X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=vswitchd%2Fvswitch.xml;h=c9d0dc438aa9209ef27fd810909f116cd25abdde;hb=7beaa082d7f422dda337780d6dd457b9a5ffe7d5;hp=5789964a8da758787ac90d8f014222bee44302f7;hpb=573c1db99dbfd655d20e0fdf27ff4d460d6529c1;p=openvswitch
diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml
index 5789964a..c9d0dc43 100644
--- a/vswitchd/vswitch.xml
+++ b/vswitchd/vswitch.xml
@@ -1,43 +1,75 @@
+
A database with this schema holds the configuration for one Open
- vSwitch daemon. The root of the configuration for the daemon is
- the table, which must have exactly one
+
+ A database with this schema holds the configuration for one Open
+ vSwitch daemon. The top-level configuration for the daemon is the
+ table, which must have exactly one
record. Records in other tables are significant only when they
- can be reached directly or indirectly from the
- table.
+ Most tables contain two special columns, named other_config
+ and external_ids
. These columns have the same form and
+ purpose each place that they appear, so we describe them here to save space
+ later.
+
other_config
: map of string-string pairs+ Key-value pairs for configuring rarely used features. Supported keys, + along with the forms taken by their values, are documented individually + for each table. +
+
+ A few tables do not have other_config
columns because no
+ key-value pairs have yet been defined for them.
+
external_ids
: map of string-string pairsEthernet address to set for this interface. If unset then the - default MAC address is used:
+ default MAC address is used:Some interfaces may not have a software-controllable MAC address.
@@ -409,454 +1154,1199 @@OpenFlow port number for this interface. Unlike most columns, this - column's value should be set only by Open vSwitch itself. Other - clients should set this column to an empty set (the default) when - creating an .
+ column's value should be set only by Open vSwitch itself. Other + clients should set this column to an empty set (the default) when + creating an .Open vSwitch populates this column when the port number becomes - known. If the interface is successfully added, - will be set to a number between 1 and 65535 - (generally either in the range 1 to 65280, exclusive, or 65534, the - port number for the OpenFlow ``local port''). If the interface - cannot be added then Open vSwitch sets this column - to -1.
+ known. If the interface is successfully added, + will be set to a number between 1 and 65535 + (generally either in the range 1 to 65279, inclusive, or 65534, the + port number for the OpenFlow ``local port''). If the interface + cannot be added then Open vSwitch sets this column + to -1. +Requested OpenFlow port number for this interface. The port + number must be between 1 and 65279, inclusive. Some datapaths + cannot satisfy all requests for particular port numbers. When + this column is empty or the request cannot be fulfilled, the + system will choose a free port. The + column reports the assigned OpenFlow port number.
+The port number must be requested in the same transaction + that creates the port.
+ The interface type, one of: +
+system
eth0
on Linux.
- Sometimes referred to as ``external interfaces'' since they are
- generally connected to hardware external to that on which the Open
- vSwitch is running. The empty string is a synonym for
- system
.system
.
+
internal
tap
gre
remote_ip
, local_ip
, and
- in_key
. Note that if two ports are defined that are
- the same except one has an optional identifier and the other does
- not, the more specific one is matched first. in_key
- is considered more specific than local_ip
if a port
- defines one and another port defines the other. The following
- options may be specified in the column:
- remote_ip
local_ip
in_key
flow
. If
- flow
is specified then any key will be accepted
- and the key will be placed in the tun_id
field
- for matching in the flow table. The ovs-ofctl manual page
- contains additional information about matching fields in
- OpenFlow flows. Default is no key.out_key
flow
. If
- flow
is specified then the key may be set using
- the set_tunnel
Nicira OpenFlow vendor extension (0
- is used in the absence of an action). The ovs-ofctl manual
- page contains additional information about the Nicira OpenFlow
- vendor extensions. Default is no key.key
in_key
and
- out_key
at the same time.tos
inherit
, in which case the ToS will be copied from
- the inner packet if it is IPv4 or IPv6 (otherwise it will be
- 0). Note that the ECN fields are always inherited. Default is
- 0.ttl
inherit
, in which case the
- TTL will be copied from the inner packet if it is IPv4 or IPv6
- (otherwise it will be the system default, typically 64).
- Default is the system default TTL.csum
true
to enable.pmtud
false
to disable.capwap
remote_ip
and
- local_ip
. If two ports are defined that are the same
- except one includes local_ip
and the other does not,
- the more specific one is matched first. CAPWAP support is not
- available on all platforms. Currently it is only supported in the
- Linux kernel module with kernel versions >= 2.6.25. The following
- options may be specified in the column:
- remote_ip
local_ip
tos
inherit
, in which case the ToS will be copied from
- the inner packet if it is IPv4 or IPv6 (otherwise it will be
- 0). Note that the ECN fields are always inherited. Default is
- 0.ttl
inherit
, in which case the
- TTL will be copied from the inner packet if it is IPv4 or IPv6
- (otherwise it will be the system default, typically 64).
- Default is the system default TTL.pmtud
false
to disable.ipsec_gre
gre64
ipsec_gre64
capwap
patch
- A pair of virtual devices that act as a patch cable. The column must have the following key-value pair: -
- + A pair of virtual devices that act as a patch cable.null
+ These options apply to interfaces with of
+ gre
, ipsec_gre
, gre64
,
+ ipsec_gre64
, and capwap
.
+
+ Each tunnel must be uniquely identified by the combination of , , , and . If two ports are defined that are the same except one + has an optional identifier and the other does not, the more specific + one is matched first. is + considered more specific than if + a port defines one and another port defines the other. +
+ ++ Required. The tunnel endpoint. Unicast and multicast endpoints are + both supported. +
+ ++ When a multicast endpoint is specified, a routing table lookup occurs + only when the tunnel is created. Following a routing change, delete + and then re-create the tunnel to force a new routing table lookup. +
Optional. The key that received packets must contain, one of:
+ +0
. The tunnel receives packets with no key or with a
+ key of 0. This is equivalent to specifying no at all.
+ flow
. The tunnel accepts packets with any
+ key. The key will be placed in the tun_id
field for
+ matching in the flow table. The ovs-ofctl
manual page
+ contains additional information about matching fields in OpenFlow
+ flows.
+
- Key-value pairs that report port status. Supported status
- values are type
-dependent.
The only currently defined key-value pair is:
-source_ip
gre
or capwap
. Not
- supported by all implementations.Optional. The key to be set on outgoing packets, one of:
+ +0
. Packets sent through the tunnel will have no key.
+ This is equivalent to specifying no at all.
+ flow
. Packets sent through the tunnel will
+ have the key set using the set_tunnel
Nicira OpenFlow
+ vendor extension (0 is used in the absence of an action). The
+ ovs-ofctl
manual page contains additional information
+ about the Nicira OpenFlow vendor extensions.
+ in_key
and
+ out_key
at the same time.
+ inherit
, in which case
+ the ToS will be copied from the inner packet if it is IPv4 or IPv6
+ (otherwise it will be 0). The ECN fields are always inherited.
+ Default is 0.
+ inherit
, in which case the TTL will be copied
+ from the inner packet if it is IPv4 or IPv6 (otherwise it will be the
+ system default, typically 64). Default is the system default TTL.
+ true
to
+ enable.
+ df_inherit
option is not set, or if
+ the encapsulated packet is not IP. Default is enabled; set to
+ false
to disable.
+ true
to enable. This feature is
+ deprecated and will be removed soon.
+
+ Only gre
interfaces support these options.
+
+ Only gre
and ipsec_gre
interfaces support
+ these options.
+
+ Optional. Compute GRE checksums on outgoing packets. Default is
+ disabled, set to true
to enable. Checksums present on
+ incoming packets will be validated regardless of this setting.
+
+ GRE checksums impose a significant performance penalty because they + cover the entire packet. The encapsulated L3, L4, and L7 packet + contents typically have their own checksums, so this additional + checksum only adds value for the GRE and encapsulated L2 headers. +
+ +
+ This option is supported for ipsec_gre
, but not useful
+ because GRE checksums are weaker than, and redundant with, IPsec
+ payload authentication.
+
+ Only ipsec_gre
interfaces support these options.
+
certificate
+ option.
+ certificate
.
+ If certificate
contains the private key, this option may
+ be omitted.
+
+ Only patch
interfaces support these options.
+
peer
option must specify this 's
+ name. That is, the two patch interfaces must have reversed and peer
values.
+ Status information about interfaces attached to bridges, updated every + 5 seconds. Not all interfaces have all of these properties; virtual + interfaces don't have a link speed, for example. Non-applicable + columns will have empty values. +
++ The administrative state of the physical network link. +
++ The observed state of the physical network link. This is ordinarily + the link's carrier status. If the interface's is + a bond configured for miimon monitoring, it is instead the network + link's miimon status. +
++ The number of times Open vSwitch has observed the + of this change. +
++ The negotiated speed of the physical network link. + Valid values are positive integers greater than 0. +
++ The duplex mode of the physical network link. +
++ The MTU (maximum transmission unit); i.e. the largest + amount of data that can fit into a single Ethernet frame. + The standard Ethernet MTU is 1500 bytes. Some physical media + and many kinds of virtual interfaces can be configured with + higher MTUs. +
++ This column will be empty for an interface that does not + have an MTU as, for example, some kinds of tunnels do not. +
+gre
or capwap
.
+
+ Key-value pairs that report interface statistics. The current
+ implementation updates these counters periodically. Future
+ implementations may update them when an interface is created, when they
+ are queried (e.g. using an OVSDB select
operation), and
+ just before an interface is deleted due to virtual interface hot-unplug
+ or VM shutdown, and perhaps at other times, but not on any regular
+ periodic basis.
+
+ These are the same statistics reported by OpenFlow in its struct
+ ofp_port_stats
structure. If an interface does not support a
+ given statistic, then that pair is omitted.
+
+ These settings control ingress policing for packets received on this + interface. On a physical interface, this limits the rate at which + traffic is allowed into the system from the outside; on a virtual + interface (one connected to a virtual machine), this limits the rate at + which the VM is able to transmit. +
++ Policing is a simple form of quality-of-service that simply drops + packets received in excess of the configured rate. Due to its + simplicity, policing is usually less accurate and less effective than + egress QoS (which is configured using the and tables). +
++ Policing is currently implemented only on Linux. The Linux + implementation uses a simple ``token bucket'' approach: +
++ Policing interacts badly with some network protocols, and especially + with fragmented IP packets. Suppose that there is enough network + activity to keep the bucket nearly empty all the time. Then this token + bucket algorithm will forward a single packet every so often, with the + period depending on packet size and on the configured rate. All of the + fragments of an IP packets are normally transmitted back-to-back, as a + group. In such a situation, therefore, only one of these fragments + will be forwarded and the rest will be dropped. IP does not provide + any way for the intended recipient to ask for only the remaining + fragments. In such a case there are two likely possibilities for what + will happen next: either all of the fragments will eventually be + retransmitted (as TCP will do), in which case the same problem will + recur, or the sender will not realize that its packet has been dropped + and data will simply be lost (as some UDP-based protocols will do). + Either way, it is possible that no forward progress will ever occur. +
+
+ Maximum rate for data received on this interface, in kbps. Data
+ received faster than this rate is dropped. Set to 0
+ (the default) to disable policing.
+
Maximum burst size for data received on this interface, in kb. The
- default burst size if set to 0
is 1000 kb. This value
- has no effect if
- is 0
.
The burst size should be at least the size of the interface's - MTU.
+ default burst size if set to0
is 1000 kb. This value
+ has no effect if
+ is 0
.
+ + Specifying a larger burst size lets the algorithm be more forgiving, + which is important for protocols like TCP that react severely to + dropped packets. The burst size should be at least the size of the + interface's MTU. Specifying a value that is numerically at least as + large as 10% of helps TCP come + closer to achieving the full rate. +
Maximum rate for data received on this interface, in kbps. Data
- received faster than this rate is dropped. Set to 0
to
- disable policing.
The meaning of ``ingress'' is from Open vSwitch's perspective. If - configured on a physical interface, then it limits the rate at which - traffic is allowed into the system from the outside. If configured - on a virtual interface that is connected to a virtual machine, then - it limits the rate at which the guest is able to transmit.
++ 802.1ag Connectivity Fault Management (CFM) allows a group of + Maintenance Points (MPs) called a Maintenance Association (MA) to + detect connectivity problems with each other. MPs within a MA should + have complete and exclusive interconnectivity. This is verified by + occasionally broadcasting Continuity Check Messages (CCMs) at a + configurable transmission interval. +
+ ++ According to the 802.1ag specification, each Maintenance Point should + be configured out-of-band with a list of Remote Maintenance Points it + should have connectivity to. Open vSwitch differs from the + specification in this area. It simply assumes the link is faulted if + no Remote Maintenance Points are reachable, and considers it not + faulted otherwise. +
+ +
+ When operating over tunnels which have no in_key
, or an
+ in_key
of flow
. CFM will only accept CCMs
+ with a tunnel key of zero.
+
+ Indicates a connectivity fault triggered by an inability to receive + heartbeats from any remote endpoint. When a fault is triggered on + s participating in bonds, they will be + disabled. +
++ Faults can be triggered for several reasons. Most importantly they + are triggered when no CCMs are received for a period of 3.5 times the + transmission interval. Faults are also triggered when any CCMs + indicate that a Remote Maintenance Point is not receiving CCMs but + able to send them. Finally, a fault is triggered if a CCM is + received which indicates unexpected configuration. Notably, this + case arises when a CCM is received which advertises the local MPID. +
+ovs-appctl
command.
+ When in extended mode, indicates the operational state of the
+ remote endpoint as either up
or down
. See
+ .
+
+ Indicates the health of the interface as a percentage of CCM frames + received over 21 s. + The health of an interface is undefined if it is communicating with + more than one . It reduces if + healthy heartbeats are not received at the expected rate, and + gradually improves as healthy heartbeats are received at the desired + rate. Every 21 s, the + health of the interface is refreshed. +
++ As mentioned above, the faults can be triggered for several reasons. + The link health will deteriorate even if heartbeats are received but + they are reported to be unhealthy. An unhealthy heartbeat in this + context is a heartbeat for which either some fault is set or is out + of sequence. The interface health can be 100 only on receiving + healthy heartbeats at the desired rate. +
++ The interval, in milliseconds, between transmissions of CFM + heartbeats. Three missed heartbeat receptions indicate a + connectivity fault. +
+ ++ In standard operation only intervals of 3, 10, 100, 1,000, 10,000, + 60,000, or 600,000 ms are supported. Other values will be rounded + down to the nearest value on the list. Extended mode (see ) supports any interval up + to 65,535 ms. In either mode, the default is 1000 ms. +
+ +We do not recommend using intervals less than 100 ms.
+true
, the CFM module operates in extended mode. This
+ causes it to use a nonstandard destination address to avoid conflicting
+ with compliant implementations which may be running concurrently on the
+ network. Furthermore, extended mode increases the accuracy of the
+ cfm_interval
configuration parameter by breaking wire
+ compatibility with 802.1ag compliant implementations. Defaults to
+ false
.
+ down
, the CFM module marks all CCMs it generates as
+ operationally down without triggering a fault. This allows remote
+ maintenance points to choose not to forward traffic to the
+ on which this CFM module is running.
+ Currently, in Open vSwitch, the opdown bit of CCMs affects
+ s participating in bonds, and the bundle
+ OpenFlow action. This setting is ignored when CFM is not in extended
+ mode. Defaults to up
.
+ random
in which
+ case each CCM will be tagged with a different randomly generated VLAN.
+ stable
bond mode to make slave
+ selection decisions. Allocating values consistently across interfaces
+ participating in a bond will guarantee consistent slave selection
+ decisions across ovs-vswitchd
instances when using
+ stable
bonding mode.
+
+ These key-value pairs specifically apply to an interface that
+ represents a virtual Ethernet interface connected to a virtual
+ machine. These key-value pairs should not be present for other types
+ of interfaces. Keys whose names end in -uuid
have
+ values that uniquely identify the entity in question. For a Citrix
+ XenServer hypervisor, these values are UUIDs in RFC 4122 format.
+ Other hypervisors may use other formats.
+
MAC
field
+ in the VIF record for this interface.
+
- Key-value pairs for use by external frameworks that integrate
- with Open vSwitch, rather than by Open vSwitch itself. System
- integrators should either use the Open vSwitch development
- mailing list to coordinate on common key-value definitions, or
- choose key names that are likely to be unique.
+ Hypervisors may sometimes have more than one interface associated
+ with a given , only one of
+ which is actually in use at a given time. For example, in some
+ circumstances XenServer has both a ``tap'' and a ``vif'' interface
+ for a single , but only
+ uses one of them at a time. A hypervisor that behaves this way must
+ mark the currently in use interface active
and the
+ others inactive
. A hypervisor that never has more than
+ one interface for a given
+ may mark that interface active
or omit entirely.
- All of the currently defined key-value pairs specifically
- apply to an interface that represents a virtual Ethernet interface
- connected to a virtual machine. These key-value pairs should not be
- present for other types of interfaces. Keys whose names end
- in -uuid
have values that uniquely identify the entity
- in question. For a Citrix XenServer hypervisor, these values are
- UUIDs in RFC 4122 format. Other hypervisors may use other
- formats.
+ During VM migration, a given might transiently be marked active
on
+ two different hypervisors. That is, active
means that
+ this is the active
+ instance within a single hypervisor, not in a broader scope.
The currently defined key-value pairs are:
-vif-uuid
network-uuid
vm-uuid
vif-mac
MAC
- field in the VIF record for this interface.+ The ``VLAN splinters'' feature increases Open vSwitch compatibility + with buggy network drivers in old versions of Linux that do not + properly support VLANs when VLAN devices are not used, at some cost + in memory and performance. +
+ ++ When VLAN splinters are enabled on a particular interface, Open vSwitch + creates a VLAN device for each in-use VLAN. For sending traffic tagged + with a VLAN on the interface, it substitutes the VLAN device. Traffic + received on the VLAN device is treated as if it had been received on + the interface on the particular VLAN. +
+ ++ VLAN splinters consider a VLAN to be in use if: +
+ ++ The same set of in-use VLANs applies to every interface on which VLAN + splinters are enabled. That is, the set is not chosen separately for + each interface but selected once as the union of all in-use VLANs based + on the rules above. +
+ ++ It does not make sense to enable VLAN splinters on an interface for an + access port, or on an interface that is not a physical port. +
+ ++ VLAN splinters are deprecated. When broken device drivers are no + longer in widespread use, we will delete this feature. +
+ +
- Key-value pairs that report interface statistics. The current
- implementation updates these counters periodically. In the future,
- we plan to, instead, update them when an interface is created, when
- they are queried (e.g. using an OVSDB select
operation),
- and just before an interface is deleted due to virtual interface
- hot-unplug or VM shutdown, and perhaps at other times, but not on any
- regular periodic basis.
true
to enable VLAN splinters on this interface.
+ Defaults to false
.
+
+
- The currently defined key-value pairs are listed below. These are
- the same statistics reported by OpenFlow in its struct
- ofp_port_stats
structure. If an interface does not support a
- given statistic, then that pair is omitted.
rx_packets
rx_bytes
tx_packets
tx_bytes
rx_dropped
rx_frame_err
rx_over_err
rx_crc_err
rx_errors
tx_dropped
collisions
tx_errors
+ VLAN splinters do not support 802.1p priority tags. Received + priorities will appear to be 0, regardless of their actual values, + and priorities on transmitted packets will also be cleared to 0. +
Common
+ Columns
at the beginning of this document.
+
+