Network attributes
Find instructions for managing network attributes.
You can use several of the Virtual I/O Server (VIOS) commands, including chdev, mkvdev, and cfglnagg, to change device or network attributes. This section defines attributes that can be modified.
Ethernet attributes
| Attribute | Description |
|---|---|
| Maximum Transmission Unit (mtu) | Specifies maximum transmission unit (MTU). This value can be any number from 60 through 65535, but it is media dependent. |
| Interface State (state) |
|
| Network Mask (netmask) | Specifies how much of the address to reserve
for subdividing networks into subnetworks. The mask includes both the network part of the local address and the subnet part, which is taken from the host field of the address. The mask can be specified as a single hexadecimal number beginning with 0x, in standard Internet dotted-decimal notation. In the 32-bit address, the mask contains bits with a value of 1 for the bit positions reserved for the network and subnet parts, and a bit with the value of 0 for the bit positions that specify the host. The mask contains the standard network portion, and the subnet segment is contiguous with the network segment. |
Shared Ethernet Adapter attributes
| Attribute | Description |
|---|---|
| PVID (pvid) | Port VLAN ID (PVID). Specifies the PVID to use for the Shared Ethernet Adapter. PVID specifies the VLAN ID that is used for the
non-VLAN tagged packets. PVID must match the PVID of the adapter that is specified in the
pvid_adapter attribute.
The PVID of trunk adapters other than the default virtual adapter (pvid_adapter), cannot be used by any client LPARs. This is because packets that have the PVID of other trunk adapters, instead of the PVID of the default virtual adapter, have their VLAN tag removed and sent out as untagged packets to comply with the IEEE VLAN specification. |
| PVID adapter (pvid_adapter) | Specifies the default virtual adapter to use for non-VLAN tagged packets. PVID of the pvid_adapter attribute must be specified as the value for the pvid attribute. |
| Physical adapter (real_adapter) | Specifies the physical adapter associated with the Shared Ethernet Adapter. |
| Thread (thread) | Activates or deactivates threading on the Shared Ethernet Adapter. Activating
this option adds approximately 16 - 20% more machine cycles per transaction
for MTU 1500 streaming, and approximately 31 – 38% more machine
cycles per transaction for MTU 9000. The threading option adds more
machine cycles per transaction at lesser workloads due to the threads
being started for each packet. At higher workload rates, such as full
duplex or the request/response workloads, the threads can run longer
without waiting and being redispatched. Threaded mode should be used when virtual Small Computer Serial Interface (SCSI) will be run on the same Virtual I/O Server logical partition as Shared Ethernet Adapter. Threaded mode helps ensure that virtual SCSI and the Shared Ethernet Adapter can share the processor resource appropriately. However, threading adds more instruction path length, which uses additional processor cycles. If the Virtual I/O Server logical partition will be dedicated to running shared Ethernet devices (and associated virtual Ethernet devices) only, the adapters should be configured with threading disabled. You can enable or disable
threading using the -attr thread option of
the mkvdev command. To enable threading, use the -attr
thread=1 option. To disable threading, use the -attr
thread=0 option. For example, the following command disables
threading for Shared Ethernet Adapter ent1:
|
| Virtual adapters (virt_adapter) | Lists the virtual Ethernet adapters associated with the Shared Ethernet Adapter. |
| TCP segmentation offload (largesend) | Enables TCP largesend capability (also known
as segmentation offload) from logical partitions to the physical adapter.
The physical adapter must be enabled for TCP largesend for the segmentation
offload from the logical partition to the Shared Ethernet Adapter to work.
Also, the logical partition must be capable of performing a largesend
operation. On AIX®, largesend can be enabled
on a logical partition using the ifconfig command. You can enable or disable TCP largesend using the -a largesend option of the chdev command. To enable it, use the '-a largesend=1' option. To disable it, use the '-a largesend=0' option. For
example, the following command enables largesend for Shared Ethernet Adapter ent1:
By default the setting is disabled (largesend=0). Note: Largesend
is enabled by default (largesend=1) on VIOS 2.2.3.0 and higher. For
VIOS 2.2.3.0 and higher, network interface that is configured over Shared Ethernet Adapter device supports
largesend operation.
|
| TCP large receive offload (large_receive) | Enables the TCP large receive offload capability on the real adapter. When it is set and if the real adapter supports it, packets received by the real adapter is aggregated before they are passed to the upper layer, resulting in better performance. This parameter should be enabled only if all the partitions that are connected to the shared Ethernet adapter can handle packets larger than their MTU. This is not the same for Linux partitions. If all the logical partitions that are connected to the shared Ethernet adapter are AIX systems, this parameter can be enabled. |
| Jumbo frames (jumbo_frames) | Allows the interface configured over the Shared Ethernet Adapter to increase its MTU to 9000 bytes (the default is 1500). If the underlying physical adapter does not support jumbo frames and the jumbo_frames attribute is set to yes, then configuration fails. The underlying physical adapter must support jumbo frames. The Shared Ethernet Adapter automatically enables jumbo frames on its underlying physical adapter if jumbo_frames is set to yes. You cannot change the value of jumbo_frames at run time. |
| GARP VLAN Registration Protocol (GVRP) (gvrp) | Enables and disables GVRP on a Shared Ethernet Adapter. |
| Quality of service (qos_mode) | Allows the shared Ethernet adapter to prioritize the traffic based on the IEEE 802.1Q (VLAN) Priority Code Point. When it is disabled, VLAN traffic is not inspected for priority and all frames are treated equally. In strict mode, the high priority traffic is sent preferentially over less priority traffic. This mode provides better performance and more bandwidth to more important traffic. This can result in substantial delays for less priority traffic. In loose mode, a cap is placed on each priority level so that after a number of bytes is sent for each priority level, the following level is serviced. This method ensures that all packets are eventually sent. The high priority traffic is given less bandwidth with this mode than with strict mode. The caps in loose mode are such that more bytes are sent for the high priority traffic, so it gets more bandwidth than less priority traffic. |
| Number of threads (nthreads) | Specifies the number of threads in threaded mode, where the value of the thread parameter is 1. This value applies only when the thread mode is enabled. The nthreads attribute can be set to any value between 1 - 128 and has a default value of 7. |
| Queue size (queue_size) | Specifies the queue size for the Shared Ethernet Adapter threads in threaded mode where the value of the thread parameter is 1. This attribute indicates the number of packets that can be accommodated in each thread queue. This value applies only when the thread mode is enabled. When you change this value, the change does not take effect until the system restarts. |
| Hash algorithms (hash_algo) | Specifies the hash algorithm that is used to assign connections to Shared Ethernet Adapter threads in threaded mode, where the value of the thread parameter is 1. When the hash_algo parameter is set to 0, an addition operation of the source and destination Media Access Control (MAC) addresses, IP addresses, and port numbers is done. When the hash_algo parameter is set to 1, a murmur3 hash function is done instead of an addition operation. The murmur3 hash function is slower, but it achieves better distribution. This value applies only when the thread mode is enabled. |
| Virtual server network (VSN) (lldpsvc) | Activates the VSN capability on the Shared Ethernet Adapter when you
set the attribute to yes. The VSN capability can
be enabled on the Hardware Management Console (HMC) Version 7 Release
7.7.0, or later. The default value of the lldpsvc attribute
is no. This attribute must be set to no before
you remove the Shared Ethernet Adapter.
For example, the following command enables the VSN capability for
the Shared Ethernet Adapter ent1: |
| Accounting (accounting) | When enabled, the Shared Ethernet Adapter (SEA) keeps a count of the number of bytes and packets that are bridged to and from each client LPAR. Use the seastat command to see those statistics. |
| Detect flip flops (ff_detect) | When enabled, the system can detect flip flops. By default, this setting is disabled. Flip flop indicates a situation in which two SEAs are constantly switching between failover and failback events. |
| Flip flops action (ff_action) | When enabled, you can specify what action the system must take when a flip
flop state is detected. This attribute is not supported when the ff_detect
attribute is disabled. The ff_action attribute can have the following
values: standby - Specifies that the SEA must be placed in the standby mode. You can use this mode to manually fix the SEA related system issues. recover - Specifies that the SEA must recover by itself. Only an SEA that is of higher priority detects the flip flop state and takes subsequent actions. The SEA changes into a flip flop state when the following conditions are met:
When a flip flop state is detected, and if the value of the ff_action attribute is set to standby, the SEA goes into a standby mode and you can manually fix the SEA related system issues. If the ff_action attribute is set to recover, the system will try to recover by itself. During the time interval that is set in the health_time attribute, the SEA remains in a flip flop state while monitoring the link status and keep alive packets. If the link remains up and if the keep alive packets are received regularly, the recover action starts after the time interval that is set in the health_time attribute has lapsed. If the SEA receives keep alive packets from an SEA that is of lower priority, it becomes a primary SEA. If the SEA receives keep alive packets from an SEA that is of higher priority, it becomes a backup SEA. Note: If the value of the health_time
attribute is 0, the recover action will be attempted immediately without
monitoring the link status and keep alive packets.
|
| Platform Large Send (plso_bridge) | One of the requirements for a Linux client is that the Maximum Segment Size (MSS) value must be known to receive large send packets. When the plso_bridge attribute is enabled, and when large send packets are received by the SEA, the Shared Ethernet Adapter can convey the MSS values to the Linux client through the receive descriptor in the hypervisor. By default, the plso_bridge attribute is enabled. |
Shared Ethernet Adapter failover attributes
You can modify the following Shared Ethernet Adapter failover attributes.
| Attribute | Description |
|---|---|
| High availability mode (ha_mode) | Determines whether the devices participate in
a failover setup. The default is disabled. Typically,
a Shared Ethernet Adapter in
a failover setup is operating in auto mode, and the
primary adapter is decided based on which adapter has the highest
priority (lowest numerical value). A shared Ethernet device can be
forced into the standby mode, where it will behave as the backup device
as long as it can detect the presence of a functional primary. The
following are the possible values for the High availability
mode attribute:
|
| Control Channel (ctl_chan) | Sets the virtual Ethernet device that is required for a Shared Ethernet Adapter in a failover
setup so that it can communicate with the other adapter. There is
no default value for this attribute, and it is required when the ha_mode is
not set to disabled. Note: The Control Channel attribute
is an optional attribute with the Power Hypervisor Version 780, or
later and with the VIOS Version
2.2.3.0, or later.
|
| Internet address to ping (netaddr) | Optional attribute that can be specified for a Shared Ethernet Adapter that has been configured in a failover setup. When this attribute is specified, a shared Ethernet device will periodically ping the IP address to verify connectivity (in addition to checking for link status of the physical devices). If it detects a loss of connectivity to the specified ping host, it will initiate a failover to the backup Shared Ethernet Adapter. This attribute is not supported when you use a Shared Ethernet Adapter with a Host Ethernet Adapter (or Integrated Virtual Ethernet). |
| Adapter reset (adapter_reset) | When enabled, the shared Ethernet adapter disables and re-enables its physical adapter whenever it becomes inactive. It might help the external switch to direct the traffic to the new server. By default the setting is disabled. |
| Enable Reverse ARP transmit (send_RARP) | When enabled, the shared Ethernet adapter sends a reverse ARP after the Shared Ethernet Adapter failover. The reverse ARP is sent by a new primary Shared Ethernet Adapter to notify the switches of routing change. By default, the setting is enabled. |
| Health Time (health_time) | Sets the time that is required to elapse before a system is considered "healthy" after a system failover. After a Shared Ethernet Adapter moves to an "unhealthy" state, the Health Time attribute specifies an integer that indicates the number of seconds for which the system must maintain a "healthy" state before it is allowed to return into the Shared Ethernet Adapter protocol. You can use the following command to display the default values for this attribute: lsattr -D -c adapter -s pseudo -t sea -a health_time |
| Link Time (link_time) | Note: Currently, the link status check is effectively disabled in levels that contain this fix due
to APAR IV97991.
|
| Failback delay (fb_delay) | Sets the time that is required to elapse before a higher priority Shared Ethernet Adapter begins the failback process to take over as the primary SEA after a failover event. The Failback delay attribute is a dynamic attribute that can be changed at run time. The new value governs the time delay in subsequent failover/failback events. You can use the following command to display the default values for this attribute: lsattr -D -c adapter -s pseudo -t sea -a fb_delay |
| No automatic failback (noauto_failback) | When enabled, the higher priority Shared Ethernet Adapter does not attempt to automatically take over the system after a failover event. Instead, it remains as the backup Shared Ethernet Adapter. When the No automatic failback attribute is disabled, the higher priority SEA begins the failback process to take over as the primary SEA. This attribute can be changed during run time. The change affects the behavior of the Shared Ethernet Adapter for subsequent failover/failback events. By default, this attribute is disabled. |
INET attributes
| Attribute | Description |
|---|---|
| Host Name (hostname) | Specify the host name that you want to assign
to the current machine. When specifying the host name, use ASCII characters, preferably alphanumeric only. Do not use a period in the host name. Avoid using hexadecimal or decimal values as the first character (for example 3Comm, where 3C might be interpreted as a hexadecimal character). For compatibility with earlier hosts, use an unqualified host name of fewer than 32 characters. If the host uses a domain name server for name resolution, the host name must contain the full domain name. In the hierarchical domain
naming system, names consist of a sequence of subnames that are not
case-sensitive and that are separated by periods with no embedded
blanks. The DOMAIN protocol specifies that a local domain name must
be fewer than 64 characters, and that a host name must be fewer than
32 characters in length. The host name is given first. Optionally,
the full domain name can be specified; the host name is followed by
a period, a series of local domain names separated by periods, and
finally by the root domain. A fully specified domain name for a host,
including periods, must be fewer than 255 characters in length and
in the following form:
In a hierarchical network, certain hosts are designated as name servers that resolve names into Internet addresses for other hosts. This arrangement has two advantages over the flat name space: resources of each host on the network are not consumed in resolving names, and the person who manages the system does not need to maintain name-resolution files on each machine on the network. The set of names managed by a single name server is known as its zone of authority. |
| Gateway (gateway) | Identifies the gateway to which packets are addressed. The Gateway parameter can be specified either by symbolic name or numeric address. |
| Route (route) | Specifies the route. The format of the Route attribute
is: route=destination, gateway, [metric].
|
Adapter attributes
| Attribute | Adapters/Drivers | Description |
|---|---|---|
| Media Speed (media_speed) |
|
The media speed attribute indicates the speed
at which the adapter attempts to operate. The available speeds are
10 Mbps half-duplex, 10 Mbps full-duplex, 100 Mbps half-duplex, 100
Mbps full-duplex and autonegotiation, with a default of autonegotiation.
Select auto-negotiate when the adapter should use autonegotiation
across the network to determine the speed. When the network will not
support autonegotiation, select the specific speed. 1000 MBps half and full duplex are not valid values. According to the IEEE 802.3z specification, gigabit speeds of any duplexity must be autonegotiated for copper (TX)-based adapters. If these speeds are desired, select auto-negotiate. |
| Media Speed (media_speed) |
|
The media speed attribute indicates the speed at which the adapter attempts to operate. The available speeds are 1000 Mbps full-duplex and autonegotiation. The default is autonegotiation. Select auto-negotiate when the adapter should use autonegotiation across the network to determine the duplexity. When the network does not support autonegotiation, select 1000 Mbps full-duplex. |
| Media Speed (media_speed) |
|
The media speed attribute indicates the speed
at which the adapter attempts to operate. The available speeds are
10 Mbps half-duplex, 10 Mbps full-duplex, 100 Mbps half-duplex, 100
Mbps full-duplex and autonegotiation, with a default of autonegotiation.
When the adapter should use autonegotiation across the network to
determine the speed, select autonegotiate. When the network will not
support autonegotiation, select the specific speed. If autonegotiation is selected, the remote link device must also be set to autonegotiate to ensure the link works correctly. |
| Media Speed (media_speed) |
|
The media speed attribute indicates the speed
at which the adapter attempts to operate. The available speeds are
10 Mbps half-duplex, 10 Mbps full-duplex, 100 Mbps half-duplex, 100
Mbps full-duplex and autonegotiation, with a default of autonegotiation.
Select autonegotiate when the adapter should use autonegotiation across
the network to determine the speed. When the network will not support
autonegotiation, select the specific speed. For the adapter to run at 1000 Mbit/s, the autonegotiation setting must be selected. Note: For
the Gigabit Ethernet-SX PCI Adapter, the only selection available
is autonegotiation.
|
| Enable Alternate Ethernet Address (use_alt_addr) | Setting this attribute to yes indicates that the address of the adapter, as it appears on the network, is the one specified by the Alternate Ethernet Address attribute. If you specify the no value, the unique adapter address written in a ROM on the adapter card is used. The default value is no. | |
| Alternate Ethernet Address (alt_addr) | Allows the adapter unique address, as it appears on the LAN network, to be changed. The value entered must be an Ethernet address of 12 hexadecimal digits and must not be the same as the address of any other Ethernet adapter. There is no default value. This field has no effect unless the Enable Alternate Ethernet Address attribute is set to yes value, in which case this field must be filled in. A typical Ethernet address is 0x02608C000001. All 12 hexadecimal digits, including leading zeros, must be entered. | |
| Enable Link Polling (poll_link) |
|
Select no to cause the device driver to poll the adapter to determine the status of the link at a specified time interval. The time interval value is specified in the Poll Link Time Interval field. If you select no, the device driver will not poll the adapter for its link status. The default value is no. |
| Poll Link Time Interval (poll_link_time) |
|
The amount of time, in milliseconds, between polls to the adapter for its link status that the device driver is allowed. This value is required when the Enable Link Polling option is set to yes. A value between 100 through 1000 can be specified. The incremental value is 10. The default value is 500. |
| Flow Control (flow_ctrl) |
|
This attribute specifies whether the adapter should enable transmit and receive flow control. The default value is no. |
| Transmit Jumbo Frames (jumbo_frames) |
|
Setting this attribute to yes indicates that frames up to 9018 bytes in length might be transmitted on this adapter. If you specify no, the maximum size of frames transmitted is 1518 bytes. Frames up to 9018 bytes in length can always be received on this adapter. |
| Checksum Offload (chksum_offload) |
|
Setting this attribute to yes indicates
that the adapter calculates the checksum for transmitted and received
TCP frames. If you specify no, the checksum will
be calculated by the appropriate software. When a virtual Ethernet adapter has checksum offload enabled, the adapter advertises it to the hypervisor. The hypervisor tracks which virtual Ethernet adapters have checksum offload enabled and manages inter-partition communication accordingly. When network packets are routed through the Shared Ethernet Adapter, there is
a potential for link errors. In this environment, the packets must
traverse the physical link with a checksum. Communication works in
the following way:
To enable checksum offload for a Shared Ethernet Adapter, all constituent devices must have it enabled as well. The shared Ethernet device will fail if the underlying devices do not have the same checksum offload settings. |
| Enable Hardware Transmit TCP Resegmentation (large_send) |
|
This attribute specifies whether the adapter is to perform transmit TCP resegmentation for TCP segments. The default value is no. |
Link Aggregation (EtherChannel) device attributes
| Attribute | Description |
|---|---|
| Link Aggregation adapters (adapter_names) | The adapters that currently make up the Link Aggregation device. If you want to modify these adapters, modify this attribute and select all the adapters that should belong to the Link Aggregation device. When you use this attribute to select all of the adapters that should belong to the Link Aggregation device, its interface must not have an IP address configured. |
| Mode (mode) | The type of channel that is configured. In standard
mode, the channel sends the packets to the adapter based on an algorithm
(the value used for this calculation is determined by the Hash Mode
attribute). In round_robin mode, the channel gives one packet to each
adapter before repeating the loop. The default mode is standard. Using the 802.3ad mode, the Link Aggregation Control Protocol (LACP) negotiates the adapters in the Link Aggregation device with an LACP-enabled switch. If the Hash Mode attribute is set to anything other than the default, this attribute must be set to standard or 802.3ad. Otherwise, the configuration of the Link Aggregation device will fail. |
| Hash Mode (hash_mode) | If operating under standard or IEEE 802.3ad
mode, the hash mode attribute determines how the outgoing adapter
for each packet is chosen. Following are the different modes:
You cannot use round-robin mode with any hash mode value other than default. The Link Aggregation device configuration will fail if you attempt this combination. If the packet is not TCP or UDP, it uses the default hashing mode (destination IP address). Using TCP or UDP ports for hashing can make better use of the adapters in the Link Aggregationdevice, because connections to the same destination IP address can be sent over different adapters (while still retaining the order of the packets), thus increasing the bandwidth of the Link Aggregation device. |
| Internet Address to Ping (netaddr) | This field is optional. The IP address that the Link Aggregation device should ping to verify that the network is up. This is only valid when there is a backup adapter and when there are one or more adapters in the Link Aggregation device. An address of zero (or all zeros) is ignored and disables the sending of ping packets if a valid address was previously defined. The default is to leave this field blank. |
| Retry Timeout (retry_time) | This field is optional. It controls how often the Link Aggregation device sends out a ping packet to poll the current adapter for link status. This is valid only when the Link Aggregation device has one or more adapters, a backup adapter is defined, and the Internet Address to Ping field contains a non-zero address. Specify the timeout value in seconds. The range of valid values is 1 to 100 seconds. The default value is 1 second. |
| Number of Retries (num_retries) | This field is optional. It specifies the number of lost ping packets before the Link Aggregation device switches adapters. This is valid only when the Link Aggregation device has one or more adapters, a backup adapter is defined, and the Internet Address to Ping field contains a non-zero address. The range of valid values is 2 to 100 retries. The default value is 3. |
| Enable Gigabit Ethernet Jumbo Frames (use_jumbo_frame) | This field is optional. To use this attribute, all of the underlying adapters, as well as the switch, must support jumbo frames. This will work only with a Standard Ethernet (en) interface, not an IEEE 802.3 (et) interface. |
| Enable Alternate Address (use_alt_addr) | This field is optional. If you set this to yes, you can specify a MAC address that you want the Link Aggregation device to use. If you set this option to no, the Link Aggregation device uses the MAC address of the first adapter. |
| Alternate Address (alt_addr) | If Enable Alternate Address is set to yes, specify the MAC address that you want to use. The address you specify must start with 0x and be a 12-digit hexadecimal address. |
VLAN attributes
| Attribute | Value |
|---|---|
| VLAN Tag ID (vlan_tag_id) | The unique ID associated with the VLAN driver. You can specify from 1 to 4094. |
| Base Adapter (base_adapter) | The network adapter to which the VLAN device driver is connected. |
Shared Ethernet Adapter QoS attribute
- disabled mode
- This is the default mode. VLAN traffic is not inspected for the
priority field. For example,
chdev -dev <sea device name> -attr qos_mode=disabled - strict mode
- More important traffic is bridged over less important traffic.
This mode provides better performance and more bandwidth to more important
traffic; however, it can result in substantial delays for less important
traffic. For example,
chdev -dev <sea device name> -attr qos_mode=strict - loose mode
- A cap is placed on each priority level, so that after a number
of bytes are sent for each priority level, the next level is serviced.
This method ensures that all packets will eventually be sent. More
important traffic is given less bandwidth with this mode than with
strict mode; however, the caps in loose mode are such that more bytes
are sent for the more important traffic, so it still gets more bandwidth
than less important traffic. For example,
chdev -dev <sea device name> -attr qos_mode=loose
Client-specific Shared Ethernet Adapter statistics
chdev -dev <sea device name> -attr accounting=enabledseastat -d <sea device name> [-n] seastat -d <sea device name> -c