1 <?xml version="1.0" encoding="utf-8"?>
2 <database title="Open vSwitch Configuration Database">
3 <p>A database with this schema holds the configuration for one Open
4 vSwitch daemon. The root of the configuration for the daemon is
5 the <ref table="Open_vSwitch"/> table, which must have exactly one
6 record. Records in other tables are significant only when they
7 can be reached directly or indirectly from the
8 <ref table="Open_vSwitch"/> table.</p>
10 <table name="Open_vSwitch" title="Open vSwitch configuration.">
11 Configuration for an Open vSwitch daemon. There must be exactly one record
12 in the <ref table="Open_vSwitch"/> table.
14 <group title="Configuration">
15 <column name="bridges">
16 Set of bridges managed by the daemon.
20 SSL used globally by the daemon.
23 <column name="external_ids">
24 Key-value pairs for use by external frameworks that integrate
25 with Open vSwitch, rather than by Open vSwitch itself. System
26 integrators should either use the Open vSwitch development
27 mailing list to coordinate on common key-value definitions, or
28 choose key names that are likely to be unique. The currently
29 defined common key-value pairs are:
31 <dt><code>system-id</code></dt>
32 <dd>A unique identifier for the Open vSwitch's physical host.
33 The form of the identifier depends on the type of the host.
34 On a Citrix XenServer, this will likely be the same as
35 <code>xs-system-uuid</code>.</dd>
36 <dt><code>xs-system-uuid</code></dt>
37 <dd>The Citrix XenServer universally unique identifier for the
38 physical host as displayed by <code>xe host-list</code>.</dd>
43 <group title="Status">
44 <column name="next_cfg">
45 Sequence number for client to increment. When a client modifies
46 any part of the database configuration and wishes to wait for
47 Open vSwitch to finish applying the changes, it may increment
51 <column name="cur_cfg">
52 Sequence number that Open vSwitch sets to the current value of
53 <ref column="next_cfg"/> after it finishes applying a set of
54 configuration changes.
57 <column name="capabilities">
58 Describes functionality supported by the hardware and software platform
59 on which this Open vSwitch is based. Clients should not modify this
60 column. See the <ref table="Capability"/> description for defined
61 capability categories and the meaning of associated
62 <ref table="Capability"/> records.
65 <column name="statistics">
67 Key-value pairs that report statistics about a system running an Open
68 vSwitch. These are updated periodically (currently, every 5
69 seconds). Key-value pairs that cannot be determined or that do not
70 apply to a platform are omitted.
74 <dt><code>cpu</code></dt>
77 Number of CPU processors, threads, or cores currently online and
78 available to the operating system on which Open vSwitch is
79 running, as an integer. This may be less than the number
80 installed, if some are not online or if they are not available to
84 Open vSwitch userspace processes are not multithreaded, but the
85 Linux kernel-based datapath is.
89 <dt><code>load_average</code></dt>
92 A comma-separated list of three floating-point numbers,
93 representing the system load average over the last 1, 5, and 15
94 minutes, respectively.
98 <dt><code>memory</code></dt>
101 A comma-separated list of integers, each of which represents a
102 quantity of memory in kilobytes that describes the operating
103 system on which Open vSwitch is running. In respective order,
108 <li>Total amount of RAM allocated to the OS.</li>
109 <li>RAM allocated to the OS that is in use.</li>
110 <li>RAM that can be flushed out to disk or otherwise discarded
111 if that space is needed for another purpose. This number is
112 necessarily less than or equal to the previous value.</li>
113 <li>Total disk space allocated for swap.</li>
114 <li>Swap space currently in use.</li>
118 On Linux, all five values can be determined and are included. On
119 other operating systems, only the first two values can be
120 determined, so the list will only have two values.
124 <dt><code>process_</code><var>name</var></dt>
127 One such key-value pair will exist for each running Open vSwitch
128 daemon process, with <var>name</var> replaced by the daemon's
129 name (e.g. <code>process_ovs-vswitchd</code>). The value is a
130 comma-separated list of integers. The integers represent the
131 following, with memory measured in kilobytes and durations in
136 <li>The process's virtual memory size.</li>
137 <li>The process's resident set size.</li>
138 <li>The amount of user and system CPU time consumed by the
140 <li>The number of times that the process has crashed and been
141 automatically restarted by the monitor.</li>
142 <li>The duration since the process was started.</li>
143 <li>The duration for which the process has been running.</li>
147 The interpretation of some of these values depends on whether the
148 process was started with the <option>--monitor</option>. If it
149 was not, then the crash count will always be 0 and the two
150 durations will always be the same. If <option>--monitor</option>
151 was given, then the crash count may be positive; if it is, the
152 latter duration is the amount of time since the most recent crash
157 There will be one key-value pair for each file in Open vSwitch's
158 ``run directory'' (usually <code>/var/run/openvswitch</code>)
159 whose name ends in <code>.pid</code>, whose contents are a
160 process ID, and which is locked by a running process. The
161 <var>name</var> is taken from the pidfile's name.
165 Currently Open vSwitch is only able to obtain all of the above
166 detail on Linux systems. On other systems, the same key-value
167 pairs will be present but the values will always be the empty
172 <dt><code>file_systems</code></dt>
175 A space-separated list of information on local, writable file
176 systems. Each item in the list describes one file system and
177 consists in turn of a comma-separated list of the following:
181 <li>Mount point, e.g. <code>/</code> or <code>/var/log</code>.
182 Any spaces or commas in the mount point are replaced by
184 <li>Total size, in kilobytes, as an integer.</li>
185 <li>Amount of storage in use, in kilobytes, as an integer.</li>
189 This key-value pair is omitted if there are no local, writable
190 file systems or if Open vSwitch cannot obtain the needed
198 <group title="Version Reporting">
200 These columns report the types and versions of the hardware and
201 software running Open vSwitch. We recommend in general that software
202 should test whether specific features are supported instead of relying
203 on version number checks. These values are primarily intended for
204 reporting to human administrators.
207 <column name="ovs_version">
208 The Open vSwitch version number, e.g. <code>1.1.0pre2</code>.
209 If Open vSwitch was configured with a build number, then it is
210 also included, e.g. <code>1.1.0pre2+build4948</code>.
213 <column name="db_version">
215 The database schema version number in the form
216 <code><var>major</var>.<var>minor</var>.<var>tweak</var></code>,
217 e.g. <code>1.2.3</code>. Whenever the database schema is changed in
218 a non-backward compatible way (e.g. deleting a column or a table),
219 <var>major</var> is incremented. When the database schema is changed
220 in a backward compatible way (e.g. adding a new column),
221 <var>minor</var> is incremented. When the database schema is changed
222 cosmetically (e.g. reindenting its syntax), <var>tweak</var> is
227 The schema version is part of the database schema, so it can also be
228 retrieved by fetching the schema using the Open vSwitch database
233 <column name="system_type">
235 An identifier for the type of system on top of which Open vSwitch
236 runs, e.g. <code>XenServer</code> or <code>KVM</code>.
239 System integrators are responsible for choosing and setting an
240 appropriate value for this column.
244 <column name="system_version">
246 The version of the system identified by <ref column="system_type"/>,
247 e.g. <code>5.5.0-24648p</code> on XenServer 5.5.0 build 24648.
250 System integrators are responsible for choosing and setting an
251 appropriate value for this column.
257 <group title="Database Configuration">
259 These columns primarily configure the Open vSwitch database
260 (<code>ovsdb-server</code>), not the Open vSwitch switch
261 (<code>ovs-vswitchd</code>). The OVSDB database also uses the <ref
262 column="ssl"/> settings.
266 The Open vSwitch switch does read the database configuration to
267 determine remote IP addresses to which in-band control should apply.
270 <column name="manager_options">
271 Database clients to which the Open vSwitch database server should
272 connect or to which it should listen, along with options for how these
273 connection should be configured. See the <ref table="Manager"/> table
274 for more information.
277 <column name="managers">
279 Remote database clients to which the Open vSwitch's database server
280 should connect or to which it should listen. Adding an OVSDB target
281 to this set is equivalent to adding it to <ref
282 column="manager_options"/> with all of the default options.
286 Use of this column is deprecated and may be removed sometime in the
287 future. New applications should use and set <ref
288 column="manager_options"/> instead.
294 <table name="Bridge">
296 Configuration for a bridge within an
297 <ref table="Open_vSwitch"/>.
300 A <ref table="Bridge"/> record represents an Ethernet switch with one or
301 more ``ports,'' which are the <ref table="Port"/> records pointed to by
302 the <ref table="Bridge"/>'s <ref column="ports"/> column.
305 <group title="Core Features">
307 Bridge identifier. Should be alphanumeric and no more than about 8
308 bytes long. Must be unique among the names of ports, interfaces, and
312 <column name="ports">
313 Ports included in the bridge.
316 <column name="mirrors">
317 Port mirroring configuration.
320 <column name="netflow">
321 NetFlow configuration.
324 <column name="sflow">
328 <column name="flood_vlans">
329 VLAN IDs of VLANs on which MAC address learning should be disabled, so
330 that packets are flooded instead of being sent to specific ports that
331 are believed to contain packets' destination MACs. This should
332 ordinarily be used to disable MAC learning on VLANs used for mirroring
333 (RSPAN VLANs). It may also be useful for debugging.
337 <group title="OpenFlow Configuration">
338 <column name="controller">
339 OpenFlow controller set. If unset, then no OpenFlow controllers
343 <column name="fail_mode">
344 <p>When a controller is configured, it is, ordinarily, responsible
345 for setting up all flows on the switch. Thus, if the connection to
346 the controller fails, no new network connections can be set up.
347 If the connection to the controller stays down long enough,
348 no packets can pass through the switch at all. This setting
349 determines the switch's response to such a situation. It may be set
350 to one of the following:
352 <dt><code>standalone</code></dt>
353 <dd>If no message is received from the controller for three
354 times the inactivity probe interval
355 (see <ref column="inactivity_probe"/>), then Open vSwitch
356 will take over responsibility for setting up flows. In
357 this mode, Open vSwitch causes the bridge to act like an
358 ordinary MAC-learning switch. Open vSwitch will continue
359 to retry connecting to the controller in the background
360 and, when the connection succeeds, it will discontinue its
361 standalone behavior.</dd>
362 <dt><code>secure</code></dt>
363 <dd>Open vSwitch will not set up flows on its own when the
364 controller connection fails or when no controllers are
365 defined. The bridge will continue to retry connecting to
366 any defined controllers forever.</dd>
369 <p>If this value is unset, the default is implementation-specific.</p>
370 <p>When more than one controller is configured,
371 <ref column="fail_mode"/> is considered only when none of the
372 configured controllers can be contacted.</p>
375 <column name="datapath_id">
376 Reports the OpenFlow datapath ID in use. Exactly 16 hex
377 digits. (Setting this column will have no useful effect. Set
378 <ref column="other_config"/>:<code>other-config</code>
383 <group title="Other Features">
384 <column name="datapath_type">
385 Name of datapath provider. The kernel datapath has
386 type <code>system</code>. The userspace datapath has
387 type <code>netdev</code>.
390 <column name="external_ids">
391 Key-value pairs for use by external frameworks that integrate
392 with Open vSwitch, rather than by Open vSwitch itself. System
393 integrators should either use the Open vSwitch development
394 mailing list to coordinate on common key-value definitions, or
395 choose key names that are likely to be unique. The currently
396 defined key-value pairs are:
398 <dt><code>bridge-id</code></dt>
399 <dd>A unique identifier of the bridge. On Citrix XenServer this
400 will commonly be the same as <code>xs-network-uuids</code>.</dd>
401 <dt><code>xs-network-uuids</code></dt>
402 <dd>Semicolon-delimited set of universally unique identifier(s) for
403 the network with which this bridge is associated on a Citrix
404 XenServer host. The network identifiers are RFC 4122 UUIDs as
405 displayed by, e.g., <code>xe network-list</code>.</dd>
409 <column name="other_config">
410 Key-value pairs for configuring rarely used bridge
411 features. The currently defined key-value pairs are:
413 <dt><code>datapath-id</code></dt>
415 digits to set the OpenFlow datapath ID to a specific
416 value. May not be all-zero.</dd>
417 <dt><code>disable-in-band</code></dt>
418 <dd>If set to <code>true</code>, disable in-band control on
419 the bridge regardless of controller and manager settings.</dd>
420 <dt><code>hwaddr</code></dt>
421 <dd>An Ethernet address in the form
422 <var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>
423 to set the hardware address of the local port and influence the
425 <dt><code>in-band-queue</code></dt>
427 A queue ID as a nonnegative integer. This sets the OpenFlow queue
428 ID that will be used by flows set up by in-band control on this
429 bridge. If unset, or if the port used by an in-band control flow
430 does not have QoS configured, or if the port does not have a queue
431 with the specified ID, the default queue is used instead.
438 <table name="Port" table="Port or bond configuration.">
439 <p>A port within a <ref table="Bridge"/>.</p>
440 <p>Most commonly, a port has exactly one ``interface,'' pointed to by its
441 <ref column="interfaces"/> column. Such a port logically
442 corresponds to a port on a physical Ethernet switch. A port
443 with more than one interface is a ``bonded port'' (see
444 <ref group="Bonding Configuration"/>).</p>
445 <p>Some properties that one might think as belonging to a port are actually
446 part of the port's <ref table="Interface"/> members.</p>
449 Port name. Should be alphanumeric and no more than about 8
450 bytes long. May be the same as the interface name, for
451 non-bonded ports. Must otherwise be unique among the names of
452 ports, interfaces, and bridges on a host.
455 <column name="interfaces">
456 The port's interfaces. If there is more than one, this is a
460 <group title="VLAN Configuration">
461 <p>A bridge port must be configured for VLANs in one of two
462 mutually exclusive ways:
464 <li>A ``trunk port'' has an empty value for <ref
465 column="tag"/>. Its <ref column="trunks"/> value may be
466 empty or non-empty.</li>
467 <li>An ``implicitly tagged VLAN port'' or ``access port''
468 has an nonempty value for <ref column="tag"/>. Its
469 <ref column="trunks"/> value must be empty.</li>
471 If <ref column="trunks"/> and <ref column="tag"/> are both
472 nonempty, the configuration is ill-formed.
477 If this is an access port (see above), the port's implicitly
478 tagged VLAN. Must be empty if this is a trunk port.
481 Frames arriving on trunk ports will be forwarded to this
482 port only if they are tagged with the given VLAN (or, if
483 <ref column="tag"/> is 0, then if they lack a VLAN header).
484 Frames arriving on other access ports will be forwarded to
485 this port only if they have the same <ref column="tag"/>
486 value. Frames forwarded to this port will not have an
490 When a frame with a 802.1Q header that indicates a nonzero
491 VLAN is received on an access port, it is discarded.
495 <column name="trunks">
497 If this is a trunk port (see above), the 802.1Q VLAN(s) that
498 this port trunks; if it is empty, then the port trunks all
499 VLANs. Must be empty if this is an access port.
502 Frames arriving on trunk ports are dropped if they are not
503 in one of the specified VLANs. For this purpose, packets
504 that have no VLAN header are treated as part of VLAN 0.
509 <group title="Bonding Configuration">
510 <p>A port that has more than one interface is a ``bonded port.''
511 Bonding allows for load balancing and fail-over. Open vSwitch
512 supports ``source load balancing'' (SLB) bonding, which
513 assigns flows to slaves based on source MAC address and output VLAN,
514 with periodic rebalancing as traffic patterns change. This form of
515 bonding does not require 802.3ad or other special support from the
516 upstream switch to which the slave devices are connected.</p>
518 <p>These columns apply only to bonded ports. Their values are
519 otherwise ignored.</p>
521 <column name="bond_updelay">
522 <p>For a bonded port, the number of milliseconds for which carrier must
523 stay up on an interface before the interface is considered to be up.
524 Specify <code>0</code> to enable the interface immediately.</p>
525 <p>This setting is honored only when at least one bonded interface is
526 already enabled. When no interfaces are enabled, then the first bond
527 interface to come up is enabled immediately.</p>
530 <column name="bond_downdelay">
531 For a bonded port, the number of milliseconds for which carrier must
532 stay down on an interface before the interface is considered to be
533 down. Specify <code>0</code> to disable the interface immediately.
536 <column name="bond_fake_iface">
537 For a bonded port, whether to create a fake internal interface with the
538 name of the port. Use only for compatibility with legacy software that
543 <group title="Other Features">
545 Quality of Service configuration for this port.
549 The MAC address to use for this port for the purpose of choosing the
550 bridge's MAC address. This column does not necessarily reflect the
551 port's actual MAC address, nor will setting it change the port's actual
555 <column name="fake_bridge">
556 Does this port represent a sub-bridge for its tagged VLAN within the
557 Bridge? See ovs-vsctl(8) for more information.
560 <column name="external_ids">
562 Key-value pairs for use by external frameworks that integrate with
563 Open vSwitch, rather than by Open vSwitch itself. System integrators
564 should either use the Open vSwitch development mailing list to
565 coordinate on common key-value definitions, or choose key names that
566 are likely to be unique.
569 No key-value pairs native to <ref table="Port"/> are currently
570 defined. For fake bridges (see the <ref column="fake_bridge"/>
571 column), external IDs for the fake bridge are defined here by
572 prefixing a <ref table="Bridge"/> <ref table="Bridge"
573 column="external_ids"/> key with <code>fake-bridge-</code>,
574 e.g. <code>fake-bridge-xs-network-uuids</code>.
578 <column name="other_config">
579 Key-value pairs for configuring rarely used port features. The
580 currently defined key-value pairs are:
582 <dt><code>hwaddr</code></dt>
583 <dd>An Ethernet address in the form
584 <code><var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var></code>.</dd>
585 <dt><code>bond-rebalance-interval</code></dt>
586 <dd>For a bonded port, the number of milliseconds between
587 successive attempts to rebalance the bond, that is, to
588 move source MACs and their flows from one interface on
589 the bond to another in an attempt to keep usage of each
590 interface roughly equal. The default is 10000 (10
591 seconds), and the minimum is 1000 (1 second).</dd>
597 <table name="Interface" title="One physical network device in a Port.">
598 An interface within a <ref table="Port"/>.
600 <group title="Core Features">
602 Interface name. Should be alphanumeric and no more than about 8 bytes
603 long. May be the same as the port name, for non-bonded ports. Must
604 otherwise be unique among the names of ports, interfaces, and bridges
609 <p>Ethernet address to set for this interface. If unset then the
610 default MAC address is used:</p>
612 <li>For the local interface, the default is the lowest-numbered MAC
613 address among the other bridge ports, either the value of the
614 <ref table="Port" column="mac"/> in its <ref table="Port"/> record,
615 if set, or its actual MAC (for bonded ports, the MAC of its slave
616 whose name is first in alphabetical order). Internal ports and
617 bridge ports that are used as port mirroring destinations (see the
618 <ref table="Mirror"/> table) are ignored.</li>
619 <li>For other internal interfaces, the default MAC is randomly
621 <li>External interfaces typically have a MAC address associated with
624 <p>Some interfaces may not have a software-controllable MAC
628 <column name="ofport">
629 <p>OpenFlow port number for this interface. Unlike most columns, this
630 column's value should be set only by Open vSwitch itself. Other
631 clients should set this column to an empty set (the default) when
632 creating an <ref table="Interface"/>.</p>
633 <p>Open vSwitch populates this column when the port number becomes
634 known. If the interface is successfully added,
635 <ref column="ofport"/> will be set to a number between 1 and 65535
636 (generally either in the range 1 to 65279, inclusive, or 65534, the
637 port number for the OpenFlow ``local port''). If the interface
638 cannot be added then Open vSwitch sets this column
643 <group title="System-Specific Details">
645 The interface type, one of:
647 <dt><code>system</code></dt>
648 <dd>An ordinary network device, e.g. <code>eth0</code> on Linux.
649 Sometimes referred to as ``external interfaces'' since they are
650 generally connected to hardware external to that on which the Open
651 vSwitch is running. The empty string is a synonym for
652 <code>system</code>.</dd>
653 <dt><code>internal</code></dt>
654 <dd>A simulated network device that sends and receives traffic. An
655 internal interface whose <ref column="name"/> is the same as its
656 bridge's <ref table="Open_vSwitch" column="name"/> is called the
657 ``local interface.'' It does not make sense to bond an internal
658 interface, so the terms ``port'' and ``interface'' are often used
659 imprecisely for internal interfaces.</dd>
660 <dt><code>tap</code></dt>
661 <dd>A TUN/TAP device managed by Open vSwitch.</dd>
662 <dt><code>gre</code></dt>
663 <dd>An Ethernet over RFC 2890 Generic Routing Encapsulation over IPv4
664 tunnel. Each tunnel must be uniquely identified by the
665 combination of <code>remote_ip</code>, <code>local_ip</code>, and
666 <code>in_key</code>. Note that if two ports are defined that are
667 the same except one has an optional identifier and the other does
668 not, the more specific one is matched first. <code>in_key</code>
669 is considered more specific than <code>local_ip</code> if a port
670 defines one and another port defines the other. The following
671 options may be specified in the <ref column="options"/> column:
673 <dt><code>remote_ip</code></dt>
674 <dd>Required. The tunnel endpoint.</dd>
677 <dt><code>local_ip</code></dt>
678 <dd>Optional. The destination IP that received packets must
679 match. Default is to match all addresses.</dd>
682 <dt><code>in_key</code></dt>
683 <dd>Optional. The GRE key that received packets must contain.
684 It may either be a 32-bit number (no key and a key of 0 are
685 treated as equivalent) or the word <code>flow</code>. If
686 <code>flow</code> is specified then any key will be accepted
687 and the key will be placed in the <code>tun_id</code> field
688 for matching in the flow table. The ovs-ofctl manual page
689 contains additional information about matching fields in
690 OpenFlow flows. Default is no key.</dd>
693 <dt><code>out_key</code></dt>
694 <dd>Optional. The GRE key to be set on outgoing packets. It may
695 either be a 32-bit number or the word <code>flow</code>. If
696 <code>flow</code> is specified then the key may be set using
697 the <code>set_tunnel</code> Nicira OpenFlow vendor extension (0
698 is used in the absence of an action). The ovs-ofctl manual
699 page contains additional information about the Nicira OpenFlow
700 vendor extensions. Default is no key.</dd>
703 <dt><code>key</code></dt>
704 <dd>Optional. Shorthand to set <code>in_key</code> and
705 <code>out_key</code> at the same time.</dd>
708 <dt><code>tos</code></dt>
709 <dd>Optional. The value of the ToS bits to be set on the
710 encapsulating packet. It may also be the word
711 <code>inherit</code>, in which case the ToS will be copied from
712 the inner packet if it is IPv4 or IPv6 (otherwise it will be
713 0). Note that the ECN fields are always inherited. Default is
717 <dt><code>ttl</code></dt>
718 <dd>Optional. The TTL to be set on the encapsulating packet.
719 It may also be the word <code>inherit</code>, in which case the
720 TTL will be copied from the inner packet if it is IPv4 or IPv6
721 (otherwise it will be the system default, typically 64).
722 Default is the system default TTL.</dd>
725 <dt><code>csum</code></dt>
726 <dd>Optional. Compute GRE checksums on outgoing packets.
727 Checksums present on incoming packets will be validated
728 regardless of this setting. Note that GRE checksums
729 impose a significant performance penalty as they cover the
730 entire packet. As the contents of the packet is typically
731 covered by L3 and L4 checksums, this additional checksum only
732 adds value for the GRE and encapsulated Ethernet headers.
733 Default is disabled, set to <code>true</code> to enable.</dd>
736 <dt><code>pmtud</code></dt>
737 <dd>Optional. Enable tunnel path MTU discovery. If enabled
738 ``ICMP destination unreachable - fragmentation'' needed
739 messages will be generated for IPv4 packets with the DF bit set
740 and IPv6 packets above the minimum MTU if the packet size
741 exceeds the path MTU minus the size of the tunnel headers. It
742 also forces the encapsulating packet DF bit to be set (it is
743 always set if the inner packet implies path MTU discovery).
744 Note that this option causes behavior that is typically
745 reserved for routers and therefore is not entirely in
746 compliance with the IEEE 802.1D specification for bridges.
747 Default is enabled, set to <code>false</code> to disable.</dd>
750 <dt><code>header_cache</code></dt>
751 <dd>Optional. Enable caching of tunnel headers and the output
752 path. This can lead to a significant performance increase
753 without changing behavior. In general it should not be
754 necessary to adjust this setting. However, the caching can
755 bypass certain components of the IP stack (such as IP tables)
756 and it may be useful to disable it if these features are
757 required or as a debugging measure. Default is enabled, set to
758 <code>false</code> to disable. If IPsec is enabled through the
759 <ref column="other_config"/> parameters, header caching will be
760 automatically disabled.</dd>
763 <dt><code>capwap</code></dt>
764 <dd>Ethernet tunneling over the UDP transport portion of CAPWAP
765 (RFC 5415). This allows interoperability with certain switches
766 where GRE is not available. Note that only the tunneling component
767 of the protocol is implemented. Due to the non-standard use of
768 CAPWAP, UDP ports 58881 and 58882 are used as the source and
769 destinations ports respectivedly. Each tunnel must be uniquely
770 identified by the combination of <code>remote_ip</code> and
771 <code>local_ip</code>. If two ports are defined that are the same
772 except one includes <code>local_ip</code> and the other does not,
773 the more specific one is matched first. CAPWAP support is not
774 available on all platforms. Currently it is only supported in the
775 Linux kernel module with kernel versions >= 2.6.25. The following
776 options may be specified in the <ref column="options"/> column:
778 <dt><code>remote_ip</code></dt>
779 <dd>Required. The tunnel endpoint.</dd>
782 <dt><code>local_ip</code></dt>
783 <dd>Optional. The destination IP that received packets must
784 match. Default is to match all addresses.</dd>
787 <dt><code>tos</code></dt>
788 <dd>Optional. The value of the ToS bits to be set on the
789 encapsulating packet. It may also be the word
790 <code>inherit</code>, in which case the ToS will be copied from
791 the inner packet if it is IPv4 or IPv6 (otherwise it will be
792 0). Note that the ECN fields are always inherited. Default is
796 <dt><code>ttl</code></dt>
797 <dd>Optional. The TTL to be set on the encapsulating packet.
798 It may also be the word <code>inherit</code>, in which case the
799 TTL will be copied from the inner packet if it is IPv4 or IPv6
800 (otherwise it will be the system default, typically 64).
801 Default is the system default TTL.</dd>
804 <dt><code>pmtud</code></dt>
805 <dd>Optional. Enable tunnel path MTU discovery. If enabled
806 ``ICMP destination unreachable - fragmentation'' needed
807 messages will be generated for IPv4 packets with the DF bit set
808 and IPv6 packets above the minimum MTU if the packet size
809 exceeds the path MTU minus the size of the tunnel headers. It
810 also forces the encapsulating packet DF bit to be set (it is
811 always set if the inner packet implies path MTU discovery).
812 Note that this option causes behavior that is typically
813 reserved for routers and therefore is not entirely in
814 compliance with the IEEE 802.1D specification for bridges.
815 Default is enabled, set to <code>false</code> to disable.</dd>
818 <dt><code>header_cache</code></dt>
819 <dd>Optional. Enable caching of tunnel headers and the output
820 path. This can lead to a significant performance increase
821 without changing behavior. In general it should not be
822 necessary to adjust this setting. However, the caching can
823 bypass certain components of the IP stack (such as IP tables)
824 and it may be useful to disable it if these features are
825 required or as a debugging measure. Default is enabled, set to
826 <code>false</code> to disable.</dd>
829 <dt><code>patch</code></dt>
832 A pair of virtual devices that act as a patch cable. The <ref
833 column="options"/> column must have the following key-value pair:
836 <dt><code>peer</code></dt>
838 The <ref column="name"/> of the <ref table="Interface"/> for
839 the other side of the patch. The named <ref
840 table="Interface"/>'s own <code>peer</code> option must specify
841 this <ref table="Interface"/>'s name. That is, the two patch
842 interfaces must have reversed <ref column="name"/> and
843 <code>peer</code> values.
850 <column name="options">
851 Configuration options whose interpretation varies based on
852 <ref column="type"/>.
855 <column name="status">
857 Key-value pairs that report port status. Supported status
858 values are <code>type</code>-dependent.
860 <p>The only currently defined key-value pair is:</p>
862 <dt><code>source_ip</code></dt>
863 <dd>The source IP address used for an IPv4 tunnel end-point,
864 such as <code>gre</code> or <code>capwap</code>. Not
865 supported by all implementations.</dd>
870 <group title="Ingress Policing">
872 These settings control ingress policing for packets received on this
873 interface. On a physical interface, this limits the rate at which
874 traffic is allowed into the system from the outside; on a virtual
875 interface (one connected to a virtual machine), this limits the rate at
876 which the VM is able to transmit.
879 Policing is a simple form of quality-of-service that simply drops
880 packets received in excess of the configured rate. Due to its
881 simplicity, policing is usually less accurate and less effective than
882 egress QoS (which is configured using the <ref table="QoS"/> and <ref
883 table="Queue"/> tables).
886 Policing is currently implemented only on Linux. The Linux
887 implementation uses a simple ``token bucket'' approach:
891 The size of the bucket corresponds to <ref
892 column="ingress_policing_burst"/>. Initially the bucket is full.
895 Whenever a packet is received, its size (converted to tokens) is
896 compared to the number of tokens currently in the bucket. If the
897 required number of tokens are available, they are removed and the
898 packet is forwarded. Otherwise, the packet is dropped.
901 Whenever it is not full, the bucket is refilled with tokens at the
902 rate specified by <ref column="ingress_policing_rate"/>.
906 Policing interacts badly with some network protocols, and especially
907 with fragmented IP packets. Suppose that there is enough network
908 activity to keep the bucket nearly empty all the time. Then this token
909 bucket algorithm will forward a single packet every so often, with the
910 period depending on packet size and on the configured rate. All of the
911 fragments of an IP packets are normally transmitted back-to-back, as a
912 group. In such a situation, therefore, only one of these fragments
913 will be forwarded and the rest will be dropped. IP does not provide
914 any way for the intended recipient to ask for only the remaining
915 fragments. In such a case there are two likely possibilities for what
916 will happen next: either all of the fragments will eventually be
917 retransmitted (as TCP will do), in which case the same problem will
918 recur, or the sender will not realize that its packet has been dropped
919 and data will simply be lost (as some UDP-based protocols will do).
920 Either way, it is possible that no forward progress will ever occur.
922 <column name="ingress_policing_rate">
924 Maximum rate for data received on this interface, in kbps. Data
925 received faster than this rate is dropped. Set to <code>0</code>
926 (the default) to disable policing.
930 <column name="ingress_policing_burst">
931 <p>Maximum burst size for data received on this interface, in kb. The
932 default burst size if set to <code>0</code> is 1000 kb. This value
933 has no effect if <ref column="ingress_policing_rate"/>
934 is <code>0</code>.</p>
936 Specifying a larger burst size lets the algorithm be more forgiving,
937 which is important for protocols like TCP that react severely to
938 dropped packets. The burst size should be at least the size of the
939 interface's MTU. Specifying a value that is numerically at least as
940 large as 10% of <ref column="ingress_policing_rate"/> helps TCP come
941 closer to achieving the full rate.
946 <group title="Other Features">
948 <column name="monitor">
949 Connectivity monitor configuration for this interface.
952 <column name="external_ids">
953 Key-value pairs for use by external frameworks that integrate
954 with Open vSwitch, rather than by Open vSwitch itself. System
955 integrators should either use the Open vSwitch development
956 mailing list to coordinate on common key-value definitions, or
957 choose key names that are likely to be unique. The currently
958 defined common key-value pairs are:
960 <dt><code>attached-mac</code></dt>
962 The MAC address programmed into the ``virtual hardware'' for this
963 interface, in the form
964 <var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>:<var>xx</var>.
965 For Citrix XenServer, this is the value of the <code>MAC</code>
966 field in the VIF record for this interface.</dd>
967 <dt><code>iface-id</code></dt>
968 <dd>A system-unique identifier for the interface. On XenServer,
969 this will commonly be the same as <code>xs-vif-uuid</code>.</dd>
972 Additionally the following key-value pairs specifically
973 apply to an interface that represents a virtual Ethernet interface
974 connected to a virtual machine. These key-value pairs should not be
975 present for other types of interfaces. Keys whose names end
976 in <code>-uuid</code> have values that uniquely identify the entity
977 in question. For a Citrix XenServer hypervisor, these values are
978 UUIDs in RFC 4122 format. Other hypervisors may use other
981 <p>The currently defined key-value pairs for XenServer are:</p>
983 <dt><code>xs-vif-uuid</code></dt>
984 <dd>The virtual interface associated with this interface.</dd>
985 <dt><code>xs-network-uuid</code></dt>
986 <dd>The virtual network to which this interface is attached.</dd>
987 <dt><code>xs-vm-uuid</code></dt>
988 <dd>The VM to which this interface belongs.</dd>
992 <column name="other_config">
993 Key-value pairs for rarely used interface features. Currently,
994 the only keys are for configuring GRE-over-IPsec, which is only
995 available through the <code>openvswitch-ipsec</code> package for
996 Debian. The currently defined key-value pairs are:
998 <dt><code>ipsec_local_ip</code></dt>
999 <dd>Required key for GRE-over-IPsec interfaces. Additionally,
1000 the <ref column="type"/> must be <code>gre</code> and the
1001 <code>ipsec_psk</code> <ref column="other_config"/> key must
1002 be set. The <code>in_key</code>, <code>out_key</code>, and
1003 <code>key</code> <ref column="options"/> must not be
1005 <dt><code>ipsec_psk</code></dt>
1006 <dd>Required key for GRE-over-IPsec interfaces. Specifies a
1007 pre-shared key for authentication that must be identical on
1008 both sides of the tunnel. Additionally, the
1009 <code>ipsec_local_ip</code> key must also be set.</dd>
1013 <column name="statistics">
1015 Key-value pairs that report interface statistics. The current
1016 implementation updates these counters periodically. In the future,
1017 we plan to, instead, update them when an interface is created, when
1018 they are queried (e.g. using an OVSDB <code>select</code> operation),
1019 and just before an interface is deleted due to virtual interface
1020 hot-unplug or VM shutdown, and perhaps at other times, but not on any
1021 regular periodic basis.</p>
1023 The currently defined key-value pairs are listed below. These are
1024 the same statistics reported by OpenFlow in its <code>struct
1025 ofp_port_stats</code> structure. If an interface does not support a
1026 given statistic, then that pair is omitted.</p>
1029 Successful transmit and receive counters:
1031 <dt><code>rx_packets</code></dt>
1032 <dd>Number of received packets.</dd>
1033 <dt><code>rx_bytes</code></dt>
1034 <dd>Number of received bytes.</dd>
1035 <dt><code>tx_packets</code></dt>
1036 <dd>Number of transmitted packets.</dd>
1037 <dt><code>tx_bytes</code></dt>
1038 <dd>Number of transmitted bytes.</dd>
1044 <dt><code>rx_dropped</code></dt>
1045 <dd>Number of packets dropped by RX.</dd>
1046 <dt><code>rx_frame_err</code></dt>
1047 <dd>Number of frame alignment errors.</dd>
1048 <dt><code>rx_over_err</code></dt>
1049 <dd>Number of packets with RX overrun.</dd>
1050 <dt><code>rx_crc_err</code></dt>
1051 <dd>Number of CRC errors.</dd>
1052 <dt><code>rx_errors</code></dt>
1054 Total number of receive errors, greater than or equal
1055 to the sum of the above.
1062 <dt><code>tx_dropped</code></dt>
1063 <dd>Number of packets dropped by TX.</dd>
1064 <dt><code>collisions</code></dt>
1065 <dd>Number of collisions.</dd>
1066 <dt><code>tx_errors</code></dt>
1068 Total number of transmit errors, greater
1069 than or equal to the sum of the above.
1078 <table name="QoS" title="Quality of Service configuration">
1079 <p>Quality of Service (QoS) configuration for each Port that
1082 <column name="type">
1083 <p>The type of QoS to implement. The <ref table="Open_vSwitch"
1084 column="capabilities"/> column in the <ref table="Open_vSwitch"/> table
1085 identifies the types that a switch actually supports. The currently
1086 defined types are listed below:</p>
1088 <dt><code>linux-htb</code></dt>
1090 Linux ``hierarchy token bucket'' classifier. See tc-htb(8) (also at
1091 <code>http://linux.die.net/man/8/tc-htb</code>) and the HTB manual
1092 (<code>http://luxik.cdi.cz/~devik/qos/htb/manual/userg.htm</code>)
1093 for information on how this classifier works and how to configure it.
1097 <dt><code>linux-hfsc</code></dt>
1099 Linux "Hierarchical Fair Service Curve" classifier.
1100 See <code>http://linux-ip.net/articles/hfsc.en/</code> for
1101 information on how this classifier works.
1106 <column name="queues">
1107 <p>A map from queue numbers to <ref table="Queue"/> records. The
1108 supported range of queue numbers depend on <ref column="type"/>. The
1109 queue numbers are the same as the <code>queue_id</code> used in
1110 OpenFlow in <code>struct ofp_action_enqueue</code> and other
1111 structures. Queue 0 is used by OpenFlow output actions that do not
1112 specify a specific queue.</p>
1115 <column name="other_config">
1116 <p>Key-value pairs for configuring QoS features that depend on
1117 <ref column="type"/>.</p>
1118 <p>The <code>linux-htb</code> and <code>linux-hfsc</code> classes support
1119 the following key-value pairs:</p>
1121 <dt><code>max-rate</code></dt>
1122 <dd>Maximum rate shared by all queued traffic, in bit/s.
1123 Optional. If not specified, for physical interfaces, the
1124 default is the link rate. For other interfaces or if the
1125 link rate cannot be determined, the default is currently 100
1130 <column name="external_ids">
1131 Key-value pairs for use by external frameworks that integrate with Open
1132 vSwitch, rather than by Open vSwitch itself. System integrators should
1133 either use the Open vSwitch development mailing list to coordinate on
1134 common key-value definitions, or choose key names that are likely to be
1135 unique. No common key-value pairs are currently defined.
1139 <table name="Queue" title="QoS output queue.">
1140 <p>A configuration for a port output queue, used in configuring Quality of
1141 Service (QoS) features. May be referenced by <ref column="queues"
1142 table="QoS"/> column in <ref table="QoS"/> table.</p>
1144 <column name="other_config">
1145 <p>Key-value pairs for configuring the output queue. The supported
1146 key-value pairs and their meanings depend on the <ref column="type"/>
1147 of the <ref column="QoS"/> records that reference this row.</p>
1148 <p>The key-value pairs defined for <ref table="QoS"/> <ref table="QoS"
1149 column="type"/> of <code>min-rate</code> are:</p>
1151 <dt><code>min-rate</code></dt>
1152 <dd>Minimum guaranteed bandwidth, in bit/s. Required. The
1153 floor value is 1500 bytes/s (12,000 bit/s).</dd>
1155 <p>The key-value pairs defined for <ref table="QoS"/> <ref table="QoS"
1156 column="type"/> of <code>linux-htb</code> are:</p>
1158 <dt><code>min-rate</code></dt>
1159 <dd>Minimum guaranteed bandwidth, in bit/s. Required.</dd>
1160 <dt><code>max-rate</code></dt>
1161 <dd>Maximum allowed bandwidth, in bit/s. Optional. If specified, the
1162 queue's rate will not be allowed to exceed the specified value, even
1163 if excess bandwidth is available. If unspecified, defaults to no
1165 <dt><code>burst</code></dt>
1166 <dd>Burst size, in bits. This is the maximum amount of ``credits''
1167 that a queue can accumulate while it is idle. Optional. Details of
1168 the <code>linux-htb</code> implementation require a minimum burst
1169 size, so a too-small <code>burst</code> will be silently
1171 <dt><code>priority</code></dt>
1172 <dd>A nonnegative 32-bit integer. Defaults to 0 if
1173 unspecified. A queue with a smaller <code>priority</code>
1174 will receive all the excess bandwidth that it can use before
1175 a queue with a larger value receives any. Specific priority
1176 values are unimportant; only relative ordering matters.</dd>
1178 <p>The key-value pairs defined for <ref table="QoS"/> <ref table="QoS"
1179 column="type"/> of <code>linux-hfsc</code> are:</p>
1181 <dt><code>min-rate</code></dt>
1182 <dd>Minimum guaranteed bandwidth, in bit/s. Required.</dd>
1183 <dt><code>max-rate</code></dt>
1184 <dd>Maximum allowed bandwidth, in bit/s. Optional. If specified, the
1185 queue's rate will not be allowed to exceed the specified value, even
1186 if excess bandwidth is available. If unspecified, defaults to no
1191 <column name="external_ids">
1192 Key-value pairs for use by external frameworks that integrate with Open
1193 vSwitch, rather than by Open vSwitch itself. System integrators should
1194 either use the Open vSwitch development mailing list to coordinate on
1195 common key-value definitions, or choose key names that are likely to be
1196 unique. No common key-value pairs are currently defined.
1200 <table name="Monitor" title="Connectivity Monitor configuration">
1202 A <ref table="Monitor"/> attaches to an <ref table="Interface"/> to
1203 implement 802.1ag Connectivity Fault Management (CFM). CFM allows a
1204 group of Maintenance Points (MPs) called a Maintenance Association (MA)
1205 to detect connectivity problems with each other. MPs within a MA should
1206 have complete and exclusive interconnectivity. This is verified by
1207 occasionally broadcasting Continuity Check Messages (CCMs) at a
1208 configurable transmission interval. A <ref table="Monitor"/> is
1209 responsible for collecting data about other MPs in its MA and
1213 <group title="Monitor Configuration">
1214 <column name="mpid">
1215 A Maintenance Point ID (MPID) uniquely identifies each endpoint within
1216 a Maintenance Association (see <ref column="ma_name"/>). The MPID is
1217 used to identify this <ref table="Monitor"/> to other endpoints in the
1221 <column name="remote_mps">
1222 A set of <ref table="Maintenance_Points"/> which this
1223 <ref table="Monitor"/> should have connectivity to. If this
1224 <ref table="Monitor"/> does not have connectivity to any MPs in this
1225 set, or has connectivity to any MPs not in this set, a fault is
1229 <column name="ma_name">
1230 A Maintenance Association (MA) name pairs with a Maintenance Domain
1231 (MD) name to uniquely identify a MA. A MA is a group of endpoints who
1232 have complete and exclusive interconnectivity. Defaults to
1233 <code>ovs</code> if unset.
1236 <column name="md_name">
1237 A Maintenance Domain name pairs with a Maintenance Association name to
1238 uniquely identify a MA. Defaults to <code>ovs</code> if unset.
1241 <column name="interval">
1242 The transmission interval of CCMs in milliseconds. Three missed CCMs
1243 indicate a connectivity fault. Defaults to 1000ms.
1247 <group title="Monitor Status">
1248 <column name="unexpected_remote_mpids">
1249 A set of MPIDs representing MPs to which this <ref table="Monitor"/>
1250 has detected connectivity that are not in the
1251 <ref column="remote_mps"/> set. This <ref table="Monitor"/> should not
1252 have connectivity to any MPs not listed in <ref column="remote_mps"/>.
1253 Thus, if this set is non-empty a fault is indicated.
1256 <column name="unexpected_remote_maids">
1257 A set of MAIDs representing foreign Maintenance Associations (MAs)
1258 which this <ref table="Monitor"/> has detected connectivity to. A
1259 <ref table="Monitor"/> should not have connectivity to a Maintenance
1260 Association other than its own. Thus, if this set is non-empty a fault
1264 <column name="fault">
1265 Indicates a Connectivity Fault caused by a configuration error, a down
1266 remote MP, or unexpected connectivity to a remote MAID or remote MP.
1271 <table name="Maintenance_Point" title="Maintenance Point configuration">
1273 A <ref table="Maintenance_Point"/> represents a MP which a
1274 <ref table="Monitor"/> has or should have connectivity to.
1277 <group title="Maintenance_Point Configuration">
1278 <column name="mpid">
1279 A Maintenance Point ID (MPID) uniquely identifies each endpoint within
1280 a Maintenance Association. All MPs within a MA should have a unique
1285 <group title="Maintenance_Point Status">
1286 <column name="fault">
1287 Indicates a connectivity fault.
1292 <table name="Mirror" title="Port mirroring (SPAN/RSPAN).">
1293 <p>A port mirror within a <ref table="Bridge"/>.</p>
1294 <p>A port mirror configures a bridge to send selected frames to special
1295 ``mirrored'' ports, in addition to their normal destinations. Mirroring
1296 traffic may also be referred to as SPAN or RSPAN, depending on the
1297 mechanism used for delivery.</p>
1299 <column name="name">
1300 Arbitrary identifier for the <ref table="Mirror"/>.
1303 <group title="Selecting Packets for Mirroring">
1304 <column name="select_all">
1305 If true, every packet arriving or departing on any port is
1306 selected for mirroring.
1309 <column name="select_dst_port">
1310 Ports on which departing packets are selected for mirroring.
1313 <column name="select_src_port">
1314 Ports on which arriving packets are selected for mirroring.
1317 <column name="select_vlan">
1318 VLANs on which packets are selected for mirroring. An empty set
1319 selects packets on all VLANs.
1323 <group title="Mirroring Destination Configuration">
1324 <column name="output_port">
1325 <p>Output port for selected packets, if nonempty. Mutually exclusive
1326 with <ref column="output_vlan"/>.</p>
1327 <p>Specifying a port for mirror output reserves that port exclusively
1328 for mirroring. No frames other than those selected for mirroring
1329 will be forwarded to the port, and any frames received on the port
1330 will be discarded.</p>
1331 <p>This type of mirroring is sometimes called SPAN.</p>
1334 <column name="output_vlan">
1335 <p>Output VLAN for selected packets, if nonempty. Mutually exclusive
1336 with <ref column="output_port"/>.</p>
1337 <p>The frames will be sent out all ports that trunk
1338 <ref column="output_vlan"/>, as well as any ports with implicit VLAN
1339 <ref column="output_vlan"/>. When a mirrored frame is sent out a
1340 trunk port, the frame's VLAN tag will be set to
1341 <ref column="output_vlan"/>, replacing any existing tag; when it is
1342 sent out an implicit VLAN port, the frame will not be tagged. This
1343 type of mirroring is sometimes called RSPAN.</p>
1344 <p><em>Please note:</em> Mirroring to a VLAN can disrupt a network that
1345 contains unmanaged switches. Consider an unmanaged physical switch
1346 with two ports: port 1, connected to an end host, and port 2,
1347 connected to an Open vSwitch configured to mirror received packets
1348 into VLAN 123 on port 2. Suppose that the end host sends a packet on
1349 port 1 that the physical switch forwards to port 2. The Open vSwitch
1350 forwards this packet to its destination and then reflects it back on
1351 port 2 in VLAN 123. This reflected packet causes the unmanaged
1352 physical switch to replace the MAC learning table entry, which
1353 correctly pointed to port 1, with one that incorrectly points to port
1354 2. Afterward, the physical switch will direct packets destined for
1355 the end host to the Open vSwitch on port 2, instead of to the end
1356 host on port 1, disrupting connectivity. If mirroring to a VLAN is
1357 desired in this scenario, then the physical switch must be replaced
1358 by one that learns Ethernet addresses on a per-VLAN basis. In
1359 addition, learning should be disabled on the VLAN containing mirrored
1360 traffic. If this is not done then intermediate switches will learn
1361 the MAC address of each end host from the mirrored traffic. If
1362 packets being sent to that end host are also mirrored, then they will
1363 be dropped since the switch will attempt to send them out the input
1364 port. Disabling learning for the VLAN will cause the switch to
1365 correctly send the packet out all ports configured for that VLAN. If
1366 Open vSwitch is being used as an intermediate switch, learning can be
1367 disabled by adding the mirrored VLAN to <ref column="flood_vlans"/>
1368 in the appropriate <ref table="Bridge"/> table or tables.</p>
1372 <group title="Other Features">
1373 <column name="external_ids">
1374 Key-value pairs for use by external frameworks that integrate with Open
1375 vSwitch, rather than by Open vSwitch itself. System integrators should
1376 either use the Open vSwitch development mailing list to coordinate on
1377 common key-value definitions, or choose key names that are likely to be
1378 unique. No common key-value pairs are currently defined.
1383 <table name="Controller" title="OpenFlow controller configuration.">
1384 <p>An OpenFlow controller.</p>
1387 Open vSwitch supports two kinds of OpenFlow controllers:
1391 <dt>Primary controllers</dt>
1394 This is the kind of controller envisioned by the OpenFlow 1.0
1395 specification. Usually, a primary controller implements a network
1396 policy by taking charge of the switch's flow table.
1400 Open vSwitch initiates and maintains persistent connections to
1401 primary controllers, retrying the connection each time it fails or
1402 drops. The <ref table="Bridge" column="fail_mode"/> column in the
1403 <ref table="Bridge"/> table applies to primary controllers.
1407 Open vSwitch permits a bridge to have any number of primary
1408 controllers. When multiple controllers are configured, Open
1409 vSwitch connects to all of them simultaneously. Because
1410 OpenFlow 1.0 does not specify how multiple controllers
1411 coordinate in interacting with a single switch, more than
1412 one primary controller should be specified only if the
1413 controllers are themselves designed to coordinate with each
1414 other. (The Nicira-defined <code>NXT_ROLE</code> OpenFlow
1415 vendor extension may be useful for this.)
1418 <dt>Service controllers</dt>
1421 These kinds of OpenFlow controller connections are intended for
1422 occasional support and maintenance use, e.g. with
1423 <code>ovs-ofctl</code>. Usually a service controller connects only
1424 briefly to inspect or modify some of a switch's state.
1428 Open vSwitch listens for incoming connections from service
1429 controllers. The service controllers initiate and, if necessary,
1430 maintain the connections from their end. The <ref table="Bridge"
1431 column="fail_mode"/> column in the <ref table="Bridge"/> table does
1432 not apply to service controllers.
1436 Open vSwitch supports configuring any number of service controllers.
1442 The <ref column="target"/> determines the type of controller.
1445 <group title="Core Features">
1446 <column name="target">
1447 <p>Connection method for controller.</p>
1449 The following connection methods are currently supported for primary
1453 <dt><code>ssl:<var>ip</var></code>[<code>:<var>port</var></code>]</dt>
1455 <p>The specified SSL <var>port</var> (default: 6633) on the host at
1456 the given <var>ip</var>, which must be expressed as an IP address
1457 (not a DNS name). The <ref table="Open_vSwitch" column="ssl"/>
1458 column in the <ref table="Open_vSwitch"/> table must point to a
1459 valid SSL configuration when this form is used.</p>
1460 <p>SSL support is an optional feature that is not always built as
1461 part of Open vSwitch.</p>
1463 <dt><code>tcp:<var>ip</var></code>[<code>:<var>port</var></code>]</dt>
1464 <dd>The specified TCP <var>port</var> (default: 6633) on the host at
1465 the given <var>ip</var>, which must be expressed as an IP address
1466 (not a DNS name).</dd>
1467 <dt><code>discover</code></dt>
1469 <p>Enables controller discovery.</p>
1470 <p>In controller discovery mode, Open vSwitch broadcasts a DHCP
1471 request with vendor class identifier <code>OpenFlow</code> across
1472 all of the bridge's network devices. It will accept any valid
1473 DHCP reply that has the same vendor class identifier and includes
1474 a vendor-specific option with code 1 whose contents are a string
1475 specifying the location of the controller in the same format as
1476 <ref column="target"/>.</p>
1477 <p>The DHCP reply may also, optionally, include a vendor-specific
1478 option with code 2 whose contents are a string specifying the URI
1479 to the base of the OpenFlow PKI
1480 (e.g. <code>http://192.168.0.1/openflow/pki</code>). This URI is
1481 used only for bootstrapping the OpenFlow PKI at initial switch
1482 setup; <code>ovs-vswitchd</code> does not use it at all.</p>
1486 The following connection methods are currently supported for service
1490 <dt><code>pssl:</code>[<var>port</var>][<code>:<var>ip</var></code>]</dt>
1493 Listens for SSL connections on the specified TCP <var>port</var>
1494 (default: 6633). If <var>ip</var>, which must be expressed as an
1495 IP address (not a DNS name), is specified, then connections are
1496 restricted to the specified local IP address.
1499 The <ref table="Open_vSwitch" column="ssl"/> column in the <ref
1500 table="Open_vSwitch"/> table must point to a valid SSL
1501 configuration when this form is used.
1503 <p>SSL support is an optional feature that is not always built as
1504 part of Open vSwitch.</p>
1506 <dt><code>ptcp:</code>[<var>port</var>][<code>:<var>ip</var></code>]</dt>
1508 Listens for connections on the specified TCP <var>port</var>
1509 (default: 6633). If <var>ip</var>, which must be expressed as an
1510 IP address (not a DNS name), is specified, then connections are
1511 restricted to the specified local IP address.
1514 <p>When multiple controllers are configured for a single bridge, the
1515 <ref column="target"/> values must be unique. Duplicate
1516 <ref column="target"/> values yield unspecified results.</p>
1519 <column name="connection_mode">
1520 <p>If it is specified, this setting must be one of the following
1521 strings that describes how Open vSwitch contacts this OpenFlow
1522 controller over the network:</p>
1525 <dt><code>in-band</code></dt>
1526 <dd>In this mode, this controller's OpenFlow traffic travels over the
1527 bridge associated with the controller. With this setting, Open
1528 vSwitch allows traffic to and from the controller regardless of the
1529 contents of the OpenFlow flow table. (Otherwise, Open vSwitch
1530 would never be able to connect to the controller, because it did
1531 not have a flow to enable it.) This is the most common connection
1532 mode because it is not necessary to maintain two independent
1534 <dt><code>out-of-band</code></dt>
1535 <dd>In this mode, OpenFlow traffic uses a control network separate
1536 from the bridge associated with this controller, that is, the
1537 bridge does not use any of its own network devices to communicate
1538 with the controller. The control network must be configured
1539 separately, before or after <code>ovs-vswitchd</code> is started.
1543 <p>If not specified, the default is implementation-specific. If
1544 <ref column="target"/> is <code>discover</code>, the connection mode
1545 is always treated as <code>in-band</code> regardless of the actual
1550 <group title="Controller Failure Detection and Handling">
1551 <column name="max_backoff">
1552 Maximum number of milliseconds to wait between connection attempts.
1553 Default is implementation-specific.
1556 <column name="inactivity_probe">
1557 Maximum number of milliseconds of idle time on connection to
1558 controller before sending an inactivity probe message. If Open
1559 vSwitch does not communicate with the controller for the specified
1560 number of seconds, it will send a probe. If a response is not
1561 received for the same additional amount of time, Open vSwitch
1562 assumes the connection has been broken and attempts to reconnect.
1563 Default is implementation-specific.
1567 <group title="OpenFlow Rate Limiting">
1568 <column name="controller_rate_limit">
1569 <p>The maximum rate at which packets in unknown flows will be
1570 forwarded to the OpenFlow controller, in packets per second. This
1571 feature prevents a single bridge from overwhelming the controller.
1572 If not specified, the default is implementation-specific.</p>
1573 <p>In addition, when a high rate triggers rate-limiting, Open
1574 vSwitch queues controller packets for each port and transmits
1575 them to the controller at the configured rate. The number of
1576 queued packets is limited by
1577 the <ref column="controller_burst_limit"/> value. The packet
1578 queue is shared fairly among the ports on a bridge.</p><p>Open
1579 vSwitch maintains two such packet rate-limiters per bridge.
1580 One of these applies to packets sent up to the controller
1581 because they do not correspond to any flow. The other applies
1582 to packets sent up to the controller by request through flow
1583 actions. When both rate-limiters are filled with packets, the
1584 actual rate that packets are sent to the controller is up to
1585 twice the specified rate.</p>
1588 <column name="controller_burst_limit">
1589 In conjunction with <ref column="controller_rate_limit"/>,
1590 the maximum number of unused packet credits that the bridge will
1591 allow to accumulate, in packets. If not specified, the default
1592 is implementation-specific.
1596 <group title="Additional Discovery Configuration">
1597 <p>These values are considered only when <ref column="target"/>
1598 is <code>discover</code>.</p>
1600 <column name="discover_accept_regex">
1602 extended regular expression against which the discovered controller
1603 location is validated. The regular expression is implicitly
1604 anchored at the beginning of the controller location string, as
1605 if it begins with <code>^</code>. If not specified, the default
1606 is implementation-specific.
1609 <column name="discover_update_resolv_conf">
1610 Whether to update <code>/etc/resolv.conf</code> when the
1611 controller is discovered. If not specified, the default
1612 is implementation-specific. Open vSwitch will only modify
1613 <code>/etc/resolv.conf</code> if the DHCP response that it receives
1614 specifies one or more DNS servers.
1618 <group title="Additional In-Band Configuration">
1619 <p>These values are considered only in in-band control mode (see
1620 <ref column="connection_mode"/>) and only when <ref column="target"/>
1621 is not <code>discover</code>. (For controller discovery, the network
1622 configuration obtained via DHCP is used instead.)</p>
1624 <p>When multiple controllers are configured on a single bridge, there
1625 should be only one set of unique values in these columns. If different
1626 values are set for these columns in different controllers, the effect
1629 <column name="local_ip">
1630 The IP address to configure on the local port,
1631 e.g. <code>192.168.0.123</code>. If this value is unset, then
1632 <ref column="local_netmask"/> and <ref column="local_gateway"/> are
1636 <column name="local_netmask">
1637 The IP netmask to configure on the local port,
1638 e.g. <code>255.255.255.0</code>. If <ref column="local_ip"/> is set
1639 but this value is unset, then the default is chosen based on whether
1640 the IP address is class A, B, or C.
1643 <column name="local_gateway">
1644 The IP address of the gateway to configure on the local port, as a
1645 string, e.g. <code>192.168.0.1</code>. Leave this column unset if
1646 this network has no gateway.
1650 <group title="Other Features">
1651 <column name="external_ids">
1652 Key-value pairs for use by external frameworks that integrate with Open
1653 vSwitch, rather than by Open vSwitch itself. System integrators should
1654 either use the Open vSwitch development mailing list to coordinate on
1655 common key-value definitions, or choose key names that are likely to be
1656 unique. No common key-value pairs are currently defined.
1661 <table name="Manager" title="OVSDB management connection.">
1663 Configuration for a database connection to an Open vSwitch database
1668 This table primarily configures the Open vSwitch database
1669 (<code>ovsdb-server</code>), not the Open vSwitch switch
1670 (<code>ovs-vswitchd</code>). The switch does read the table to determine
1671 what connections should be treated as in-band.
1675 The Open vSwitch database server can initiate and maintain active
1676 connections to remote clients. It can also listen for database
1680 <group title="Core Features">
1681 <column name="target">
1682 <p>Connection method for managers.</p>
1684 The following connection methods are currently supported:
1687 <dt><code>ssl:<var>ip</var></code>[<code>:<var>port</var></code>]</dt>
1690 The specified SSL <var>port</var> (default: 6632) on the host at
1691 the given <var>ip</var>, which must be expressed as an IP address
1692 (not a DNS name). The <ref table="Open_vSwitch" column="ssl"/>
1693 column in the <ref table="Open_vSwitch"/> table must point to a
1694 valid SSL configuration when this form is used.
1697 SSL support is an optional feature that is not always built as
1698 part of Open vSwitch.
1702 <dt><code>tcp:<var>ip</var></code>[<code>:<var>port</var></code>]</dt>
1704 The specified TCP <var>port</var> (default: 6632) on the host at
1705 the given <var>ip</var>, which must be expressed as an IP address
1708 <dt><code>pssl:</code>[<var>port</var>][<code>:<var>ip</var></code>]</dt>
1711 Listens for SSL connections on the specified TCP <var>port</var>
1712 (default: 6632). If <var>ip</var>, which must be expressed as an
1713 IP address (not a DNS name), is specified, then connections are
1714 restricted to the specified local IP address.
1717 The <ref table="Open_vSwitch" column="ssl"/> column in the <ref
1718 table="Open_vSwitch"/> table must point to a valid SSL
1719 configuration when this form is used.
1722 SSL support is an optional feature that is not always built as
1723 part of Open vSwitch.
1726 <dt><code>ptcp:</code>[<var>port</var>][<code>:<var>ip</var></code>]</dt>
1728 Listens for connections on the specified TCP <var>port</var>
1729 (default: 6632). If <var>ip</var>, which must be expressed as an
1730 IP address (not a DNS name), is specified, then connections are
1731 restricted to the specified local IP address.
1734 <p>When multiple managers are configured, the <ref column="target"/>
1735 values must be unique. Duplicate <ref column="target"/> values yield
1736 unspecified results.</p>
1739 <column name="connection_mode">
1741 If it is specified, this setting must be one of the following strings
1742 that describes how Open vSwitch contacts this OVSDB client over the
1747 <dt><code>in-band</code></dt>
1749 In this mode, this connection's traffic travels over a bridge
1750 managed by Open vSwitch. With this setting, Open vSwitch allows
1751 traffic to and from the client regardless of the contents of the
1752 OpenFlow flow table. (Otherwise, Open vSwitch would never be able
1753 to connect to the client, because it did not have a flow to enable
1754 it.) This is the most common connection mode because it is not
1755 necessary to maintain two independent networks.
1757 <dt><code>out-of-band</code></dt>
1759 In this mode, the client's traffic uses a control network separate
1760 from that managed by Open vSwitch, that is, Open vSwitch does not
1761 use any of its own network devices to communicate with the client.
1762 The control network must be configured separately, before or after
1763 <code>ovs-vswitchd</code> is started.
1768 If not specified, the default is implementation-specific.
1773 <group title="Client Failure Detection and Handling">
1774 <column name="max_backoff">
1775 Maximum number of milliseconds to wait between connection attempts.
1776 Default is implementation-specific.
1779 <column name="inactivity_probe">
1780 Maximum number of milliseconds of idle time on connection to the client
1781 before sending an inactivity probe message. If Open vSwitch does not
1782 communicate with the client for the specified number of seconds, it
1783 will send a probe. If a response is not received for the same
1784 additional amount of time, Open vSwitch assumes the connection has been
1785 broken and attempts to reconnect. Default is implementation-specific.
1789 <group title="Other Features">
1790 <column name="external_ids">
1791 Key-value pairs for use by external frameworks that integrate with Open
1792 vSwitch, rather than by Open vSwitch itself. System integrators should
1793 either use the Open vSwitch development mailing list to coordinate on
1794 common key-value definitions, or choose key names that are likely to be
1795 unique. No common key-value pairs are currently defined.
1800 <table name="NetFlow">
1801 A NetFlow target. NetFlow is a protocol that exports a number of
1802 details about terminating IP flows, such as the principals involved
1805 <column name="targets">
1806 NetFlow targets in the form
1807 <code><var>ip</var>:<var>port</var></code>. The <var>ip</var>
1808 must be specified numerically, not as a DNS name.
1811 <column name="engine_id">
1812 Engine ID to use in NetFlow messages. Defaults to datapath index
1816 <column name="engine_type">
1817 Engine type to use in NetFlow messages. Defaults to datapath
1818 index if not specified.
1821 <column name="active_timeout">
1822 The interval at which NetFlow records are sent for flows that are
1823 still active, in seconds. A value of <code>0</code> requests the
1824 default timeout (currently 600 seconds); a value of <code>-1</code>
1825 disables active timeouts.
1828 <column name="add_id_to_interface">
1829 <p>If this column's value is <code>false</code>, the ingress and egress
1830 interface fields of NetFlow flow records are derived from OpenFlow port
1831 numbers. When it is <code>true</code>, the 7 most significant bits of
1832 these fields will be replaced by the least significant 7 bits of the
1833 engine id. This is useful because many NetFlow collectors do not
1834 expect multiple switches to be sending messages from the same host, so
1835 they do not store the engine information which could be used to
1836 disambiguate the traffic.</p>
1837 <p>When this option is enabled, a maximum of 508 ports are supported.</p>
1840 <column name="external_ids">
1841 Key-value pairs for use by external frameworks that integrate with Open
1842 vSwitch, rather than by Open vSwitch itself. System integrators should
1843 either use the Open vSwitch development mailing list to coordinate on
1844 common key-value definitions, or choose key names that are likely to be
1845 unique. No common key-value pairs are currently defined.
1850 SSL configuration for an Open_vSwitch.
1852 <column name="private_key">
1853 Name of a PEM file containing the private key used as the switch's
1854 identity for SSL connections to the controller.
1857 <column name="certificate">
1858 Name of a PEM file containing a certificate, signed by the
1859 certificate authority (CA) used by the controller and manager,
1860 that certifies the switch's private key, identifying a trustworthy
1864 <column name="ca_cert">
1865 Name of a PEM file containing the CA certificate used to verify
1866 that the switch is connected to a trustworthy controller.
1869 <column name="bootstrap_ca_cert">
1870 If set to <code>true</code>, then Open vSwitch will attempt to
1871 obtain the CA certificate from the controller on its first SSL
1872 connection and save it to the named PEM file. If it is successful,
1873 it will immediately drop the connection and reconnect, and from then
1874 on all SSL connections must be authenticated by a certificate signed
1875 by the CA certificate thus obtained. <em>This option exposes the
1876 SSL connection to a man-in-the-middle attack obtaining the initial
1877 CA certificate.</em> It may still be useful for bootstrapping.
1880 <column name="external_ids">
1881 Key-value pairs for use by external frameworks that integrate with Open
1882 vSwitch, rather than by Open vSwitch itself. System integrators should
1883 either use the Open vSwitch development mailing list to coordinate on
1884 common key-value definitions, or choose key names that are likely to be
1885 unique. No common key-value pairs are currently defined.
1889 <table name="sFlow">
1890 <p>An sFlow(R) target. sFlow is a protocol for remote monitoring
1893 <column name="agent">
1894 Name of the network device whose IP address should be reported as the
1895 ``agent address'' to collectors. If not specified, the IP address
1896 defaults to the <ref table="Controller" column="local_ip"/> in the
1897 collector's <ref table="Controller"/>. If an agent IP address cannot be
1898 determined either way, sFlow is disabled.
1901 <column name="header">
1902 Number of bytes of a sampled packet to send to the collector.
1903 If not specified, the default is 128 bytes.
1906 <column name="polling">
1907 Polling rate in seconds to send port statistics to the collector.
1908 If not specified, defaults to 30 seconds.
1911 <column name="sampling">
1912 Rate at which packets should be sampled and sent to the collector.
1913 If not specified, defaults to 400, which means one out of 400
1914 packets, on average, will be sent to the collector.
1917 <column name="targets">
1918 sFlow targets in the form
1919 <code><var>ip</var>:<var>port</var></code>.
1922 <column name="external_ids">
1923 Key-value pairs for use by external frameworks that integrate with Open
1924 vSwitch, rather than by Open vSwitch itself. System integrators should
1925 either use the Open vSwitch development mailing list to coordinate on
1926 common key-value definitions, or choose key names that are likely to be
1927 unique. No common key-value pairs are currently defined.
1931 <table name="Capability">
1932 <p>Records in this table describe functionality supported by the hardware
1933 and software platform on which this Open vSwitch is based. Clients
1934 should not modify this table.</p>
1936 <p>A record in this table is meaningful only if it is referenced by the
1937 <ref table="Open_vSwitch" column="capabilities"/> column in the
1938 <ref table="Open_vSwitch"/> table. The key used to reference it, called
1939 the record's ``category,'' determines the meanings of the
1940 <ref column="details"/> column. The following general forms of
1941 categories are currently defined:</p>
1944 <dt><code>qos-<var>type</var></code></dt>
1945 <dd><var>type</var> is supported as the value for
1946 <ref column="type" table="QoS"/> in the <ref table="QoS"/> table.
1950 <column name="details">
1951 <p>Key-value pairs that describe capabilities. The meaning of the pairs
1952 depends on the category key that the <ref table="Open_vSwitch"
1953 column="capabilities"/> column in the <ref table="Open_vSwitch"/> table
1954 uses to reference this record, as described above.</p>
1956 <p>The presence of a record for category <code>qos-<var>type</var></code>
1957 indicates that the switch supports <var>type</var> as the value of
1958 the <ref table="QoS" column="type"/> column in the <ref table="QoS"/>
1959 table. The following key-value pairs are defined to further describe
1960 QoS capabilities:</p>
1963 <dt><code>n-queues</code></dt>
1964 <dd>Number of supported queues, as a positive integer. Keys in the
1965 <ref table="QoS" column="queues"/> column for <ref table="QoS"/>
1966 records whose <ref table="QoS" column="type"/> value
1967 equals <var>type</var> must range between 0 and this value minus one,