult to maintain. | |
211 | Now that many environments use OVS directly, it would be rarely | |
212 | useful in any case. | |
213 | ||
214 | To use bridge compatibility, install OVS 1.9 or earlier, including | |
215 | the accompanying kernel modules (both the main and bridge | |
216 | compatibility modules), following the instructions that come with | |
217 | the release. Be sure to start the ovs-brcompatd daemon. | |
218 | ||
219 | ||
220 | Terminology | |
221 | ----------- | |
222 | ||
223 | Q: I thought Open vSwitch was a virtual Ethernet switch, but the | |
224 | documentation keeps talking about bridges. What's a bridge? | |
225 | ||
226 | A: In networking, the terms "bridge" and "switch" are synonyms. Open | |
227 | vSwitch implements an Ethernet switch, which means that it is also | |
228 | an Ethernet bridge. | |
229 | ||
230 | Q: What's a VLAN? | |
231 | ||
232 | A: See the "VLAN" section below. | |
233 | ||
234 | ||
235 | Basic Configuration | |
236 | ------------------- | |
237 | ||
238 | Q: How do I configure a port as an access port? | |
239 | ||
240 | A: Add "tag=VLAN" to your "ovs-vsctl add-port" command. For example, | |
241 | the following commands configure br0 with eth0 as a trunk port (the | |
242 | default) and tap0 as an access port for VLAN 9: | |
243 | ||
244 | ovs-vsctl add-br br0 | |
245 | ovs-vsctl add-port br0 eth0 | |
246 | ovs-vsctl add-port br0 tap0 tag=9 | |
247 | ||
248 | If you want to configure an already added port as an access port, | |
249 | use "ovs-vsctl set", e.g.: | |
250 | ||
251 | ovs-vsctl set port tap0 tag=9 | |
252 | ||
253 | Q: How do I configure a port as a SPAN port, that is, enable mirroring | |
254 | of all traffic to that port? | |
255 | ||
256 | A: The following commands configure br0 with eth0 and tap0 as trunk | |
257 | ports. All traffic coming in or going out on eth0 or tap0 is also | |
258 | mirrored to tap1; any traffic arriving on tap1 is dropped: | |
259 | ||
260 | ovs-vsctl add-br br0 | |
261 | ovs-vsctl add-port br0 eth0 | |
262 | ovs-vsctl add-port br0 tap0 | |
263 | ovs-vsctl add-port br0 tap1 \ | |
264 | -- --id=@p get port tap1 \ | |
265 | -- --id=@m create mirror name=m0 select-all=true output-port=@p \ | |
266 | -- set bridge br0 mirrors=@m | |
267 | ||
268 | To later disable mirroring, run: | |
269 | ||
270 | ovs-vsctl clear bridge br0 mirrors | |
271 | ||
272 | Q: Does Open vSwitch support configuring a port in promiscuous mode? | |
273 | ||
274 | A: Yes. How you configure it depends on what you mean by "promiscuous | |
275 | mode": | |
276 | ||
277 | - Conventionally, "promiscuous mode" is a feature of a network | |
278 | interface card. Ordinarily, a NIC passes to the CPU only the | |
279 | packets actually destined to its host machine. It discards | |
280 | the rest to avoid wasting memory and CPU cycles. When | |
281 | promiscuous mode is enabled, however, it passes every packet | |
282 | to the CPU. On an old-style shared-media or hub-based | |
283 | network, this allows the host to spy on all packets on the | |
284 | network. But in the switched networks that are almost | |
285 | everywhere these days, promiscuous mode doesn't have much | |
286 | effect, because few packets not destined to a host are | |
287 | delivered to the host's NIC. | |
288 | ||
289 | This form of promiscuous mode is configured in the guest OS of | |
290 | the VMs on your bridge, e.g. with "ifconfig". | |
291 | ||
292 | - The VMware vSwitch uses a different definition of "promiscuous | |
293 | mode". When you configure promiscuous mode on a VMware vNIC, | |
294 | the vSwitch sends a copy of every packet received by the | |
295 | vSwitch to that vNIC. That has a much bigger effect than just | |
296 | enabling promiscuous mode in a guest OS. Rather than getting | |
297 | a few stray packets for which the switch does not yet know the | |
298 | correct destination, the vNIC gets every packet. The effect | |
299 | is similar to replacing the vSwitch by a virtual hub. | |
300 | ||
301 | This "promiscuous mode" is what switches normally call "port | |
302 | mirroring" or "SPAN". For information on how to configure | |
303 | SPAN, see "How do I configure a port as a SPAN port, that is, | |
304 | enable mirroring of all traffic to that port?" | |
305 | ||
306 | Q: How do I configure a VLAN as an RSPAN VLAN, that is, enable | |
307 | mirroring of all traffic to that VLAN? | |
308 | ||
309 | A: The following commands configure br0 with eth0 as a trunk port and | |
310 | tap0 as an access port for VLAN 10. All traffic coming in or going | |
311 | out on tap0, as well as traffic coming in or going out on eth0 in | |
312 | VLAN 10, is also mirrored to VLAN 15 on eth0. The original tag for | |
313 | VLAN 10, in cases where one is present, is dropped as part of | |
314 | mirroring: | |
315 | ||
316 | ovs-vsctl add-br br0 | |
317 | ovs-vsctl add-port br0 eth0 | |
318 | ovs-vsctl add-port br0 tap0 tag=10 | |
319 | ovs-vsctl \ | |
320 | -- --id=@m create mirror name=m0 select-all=true select-vlan=10 \ | |
321 | output-vlan=15 \ | |
322 | -- set bridge br0 mirrors=@m | |
323 | ||
324 | To later disable mirroring, run: | |
325 | ||
326 | ovs-vsctl clear bridge br0 mirrors | |
327 | ||
328 | Mirroring to a VLAN can disrupt a network that contains unmanaged | |
329 | switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a | |
330 | GRE tunnel has fewer caveats than mirroring to a VLAN and should | |
331 | generally be preferred. | |
332 | ||
333 | Q: Can I mirror more than one input VLAN to an RSPAN VLAN? | |
334 | ||
335 | A: Yes, but mirroring to a VLAN strips the original VLAN tag in favor | |
336 | of the specified output-vlan. This loss of information may make | |
337 | the mirrored traffic too hard to interpret. | |
338 | ||
339 | To mirror multiple VLANs, use the commands above, but specify a | |
340 | comma-separated list of VLANs as the value for select-vlan. To | |
341 | mirror every VLAN, use the commands above, but omit select-vlan and | |
342 | its value entirely. | |
343 | ||
344 | When a packet arrives on a VLAN that is used as a mirror output | |
345 | VLAN, the mirror is disregarded. Instead, in standalone mode, OVS | |
346 | floods the packet across all the ports for which the mirror output | |
347 | VLAN is configured. (If an OpenFlow controller is in use, then it | |
348 | can override this behavior through the flow table.) If OVS is used | |
349 | as an intermediate switch, rather than an edge switch, this ensures | |
350 | that the RSPAN traffic is distributed through the network. | |
351 | ||
352 | Mirroring to a VLAN can disrupt a network that contains unmanaged | |
353 | switches. See ovs-vswitchd.conf.db(5) for details. Mirroring to a | |
354 | GRE tunnel has fewer caveats than mirroring to a VLAN and should | |
355 | generally be preferred. | |
356 | ||
357 | Q: How do I configure mirroring of all traffic to a GRE tunnel? | |
358 | ||
359 | A: The following commands configure br0 with eth0 and tap0 as trunk | |
360 | ports. All traffic coming in or going out on eth0 or tap0 is also | |
361 | mirrored to gre0, a GRE tunnel to the remote host 192.168.1.10; any | |
362 | traffic arriving on gre0 is dropped: | |
363 | ||
364 | ovs-vsctl add-br br0 | |
365 | ovs-vsctl add-port br0 eth0 | |
366 | ovs-vsctl add-port br0 tap0 | |
367 | ovs-vsctl add-port br0 gre0 \ | |
368 | -- set interface gre0 type=gre options:remote_ip=192.168.1.10 \ | |
369 | -- --id=@p get port gre0 \ | |
370 | -- --id=@m create mirror name=m0 select-all=true output-port=@p \ | |
371 | -- set bridge br0 mirrors=@m | |
372 | ||
373 | To later disable mirroring and destroy the GRE tunnel: | |
374 | ||
375 | ovs-vsctl clear bridge br0 mirrors | |
376 | ovs-vcstl del-port br0 gre0 | |
377 | ||
378 | Q: Does Open vSwitch support ERSPAN? | |
379 | ||
380 | A: No. ERSPAN is an undocumented proprietary protocol. As an | |
381 | alternative, Open vSwitch supports mirroring to a GRE tunnel (see | |
382 | above). | |
383 | ||
384 | Q: How do I connect two bridges? | |
385 | ||
386 | A: First, why do you want to do this? Two connected bridges are not | |
387 | much different from a single bridge, so you might as well just have | |
388 | a single bridge with all your ports on it. | |
389 | ||
390 | If you still want to connect two bridges, you can use a pair of | |
391 | patch ports. The following example creates bridges br0 and br1, | |
392 | adds eth0 and tap0 to br0, adds tap1 to br1, and then connects br0 | |
393 | and br1 with a pair of patch ports. | |
394 | ||
395 | ovs-vsctl add-br br0 | |
396 | ovs-vsctl add-port br0 eth0 | |
397 | ovs-vsctl add-port br0 tap0 | |
398 | ovs-vsctl add-br br1 | |
399 | ovs-vsctl add-port br1 tap1 | |
400 | ovs-vsctl \ | |
401 | -- add-port br0 patch0 \ | |
402 | -- set interface patch0 type=patch options:peer=patch1 \ | |
403 | -- add-port br1 patch1 \ | |
404 | -- set interface patch1 type=patch options:peer=patch0 | |
405 | ||
406 | Bridges connected with patch ports are much like a single bridge. | |
407 | For instance, if the example above also added eth1 to br1, and both | |
408 | eth0 and eth1 happened to be connected to the same next-hop switch, | |
409 | then you could loop your network just as you would if you added | |
410 | eth0 and eth1 to the same bridge (see the "Configuration Problems" | |
411 | section below for more information). | |
412 | ||
413 | If you are using Open vSwitch 1.9 or an earlier version, then you | |
414 | need to be using the kernel module bundled with Open vSwitch rather | |
415 | than the one that is integrated into Linux 3.3 and later, because | |
416 | Open vSwitch 1.9 and earlier versions need kernel support for patch | |
417 | ports. This also means that in Open vSwitch 1.9 and earlier, patch | |
418 | ports will not work with the userspace datapath, only with the | |
419 | kernel module. | |
420 | ||
421 | ||
422 | Implementation Details | |
423 | ---------------------- | |
424 | ||
425 | Q: I hear OVS has a couple of kinds of flows. Can you tell me about them? | |
426 | ||
427 | A: Open vSwitch uses different kinds of flows for different purposes: | |
428 | ||
429 | - OpenFlow flows are the most important kind of flow. OpenFlow | |
430 | controllers use these flows to define a switch's policy. | |
431 | OpenFlow flows support wildcards, priorities, and multiple | |
432 | tables. | |
433 | ||
434 | When in-band control is in use, Open vSwitch sets up a few | |
435 | "hidden" flows, with priority higher than a controller or the | |
436 | user can configure, that are not visible via OpenFlow. (See | |
437 | the "Controller" section of the FAQ for more information | |
438 | about hidden flows.) | |
439 | ||
440 | - The Open vSwitch software switch implementation uses a second | |
441 | kind of flow internally. These flows, called "datapath" or | |
442 | "kernel" flows, do not support priorities and comprise only a | |
443 | single table, which makes them suitable for caching. (Like | |
444 | OpenFlow flows, datapath flows do support wildcarding, in Open | |
445 | vSwitch 1.11 and later.) OpenFlow flows and datapath flows | |
446 | also support different actions and number ports differently. | |
447 | ||
448 | Datapath flows are an implementation detail that is subject to | |
449 | change in future versions of Open vSwitch. Even with the | |
450 | current version of Open vSwitch, hardware switch | |
451 | implementations do not necessarily use this architecture. | |
452 | ||
453 | Users and controllers directly control only the OpenFlow flow | |
454 | table. Open vSwitch manages the datapath flow table itself, so | |
455 | users should not normally be concerned with it. | |
456 | ||
457 | Q: Why are there so many different ways to dump flows? | |
458 | ||
459 | A: Open vSwitch has two kinds of flows (see the previous question), so | |
460 | it has commands with different purposes for dumping each kind of | |
461 | flow: | |
462 | ||
463 | - "ovs-ofctl dump-flows <br>" dumps OpenFlow flows, excluding | |
464 | hidden flows. This is the most commonly useful form of flow | |
465 | dump. (Unlike the other commands, this should work with any | |
466 | OpenFlow switch, not just Open vSwitch.) | |
467 | ||
468 | - "ovs-appctl bridge/dump-flows <br>" dumps OpenFlow flows, | |
469 | including hidden flows. This is occasionally useful for | |
470 | troubleshooting suspected issues with in-band control. | |
471 | ||
472 | - "ovs-dpctl dump-flows [dp]" dumps the datapath flow table | |
473 | entries for a Linux kernel-based datapath. In Open vSwitch | |
474 | 1.10 and later, ovs-vswitchd merges multiple switches into a | |
475 | single datapath, so it will show all the flows on all your | |
476 | kernel-based switches. This command can occasionally be | |
477 | useful for debugging. | |
478 | ||
479 | - "ovs-appctl dpif/dump-flows <br>", new in Open vSwitch 1.10, | |
480 | dumps datapath flows for only the specified bridge, regardless | |
481 | of the type. | |
482 | ||
483 | ||
484 | Performance | |
485 | ----------- | |
486 | ||
487 | Q: I just upgraded and I see a performance drop. Why? | |
488 | ||
489 | A: The OVS kernel datapath may have been updated to a newer version than | |
490 | the OVS userspace components. Sometimes new versions of OVS kernel | |
491 | module add functionality that is backwards compatible with older | |
492 | userspace components but may cause a drop in performance with them. | |
493 | Especially, if a kernel module from OVS 2.1 or newer is paired with | |
494 | OVS userspace 1.10 or older, there will be a performance drop for | |
495 | TCP traffic. | |
496 | ||
497 | Updating the OVS userspace components to the latest released | |
498 | version should fix the performance degradation. | |
499 | ||
500 | To get the best possible performance and functionality, it is | |
501 | recommended to pair the same versions of the kernel module and OVS | |
502 | userspace. | |
503 | ||
504 | ||
505 | Configuration Problems | |
506 | ---------------------- | |
507 | ||
508 | Q: I created a bridge and added my Ethernet port to it, using commands | |
509 | like these: | |
510 | ||
511 | ovs-vsctl add-br br0 | |
512 | ovs-vsctl add-port br0 eth0 | |
513 | ||
514 | and as soon as I ran the "add-port" command I lost all connectivity | |
515 | through eth0. Help! | |
516 | ||
517 | A: A physical Ethernet device that is part of an Open vSwitch bridge | |
518 | should not have an IP address. If one does, then that IP address | |
519 | will not be fully functional. | |
520 | ||
521 | You can restore functionality by moving the IP address to an Open | |
522 | vSwitch "internal" device, such as the network device named after | |
523 | the bridge itself. For example, assuming that eth0's IP address is | |
524 | 192.168.128.5, you could run the commands below to fix up the | |
525 | situation: | |
526 | ||
527 | ifconfig eth0 0.0.0.0 | |
528 | ifconfig br0 192.168.128.5 | |
529 | ||
530 | (If your only connection to the machine running OVS is through the | |
531 | IP address in question, then you would want to run all of these | |
532 | commands on a single command line, or put them into a script.) If | |
533 | there were any additional routes assigned to eth0, then you would | |
534 | also want to use commands to adjust these routes to go through br0. | |
535 | ||
536 | If you use DHCP to obtain an IP address, then you should kill the | |
537 | DHCP client that was listening on the physical Ethernet interface | |
538 | (e.g. eth0) and start one listening on the internal interface | |
539 | (e.g. br0). You might still need to manually clear the IP address | |
540 | from the physical interface (e.g. with "ifconfig eth0 0.0.0.0"). | |
541 | ||
542 | There is no compelling reason why Open vSwitch must work this way. | |
543 | However, this is the way that the Linux kernel bridge module has | |
544 | always worked, so it's a model that those accustomed to Linux | |
545 | bridging are already used to. Also, the model that most people | |
546 | expect is not implementable without kernel changes on all the | |
547 | versions of Linux that Open vSwitch supports. | |
548 | ||
549 | By the way, this issue is not specific to physical Ethernet | |
550 | devices. It applies to all network devices except Open vswitch | |
551 | "internal" devices. | |
552 | ||
553 | Q: I created a bridge and added a couple of Ethernet ports to it, | |
554 | using commands like these: | |
555 | ||
556 | ovs-vsctl add-br br0 | |
557 | ovs-vsctl add-port br0 eth0 | |
558 | ovs-vsctl add-port br0 eth1 | |
559 | ||
560 | and now my network seems to have melted: connectivity is unreliable | |
561 | (even connectivity that doesn't go through Open vSwitch), all the | |
562 | LEDs on my physical switches are blinking, wireshark shows | |
563 | duplicated packets, and CPU usage is very high. | |
564 | ||
565 | A: More than likely, you've looped your network. Probably, eth0 and | |
566 | eth1 are connected to the same physical Ethernet switch. This | |
567 | yields a scenario where OVS receives a broadcast packet on eth0 and | |
568 | sends it out on eth1, then the physical switch connected to eth1 | |
569 | sends the packet back on eth0, and so on forever. More complicated | |
570 | scenarios, involving a loop through multiple switches, are possible | |
571 | too. | |
572 | ||
573 | The solution depends on what you are trying to do: | |
574 | ||
575 | - If you added eth0 and eth1 to get higher bandwidth or higher | |
576 | reliability between OVS and your physical Ethernet switch, | |
577 | use a bond. The following commands create br0 and then add | |
578 | eth0 and eth1 as a bond: | |
579 | ||
580 | ovs-vsctl add-br br0 | |
581 | ovs-vsctl add-bond br0 bond0 eth0 eth1 | |
582 | ||
583 | Bonds have tons of configuration options. Please read the | |
584 | documentation on the Port table in ovs-vswitchd.conf.db(5) | |
585 | for all the details. | |
586 | ||
587 | - Perhaps you don't actually need eth0 and eth1 to be on the | |
588 | same bridge. For example, if you simply want to be able to | |
589 | connect each of them to virtual machines, then you can put | |
590 | each of them on a bridge of its own: | |
591 | ||
592 | ovs-vsctl add-br br0 | |
593 | ovs-vsctl add-port br0 eth0 | |
594 | ||
595 | ovs-vsctl add-br br1 | |
596 | ovs-vsctl add-port br1 eth1 | |
597 | ||
598 | and then connect VMs to br0 and br1. (A potential | |
599 | disadvantage is that traffic cannot directly pass between br0 | |
600 | and br1. Instead, it will go out eth0 and come back in eth1, | |
601 | or vice versa.) | |
602 | ||
603 | - If you have a redundant or complex network topology and you | |
604 | want to prevent loops, turn on spanning tree protocol (STP). | |
605 | The following commands create br0, enable STP, and add eth0 | |
606 | and eth1 to the bridge. The order is important because you | |
607 | don't want have to have a loop in your network even | |
608 | transiently: | |
609 | ||
610 | ovs-vsctl add-br br0 | |
611 | ovs-vsctl set bridge br0 stp_enable=true | |
612 | ovs-vsctl add-port br0 eth0 | |
613 | ovs-vsctl add-port br0 eth1 | |
614 | ||
615 | The Open vSwitch implementation of STP is not well tested. | |
616 | Please report any bugs you observe, but if you'd rather avoid | |
617 | acting as a beta tester then another option might be your | |
618 | best shot. | |
619 | ||
620 | Q: I can't seem to use Open vSwitch in a wireless network. | |
621 | ||
622 | A: Wireless base stations generally only allow packets with the source | |
623 | MAC address of NIC that completed the initial handshake. | |
624 | Therefore, without MAC rewriting, only a single device can | |
625 | communicate over a single wireless link. | |
626 | ||
627 | This isn't specific to Open vSwitch, it's enforced by the access | |
628 | point, so the same problems will show up with the Linux bridge or | |
629 | any other way to do bridging. | |
630 | ||
631 | Q: I can't seem to add my PPP interface to an Open vSwitch bridge. | |
632 | ||
633 | A: PPP most commonly carries IP packets, but Open vSwitch works only | |
634 | with Ethernet frames. The correct way to interface PPP to an | |
635 | Ethernet network is usually to use routing instead of switching. | |
636 | ||
637 | Q: Is there any documentation on the database tables and fields? | |
638 | ||
639 | A: Yes. ovs-vswitchd.conf.db(5) is a comprehensive reference. | |
640 | ||
641 | Q: When I run ovs-dpctl I no longer see the bridges I created. Instead, | |
642 | I only see a datapath called "ovs-system". How can I see datapath | |
643 | information about a particular bridge? | |
644 | ||
645 | A: In version 1.9.0, OVS switched to using a single datapath that is | |
646 | shared by all bridges of that type. The "ovs-appctl dpif/*" | |
647 | commands provide similar functionality that is scoped by the bridge. | |
648 | ||
649 | ||
650 | Quality of Service (QoS) | |
651 | ------------------------ | |
652 | ||
653 | Q: How do I configure Quality of Service (QoS)? | |
654 | ||
655 | A: Suppose that you want to set up bridge br0 connected to physical | |
656 | Ethernet port eth0 (a 1 Gbps device) and virtual machine interfaces | |
657 | vif1.0 and vif2.0, and that you want to limit traffic from vif1.0 | |
658 | to eth0 to 10 Mbps and from vif2.0 to eth0 to 20 Mbps. Then, you | |
659 | could configure the bridge this way: | |
660 | ||
661 | ovs-vsctl -- \ | |
662 | add-br br0 -- \ | |
663 | add-port br0 eth0 -- \ | |
664 | add-port br0 vif1.0 -- set interface vif1.0 ofport_request=5 -- \ | |
665 | add-port br0 vif2.0 -- set interface vif2.0 ofport_request=6 -- \ | |
666 | set port eth0 qos=@newqos -- \ | |
667 | --id=@newqos create qos type=linux-htb \ | |
668 | other-config:max-rate=1000000000 \ | |
669 | queues:123=@vif10queue \ | |
670 | queues:234=@vif20queue -- \ | |
671 | --id=@vif10queue create queue other-config:max-rate=10000000 -- \ | |
672 | --id=@vif20queue create queue other-config:max-rate=20000000 | |
673 | ||
674 | At this point, bridge br0 is configured with the ports and eth0 is | |
675 | configured with the queues that you need for QoS, but nothing is | |
676 | actually directing packets from vif1.0 or vif2.0 to the queues that | |
677 | we have set up for them. That means that all of the packets to | |
678 | eth0 are going to the "default queue", which is not what we want. | |
679 | ||
680 | We use OpenFlow to direct packets from vif1.0 and vif2.0 to the | |
681 | queues reserved for them: | |
682 | ||
683 | ovs-ofctl add-flow br0 in_port=5,actions=set_queue:123,normal | |
684 | ovs-ofctl add-flow br0 in_port=6,actions=set_queue:234,normal | |
685 | ||
686 | Each of the above flows matches on the input port, sets up the | |
687 | appropriate queue (123 for vif1.0, 234 for vif2.0), and then | |
688 | executes the "normal" action, which performs the same switching | |
689 | that Open vSwitch would have done without any OpenFlow flows being | |
690 | present. (We know that vif1.0 and vif2.0 have OpenFlow port | |
691 | numbers 5 and 6, respectively, because we set their ofport_request | |
692 | columns above. If we had not done that, then we would have needed | |
693 | to find out their port numbers before setting up these flows.) | |
694 | ||
695 | Now traffic going from vif1.0 or vif2.0 to eth0 should be | |
696 | rate-limited. | |
697 | ||
698 | By the way, if you delete the bridge created by the above commands, | |
699 | with: | |
700 | ||
701 | ovs-vsctl del-br br0 | |
702 | ||
703 | then that will leave one unreferenced QoS record and two | |
704 | unreferenced Queue records in the Open vSwich database. One way to | |
705 | clear them out, assuming you don't have other QoS or Queue records | |
706 | that you want to keep, is: | |
707 | ||
708 | ovs-vsctl -- --all destroy QoS -- --all destroy Queue | |
709 | ||
710 | If you do want to keep some QoS or Queue records, or the Open | |
711 | vSwitch you are using is older than version 1.8 (which added the | |
712 | --all option), then you will have to destroy QoS and Queue records | |
713 | individually. | |
714 | ||
715 | Q: I configured Quality of Service (QoS) in my OpenFlow network by | |
716 | adding records to the QoS and Queue table, but the results aren't | |
717 | what I expect. | |
718 | ||
719 | A: Did you install OpenFlow flows that use your queues? This is the | |
720 | primary way to tell Open vSwitch which queues you want to use. If | |
721 | you don't do this, then the default queue will be used, which will | |
722 | probably not have the effect you want. | |
723 | ||
724 | Refer to the previous question for an example. | |
725 | ||
726 | Q: I configured QoS, correctly, but my measurements show that it isn't | |
727 | working as well as I expect. | |
728 | ||
729 | A: With the Linux kernel, the Open vSwitch implementation of QoS has | |
730 | two aspects: | |
731 | ||
732 | - Open vSwitch configures a subset of Linux kernel QoS | |
733 | features, according to what is in OVSDB. It is possible that | |
734 | this code has bugs. If you believe that this is so, then you | |
735 | can configure the Linux traffic control (QoS) stack directly | |
736 | with the "tc" program. If you get better results that way, | |
737 | you can send a detailed bug report to bugs@openvswitch.org. | |
738 | ||
739 | It is certain that Open vSwitch cannot configure every Linux | |
740 | kernel QoS feature. If you need some feature that OVS cannot | |
741 | configure, then you can also use "tc" directly (or add that | |
742 | feature to OVS). | |
743 | ||
744 | - The Open vSwitch implementation of OpenFlow allows flows to | |
745 | be directed to particular queues. This is pretty simple and | |
746 | unlikely to have serious bugs at this point. | |
747 | ||
748 | However, most problems with QoS on Linux are not bugs in Open | |
749 | vSwitch at all. They tend to be either configuration errors | |
750 | (please see the earlier questions in this section) or issues with | |
751 | the traffic control (QoS) stack in Linux. The Open vSwitch | |
752 | developers are not experts on Linux traffic control. We suggest | |
753 | that, if you believe you are encountering a problem with Linux | |
754 | traffic control, that you consult the tc manpages (e.g. tc(8), | |
755 | tc-htb(8), tc-hfsc(8)), web resources (e.g. http://lartc.org/), or | |
756 | mailing lists (e.g. http://vger.kernel.org/vger-lists.html#netdev). | |
757 | ||
758 | ||
759 | VLANs | |
760 | ----- | |
761 | ||
762 | Q: What's a VLAN? | |
763 | ||
764 | A: At the simplest level, a VLAN (short for "virtual LAN") is a way to | |
765 | partition a single switch into multiple switches. Suppose, for | |
766 | example, that you have two groups of machines, group A and group B. | |
767 | You want the machines in group A to be able to talk to each other, | |
768 | and you want the machine in group B to be able to talk to each | |
769 | other, but you don't want the machines in group A to be able to | |
770 | talk to the machines in group B. You can do this with two | |
771 | switches, by plugging the machines in group A into one switch and | |
772 | the machines in group B into the other switch. | |
773 | ||
774 | If you only have one switch, then you can use VLANs to do the same | |
775 | thing, by configuring the ports for machines in group A as VLAN | |
776 | "access ports" for one VLAN and the ports for group B as "access | |
777 | ports" for a different VLAN. The switch will only forward packets | |
778 | between ports that are assigned to the same VLAN, so this | |
779 | effectively subdivides your single switch into two independent | |
780 | switches, one for each group of machines. | |
781 | ||
782 | So far we haven't said anything about VLAN headers. With access | |
783 | ports, like we've described so far, no VLAN header is present in | |
784 | the Ethernet frame. This means that the machines (or switches) | |
785 | connected to access ports need not be aware that VLANs are | |
786 | involved, just like in the case where we use two different physical | |
787 | switches. | |
788 | ||
789 | Now suppose that you have a whole bunch of switches in your | |
790 | network, instead of just one, and that some machines in group A are | |
791 | connected directly to both switches 1 and 2. To allow these | |
792 | machines to talk to each other, you could add an access port for | |
793 | group A's VLAN to switch 1 and another to switch 2, and then | |
794 | connect an Ethernet cable between those ports. That works fine, | |
795 | but it doesn't scale well as the number of switches and the number | |
796 | of VLANs increases, because you use up a lot of valuable switch | |
797 | ports just connecting together your VLANs. | |
798 | ||
799 | This is where VLAN headers come in. Instead of using one cable and | |
800 | two ports per VLAN to connect a pair of switches, we configure a | |
801 | port on each switch as a VLAN "trunk port". Packets sent and | |
802 | received on a trunk port carry a VLAN header that says what VLAN | |
803 | the packet belongs to, so that only two ports total are required to | |
804 | connect the switches, regardless of the number of VLANs in use. | |
805 | Normally, only switches (either physical or virtual) are connected | |
806 | to a trunk port, not individual hosts, because individual hosts | |
807 | don't expect to see a VLAN header in the traffic that they receive. | |
808 | ||
809 | None of the above discussion says anything about particular VLAN | |
810 | numbers. This is because VLAN numbers are completely arbitrary. | |
811 | One must only ensure that a given VLAN is numbered consistently | |
812 | throughout a network and that different VLANs are given different | |
813 | numbers. (That said, VLAN 0 is usually synonymous with a packet | |
814 | that has no VLAN header, and VLAN 4095 is reserved.) | |
815 | ||
816 | Q: VLANs don't work. | |
817 | ||
818 | A: Many drivers in Linux kernels before version 3.3 had VLAN-related | |
819 | bugs. If you are having problems with VLANs that you suspect to be | |
820 | driver related, then you have several options: | |
821 | ||
822 | - Upgrade to Linux 3.3 or later. | |
823 | ||
824 | - Build and install a fixed version of the particular driver | |
825 | that is causing trouble, if one is available. | |
826 | ||
827 | - Use a NIC whose driver does not have VLAN problems. | |
828 | ||
829 | - Use "VLAN splinters", a feature in Open vSwitch 1.4 and later | |
830 | that works around bugs in kernel drivers. To enable VLAN | |
831 | splinters on interface eth0, use the command: | |
832 | ||
833 | ovs-vsctl set interface eth0 other-config:enable-vlan-splinters=true | |
834 | ||
835 | For VLAN splinters to be effective, Open vSwitch must know | |
836 | which VLANs are in use. See the "VLAN splinters" section in | |
837 | the Interface table in ovs-vswitchd.conf.db(5) for details on | |
838 | how Open vSwitch infers in-use VLANs. | |
839 | ||
840 | VLAN splinters increase memory use and reduce performance, so | |
841 | use them only if needed. | |
842 | ||
843 | - Apply the "vlan workaround" patch from the XenServer kernel | |
844 | patch queue, build Open vSwitch against this patched kernel, | |
845 | and then use ovs-vlan-bug-workaround(8) to enable the VLAN | |
846 | workaround for each interface whose driver is buggy. | |
847 | ||
848 | (This is a nontrivial exercise, so this option is included | |
849 | only for completeness.) | |
850 | ||
851 | It is not always easy to tell whether a Linux kernel driver has | |
852 | buggy VLAN support. The ovs-vlan-test(8) and ovs-test(8) utilities | |
853 | can help you test. See their manpages for details. Of the two | |
854 | utilities, ovs-test(8) is newer and more thorough, but | |
855 | ovs-vlan-test(8) may be easier to use. | |
856 | ||
857 | Q: VLANs still don't work. I've tested the driver so I know that it's OK. | |
858 | ||
859 | A: Do you have VLANs enabled on the physical switch that OVS is | |
860 | attached to? Make sure that the port is configured to trunk the | |
861 | VLAN or VLANs that you are using with OVS. | |
862 | ||
863 | Q: Outgoing VLAN-tagged traffic goes through OVS to my physical switch | |
864 | and to its destination host, but OVS seems to drop incoming return | |
865 | traffic. | |
866 | ||
867 | A: It's possible that you have the VLAN configured on your physical | |
868 | switch as the "native" VLAN. In this mode, the switch treats | |
869 | incoming packets either tagged with the native VLAN or untagged as | |
870 | part of the native VLAN. It may also send outgoing packets in the | |
871 | native VLAN without a VLAN tag. | |
872 | ||
873 | If this is the case, you have two choices: | |
874 | ||
875 | - Change the physical switch port configuration to tag packets | |
876 | it forwards to OVS with the native VLAN instead of forwarding | |
877 | them untagged. | |
878 | ||
879 | - Change the OVS configuration for the physical port to a | |
880 | native VLAN mode. For example, the following sets up a | |
881 | bridge with port eth0 in "native-tagged" mode in VLAN 9: | |
882 | ||
883 | ovs-vsctl add-br br0 | |
884 | ovs-vsctl add-port br0 eth0 tag=9 vlan_mode=native-tagged | |
885 | ||
886 | In this situation, "native-untagged" mode will probably work | |
887 | equally well. Refer to the documentation for the Port table | |
888 | in ovs-vswitchd.conf.db(5) for more information. | |
889 | ||
890 | Q: I added a pair of VMs on different VLANs, like this: | |
891 | ||
892 | ovs-vsctl add-br br0 | |
893 | ovs-vsctl add-port br0 eth0 | |
894 | ovs-vsctl add-port br0 tap0 tag=9 | |
895 | ovs-vsctl add-port br0 tap1 tag=10 | |
896 | ||
897 | but the VMs can't access each other, the external network, or the | |
898 | Internet. | |
899 | ||
900 | A: It is to be expected that the VMs can't access each other. VLANs | |
901 | are a means to partition a network. When you configured tap0 and | |
902 | tap1 as access ports for different VLANs, you indicated that they | |
903 | should be isolated from each other. | |
904 | ||
905 | As for the external network and the Internet, it seems likely that | |
906 | the machines you are trying to access are not on VLAN 9 (or 10) and | |
907 | that the Internet is not available on VLAN 9 (or 10). | |
908 | ||
909 | Q: I added a pair of VMs on the same VLAN, like this: | |
910 | ||
911 | ovs-vsctl add-br br0 | |
912 | ovs-vsctl add-port br0 eth0 | |
913 | ovs-vsctl add-port br0 tap0 tag=9 | |
914 | ovs-vsctl add-port br0 tap1 tag=9 | |
915 | ||
916 | The VMs can access each other, but not the external network or the | |
917 | Internet. | |
918 | ||
919 | A: It seems likely that the machines you are trying to access in the | |
920 | external network are not on VLAN 9 and that the Internet is not | |
921 | available on VLAN 9. Also, ensure VLAN 9 is set up as an allowed | |
922 | trunk VLAN on the upstream switch port to which eth0 is connected. | |
923 | ||
924 | Q: Can I configure an IP address on a VLAN? | |
925 | ||
926 | A: Yes. Use an "internal port" configured as an access port. For | |
927 | example, the following configures IP address 192.168.0.7 on VLAN 9. | |
928 | That is, OVS will forward packets from eth0 to 192.168.0.7 only if | |
929 | they have an 802.1Q header with VLAN 9. Conversely, traffic | |
930 | forwarded from 192.168.0.7 to eth0 will be tagged with an 802.1Q | |
931 | header with VLAN 9: | |
932 | ||
933 | ovs-vsctl add-br br0 | |
934 | ovs-vsctl add-port br0 eth0 | |
935 | ovs-vsctl add-port br0 vlan9 tag=9 -- set interface vlan9 type=internal | |
936 | ifconfig vlan9 192.168.0.7 | |
937 | ||
938 | Q: My OpenFlow controller doesn't see the VLANs that I expect. | |
939 | ||
940 | A: The configuration for VLANs in the Open vSwitch database (e.g. via | |
941 | ovs-vsctl) only affects traffic that goes through Open vSwitch's | |
942 | implementation of the OpenFlow "normal switching" action. By | |
943 | default, when Open vSwitch isn't connected to a controller and | |
944 | nothing has been manually configured in the flow table, all traffic | |
945 | goes through the "normal switching" action. But, if you set up | |
946 | OpenFlow flows on your own, through a controller or using ovs-ofctl | |
947 | or through other means, then you have to implement VLAN handling | |
948 | yourself. | |
949 | ||
950 | You can use "normal switching" as a component of your OpenFlow | |
951 | actions, e.g. by putting "normal" into the lists of actions on | |
952 | ovs-ofctl or by outputting to OFPP_NORMAL from an OpenFlow | |
953 | controller. In situations where this is not suitable, you can | |
954 | implement VLAN handling yourself, e.g.: | |
955 | ||
956 | - If a packet comes in on an access port, and the flow table | |
957 | needs to send it out on a trunk port, then the flow can add | |
958 | the appropriate VLAN tag with the "mod_vlan_vid" action. | |
959 | ||
960 | - If a packet comes in on a trunk port, and the flow table | |
961 | needs to send it out on an access port, then the flow can | |
962 | strip the VLAN tag with the "strip_vlan" action. | |
963 | ||
964 | Q: I configured ports on a bridge as access ports with different VLAN | |
965 | tags, like this: | |
966 | ||
967 | ovs-vsctl add-br br0 | |
968 | ovs-vsctl set-controller br0 tcp:192.168.0.10:6633 | |
969 | ovs-vsctl add-port br0 eth0 | |
970 | ovs-vsctl add-port br0 tap0 tag=9 | |
971 | ovs-vsctl add-port br0 tap1 tag=10 | |
972 | ||
973 | but the VMs running behind tap0 and tap1 can still communicate, | |
974 | that is, they are not isolated from each other even though they are | |
975 | on different VLANs. | |
976 | ||
977 | A: Do you have a controller configured on br0 (as the commands above | |
978 | do)? If so, then this is a variant on the previous question, "My | |
979 | OpenFlow controller doesn't see the VLANs that I expect," and you | |
980 | can refer to the answer there for more information. | |
981 | ||
982 | ||
983 | VXLANs | |
984 | ----- | |
985 | ||
986 | Q: What's a VXLAN? | |
987 | ||
988 | A: VXLAN stands for Virtual eXtensible Local Area Network, and is a means | |
989 | to solve the scaling challenges of VLAN networks in a multi-tenant | |
990 | environment. VXLAN is an overlay network which transports an L2 network | |
991 | over an existing L3 network. For more information on VXLAN, please see | |
992 | the IETF draft available here: | |
993 | ||
994 | http://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-03 | |
995 | ||
996 | Q: How much of the VXLAN protocol does Open vSwitch currently support? | |
997 | ||
998 | A: Open vSwitch currently supports the framing format for packets on the | |
999 | wire. There is currently no support for the multicast aspects of VXLAN. | |
1000 | To get around the lack of multicast support, it is possible to | |
1001 | pre-provision MAC to IP address mappings either manually or from a | |
1002 | controller. | |
1003 | ||
1004 | Q: What destination UDP port does the VXLAN implementation in Open vSwitch | |
1005 | use? | |
1006 | ||
1007 | A: By default, Open vSwitch will use the assigned IANA port for VXLAN, which | |
1008 | is 4789. However, it is possible to configure the destination UDP port | |
1009 | manually on a per-VXLAN tunnel basis. An example of this configuration is | |
1010 | provided below. | |
1011 | ||
1012 | ovs-vsctl add-br br0 | |
1013 | ovs-vsctl add-port br0 vxlan1 -- set interface vxlan1 | |
1014 | type=vxlan options:remote_ip=192.168.1.2 options:key=flow | |
1015 | options:dst_port=8472 | |
1016 | ||
1017 | ||
1018 | Using OpenFlow (Manually or Via Controller) | |
1019 | ------------------------------------------- | |
1020 | ||
1021 | Q: What versions of OpenFlow does Open vSwitch support? | |
1022 | ||
1023 | A: Open vSwitch 1.9 and earlier support only OpenFlow 1.0 (plus | |
1024 | extensions that bring in many of the features from later versions | |
1025 | of OpenFlow). | |
1026 | ||
1027 | Open vSwitch 1.10 and later have experimental support for OpenFlow | |
1028 | 1.2 and 1.3. On these versions of Open vSwitch, the following | |
1029 | command enables OpenFlow 1.0, 1.2, and 1.3 on bridge br0: | |
1030 | ||
1031 | ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow12,OpenFlow13 | |
1032 | ||
1033 | Open vSwitch version 1.12 and later will have experimental support | |
1034 | for OpenFlow 1.1, 1.2, and 1.3. On these versions of Open vSwitch, | |
1035 | the following command enables OpenFlow 1.0, 1.1, 1.2, and 1.3 on | |
1036 | bridge br0: | |
1037 | ||
1038 | ovs-vsctl set bridge br0 protocols=OpenFlow10,OpenFlow11,OpenFlow12,OpenFlow13 | |
1039 | ||
1040 | Use the -O option to enable support for later versions of OpenFlow | |
1041 | in ovs-ofctl. For example: | |
1042 | ||
1043 | ovs-ofctl -O OpenFlow13 dump-flows br0 | |
1044 | ||
1045 | Support for OpenFlow 1.1, 1.2, and 1.3 is still incomplete. Work | |
1046 | to be done is tracked in OPENFLOW-1.1+ in the Open vSwitch sources | |
1047 | (also via http://openvswitch.org/development/openflow-1-x-plan/). | |
1048 | When support for a given OpenFlow version is solidly implemented, | |
1049 | Open vSwitch will enable that version by default. | |
1050 | ||
1051 | Q: I'm getting "error type 45250 code 0". What's that? | |
1052 | ||
1053 | A: This is a Open vSwitch extension to OpenFlow error codes. Open | |
1054 | vSwitch uses this extension when it must report an error to an | |
1055 | OpenFlow controller but no standard OpenFlow error code is | |
1056 | suitable. | |
1057 | ||
1058 | Open vSwitch logs the errors that it sends to controllers, so the | |
1059 | easiest thing to do is probably to look at the ovs-vswitchd log to | |
1060 | find out what the error was. | |
1061 | ||
1062 | If you want to dissect the extended error message yourself, the | |
1063 | format is documented in include/openflow/nicira-ext.h in the Open | |
1064 | vSwitch source distribution. The extended error codes are | |
1065 | documented in lib/ofp-errors.h. | |
1066 | ||
1067 | Q1: Some of the traffic that I'd expect my OpenFlow controller to see | |
1068 | doesn't actually appear through the OpenFlow connection, even | |
1069 | though I know that it's going through. | |
1070 | Q2: Some of the OpenFlow flows that my controller sets up don't seem | |
1071 | to apply to certain traffic, especially traffic between OVS and | |
1072 | the controller itself. | |
1073 | ||
1074 | A: By default, Open vSwitch assumes that OpenFlow controllers are | |
1075 | connected "in-band", that is, that the controllers are actually | |
1076 | part of the network that is being controlled. In in-band mode, | |
1077 | Open vSwitch sets up special "hidden" flows to make sure that | |
1078 | traffic can make it back and forth between OVS and the controllers. | |
1079 | These hidden flows are higher priority than any flows that can be | |
1080 | set up through OpenFlow, and they are not visible through normal | |
1081 | OpenFlow flow table dumps. | |
1082 | ||
1083 | Usually, the hidden flows are desirable and helpful, but | |
1084 | occasionally they can cause unexpected behavior. You can view the | |
1085 | full OpenFlow flow table, including hidden flows, on bridge br0 | |
1086 | with the command: | |
1087 | ||
1088 | ovs-appctl bridge/dump-flows br0 | |
1089 | ||
1090 | to help you debug. The hidden flows are those with priorities | |
1091 | greater than 65535 (the maximum priority that can be set with | |
1092 | OpenFlow). | |
1093 | ||
1094 | The DESIGN file at the top level of the Open vSwitch source | |
1095 | distribution describes the in-band model in detail. | |
1096 | ||
1097 | If your controllers are not actually in-band (e.g. they are on | |
1098 | localhost via 127.0.0.1, or on a separate network), then you should | |
1099 | configure your controllers in "out-of-band" mode. If you have one | |
1100 | controller on bridge br0, then you can configure out-of-band mode | |
1101 | on it with: | |
1102 | ||
1103 | ovs-vsctl set controller br0 connection-mode=out-of-band | |
1104 | ||
1105 | Q: I configured all my controllers for out-of-band control mode but | |
1106 | "ovs-appctl bridge/dump-flows" still shows some hidden flows. | |
1107 | ||
1108 | A: You probably have a remote manager configured (e.g. with "ovs-vsctl | |
1109 | set-manager"). By default, Open vSwitch assumes that managers need | |
1110 | in-band rules set up on every bridge. You can disable these rules | |
1111 | on bridge br0 with: | |
1112 | ||
1113 | ovs-vsctl set bridge br0 other-config:disable-in-band=true | |
1114 | ||
1115 | This actually disables in-band control entirely for the bridge, as | |
1116 | if all the bridge's controllers were configured for out-of-band | |
1117 | control. | |
1118 | ||
1119 | Q: My OpenFlow controller doesn't see the VLANs that I expect. | |
1120 | ||
1121 | A: See answer under "VLANs", above. | |
1122 | ||
1123 | Q: I ran "ovs-ofctl add-flow br0 nw_dst=192.168.0.1,actions=drop" | |
1124 | but I got a funny message like this: | |
1125 | ||
1126 | ofp_util|INFO|normalization changed ofp_match, details: | |
1127 | ofp_util|INFO| pre: nw_dst=192.168.0.1 | |
1128 | ofp_util|INFO|post: | |
1129 | ||
1130 | and when I ran "ovs-ofctl dump-flows br0" I saw that my nw_dst | |
1131 | match had disappeared, so that the flow ends up matching every | |
1132 | packet. | |
1133 | ||
1134 | A: The term "normalization" in the log message means that a flow | |
1135 | cannot match on an L3 field without saying what L3 protocol is in | |
1136 | use. The "ovs-ofctl" command above didn't specify an L3 protocol, | |
1137 | so the L3 field match was dropped. | |
1138 | ||
1139 | In this case, the L3 protocol could be IP or ARP. A correct | |
1140 | command for each possibility is, respectively: | |
1141 | ||
1142 | ovs-ofctl add-flow br0 ip,nw_dst=192.168.0.1,actions=drop | |
1143 | ||
1144 | and | |
1145 | ||
1146 | ovs-ofctl add-flow br0 arp,nw_dst=192.168.0.1,actions=drop | |
1147 | ||
1148 | Similarly, a flow cannot match on an L4 field without saying what | |
1149 | L4 protocol is in use. For example, the flow match "tp_src=1234" | |
1150 | is, by itself, meaningless and will be ignored. Instead, to match | |
1151 | TCP source port 1234, write "tcp,tp_src=1234", or to match UDP | |
1152 | source port 1234, write "udp,tp_src=1234". | |
1153 | ||
1154 | Q: How can I figure out the OpenFlow port number for a given port? | |
1155 | ||
1156 | A: The OFPT_FEATURES_REQUEST message requests an OpenFlow switch to | |
1157 | respond with an OFPT_FEATURES_REPLY that, among other information, | |
1158 | includes a mapping between OpenFlow port names and numbers. From a | |
1159 | command prompt, "ovs-ofctl show br0" makes such a request and | |
1160 | prints the response for switch br0. | |
1161 | ||
1162 | The Interface table in the Open vSwitch database also maps OpenFlow | |
1163 | port names to numbers. To print the OpenFlow port number | |
1164 | associated with interface eth0, run: | |
1165 | ||
1166 | ovs-vsctl get Interface eth0 ofport | |
1167 | ||
1168 | You can print the entire mapping with: | |
1169 | ||
1170 | ovs-vsctl -- --columns=name,ofport list Interface | |
1171 | ||
1172 | but the output mixes together interfaces from all bridges in the | |
1173 | database, so it may be confusing if more than one bridge exists. | |
1174 | ||
1175 | In the Open vSwitch database, ofport value -1 means that the | |
1176 | interface could not be created due to an error. (The Open vSwitch | |
1177 | log should indicate the reason.) ofport value [] (the empty set) | |
1178 | means that the interface hasn't been created yet. The latter is | |
1179 | normally an intermittent condition (unless ovs-vswitchd is not | |
1180 | running). | |
1181 | ||
1182 | Q: I added some flows with my controller or with ovs-ofctl, but when I | |
1183 | run "ovs-dpctl dump-flows" I don't see them. | |
1184 | ||
1185 | A: ovs-dpctl queries a kernel datapath, not an OpenFlow switch. It | |
1186 | won't display the information that you want. You want to use | |
1187 | "ovs-ofctl dump-flows" instead. | |
1188 | ||
1189 | Q: It looks like each of the interfaces in my bonded port shows up | |
1190 | as an individual OpenFlow port. Is that right? | |
1191 | ||
1192 | A: Yes, Open vSwitch makes individual bond interfaces visible as | |
1193 | OpenFlow ports, rather than the bond as a whole. The interfaces | |
1194 | are treated together as a bond for only a few purposes: | |
1195 | ||
1196 | - Sending a packet to the OFPP_NORMAL port. (When an OpenFlow | |
1197 | controller is not configured, this happens implicitly to | |
1198 | every packet.) | |
1199 | ||
1200 | - Mirrors configured for output to a bonded port. | |
1201 | ||
1202 | It would make a lot of sense for Open vSwitch to present a bond as | |
1203 | a single OpenFlow port. If you want to contribute an | |
1204 | implementation of such a feature, please bring it up on the Open | |
1205 | vSwitch development mailing list at dev@openvswitch.org. | |
1206 | ||
1207 | Q: I have a sophisticated network setup involving Open vSwitch, VMs or | |
1208 | multiple hosts, and other components. The behavior isn't what I | |
1209 | expect. Help! | |
1210 | ||
1211 | A: To debug network behavior problems, trace the path of a packet, | |
1212 | hop-by-hop, from its origin in one host to a remote host. If | |
1213 | that's correct, then trace the path of the response packet back to | |
1214 | the origin. | |
1215 | ||
1216 | Usually a simple ICMP echo request and reply ("ping") packet is | |
1217 | good enough. Start by initiating an ongoing "ping" from the origin | |
1218 | host to a remote host. If you are tracking down a connectivity | |
1219 | problem, the "ping" will not display any successful output, but | |
1220 | packets are still being sent. (In this case the packets being sent | |
1221 | are likely ARP rather than ICMP.) | |
1222 | ||
1223 | Tools available for tracing include the following: | |
1224 | ||
1225 | - "tcpdump" and "wireshark" for observing hops across network | |
1226 | devices, such as Open vSwitch internal devices and physical | |
1227 | wires. | |
1228 | ||
1229 | - "ovs-appctl dpif/dump-flows <br>" in Open vSwitch 1.10 and | |
1230 | later or "ovs-dpctl dump-flows <br>" in earlier versions. | |
1231 | These tools allow one to observe the actions being taken on | |
1232 | packets in ongoing flows. | |
1233 | ||
1234 | See ovs-vswitchd(8) for "ovs-appctl dpif/dump-flows" | |
1235 | documentation, ovs-dpctl(8) for "ovs-dpctl dump-flows" | |
1236 | documentation, and "Why are there so many different ways to | |
1237 | dump flows?" above for some background. | |
1238 | ||
1239 | - "ovs-appctl ofproto/trace" to observe the logic behind how | |
1240 | ovs-vswitchd treats packets. See ovs-vswitchd(8) for | |
1241 | documentation. You can out more details about a given flow | |
1242 | that "ovs-dpctl dump-flows" displays, by cutting and pasting | |
1243 | a flow from the output into an "ovs-appctl ofproto/trace" | |
1244 | command. | |
1245 | ||
1246 | - SPAN, RSPAN, and ERSPAN features of physical switches, to | |
1247 | observe what goes on at these physical hops. | |
1248 | ||
1249 | Starting at the origin of a given packet, observe the packet at | |
1250 | each hop in turn. For example, in one plausible scenario, you | |
1251 | might: | |
1252 | ||
1253 | 1. "tcpdump" the "eth" interface through which an ARP egresses | |
1254 | a VM, from inside the VM. | |
1255 | ||
1256 | 2. "tcpdump" the "vif" or "tap" interface through which the ARP | |
1257 | ingresses the host machine. | |
1258 | ||
1259 | 3. Use "ovs-dpctl dump-flows" to spot the ARP flow and observe | |
1260 | the host interface through which the ARP egresses the | |
1261 | physical machine. You may need to use "ovs-dpctl show" to | |
1262 | interpret the port numbers. If the output seems surprising, | |
1263 | you can use "ovs-appctl ofproto/trace" to observe details of | |
1264 | how ovs-vswitchd determined the actions in the "ovs-dpctl | |
1265 | dump-flows" output. | |
1266 | ||
1267 | 4. "tcpdump" the "eth" interface through which the ARP egresses | |
1268 | the physical machine. | |
1269 | ||
1270 | 5. "tcpdump" the "eth" interface through which the ARP | |
1271 | ingresses the physical machine, at the remote host that | |
1272 | receives the ARP. | |
1273 | ||
1274 | 6. Use "ovs-dpctl dump-flows" to spot the ARP flow on the | |
1275 | remote host that receives the ARP and observe the VM "vif" | |
1276 | or "tap" interface to which the flow is directed. Again, | |
1277 | "ovs-dpctl show" and "ovs-appctl ofproto/trace" might help. | |
1278 | ||
1279 | 7. "tcpdump" the "vif" or "tap" interface to which the ARP is | |
1280 | directed. | |
1281 | ||
1282 | 8. "tcpdump" the "eth" interface through which the ARP | |
1283 | ingresses a VM, from inside the VM. | |
1284 | ||
1285 | It is likely that during one of these steps you will figure out the | |
1286 | problem. If not, then follow the ARP reply back to the origin, in | |
1287 | reverse. | |
1288 | ||
1289 | Q: How do I make a flow drop packets? | |
1290 | ||
1291 | A: To drop a packet is to receive it without forwarding it. OpenFlow | |
1292 | explicitly specifies forwarding actions. Thus, a flow with an | |
1293 | empty set of actions does not forward packets anywhere, causing | |
1294 | them to be dropped. You can specify an empty set of actions with | |
1295 | "actions=" on the ovs-ofctl command line. For example: | |
1296 | ||
1297 | ovs-ofctl add-flow br0 priority=65535,actions= | |
1298 | ||
1299 | would cause every packet entering switch br0 to be dropped. | |
1300 | ||
1301 | You can write "drop" explicitly if you like. The effect is the | |
1302 | same. Thus, the following command also causes every packet | |
1303 | entering switch br0 to be dropped: | |
1304 | ||
1305 | ovs-ofctl add-flow br0 priority=65535,actions=drop | |
1306 | ||
1307 | "drop" is not an action, either in OpenFlow or Open vSwitch. | |
1308 | Rather, it is only a way to say that there are no actions. | |
1309 | ||
1310 | Q: I added a flow to send packets out the ingress port, like this: | |
1311 | ||
1312 | ovs-ofctl add-flow br0 in_port=2,actions=2 | |
1313 | ||
1314 | but OVS drops the packets instead. | |
1315 | ||
1316 | A: Yes, OpenFlow requires a switch to ignore attempts to send a packet | |
1317 | out its ingress port. The rationale is that dropping these packets | |
1318 | makes it harder to loop the network. Sometimes this behavior can | |
1319 | even be convenient, e.g. it is often the desired behavior in a flow | |
1320 | that forwards a packet to several ports ("floods" the packet). | |
1321 | ||
1322 | Sometimes one really needs to send a packet out its ingress port. | |
1323 | In this case, output to OFPP_IN_PORT, which in ovs-ofctl syntax is | |
1324 | expressed as just "in_port", e.g.: | |
1325 | ||
1326 | ovs-ofctl add-flow br0 in_port=2,actions=in_port | |
1327 | ||
1328 | This also works in some circumstances where the flow doesn't match | |
1329 | on the input port. For example, if you know that your switch has | |
1330 | five ports numbered 2 through 6, then the following will send every | |
1331 | received packet out every port, even its ingress port: | |
1332 | ||
1333 | ovs-ofctl add-flow br0 actions=2,3,4,5,6,in_port | |
1334 | ||
1335 | or, equivalently: | |
1336 | ||
1337 | ovs-ofctl add-flow br0 actions=all,in_port | |
1338 | ||
1339 | Sometimes, in complicated flow tables with multiple levels of | |
1340 | "resubmit" actions, a flow needs to output to a particular port | |
1341 | that may or may not be the ingress port. It's difficult to take | |
1342 | advantage of OFPP_IN_PORT in this situation. To help, Open vSwitch | |
1343 | provides, as an OpenFlow extension, the ability to modify the | |
1344 | in_port field. Whatever value is currently in the in_port field is | |
1345 | the port to which outputs will be dropped, as well as the | |
1346 | destination for OFPP_IN_PORT. This means that the following will | |
1347 | reliably output to port 2 or to ports 2 through 6, respectively: | |
1348 | ||
1349 | ovs-ofctl add-flow br0 in_port=2,actions=load:0->NXM_OF_IN_PORT[],2 | |
1350 | ovs-ofctl add-flow br0 actions=load:0->NXM_OF_IN_PORT[],2,3,4,5,6 | |
1351 | ||
1352 | If the input port is important, then one may save and restore it on | |
1353 | the stack: | |
1354 | ||
1355 | ovs-ofctl add-flow br0 actions=push:NXM_OF_IN_PORT[],\ | |
1356 | load:0->NXM_OF_IN_PORT[],\ | |
1357 | 2,3,4,5,6,\ | |
1358 | pop:NXM_OF_IN_PORT[] | |
1359 | ||
1360 | Q: My bridge br0 has host 192.168.0.1 on port 1 and host 192.168.0.2 | |
1361 | on port 2. I set up flows to forward only traffic destined to the | |
1362 | other host and drop other traffic, like this: | |
1363 | ||
1364 | priority=5,in_port=1,ip,nw_dst=192.168.0.2,actions=2 | |
1365 | priority=5,in_port=2,ip,nw_dst=192.168.0.1,actions=1 | |
1366 | priority=0,actions=drop | |
1367 | ||
1368 | But it doesn't work--I don't get any connectivity when I do this. | |
1369 | Why? | |
1370 | ||
1371 | A: These flows drop the ARP packets that IP hosts use to establish IP | |
1372 | connectivity over Ethernet. To solve the problem, add flows to | |
1373 | allow ARP to pass between the hosts: | |
1374 | ||
1375 | priority=5,in_port=1,arp,actions=2 | |
1376 | priority=5,in_port=2,arp,actions=1 | |
1377 | ||
1378 | This issue can manifest other ways, too. The following flows that | |
1379 | match on Ethernet addresses instead of IP addresses will also drop | |
1380 | ARP packets, because ARP requests are broadcast instead of being | |
1381 | directed to a specific host: | |
1382 | ||
1383 | priority=5,in_port=1,dl_dst=54:00:00:00:00:02,actions=2 | |
1384 | priority=5,in_port=2,dl_dst=54:00:00:00:00:01,actions=1 | |
1385 | priority=0,actions=drop | |
1386 | ||
1387 | The solution already described above will also work in this case. | |
1388 | It may be better to add flows to allow all multicast and broadcast | |
1389 | traffic: | |
1390 | ||
1391 | priority=5,in_port=1,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=2 | |
1392 | priority=5,in_port=2,dl_dst=01:00:00:00:00:00/01:00:00:00:00:00,actions=1 | |
1393 | ||
1394 | Contact | |
1395 | ------- | |
1396 | ||
1397 | bugs@openvswitch.org | |
1398 | http://openvswitch.org/ |