5.5.4 Reset path configuration

A reset path is used for the connection when you need to issue a host input/output reset instruction because of a failure in the active system. A reset path is not used when SCSI reservation for shared disk is used or when the function for controlling hot standby based on the availability of LAN communications is used.

For a reset path, use a TCP/IP LAN that is for only resetting purposes (reset-only LAN). Make sure that the LAN allows communication via broadcast.

If the server is BladeSymphony, or model HA8000xM or earlier, connect the failure management processor of each host to the reset-only LAN. If the server is model HA8000xN or later, connect the BMC and the management server that will be used as a failure management processor on each host to the reset-only LAN.

Also, to issue reset instructions from HA Monitor to the failure management processor, connect each host's LAN adapter to the LAN dedicated to resetting.

Make sure that you provide a LAN dedicated to resetting separate from LANs used for other jobs in order to avoid the delay caused by host reset or host reset errors that might result from an increase in traffic. You can also protect the failure management processors from unauthorized access.

Important

Do not share a reset path as a monitoring path. If a reset path is shared in this manner, both functions might be disabled by a single point of failure.

Note that if you use BladeSymphony, there is a limitation to the supported number of LAN adapters. Share the reset path only for the purpose of providing multiple monitoring paths.

This subsection discusses the considerations for the reset path configuration and provides notes about sharing a reset path among multiple hot-standby configurations.

Organization of this subsection

(1) Considerations for the reset path configuration
(2) Notes about sharing a reset path among hot-standby configurations

(1) Considerations for the reset path configuration

The reset path configuration depends on the system being used and the hot-standby configuration. This subsection explains the reset path configuration. For details about the system being used, see (1) Processors. For details about types of hot-standby configurations, see 2.1 Hot-standby configurations.

The following table describes the considerations for the reset path configuration.

Table 5‒8: Considerations for the reset path configuration
Consideration	Description
How to connect failure management processors and the reset path	A failure management processor has a separate LAN port from the system LAN port. You connect the failure management processor to the reset path by connecting that LAN port to the reset path. The location of the LAN port depends on the system. For details, see your hardware documentation.
Providing dual reset paths	The reset path itself cannot be made dual. You must use a kernel extension package in Linux, such as a bonding package, to provide dual LAN adapters that are connected to the reset path. If you use dual LAN adapters and a failure occurs on one of the LAN adapters, HA Monitor can use the other LAN adapter to perform hot standby processing. If you use VMware ESXi-based virtualization, we recommend that you make LAN adapters redundant by using the VMware ESXi functions.
Whether Hitachi server virtualization is used (BladeSymphony)	If you use Hitachi server virtualization to perform hot standby processing between LPARs, one LPAR is treated as a host. Therefore, you must connect the reset path from each LPAR.

Note:: If the model of the server in use is HA8000xN or later, interpret the failure management processor as BMC.

Examples of reset path connection configurations are shown below.

To create a 1-to-1 switchover configuration (with model HA8000xM or earlier)

Figure 5‒8: Reset path connection configuration (1-to-1 switchover configuration (with model HA8000xM or earlier))

To create a 1-to-1 switchover configuration (with model HA8000xN or later)

Figure 5‒9: Reset path connection configuration (1-to-1 switchover configuration in which a management server is used as a failure management processor (with model HA8000xN or later))

Figure 5‒10: Reset path connection configuration (1-to-1 switchover configuration in which a management server is not used as a failure management processor (with model HA8000xN or later))

1-to-1 switchover configuration in the same chassis (BladeSymphony)

Figure 5‒11: Reset path connection configuration (1-to-1 switchover configuration in the same chassis) (BladeSymphony)

If the system being used is BladeSymphony, you can employ a hot-standby configuration within the same chassis. If you use Hitachi server virtualization to perform hot standby processing between LPAR, you can also employ a hot-standby configuration within the same chassis.

To Page Top

(2) Notes about sharing a reset path among hot-standby configurations

If you use BladeSymphony or HA8000 and create multiple hot-standby configurations that do not monitor each other in the system, the reset path might be shared among the hot-standby configurations due to hardware limitations.

If the environment is set up incorrectly in such a network configuration, HA Monitor will not be able to identify hosts correctly. As a result, a host that has nothing to do with a host failure that has occurred might be reset or the HA Monitor might terminate abnormally. This subsection describes the settings required for HA Monitor to identify hosts correctly in the following cases:

When sharing failure management processors (BladeSymphony)
When sharing a failure management processor (model HA8000xN or later)
When not sharing failure management processors (BladeSymphony or HA8000)
When sharing standby server modules in an N+M cold standby configuration (BladeSymphony)

When you configure your system, you must check the IP addresses specified here for any duplications. For details about how to perform an IP address duplication check, see 6.23.1 Checking the settings and definitions between multiple hot-standby configurations.

Note

This applies to hot-standby configurations that use the following OSs and systems supported by HA Monitor:

AIX (EP8000)
HP-UX (IPF) (BladeSymphony)

(a) When sharing failure management processors (BladeSymphony)

Depending on the model of BladeSymphony, it might or might not be possible to set a separate port number for each hot-standby switchover configuration by using the HA Monitor linkage settings of SVP. The following describes how to specify the settings in both cases.

For models that allow you to use a separate port number for each hot-standby switchover configuration

We recommend that you use a separate port number for each hot-standby switchover configuration to prevent different hot-standby switchover configurations from communicating with each other. You can also use the same port number for multiple configurations. In this case, if you set the same host address for multiple configurations by mistake, a problem might occur in the event of a host failure. For example, a host that is not related to the failure might be reset, or HA Monitor might terminate abnormally.

Figure 5‒12: Network configuration that shares a reset path among multiple hot-standby configurations (when sharing SVPs)

As shown in the figure, if two hot-standby configurations are configured with one pair of hosts 1 and 3 and another pair of hosts 2 and 4, an SVP in each chassis is shared between hot-standby configurations because it is connected to two hosts in different hot-standby configurations. In this case, you must specify the following settings:

Specify a local host address that is unique in all hosts. This host address must also be unique among the hosts in different chassis so that each host can be identified.

Use the address operand in the HA Monitor environment settings to specify a host address. Note that if the specified host address is duplicated, the KAMN414-E message is issued. In such a case, see 7.5.14 Handling a duplicated address operand value and take appropriate action.
Make sure that the port number of the reset path differs from the port numbers of the SVPs.
Make sure that the port number of the reset path is different for each hot-standby switchover configuration. However, to use virtualization, set the same value for the hosts on the same device.
For SVP port numbers, set a different value for each hot-standby switchover configuration. However, to use virtualization, you must set the same value for the hosts on the same device.

For an example of the settings, see (3) Configuration consisting of two chassis and four hosts (Basic mode).

For an N+M cold standby configuration, also note that an assigned partition name must be unique within the entire host. For specification examples, see (6) N+M cold standby configuration.

For models that do not allow you to use a separate port number for each hot-standby switchover configuration

The following shows an example of a network configuration that shares a reset path among multiple hot-standby configurations and in which the SVPs that serve as failure management processors are shared by the hot-standby configurations.

Figure 5‒13: Network configuration that shares a reset path among multiple hot-standby configurations (when sharing SVPs)

Specify a local host address that is unique in all hosts. This host address must also be unique among the hosts in different chassis so that each host can be identified.

Use the address operand in the HA Monitor environment settings to specify a host address. Note that if the specified host address is duplicated, the KAMN414-E message is issued. In such a case, see 7.5.14 Handling a duplicated address operand value and take appropriate action.
Specify the same port number for the reset path in all hosts. Even in a separate chassis, specify the same value for all the hosts that share the SVP.
For the port number of an SVP, specify the same value.
Make sure that the port number of the reset path differs from the port numbers of the SVPs.

For an example of the settings, see (3) Configuration consisting of two chassis and four hosts (Basic mode).

For an N+M cold standby configuration, also note that an assigned partition name must be unique within the entire host. For specification examples, see (6) N+M cold standby configuration.

(b) When sharing a failure management processor (model HA8000xN or later)

The following shows an example of network connections in which multiple hot-standby switchover configurations share a reset path and failure management processor.

Figure 5‒14: Network connections in which a reset path is shared by hot-standby switchover configurations (in the case where a failure management processor is shared)

To create two hot-standby switchover configurations with two host pairs (hosts 1 and 3, and hosts 2 and 4) in a network shown in the preceding figure, the hot-standby switchover configurations need to share a failure management processor. In this case, you must specify the following settings:

Specify a local host address that is unique in all hosts. Use the address operand in the HA Monitor environment settings to specify a host address. Note that if the specified host address is duplicated, the KAMN414-E message is issued. In such a case, see 7.5.14 Handling a duplicated address operand value and take appropriate action.
Specify the port number of the reset path and the port number of the failure management processor as follows:
- Set the same value for all hosts in one hot-standby switchover configuration.
- Set a different value for each hot-standby switchover configuration. Note that if you use VMware to perform virtualization, the same value must be set for the hosts on the same device.
Specify different values for the port number of the reset path and the port number of the failure management processor.

For an example of the settings, see (2) When there are two pairs of hot-standby configurations.

(c) When not sharing failure management processors (BladeSymphony or HA8000)

The following shows an example of a network configuration that shares a reset path among multiple hot-standby configurations but in which the SVPs that serve as failure management processors are not shared by the hot-standby configurations.

Figure 5‒15: Network configuration that shares a reset path among multiple hot-standby configurations (when not sharing failure management processors)

As shown in the figure, if two hot-standby configurations are configured with one pair of hosts 1 and 3 and another pair of hosts 2 and 4, SVPs are not shared between different hot-standby configurations. In this case, you must specify one of the following (specifying both will not cause a problem):

Specify a local host address that is unique in all hosts. This host address must also be unique among the hosts in different chassis so that each host can be identified.

Use the address operand in the HA Monitor environment settings to specify a host address.
Specify a different port number for the reset path in each hot-standby configuration. In a pair of hot-standby configurations, specify the same value for all hosts. Also specify a different port number for the SVP in each hot-standby configuration.

Make sure that the port number of the reset path differs from the port numbers of the SVPs.

For an example of the settings, see (4) Configuration consisting of four chassis and four hosts (Basic mode) for BladeSymphony and (2) When there are two pairs of hot-standby configurations for HA8000.

For an N+M cold standby configuration, also note that an assigned partition name must be unique within the entire host. For specification examples, see (6) N+M cold standby configuration.

(d) When sharing standby server modules in an N+M cold standby configuration (BladeSymphony)

This subsection explains a network configuration that satisfies the following conditions:

A set path is shared among multiple hot-standby configurations.
A standby server module is shared among multiple N+M cold standby configurations.

The following figure shows an example of such a network configuration.

Figure 5‒16: Network configuration that shares a reset path among multiple hot-standby configurations (a standby server module is shared among multiple N+M cold standby configurations)

To configure a total of two sets of hot-standby configurations, where Hosts 1 and 3 make up one set and Hosts 2 and 4 make up another set, specify the following settings:

Set the local host's host address to a value that is unique among all hosts so that a host in a separate chassis can also be identified by its host address.

Specify a host address in the address operand in the HA Monitor environment settings. If the specified host address is duplicated, the KAMN414-E message will be issued; in such a case, see 7.5.14 Handling a duplicated address operand value and take appropriate action.
Specify the same port number for the reset path on all hosts. Also specify the same SVP port number on all hosts.
Specify a different partition name for each host.

For a specification example, see (6) N+M cold standby configuration.

To Page Top