C H A P T E R 4 |
Network Configuration |
This chapter describes how to manually configure the network interface and attributes, including optional methods to automate network configuration.
This chapter contains the following topics:
After installing the driver software as described in Chapter 3, you can create a hostname.nxge file to configure each of the Netra CP32x0 ARTM-10G interfaces.
You must designate both an IP address and a host name for each Netra CP32x0 ARTM-10G Ethernet interface in the /etc/hosts file.
Adhere to the following guidelines when assigning host names:
The following example shows the /etc/hostname.nxge file required for a system called zardoz that has an nxge driver (zardoz-11).
To Manually Configure Network Host Files |
This method enables you to manually manipulate a network interface via the command line interface. Note that any changes made using this method will be lost when the system reboots.
1. At the command line, search the /etc/path_to_inst file for nxge interfaces.
The ARTM reports two nxge interfaces, and for each, includes three descriptive parameters:
2. Write down the device path and instance, which in the example is /pci@7c0/pci@0/pci@9/network@0 0.
Your Netra CP32x0 ARTM-10G device path and instance will be similar. You need this information later to make changes to the nxge.conf file.
3. Set up the Netra CP32x0 ARTM-10G’s nxge interface.
4. Use the ifconfig command to assign an IP address to the network interface.
Type the following at the command line, replacing ip-address with the Netra CP32x0 ARTM-10Gs IP address:
Refer to the ifconfig(1M) man page and the Solaris documentation for more information.
5. Create an appropriate entry in the /etc/hosts file for each active nxge interface.
Follow the guidelines for assigning IP addresses and host names, described at the beginning of this Section 4.1, Configuring Network Host Files.
# cat /etc/hosts # # Internet host table # 127.0.0.1 localhost 129.144.10.57 zardoz loghost 129.144.11.83 zardoz-11 |
To Automatically Configure Network Host Files |
This method enables you to create a file that provides a persistent setup, which is automatically loaded at each system reboot.
1. Create an /etc/hostname.nxgex file.
Where x is the instance number of the nxge interface you plan to use. For example, to auto-configure a Netra CP32x0 ARTM-10Gs nxge interface with instance 0, create an /etc/hostname.nxge0 file, where 0 is the number of the nxge interface. If the instance number were 1, the file name would be /etc/hostname.nxge1.
Note - Do not create an /etc/hostname.nxge file for any Netra CP32x0 ARTM-10Gs you plan to leave unused. |
2. Create an appropriate entry in the /etc/hosts file for each active nxge interface.
Follow the guidelines for assigning IP addresses and host names, described at the beginning of Section 4.1, Configuring Network Host Files.
# cat /etc/hosts # # Internet host table # 127.0.0.1 localhost 129.144.10.57 zardoz loghost 129.144.11.83 zardoz-11 |
Configuring Jumbo Frames enables Ethernet interfaces to send and receive packets larger than the standard 1500 bytes. However, the actual transfer size depends on the switch capability and the device driver capability.
Note - Refer to the documentation that came with your switch for commands to configure Jumbo Frames. |
The Jumbo Frames checking occurs at Layer 2 or Layer 3, depending on the configuration method.
View the maximum transmission unit (MTU) configuration of a nxge instance at any time with the kstat command:
The kstat mac_mtu variable represents the complete size of the Ethernet frame, which includes the Ethernet header, maximum payload, and CRC. The value of this variable should be equal to or less than the MTU configured on the switch.
View the Layer 3 configuration by using the dladm command with the show-link option.
This section describes how to display all the statistics that the driver tracks for an interface.
1. Use the kstat command to display driver statistics.
The output can be further filtered to display specific data of interest. The following example displays the receive packet counts on all eight receive DMA channels on nxgeinterface 1.
2. Use the kstat command to display driver statistics of a VLAN interface.
1. Use the ifconfig utility to display driver statistics.
In the previous example output, MTU indicates the complete size of the Ethernet frame, which includes the Ethernet header, maximum payload, and CRC. The value of MTU should be equal to or less than the MTU configured on the switch.
2. If you want to see only MTU, use the following command:
3. Use the ethtool utility with the -S option to get more detailed information.
This option displays a large amount of statistics maintained by the driver. Output can be further filtered to display specific data of interest using grep.
# ethtool -S eth9 |grep rx_pac rx_packets: 748274 rx_packets: 828 rx_packets: 112 rx_packets: 189088 rx_packets: 134 rx_packets: 196085 rx_packets: 177884 rx_packets: 93 rx_packets: 184050 |
Enable Jumbo Frames if you want to configure Ethernet interfaces to send and receive packets larger than the standard 1500 bytes.
This section describes how to enable Jumbo Frames in both a SPARC and an x86 environment.
1. Enable Jumbo Frames for a port using the nxge.conf file.
name = "pciex108e,abcd" parent = "/pci@780/pci@0/pci@8/network@0" unit-address = "0" accept-jumbo=1; |
3. Set the maximum transmission unit (MTU) for maximum performance.
This section describes how to enable Jumbo Frames in Linux environment.
1. Ensure that the nxge software is installed.
2. Plumb the Netra CP32x0 ARTM-10G 10GbE SFP+ interface.
The xxx.xxx.xx.xxx represents the IP address of the interface.
3. Set the maximum transmission unit (MTU) to specify jumbo frames size.
Where x designates the instance number of the interface.
Link aggregation enables one or more network links to be aggregated together to form a link-aggregation group. This link-aggregation group appears to MAC clients as a regular link. Link aggregation is defined by IEEE 802.3ad and provides the following benefits:
This section explains how to configure link aggregation.
1. Aggregate nxge0 and nxge1 to form an aggregation and use a random number as key.
2. Unplumb the interfaces to be aggregated:
3. Create a link aggregation group with a random number as a key, without specifying mode.
When the command returns, one line appears in the /etc/aggregation.conf file and indicates that the default mode is off. For example:
4. Plumb up the interface aggrkey, which is aggr33 in the following example:
5. Show link aggregation status again.
6. Use the dladm show-aggr -s command to display statistics:
# dladm show-aggr -s key: 33 ipackets rbytes opackets obytes %ipkts %opkts Total 0 0 16 1182 nxge0 0 0 16 1182 - 100.0 nxge1 0 0 0 0 - 0.0 |
7. Use the dladm show-aggr -L command to display LACP specific information:
For more information, refer to the man pages for the dladm command.
Many administrators familiar with Linux refer to Link Aggregation as “NIC Bonding.” Both refer to the IEEE802.3ad standard that defines Dynamic link aggregation.
1. Modify the /etc/modprobe.conf file for the 2.6 kernels file to add the following lines:
Refer to Linux documentation for more information.
3. Configure the bond0 interface.
In this example, bond0 is the master of the two interfaces eth4 and eth5.
# ip addr add 192.12.38.64/24 brd + dev bond0 # ip link set dev bond0 up # ifenslave bond0 eth4 eth5 |
VLANs enable you to split your physical LAN into logical subparts, providing an essential tool for increasing the efficiency and flexibility of your network.
VLANs are commonly used to separate groups of network users into manageable broadcast domains, to create logical segmentation of workgroups, and to enforce security policies among each logical segment.
Each defined VLAN behaves as its own separate network. The traffic and broadcasts of each VLAN are isolated from the others, increasing the bandwidth efficiency within each logical group.
Although VLANs are commonly used to create individual broadcast domains or separate IP subnets, a server can have a presence on more than one VLAN simultaneously.
The Netra CP32x0 ARTM-10G supports multiple VLANs on a per port or per interface basis, allowing very flexible network configurations. With multiple VLANs on an ARTM, a single physical port can have a logical presence on multiple IP subnets.
By default, 128 VLANs can be defined for each VLAN-aware ARTM port in the chassis. However, you can increase this number by changing the system parameters. If your network does not require multiple VLANs, you can use the default configuration, in which case no further configuration is necessary.
VLANs can be created according to various criteria, but each VLAN must be assigned a VLAN tag or VLAN ID (VID). The VID is a 12-bit identifier between 1 and 4094 that identifies a unique VLAN. For each network interface (nxge0 and nxge1), 4094 possible VLAN IDs can be selected per port for up to 4 ports.
Tagging an Ethernet frame requires adding a tag header to the frame. Insert the header immediately following the destination MAC address and the source MAC address. The tag header consists of two bytes of Ethernet Tag Protocol identifier (TPID, 0x8100) and two bytes of Tag Control Information (TCI). TABLE 4-1 shows the Ethernet tag header format.
By default, a single VLAN is configured for every port. This configuration groups all ports into the same broadcast domain, just as if there were no VLANs at all. VLAN tagging for the switch port is turned off.
Note - If you configure a VLAN virtual device for an ARTM, all traffic sent or received by that ARTM must be in VLAN-tagged format. |
This section describes how to configure VLANs in a Solaris environment.
1. Create one /etc/hostname.nxgex file for each VLAN that will be configured for the installed ARTM.
Where x is the instance number of the nxge interface you plan to use. For example, to configure a Netra CP32x0 ARTM-10Gs nxge interface with instance 0, create a hostname.nxge0 file, where 0 is the number of the nxge interface. If the instance number were 1, the file name would be hostname.nxge1.
2. Enter the VLAN ID (VID) and the physical point of attachment (PPA), as in the following example.
This format limits the maximum number of PPAs (instances) you can configure to 1000 in the /etc/path_to_inst file.
For example, on a server blade with an Netra CP32x0 ARTM-10G having an instance of 0, belonging to a member of two VLANs (with VID 123 and 224), you would use nxge123000 and nxge224000, respectively, as the two VLAN PPAs.
For more information about VID formats, see the introduction to this Section 4.4, Configuring VLANs.
3. Use ifconfig to configure a VLAN virtual device.
The output of ifconfig -a on a system having VLAN devices nxge123000 and nxge224000 would be as follows:
4. On the switch, set VLAN tagging and set VLAN ports to coincide with the VLANs you have set up on the server.
For formats, see the introduction to Section 4.4, Configuring VLANs.
Continuing the examples used in Step 2, you would set up VLAN ports 123 and 224 on the switch.
Refer to the documentation that came with your switch for specific instructions for setting VLAN tagging and ports.
This section describes how to configure VLANs in a Linux environment.
1. Ensure that the nxge driver is installed:
2. Plumb the Sun Dual 10GbE SFP+ PCIe ExpressModule interface.
Where xxx.xxx.xx.xxx = the IP address of the interface.
4. Add the VLAN instance (VID):
For more information about VID formats, see the introduction to Section 4.4, Configuring VLANs.
5. Configure the nxge VLAN (eth2 in the following example):
Copyright © 2008 Sun Microsystems, Inc. All Rights Reserved.