5.2.5 Considerations for using a hot-standby configuration in a virtualization environment (when using VMware ESXi)
If you use VMware ESXi-based virtualization, you can divide a single physical partition into multiple logical partitions (virtual machines) and perform hot standby processing for each virtual machine. For details about hot standby processing in units of LPARs when host reset is performed, see 3.3.6 Physical partition reset function in a virtualization environment. For details about the setup procedure for running virtual machines using VMware ESXi, see the VMware ESXi documentation.
This subsection discusses considerations on using virtual machines as the hot standby units.
- Important
-
If you reset hosts in a VMware ESXi-based virtualization environment, do not suspend the virtual machines while HA Monitor is running. For details, see Release Notes.
- Organization of this subsection
(1) Evaluating the virtual machines to be used in a hot-standby configuration
Between a VMware ESXi-based virtualization environment and the normal environment, there is no difference in HA Monitor operations other than whether the reset target is a virtual machine or a processor. If host reset is performed, you can perform hot standby processing between virtual machines. If SCSI reservation or the function for controlling hot standby based on the availability of LAN communications is used, you can perform hot standby processing between virtual machines or between a virtual machine and a processor.
However, in terms of reliability, we recommend that you use separate processors for the hosts that constitute a hot-standby configuration.
(2) Checking the version of HA Monitor
If host reset is performed in a hot-standby configuration that includes a VMware ESXi-based virtualization environment, the versions of all HA Monitors in the hot-standby configuration must be 01-50 or later. If the configuration contains an HA Monitor whose version is earlier than 01-50, it will not be possible to reset hosts in the virtualization environment.
If you use the same monitoring paths as in the existing system, use HA Monitor version 01-50 or later in the existing system, or set the monitoring paths so that the HA Monitor in the existing system is not connected to a host in the hot-standby switchover configuration.
(3) Checking the machine environment
If you use VMware ESXi-based virtualization, make sure that the machine environment satisfies the following conditions:
-
The /usr/bin/ssh command has been installed (when host reset is performed).
To reset a host on a virtual machine, HA Monitor uses the /usr/bin/ssh command to connect to VMware ESXi. If this command has not been installed, install the openssh-clients package provided with Linux.
-
SSH connection can be established with VMware ESXi (when host reset is performed).
The root user must be able to connect to VMware ESXi by using SSH.
We also recommend that at least one physical CPU core is allocated for each virtual machine running on a physical partition.
If this condition is not satisfied, HA Monitor might become unavailable temporarily because no core is allocated to the virtual machines during certain periods of time. This might result in a delay of host failure detection and alive message notification.
(4) Checking the functions to be used
-
If host reset is performed in a VMware ESXi-based virtualization environment, the hot standby processing time cannot be reduced by using a mutual hot-standby configuration (system cannot be specified in the cpudown operand in the HA Monitor environment settings).
-
If all of the following conditions are met, before starting HA Monitor, start the virtual machine on which HA Monitor Connector operates.
-
The server model is HA8000xN or later.
-
A management server is not used as a failure management processor.
-
HA Monitor and HA Monitor Connector operate on different virtual machines.
-