The following conditions must be satisfied for manual server switchover, failback, and Auto-Recovery to function correctly.
Conditions for Spare Servers
The spare server must be identical to the primary server regarding all of the following conditions.
If these conditions are not satisfied, allocation of a spare server may not be possible, server switchover may fail, or the server may malfunction after the switchover is completed.
Server Model
Server Hardware Configuration
The following settings or configurations of option cards or expansion cards must be the same.
Model numbers
The locations they are mounted in
The number and sizes of local disks
RAID settings
There are no other hardware conditions for the spare server (such as memory capacity, number of CPUs and CPU clock speed). However, the hardware configuration of the spare server must be capable of running the operating system and applications running on the primary server.
BIOS Settings
The same BIOS settings must have been made for all servers according to the procedure in "6.2.7 Configuring BIOS Settings of Managed Servers" in the "Design Guide VE".
OBP Settings
For SPARC M10/M12 and SPARC Enterprise, the OBP settings to start automatically from SAN servers of the same target disk must have been made according to the procedure in "6.2.9 Configuring OBP (Open Boot Prom) Settings (SPARC M10/M12 and SPARC Enterprise Servers)" in the "Design Guide VE".
With the storage affinity switchover method, as setting changes are only performed for Fibre Channel switch storage devices, it is necessary to perform configuration of servers that will be changed during switchover, and HBA and OBP settings.
LAN and SAN Access Scope
The spare server must use the same network redundancy method, have the same redundancy paths, and have access to the same network and storage devices. Note that LAN or fibre channel switches connected in a cascade configuration are viewed as a single device.
Firewall
There must be no firewall between the primary server and the spare server.
Subnet
For servers other than blade servers, the primary server and the spare server must belong to the same subnet.
VLAN
For the following configurations, the primary server and the spare server must belong to the same VLAN.
LAN switch blades connected to managed servers are as follows:
Operating in Converged Fabric mode
PY CB 10Gb FEX Nexus B22
Managed servers are not blade servers
If a spare server is shared by a physical OS and one or more VM guests (using the high-availability feature provided with their server virtualization software), the spare server must be configured for SAN boot using I/O virtualization.
For details, refer to "Figure 9.3 Sharing a Spare Server Between Physical OSs and VM Guests (High-availability Function of Server Virtualization Software)" in "9.2 Configuration".
When using a server on which a server OS is operating as a spare server, both the primary server and spare server should be in a configuration that uses I/O virtualization.
For details, refer to "Figure 9.4 Configuration Using a Server on which a Server OS is Operating as a Spare Server" in "9.2 Configuration".
Also, if more than two local disks are connected to a spare server in a backup/restore configuration, and if the partitioning of the disks coming after the first in the boot order is different from the first disk, a warning message may be displayed at restart, or the operating system may not restart, causing any switchovers initiated to fail. After configuring the spare server, verify that it is operating properly by performing a switchover and failback.
If the operation fails, configure the partitions other than the first boot disk of the spare server to match the configuration of the primary server, or set up the primary server configuration so that it does not depend on any disk other than the first, using automatic service startups, and re-labeling the Windows drives as necessary.
With blade servers, if the primary server and spare server do not belong to the same subnet, if server switchover is performed to the spare server it is necessary that the VLAN ID or port group settings of the internal LAN switch ports are adjusted automatically.
For details on how to configure the settings, refer to "18.2 Settings for Server Switchover" in the "User's Guide VE".
Conditions for Server Switchover
All of the following conditions must be satisfied for server switchover or Auto-Recovery to succeed:
The server configuration cannot already be switched over (the spare server must not be being used for operations)
The status of the spare server must be "normal", "warning", or "stop". If a VM host has been installed on the spare server(for VMware HA), its status must be either "normal" or "warning"
If a spare server is shared by more than one active server, none of the other primary servers may be switched over to that spare server
If the server is in a local boot environment, its system image must have been backed up
Note
When performing server switchover using the backup and restore method in a SAN data server environment that uses local boot, configure a target disk for image operations on both the primary server and the spare server.
If the switchover is performed without configuring the target disk for image operations, the data may be overwritten on an unintended disk.
For details, refer to "9.1.13 Changing Target Disks of Image Operations" in the "User's Guide VE".
Conditions for Server Failback
All of the following conditions must be satisfied for server failback to succeed:
The active server must have been switched over to the spare server
The status of the primary server must be "stop"
If the server is in a local boot environment, its system image must have been backed up
Conditions for Server Switchover of OVM for SPARC Servers
When the OS of the server is OVM for SPARC, it is also necessary to satisfy the following conditions as well as those above.
The XCP firmware version of the primary server and the spare server are the same
The startup settings of the primary server are configured
For details, refer to "6.1.6 Settings when Switching Over SPARC M10/M12 or SPARC Enterprise Servers" in the "Design Guide VE".
There are seven sets of domain configuration information of the spare server (hereinafter configuration information) or less, including the factory-default
In SPARC M10/M12s, a maximum of eight sets of configuration information can be saved, including the factory-default.
After switchover, the configuration information is saved on the service processor using the domain configuration name used before the switchover. Therefore, if there is configuration information with the same name on the spare server, save the configuration information after deleting the configuration information with the same name during switchover.
Configuration information other than the factory-default has been saved on the service processor of the primary server
When performing server switchover while the configuration information is factory-default, the configuration information cannot be saved on the switchover destination server.
When registering the server in a VM pool, the primary server and the spare server are registered in the same VM pool
If the servers are registered in different VM pools, the displayed information of the server in the VM pool is changed after switchover.
An alias is not used in the boot-device settings of the OBP of the I/O domain or the control domain for the primary server
When server switchover is performed, only boot-device settings of the primary server are carried over to the spare server.
The admin LAN (IP address) of the control domain is allocated to a physical NIC
The admin LAN (IP address) cannot be allocated to a virtual switch (vsw).
CPU core activation has been applied to the spare server
CPU core activation is not carried over during server switchover.
No network interface which cannot communicate in the factory-default state exists in the control domain
When a network interface created on a virtual switch (vsw) is used in the control domain, it is considered that the server switchover conditions are not satisfied.
For example, a configuration where a virtual switch (vsw) is used for IPMP redundancy.
Note
When using an OVM for SPARC server as a spare server, only OVM for SPARC environments are supported for the primary server.
