(1) Verification of the physical interface
Verify if the physical interface is inserted into the system using prtconf (1M) command.
# prtconf -D | grep "name of the physical interface" |
For example, to use qfe, execute the command as below:
# prtconf -D | grep qfe
SUNW,qfe, instance #0 (driver name: qfe)
SUNW,qfe, instance #1 (driver name: qfe)
SUNW,qfe, instance #2 (driver name: qfe)
SUNW,qfe, instance #3 (driver name: qfe) |
In the above example, it is possible to use qfe0, qfe1, qfe2, and qfe3. For details regarding prtconf (1M) command, refer to the Solaris manual.
If the system has no NIC installed, install a NIC. After adding a new NIC on the system, run "boot -r" command at the ok prompt, and then verify the physical interface as above.
Information
When using Tagged VLAN, ensure that the NIC supports tagged VLAN functionality (IEEE802.1Q). Refer to the documents of individual ethernet driver for configuring tagged VLAN interface. In addition, in a Redundant Line Control function, the effective range of VLAN-ID which can be specified is from 1 to 4094.
(2) Checking the name service
When using name services such as DNS or NIS, define keywords such as hosts, netmasks, and ipnodes in /etc/nsswitch.conf file to first refer to the local file. This allows to solve the address even if the DNS, NIS or LDAP sever is unreachable. The following is an example of /etc/nsswitch.conf.
# # /etc/nsswitch.files: # # An example file that could be copied over to /etc/nsswitch.conf; it # does not use any naming service. # # "hosts:" and "services:" in this file are used only if the # /etc/netconfig file has a "-" for nametoaddr_libs of "inet" transports. passwd: files group: files hosts: files dns ipnodes: files networks: files protocols: files rpc: files ethers: files netmasks: files bootparams: files |
Note
Even when using only IPv4 address in Redundant Line Control function, please define a host name as both /etc/inet/hosts file and /etc/inet/ipnodes file.
(3) Checking the network configuration files of the OS
When setting IP addresses using the network configuration files of the OS (/etc/hostname.interface file or /etc/hostname6.interface file), it is recommended to set up as follows for hardware failure such as NIC or SB (System Board).
Fast switching mode
Create either or both of the following files as the network configuration files for both physical interfaces for redundancy:
/etc/hostname.interface
/etc/hostname6.interface
RIP mode
Create the following file as the network configuration file for both physical interfaces for redundancy:
/etc/hostname.interface
NIC switching mode
Create either or both of the following files as the network configuration files for either of the physical interfaces for redundancy:
/etc/hostname.interface
/etc/hostname6.interface
GS/SURE linkage mode
It is not necessary to create any network configuration files of physical interfaces for redundancy.
Note
On the Solaris 10 OS, services related to the network including Redundant Line Control function will not start if all IP addresses entered in hostname.interface cannot be allocated due to a hardware failure on system startup. To activate Redundant Line Control function even in the case of a hardware failure, create hostname.interface as shown in the following figure:
For NIC switching mode, it is recommended to create the network configuration files so that at least one physical interface is to be activated on system startup as shown in the following figure:
Figure 3.2 Setup example of the OS setting file in the NIC switching mode
In addition, if the a system board is multiplexed, it is recommend to create one or more network configuration files on each system board so that at least one physical interface is to be activated on system startup as shown in the following figure:
If you cannot follow the procedures above due to reasons such as the small number of equipped NICs, check that the physical interface for the network configuration file has no failure before rebooting the system.
Figure 3.3 Setup example of the OS setting file in the environment with redundant system boards
In the NIC switching mode (physical IP takeover II), it is necessary to avoid IP duplication between nodes in the cluster. Create an empty network configuration file to avoid IP addresses being set. When network configuration files are required, create network configuration files so that at least one physical interface other than the physical interface for redundancy is to be activated on system startup
When using an IPv4 address, define in the /etc/inet/hosts file the host names (host names to be attached to virtual IP, monitored host names to be specified in monitoring destination information, etc.) to be specified in environment definitions of Redundant Line Control function. These host names must be specified in the /etc/inet/hosts file even if no host names but IP addresses are directly specified in environment definitions.
If an IPv6 address is used, define the IPv6 address and a host name in /etc/inet/ipnodes file.
When using an IPv4 address, define a configured physical interface to use in IPv4 before defining a virtual interface. (Check whether or not an /etc/hostname.interface file exists. If not, create it and reboot the system.)
When using an IPv6 address, define a configured physical interface to use in IPv6 before defining a virtual interface. (Check whether or not an /etc/hostname6.interface file exists. If not, create it and reboot the system. When creating a /etc/hostname6.interface file, make sure it is an empty file.)
If IPv6 address is used, it is recommended to setup at least two Solaris servers running in Fast switching mode as IPv6 routers just in case an IPv6 router fails and communication cannot be achieved using a site local address. Note that if configuring IPv6 router for multiple servers, make sure these servers use the same prefix information for the virtual interface configured in /etc/inet/ndpd.conf.
An example of setting a /etc/inet/ndpd.conf file when using a Solaris server as an IPv6 router is shown below. (See a Solaris manual for the detail of a /etc/inet/ndpd.conf file.)
ifdefault AdvSendAdvertisements true # Every interface sends a router advertisement. prefix fec0:1::0/64 sha0 # sha0 sends "Prefix fec0:1::0/64". |
After configuring a GLS environment, hostname information on a host database (such as /etc/inet/hosts file) should not be modified, if it has been used for GLS. To change the information on the host database, it is necessary to delete the GLS configuration first, and configure it again after modifying the hostname information.
As for configuring physical interface, be sure to define to use in TCP/IP before defining a virtual interface. (Check whether or not an /etc/hostname.interface file exists. If not, create it and reboot the system.)
Set to activate a routing daemon because it is necessary to change the route information dynamically. If you are using Solaris 8 or Solaris 9 as a basic OS, you must not create the "/etc/defaultrouter" or "/etc/notrouter" file. Check whether the files exit, and if they do, change the file name or delete them. If you are using Solaris 10 for a basic OS, use the "routeadm(1M)" command to set up a routing daemon.
Routing daemon setup (Solaris 10 only):
# routeadm -e ipv4-forwarding
# routeadm -e ipv4-routing
# routeadm -s ipv4-routing-daemon="/usr/sbin/in.routed"
# routeadm -s ipv4-routing-daemon-args="-s"
# routeadm
Configuration Current Current
Option Configuration System State
---------------------------------------------------------------
IPv4 forwarding enabled disabled
IPv4 routing enabled disabled
IPv6 forwarding disabled disabled
IPv6 routing disabled disabled
IPv4 routing daemon "/usr/sbin/in.routed"
IPv4 routing daemon args "-s"
IPv4 routing daemon stop "kill -TERM `cat /var/tmp/in.routed.pid`"
IPv6 routing daemon "/usr/lib/inet/in.ripngd"
IPv6 routing daemon args "-s"
IPv6 routing daemon stop "kill -TERM `cat /var/tmp/in.ripngd.pid`"
# routeadm -u |
For Redundant Line Control function, the path information must be initialized and the routing daemon must be restarted. If path information is statically specified, the static paths must be described in /etc/gateways.
Because this mode uses dynamic routing by RIP, do not start RDISC (search protocol for ICMP router) during system startup. In order to restrain RDISC from starting, rename /usr/sbin/in.rdisc then reboot the system.
For the setup procedure, refer to "3.2.2.2 System setup in Fast switching mode" and "3.2.2.3 System setup in RIP mode".
When using IPv4 address:
When using an IPv4 address, define in the /etc/inet/hosts file the host names (host names to be attached to virtual IP, monitored host names to be specified in monitoring destination information, etc.) to be specified in environment definitions of Redundant Line Control function. These host names must be specified in the /etc/inet/hosts file even if no host names but IP addresses are directly specified in environment definitions.
When using an IPv4 address, define a configured primary physical interface to use in IPv4 before defining a virtual interface. (Check whether or not an /etc/hostname.interface file exists. If not, create it and reboot the system.)
For Redundant Line Control function, the path information must be initialized and the routing daemon must be restarted. If path information is statically specified, the static paths must be described in /etc/gateways.
When using IPv6 address:
If an IPv6 address is used, define the IPv6 address and a host name in /etc/inet/ipnodes file.
When using an IPv6 address, define a configured physical interface to use in IPv6 before defining a virtual interface. (Check whether or not an /etc/hostname6.interface file exists. If not, create it and reboot the system. When creating a /etc/hostname6.interface file, make sure it is an empty file.)
When using an IPv6 address, set an IPv6 router on a network to be connected without fail. Specify the same prefix and the same length of a prefix for an IPv6 address to be set in a Redundant Line Control function as those set in an IPv6 router.
In addition, when you use a Solaris server as an IPv6 router, please define two or more server as an IPv6 router. When abnormalities occur in an IPv6 router, it becomes impossible to perform communication which used a site local address. Therefore, it recommends defining two or more IPv6 routers. When you define an IPv6 router as two or more servers, the prefix information on the virtual interface defined as /etc/inet/ndpd.conf should define the same value in each server.
An example of setting a /etc/inet/ndpd.conf file when using a Solaris server as an IPv6 router is shown below. (See a Solaris manual for the detail of a /etc/inet/ndpd.conf file.)
ifdefault AdvSendAdvertisements true # Every interface sends a router advertisement. prefix fec0:1::0/64 hme0 # hme0 sends "Prefix fec0:1::0/64". prefix fec0:2::0/64 hme1 # hme1 sends "Prefix fec0:2::0/64". |
When using an IPv4 address, define in the /etc/inet/hosts file the host names (host names to be attached to virtual IP, monitored host names to be specified in monitoring destination information, etc.) to be specified in environment definitions of Redundant Line Control function. These host names must be specified in the /etc/inet/hosts file even if no host names but IP addresses are directly specified in environment definitions.
The physical interface to be specified must not be defined for normal use in TCP/IP.
(Check whether or not an /etc/hostname.interface file exists. If it does, rename the file or delete it, and execute "/usr/sbin/ifconfig <interface> unplumb" command.)
GS/SURE linkage mode requires dynamic routings, so do not create /etc/defaultrouter file. (Check for existence of /etc/defaultrouter. If the file exists, either delete the file or rename the file.)
It is necessary to block leaks of routing information from the system. If you are using Solaris 8 or Solaris 9 for a basic OS, create the "/etc/notrouter" file. If you are using Solaris 10 for a basic OS, use the "routeadm(1M)" command to set up a routing daemon.
Routing daemon setup (Solaris 10 only):
# routeadm -e ipv4-routing
# routeadm -s ipv4-routing-daemon="/usr/sbin/in.routed"
# routeadm -s ipv4-routing-daemon-args="-q"
# routeadm
Configuration Current Current
Option Configuration System State
---------------------------------------------------------------
IPv4 forwarding disabled disabled
IPv4 routing enabled disabled
IPv6 forwarding disabled disabled
IPv6 routing disabled disabled
IPv4 routing daemon "/usr/sbin/in.routed"
IPv4 routing daemon args "-q"
IPv4 routing daemon stop "kill -TERM `cat /var/tmp/in.routed.pid`"
IPv6 routing daemon "/usr/lib/inet/in.ripngd"
IPv6 routing daemon args "-s"
IPv6 routing daemon stop "kill -TERM `cat /var/tmp/in.ripngd.pid`"
# routeadm -u |
If path information is statically specified, the static paths must be described in /etc/gateways.
Because this mode uses dynamic routing by RIP, do not start RDISC (search protocol for ICMP router) during system startup. In order to restrain RDISC from starting, rename /usr/sbin/in.rdisc then reboot the system.
