set format

[set name_total_size=service-information-area-size]
[set name_cache_size=service-information-cache-area-size]
[set max_socket_descriptors=maximum-number-of-file-descriptors-for-
                            sockets]
[set name_global_lookup=Y|N]
[set name_service_extend=1|0]
[set name_audit_conf=2|1|0]
[set name_audit_interval=monitoring-interval]
[set name_audit_watch_time=maximum-time-to-wait-until-a-node-
                            failure-is-detected]
[set name_rpc_control_list=Y|N]
[set name_nodeid_check_message=Y|N]
[set name_cache_validity_time=validity-duration-of-service-information-of-other-nodes]
[set watch_time=maximum-time-to-wait-for-a-response]

command format

None

The name service definition defines the execution environment to manage the service name and corresponding address which enables RPC. A name service secures storage equal to the sum of the size of the service information area and service information cache area as specified here, from the static shared memory specified with the system environment definition.

name_total_size=service-information-area-size~<unsigned integer> ((1-32767)) <<64>> (Unit: kilobytes)

Specify the size of the service information area to be secured by the name service on the local node. Compute this size based on the number of service information to be registered with the system services and user servers. For details about the formula, see B.1 Estimating the shared memory requirements for TP1/Server Base.

name_cache_size=service-information-cache-area-size~<unsigned integer> ((1-32767)) <<64>> (Unit: kilobytes)

Specify the size of the service information cache area to be secured by the name service on the local node. Compute this size based on the number of service information retrievals to be requested by the local node from the service information registered in other nodes. For details about the formula, see B.1 Estimating the shared memory requirements for TP1/Server Base.

A name service could use this service information cache area as the service information area if the latter is insufficient. Conversely however, the service information area will never be used as the service information cache area.

A larger service information cache area will improve the response time of service information retrievals from other nodes. However, an area in excess of what is required will cause the memory of the OpenTP1 system as a whole to be compressed.

max_socket_descriptors=maximum-number-of-file-descriptors-for-sockets~<unsigned integer> ((32-2032))

Specify the maximum number of file descriptors to be used for sockets by the processes under control of OpenTP1^#.

The processes under control of OpenTP1^# exchange the process information with the system servers or user servers through the TCP/IP communication using sockets. Therefore, you must change the maximum number of file descriptors for sockets depending on the number of UAP processes that run concurrently and the number of other nodes to communicate with.

#1: OpenTP1 processes other than the MCF services (MCF manager service, MCF communication service, and application startup service). For the MCF services, see the sections on the system service information definition and the system service common information definition.

Use the following formula for calculating the maximum number of file descriptors for sockets:

(Number of UAP processes in the local node^#1 + number of nodes that request a service of the name service^#2 + number of system service processes^#3)/0.8

: Rounded up to the nearest whole integer.

#1: The number of UAP processes in the local node is the sum of the following values:

• Number of UAP processes in the local OpenTP1
• Number of transactions to be started concurrently by the CUP (value of the parallel_count operand specified in the client service definition)

#2: The number of nodes that request a service of the name service is the sum of the following values. Count any duplicated nodes as one.

• Number of node names specified in the all_node operand for the local OpenTP1
• Number of other nodes that specify the local node name in the all_node operand for OpenTP1

#3: The number of system service processes in the local OpenTP1.

If the value specified for this operand is too small, the connection cannot be set with other processes under control of OpenTP1. The process terminates abnormally after outputting the KFCA00307-E error message.

The order of priority of the specified values is 1 > 2.

1. Name service definition
2. System common definition

If this operand is omitted, the system assumes the value in the system common definition.

name_global_lookup=Y|N~<<N>>

Specify whether to use the global search facility.

The global search facility allows you to retrieve information about the services in other nodes cached by the name service on each node specified in the all_node operand.

Y

The global search facility is used.

N

The global search facility is not used.

Figure 3-3 shows the system configuration example when the global search facility is used.

Figure 3-3 System configuration when the global search facility is used

The following table shows the operand values to be specified on each OpenTP1 node in the system configuration shown in the above figure.

Table 3-1 Values specified in the operands in the system using the global search facility

The range of search from OpenTP1-A is OpenTP1-B to OpenTP1-G. In other words, OpenTP1-A can communicate with the services in OpenTP1 systems 2 and 3 as well as OpenTP1 system 1.

In the environment where the global search facility is enabled, the dc_rpc_call_to function, which specifies the port number of the name service in parameter portno in the DCRPC_BINDTBL_SET function, is not available.

The dc_rpc_call_to function, which specifies a node identifier in parameter nid in the DCRPC_BINDTBL_SET function, can be used only when each of the node identifiers of all the OpenTP1 nodes in the search range of the global search facility (OpenTP1 systems 1 to 3 in the above figure) is unique.

The information about services (such as shutdown status or load status) is not reported to the search source OpenTP1 (OpenTP1-A in the above figure). Therefore, we recommend that you request services using the horizontal distribution shown in the figure.

When calculating the value specified in the name_cache_size operand in the name service definition, the number of service information items cached to the node specified in the all_node operand must be added to the number of service information items which the local node requests to search for.

When TP1/Client/P, TP1/Client/W, or TP1/Client/J sends a service request to OpenTP1 (OpenTP1-A in the above figure) using the global search facility, information about the services in OpenTP1 systems 2 and 3 as well as OpenTP1 system 1 is acquired.

For details about the global search facility, see the manual OpenTP1 Description.

name_service_extend=1|0~<<0>>

Increase the number of server UAPs (managed by the name service) to be obtained. When you specify 1 for this operand, an RPC can be sent to a maximum of 512 server UAPs that have the same service group name and that are started on the node the RPC is sent to. However, RPC destination nodes need to be included in the OpenTP1 system specified in the all_node operand. When you specify 0 for this operand or do not specify any value, an RPC can be sent to up to 128 server UAPs.

name_audit_conf=2|1|0~<<0>>

Specify whether to use the node monitoring feature. For details about the node monitoring feature, see the manual OpenTP1 Description.

2

Uses the node monitoring feature in the two-way (send and receive) mode.

1

Uses the node monitoring feature in the one-way (send-only) mode.

0

Does not use the node monitoring feature.

Note the following points when you specify this operand:

• The node monitoring feature is unavailable when multiple instances of OpenTP1 are running on the host for which you want to monitor activation of OpenTP1. The facility is also unavailable in an environment in which multiple instances of OpenTP1 will start with the same IP address (when only one LAN board is used) after a system switchover occurs.
• For the communication that the node monitoring feature uses for monitoring, you can tune the sensitivity of node failure detection by adjusting the value of an operand. Which operand you adjust depends on the value of the name_audit_conf operand.
  When 1 is specified in the name_audit_conf operand
  Adjust the value of the ipc_conn_interval operand in the system common definition.
  When 2 is specified in the name_audit_conf operand
  Adjust the value of the name_audit_watch_time operand in the name service definition.
  
• The node monitoring feature can concurrently monitor a maximum of 60 nodes. If the total number of nodes specified in the all_node and all_node_ex operands of the system common definition exceeds 60, the feature performs monitoring for 60 nodes at a time.
• If you specify many nodes in the all_node and all_node_ex operands of the system common definition, the operation of the node monitoring feature may affect the RPCs issued by UAPs. If you need to specify many nodes in the operands, do not specify a small value in the name_audit_interval operand. Also, if the namalivechk command is repeatedly executed, make sure that the command is not executed at short intervals.
• If the name_rpc_control_list operand is omitted or Y is specified in the operand, a node recovered from a failure may be removed from the RPC control list before the time specified in the name_audit_interval operand elapses. If this happens, the KFCA00651-I message is not output.
• If 2 is specified in the name_audit_conf operand when the name_rpc_control_list operand is omitted or Y is specified in the name_rpc_control_list operand, the KFCA00650-I message may be output periodically.
• When 1 or 2 is specified in the name_audit_conf operand and 180 or a smaller value is specified in the name_audit_interval operand, Hitachi recommends that you specify N in the name_rpc_control_list operand.
• If you specify N in the name_rpc_control_list operand when the name_audit_conf operand is omitted or 0 is specified in the name_audit_conf operand, the node monitoring feature and the feature for monitoring the nodes registered in the RPC control list are disabled. Note the following when both features are disabled:
  
  • A node is removed from the RPC control list only when communication from the node to the local node occurs.
  • If the all_node operand for a node registered in the RPC control list does not specify the local node, the node is not removed from the list until OpenTP1 on the local node is restarted.
  Hitachi recommends that you omit the name_rpc_control_list operand or specify Y in the operand when the name_audit_conf operand is omitted or 0 is specified in the name_audit_conf operand.
  

name_audit_interval=monitoring-interval~<unsigned integer> ((1-65535)) <<60>> (Units: seconds)

Specify the length of time between the end of node monitoring by the monitoring service and the beginning of the next monitoring.

name_audit_watch_time=maximum-time-to-wait-until-a-node-failure-is-detected~<unsigned integer> ((8-65535))<<8>>(units: seconds)

When the name_audit_conf operand is set to 2 (node monitoring in the two-way mode), specify the maximum time to wait until the monitoring-target node responds. Note that if the name_audit_conf operand is set to 1, the value of the ipc_conn_interval operand in the system common definition is set as the time to wait until a node failure is detected.

name_rpc_control_list=Y|N~<<Y>>

Specify whether to enable monitoring of the nodes registered in the RPC control list.

Y

Performs monitoring of the nodes registered in the RPC control list at 180-second intervals.

N

Does not monitor the nodes registered in the RPC control list.

The name service provides a feature separate from the node monitoring feature for checking the activation status of the nodes registered in the RPC control list at 180-second intervals. This feature is enabled using the name_rpc_control_list operand.

Decide whether to use this feature after considering the settings of the node monitoring feature. For example, you must disable the feature for monitoring the nodes registered in the RPC control list in the following cases:

• A node recovered from a failure is removed from the RPC control list before the time specified in the name_audit_interval operand elapses, and the KFCA00651-I message is not output.
• 2 is specified in the name_audit_conf operand, and the KFCA00650-I message is output periodically.

When the feature for monitoring the nodes registered in the RPC control list is disabled and the name_audit_interval operand is set to 180 or more seconds, removal of a recovered node from the list takes longer than usual.

Hitachi recommends that you specify N in the name_rpc_control_list operand when both the following conditions are satisfied:

• The name_audit_conf operand is set to 1 or 2.
• The name_audit_interval operand is set to 180 or a smaller value.

If name_audit_conf is omitted or is 0, and name_rpc_control_list is N, the node monitoring feature and the feature for monitoring the nodes registered in the RPC control list are disabled. Note the following when both features are disabled:

• A node is removed from the RPC control list only when communication from the node to the local node occurs.
• If the all_node operand for a node registered in the RPC control list does not specify the local node, the node is not removed from the list until OpenTP1 on the local node is restarted.

Hitachi recommends that you omit the name_rpc_control_list operand or specify Y in the operand when the name_audit_conf operand is omitted or 0 is specified in the name_audit_conf operand.

name_nodeid_check_message=Y|N ~<<Y>>

Specify whether the KFCA00677-W message is output when communication is received from a node whose node identifier is the same as the local node.

Y

The KFCA00677-W message is output.

N

The KFCA00677-W message is not output.

name_cache_validity_time= validity-duration-of-service-information-of-other-nodes

~<unsigned integer>((0-65535)) <<1800>> (Unit: seconds)

Specify the validity duration of the server UAP service information acquired from other nodes. The name service acquires the service information of the server UAPs that are active in other nodes at the times listed below.

• When an RPC is executed at the node of an RPC-target service group for the first time
• When a server UAP is started at another node

During the time period specified in this operand following the acquisition of service information from other nodes, the RPC is executed without performing a service information search on other nodes. If the validity duration of the service information has exceeded the time specified in this operand when the RPC is executed, service information is newly acquired from the nodes specified in the all_node operand of the system common definition, and the registered service information is updated. Even within the validity duration of the service information, the service information is updated if a service information update notice is received from another node.

Note the following when you change the specification value in the name_cache_validity_time operand:

• When 0 is specified:
  Even when the validity duration for service information expires, service information is not acquired.
  
• When a value smaller than the default value (1-1799) is specified:
  Since the validity duration for service information becomes shorter, service information is exchanged with the name services of other nodes frequently during RPC execution, increasing communications traffic with other nodes. In this case, shortening the interval for updating the global cache can accelerate the detection of service information changes at the nodes specified in the all_node operand of the system common definition.
  If an extremely small value is specified, the load on the name service of each node increases and may make it impossible to use name service functions such as service information searching.
  
• When a value greater than the default value (1801-65535) is specified
  Since the validity duration for service information becomes longer, service information is exchanged with the name services of other nodes less frequently during RPC execution, decreasing communications traffic with other nodes. Since there are fewer opportunities to acquire service information from server UAPs that are active on other nodes, a phenomenon may occur in which the RPC cannot be executed on a server UAP that is active on a given node.
  

watch_time=maximum-time-to-wait-for-a-response ~<unsigned integer> ((0-65535)) (units: seconds))

Specify the maximum time to wait for a response to a service request sent for inter-process communication that uses RPCs.

Specify this operand in an environment in which a client UAP requests a service from a server UAP on a remote node. The following figure provides an overview of requesting a service from a server UAP on a remote node, and shows the points in the RPC communication to which this operand applies.

Figure 3-4 Overview of requesting a service from a server UAP on a remote node

Make sure that the value of this operand is smaller than the value for the maximum time to wait for a response specified in the client UAP. If the value of this operand is equal to or larger than the value specified in the client UAP, a problem might occur. That is, the service search request from the client UAP process might time out before the transferred request times out if a delay in name service processing occurs on the remote node.

Since OpenTP1 might wait for the time specified in this operand during termination processing. OpenTP1 termination processing might take time if you specify a large value in this operand.

If there is no response for the specified period of time, the RPC returns a timeout error.

If 0 is specified in this operand, OpenTP1 waits until a response is received. Note also that if 0 is specified, OpenTP1 might not be able to terminate.

If this operand is omitted, the value of the watch_time operand in the system common definition is assumed.

Note that if the value of this operand is very much larger or smaller than the default value of the watch_time operand in the system common definition, difficulties that can cause OpenTP1 to fail might occur.

None