Range: a string of up to 15 characters.

The DOMAIN element configures settings for the overall Tmax environment.

Range: 32768 ~ 262143

Value of the pointer that points to a shared memory segment. Tmax has four main processes: Tmax Manager TMM that manages other processes, TMS that manages databases and resources, CLL that manages client connections, and CLH that processes requests and responses between clients and server processes.

SHMKEY configures shared memory key value for managing the information shared by the four processes. Tmax uses the four key values, SHMKEY, SHMKEY+1, SHMKEY+2, and SHMKEY+3, in the order listed.

Can be used in all elements of the environment file. For details, refer to "DEFAULT in 3.2.2 NODE".

Range: 1 ~ MAX_INT

The maximum number of clients that can be simultaneously connected to the domain. The actual number of connections allowed is MAXUSER +1.

Range: 1 ~ 10

Default: 1

The minimum number of startup CLH processes.

The CLH is used to forward client requests to the appropriate server process, and to return the results back to the client. Every client must be connected to CLH, but the number of clients that can connect to a single CLH is limited. Thus, a large domain with a large volume of clients should run more than one CLH instance by setting a suitable MINCLH value.

[Figure 3.2] MINCLH = 2

Range: 1 ~ 10

Default: 10

The maximum number of active concurrent CLH processes allowed. When Tmax is started, MINCLH number of CLH processes are activated. Additional CLHs can be activated if the number of client requests exceeds the capacity of the active CLHs.

The total number of active CLH instances cannot exceed the MAXCLH value. Currently, Tmax does not automatically increase the number of MAXCLH processes according to the number of client requests.

Default: 8888

Port number used by clients to connect with Tmax. Two port numbers are used: TPORTNO and TPORTNO+1.

The TPORTNO value will be used for connections between clients and servers, while TPORTNO+1 will be used for inter-node communication. In a client/server environment, clients need to know the server address as well as the port number of Tmax.

The address (TMAX_HOST_ADDR) and port number (TMAX_HOST_PORT) must be set on the client side. The port number must be set to TPORTNO. If the client and server are both on the same node, it is recommended to use the domain socket instead of the TCP/IP socket to achieve more efficient connections.

Servers are capable of receiving two types of client connections: TPORTNO is used for TCP/IP socket connections. If the client and server exist on the same node, stream pipe (cllrcad) is used according to the PATHDIR setting.

If the client and server are both on the same node, TMAX_HOST_PORT must be set to PATHDIR instead of TPORTNO. For example, set TMAX_HOST_PORT=/home/tmax/path in .profile or tmax.env.

TMAX_HOST_PORT=/home/tmax/path

TPORTNO+1 can also be used for inter-node communication.

Nodes in the Tmax system are connected in a peer-to-peer relationship. By using the port number TPORTNO+1, subsequent nodes will be checked in the order they are configured. Port numbers less than 1024 should not be used since they are reserved for system use.

Since the port numbers, TPORTNO and TPORTNO+1, are used for communication between a client and Tmax as well as between nodes, check that these values are not used anywhere else. In addition, for multi-node systems, the number of specified port numbers must match the number of CLHs. There must be a separate port number for each CLH.

To use the IPv6 protocol, CLIENT_IPV6 must be set to Y.

Default: 3333

Port number used by racd for inter-node communication and centralized node management in a multi-node domain. The port number must be same as the TMAX_RAC_PORT port number. Port numbers less than 1024 should not be used since they are reserved for system use.

If the values specified here and in TMAX_RAC_PORT are different, only the TMAX_RAC_PORT value will be used. It is recommended to use a lower value for RACPORT than TPORTNO. Make sure that the port numbers specified in RACPORT are not used anywhere else.

Range: 1 ~ MAX_INT

Default: 0 (seconds)

If no value is set, by default requests will infinitely wait for a response.

Time-out value for service requests. If a service request issued by a client fails to receive a response within the specified time frame, the request will time-out on the client side. However, the server process will continue to process the request.

To terminate the service on the server side, set the desired time-out period in SVCTIME of the SERVICE element. BLOCKTIME is strongly linked to SVCTIME. The shorter time value between BLOCKTIME and SVCTIME is used as timeout for client tpcall. BLOCKTIME is the tpcall timeout value, and SVCTIME is the server process timeout value. In general, BLOCKTIME value is greater than SVCTIME value.

Range: 1 ~ 128

Default: 1

The number of channels of CLH processes on a node in multi-node systems. In systems with large volumes of inter-node service requests or that process large volumes of distributed transactions, too few number of channels may slow down service processing. For optimal system performance, use an adequate number of parallel communication channels.

Range: 1 ~ MAX_INT

Default: 0

If not specified, the default value is set, and the function is not used.

Size of the function table require for TopEnd system functions in Tmax system when converting a TopEnd system to a Tmax system.

Registers services to be processed in normal or abnormal state with the client logout process of Tmax server.

Used for normal client logouts when logging in or out of a service. Handles the case where clients are automatically logged out when a preset time (CLICHKINT, IDLETIME) expires.

Note that this variable only runs on Tmax clients. It does not run when using a general socket connection. For programs that use the service name, client IDs can be obtained by calling tpgetclid(). Refer to CD of tpsvcinfo structure to determine the cause of the logout.

Range: 1 ~ MAX_INT

Default: 0

If not specified, the default value is set, and the function is not used.

The time interval at which clients are monitored. CLICHKINT must be used in conjunction with IDLETIME to minimize resource waste.

If CLICHKINT is configured in the DOMAIN element, all clients connected to the domain are monitored. However, if it is configured in the NODE element, only those clients that are connected to the node are monitored.

Range: 1 ~ MAX_INT

Default: 0

If not specified, the default value is set, and the function is not used.

Time period that clients can remain idle before being automatically disconnected from the system. If CLICHKINT is set to 5 and IDLETIME is set to 30, clients will be monitored every 5 seconds (CLICHKINT) and if they do not issue any service requests within 30 seconds (IDLETIME), they will be disconnected from Tmax system.

Clients will be monitored from the time tpstart() is called to make a connection, or when tpcall()/tpacall() is requested. The value of IDLETIME must be set to a value greater than BLOCKTIME and SVCTIME. If IDLETIME is set to a value less than either BLOCKTIME or SVCTIME value, clients will automatically be disconnected before they receive a response.

Range: 1 ~ 1024

Default: 8

The number of times tpacall() can be requested by a server process without a tpgetrply() call. Since the server library follows each tpacall() call, the table needs to be created in the initial MAXSACALL that configures the table size.

Range: 1 ~ 1024

Default: 16

The number of times tpacall() can be requested by a client process without a tpgetrply() call. Since the client library follows each tpacall(), the table needs to be created in the initial MAXCACALL that configures the table size.

Range: 1 ~ MAX_INT

Default: 0 (seconds)

If not specified, by default requests will infinitely wait for a response.

Time-out period for service processing in a transaction (in seconds). When the time-out period is exceeded, the transaction is cancelled.

Internal time-out setting that is applied to prevent infinite blocking that can be caused by DB or network failure during XA processing before/after service processing.

Range: 1 ~ MAX_INT

Default: 30 seconds

Node monitoring interval (in seconds) for a multi-node Tmax system. To ensure reliable inter-node communications at all times, the nodes of a domain are connected in a peer-to-peer relationship, and each node regularly monitor the subsequent node in the order they are configured.

If the interval period is short, nodes are monitored frequently and node failures are detected quickly, but inter-node traffic may increase to an unacceptable level. On the other hand, if the interval period is long, nodes are monitored less frequently but inter-node traffic is reduced. Select an appropriate value by considering both the network traffic and need for quick error detection.

Range: 1 ~ MAX_INT

Default: 0 (seconds)

Time-out period (in seconds) with no data flow that triggers CLHs of each node to send a monitoring signal to each other.

This feature must be used with care since after a time-out period with no data flow, the firewall will terminate all processes and this will affect CLH processes of Tmax.

Range: 1 ~ 2048

Default: 16

Maximum number of interactive services that can be simultaneously executed for each CLH. If all clients and server processes can simultaneously call up to 100 interactive services, use a value that is greater than or equal to 100.

If the number of services exceeds this value, a TPELIMIT error will occur.

Range: 1 ~ 256

Default: 8

Similar to MAXCACALL and MAXSACALL, this variable specifies the maximum number of interactive services that can be simultaneously executed by a client or a server process.

If the number of services exceeds this value, a TPELIMIT error will occur.

Default: "NO_SECURITY"

Security setting whose value is one of "NO_SECURITY" | "DOMAIN_SEC" | "USER_AUTH".

For more details about Tmax security settings, refer to "3.9. Security Settings".

Range: string within 15 characters

For details, refer to "3.9. Security Settings".

Range: 0600 ~ 0777

Default: 0600 (If undefined, the default value for file access control is 600.)

IPCPERM (Inter-Process Communication Permission Mask) specifies the control unit of the pipe file located in $(TMAXDIR)/path and in shared memory segment. It also sets the control unit of the system log file ($(TMAXDIR)/log/slog).

In a UNIX environment, the administrator can assign file access control (to read/record/execute files) to each individual, group, etc. Read and write accesses can be assigned to individuals, but not to groups.

This value can be used to grant users the permission to start and terminate server processes. Without IPCPERM, users cannot start or terminate server processes.

Range: 0 ~ 255

Default: 0

Key value that identifies the domain in a multi-domain environment. Required setting for multi-domain systems.

Default: Y

Refer to "3.4. Distributed Transaction Settings".

Range: 1 ~ MAX_INT

Default: 32

Maximum number of nodes allowed per domain.

Range: 1 ~ MAX_INT

Default: 32

Maximum number of server groups allowed per node.

Range: 1 ~ MAX_INT

Default: 64

Maximum number of server processes that can be started on the node. It is calculated by taking the sum of maximum values of each server processes.

Range: 1 ~ MAX_INT

Default: 64

Maximum number of servers that can be registered per node.

For example, if there are three servers (svr1, svr2, and svr3) assigned to the node and the MAX value for each server is 50 (each server can generate up to 50 server processes), the MAXSPR is 150 and MAXSVR is 3.

Range: 1 ~ MAX_INT

Default: 512

Maximum number of services that can be assigned to each domain.

This value may be different from the result of executing "cfg -d" in tmadmin because this includes space reserved for dynamically adding services and nodes later.

Range: 0 ~ 65535

Default: 32

Maximum number of CPCs allowed per node.

Range: 1 ~ MAX_INT

Default: 32

Maximum number of active TMSs allowed per node.

Range: 1 ~ MAX_INT

Default: 16

Maximum number of services that can configure the ROUTING setting.

For example, if 10 out of 30 services can configure the ROUTING setting, then set MAXROUT to 10.

Range: 1 ~ MAX_INT

Default: 32

Maximum total number of groups (server groups or domain gateways) configured in the ROUTING setting of the SERVICE element.

For instance, assume that the ROUTING setting of Service A is RT1 and that of Service B is RT2. If the number of groups set in RANGES of RT1 is 2, and that of RT2 is 4, the total number of ROUT SVGs is 6. In this case, MAXROUTSVG should be set to at least 6.

*SERVICE
A        ROUTING = RT1
B        ROUTING = RT2

*ROUTING
RT1     RANGES = "'a':svg1,*:DGW1"
RT2     RANGES = "'a':svg1,'b':svg2,'c':DGW2,*:DGW1"

Range: 1 ~ MAX_INT

Default: 2

Maximum number of active RQs per node.

Range: 1 ~ MAX_INT

Default: 2

Maximum number of active gateways per node.

Range: 1 ~ MAX_INT

Default: 16

Maximum number of server groups that can configure the COUSIN setting. For example, if 10 out of 30 server groups can configure the COUSIN setting, set MAXCOUSIN to 10.

Range: 1 ~ MAX_INT

Default: 32

Maximum total number of server groups that can configure the COUSIN setting of SVRGROUP element. For example, if the MAXCOUSIN value is 10 and COUSIN setting can be configured in two server groups, set MAXCOUSINSVG to 20.

Range: 1 ~ MAX_INT

Default: 16

Maximum number of server groups that can configure the BACKUP setting. For example, if 10 out of 30 server groups can configure the BACKUP setting, set MAXBACKUP to 10.

Range: 1 ~ MAX_INT

Default: 32

Maximum total number of the server groups that can configure the BACKUP setting of SVRGROUP element. For example, if the MAXBACKUP value is 10 and each BACKUP setting can include two server groups, set MAXBACKUPSVG to 20.

Range: 1 ~ MAX_INT

Default: 64

Maximum number of server groups that can be configured.

Range: 1 ~ MAX_INT

Default: 0

If not set, the default value is used, and this function is not used.

This item indicates the maximum value of product name that can be specified as the item that is used to convert a system, which was made for TopEnd system, into the Tmax system. (for TopEnd use)

Range: 1 ~ MAX_INT (seconds)

If a gateway to a backup node is available within a multi-domain system, service requests will be transferred to the backup node if a fault occurs in the original remote node. Service requests will maintain their connection to the backup node even after the original node has been recovered. The gateway to the backup node must be shut down before the service request can re-connect to the original node. However, if the GWCHKINT variable is properly configured, service requests will automatically reconnect to the original node after recovery.

Note

Range: 1 ~ MAX_INT (seconds)

The domain gateway tries to connect to the remote gateway at system startup.

If the GWCHKINT item is not specified in the DOMAIN element of the configuration file and the machine where the remote gateway is set is currently not running, then it takes a long time to show a connection failure message to the user who requested the service.

If the domain gateway does not attempt to reconnect when there are no requests, it will attempt to reconnect upon a service request, which will take a long time if the remote gateway is not ready. Such issue can be prevented by specifying the connection timeout.

Starting from Tmax v5.0 SP2, the domain gateway uses the non-blocking method to establish a connection. Since the interval to check connection timeout is determined according to environment settings such as NLIVEINQ and GWCHKINT, timeout can occur after the GWCONNECT_TIMEOUT.

If the GWCONNECT_TIMEOUT is not specified, default procedures will be used.

Default: DETAIL

Specifies the logging level for the TMM. (COMPACT, BASIC, (DETAIL), DEBUG1, DEBUG2, DEBUG3, or DEBUG4)

Range: Same as that of the TMM

Specifies the logging level for the CLH.

Range: Same as that of the TMM

Specifies the logging level for the TMS.

Specifies the logging level of the servers.

Default: N

Tmax v4.0 and later is capable of encrypting data transmitted between clients and the Tmax system to ensure extra protection. For this feature to be used, it must be configured for both the client and Tmax system.

To set this encryption function for Tmax system, set CRYPT to Y. To not use the encryption function, set CRYPT to N, default value. Refer to element "2.2.2. Tmax Client Environment Variables" TMAX_SEMANTICS environment variables for an explanation of how to set the encryption function for a client.

Note

Range: 1 ~ MAX_INT

Default: 0

If not specified, the default value is set, and the function is not used.

In Tmax version v4.0 and later, TMS can be configured as a multi-thread transaction manager to improve transaction processing efficiency. To enable this feature, set this value to the maximum number of threads for TMS. Additionally, the TMSTYPE, TMSTHREAD, and TMSOPT variables of the SVRGROUP element must also be configured in conjunction with this variable.

For a description of each variable, refer to "3.2.3 SVRGROUP" TMSTYPE, TMSTHREAD, and TMSOPT.

Note

Default: N

Option to use (Y|N) the TDL (Tmax Dynamic Library) Implicit Version Protection will be used.

Range: 1 ~ 65535

Default: 1024

Maximum number of sessions that HMS can create per node. For details, refer to "3.7. HMS Settings".

Range: 1 ~ MAX_INT

Default: 0

Time-out value for pending transactions. If no response is issued from a pending transaction for this time period, the transaction will be rolled back. If no value is set, no checks are performed.

Default: N

Option to use the IPv6 protocol when creating a port for client connections.

Default: N

Option to use the IPv6 protocol when creating a port for inter-node communication in a multi-node environment.

Default: N

Option to use the IPv6 protocol when creating a port for connection to the Extern Server from Tmax system.

Default: Y

Option to use the interface of ip address configured in IP setting of the NODE element when creating a port for client connection.

If set to N, or no IP is specified, connections from all interfaces are allowed.

Default: 1

Range: 1 ~ 2

If set to 2, it uses the version with enhanced search performance and is not compatible with existing Tmax system.

This setting affects server applications and the CLH scheduling that follows the ROUTING configuration section. When using fb* API, it returns the FBESETVER error if FDL_VERSION setting does not match with its environment.

Size: Up to 15 characters

Name of the physical machine. This value must match the name returned by the UNIX command, uname -n.

Node name must be configured in the "/etc/hosts" file. Since a domain consists of one or more nodes, there must be at least one node configured in the NODE element.

Size: Up to 255 characters

Full path of the directory where Tmax is installed. The path must be the same as the value of TMAXHOME. If TMAXHOME is not specified, all Tmax related tasks are performed from the TMAXDIR directory.

Size: Up to 255 characters

Full directory path for server application program executable files.

Tmax system processes as well as application processes for client requests start and terminate with the system.

APPDIR is also used as a working directory of the server. Core files created in a server program is stored in the APPDIR.

This variable can be defined in all elements of the configuration file (NODE, DOMAIN, etc.).

In the NODE element, it is useful for configuring items with the same value for all nodes in a multi-node configuration. When a default value is set using this label, it takes effect until the end of the element is reached or is overridden by another default setting.

If the default settings defined for "DEFAULT:" are not reconfigured in a NODE element, they are applied to all the nodes.

If a default setting is reconfigured in a NODE element, the new value is applied to that node.

Size: Up to 255 characters

This item is used to assign an actual hostname when the hostname is different from the configured node name. By setting the HOSTNAME, multiple logical nodes can be configured on a single machine. Multiple node names can be used per hostname.

TMAXHOME, NODETYPE, RACPORT, TPORTNO, and SHMKEY items must be configured in the NODE element. For details, refer to the following "Example" of the NODE element.

Size: Up to 255 characters

This item can be used to separately configure the Tmax installation and working directories. TMAXHOME is the absolute path of the directory where Tmax is installed, and must be same as the value of the TMAXHOME environment variable. If TMAXHOME is configured, TMAXDIR is the working directory path. Otherwise, TMAXDIR is the path for both directories.

The following is description of Tmax installation and working directories.

If HOSTNAME is also configured, each TMAXDIR becomes a single logical node, and TMAXHOME is Tmax installation directory shared by all logical nodes. Thus, TMAXDIR, SHMKEY, and TPORTNO must be separately configured for each logical node of the physical node.

Default: SHM_RACD

Method for managing individual nodes when there are multiple logical nodes on a single machine. Each local node of a machine must have a different RACPORT value. Where there are multiple local nodes, each local node must be set to a different RACPORT value.

Note that this setting is only used for managing nodes that are on the same host node as the logical node that executes tmboot/tmdown/tmadmin/cfl. Nodes on another host node will be managed by using RACPORT setting regardless of the NODETYPE setting.

This setting is used to determine whether to use RACD when executing tmboot/tmdown/tmadmin/cfl on node B, which is on the same host node as the logical node (node A) that has already started executing tmboot/tmdown/tmadmin/cfl.

Management options are:

Size: Up to 255 characters

Default: (TMAXDIR)/path

The directory path of the named stream pipe used for communication between Tmax processes. Communication methods include using a pipe (unnamed/named pipe), message queue, or shared memory.

Tmax uses a shared memory and stream pipe for inter-process communication (refer to the description in "3.2.1 DOMAIN" SHMKEY). A pathname is required to use a stream pipe.

PATHDIR is the directory path where pathnames, which are required for communication between Tmax processes through stream pipes, are created. Named pipes are created in this directory. If PATHDIR is not defined, pipes will be created in a new directory inside the directory configured in TMAXDIR.

If PATHDIR is set to another(not $TMAXDIR/path) directory, the system may not terminate properly when Tmax system shuts down. This is because the system uses the stream pipe(named pipe) in $TMAXDIR/path directory when the system terminates. In order to resolve this problem, the user should configure the environment variable, TMAX_PATHDIR, with the PATHDIR value in the configuration file. Now if the system terminates, it will shut down properly by referencing the environment variable, TMAX_PATHDIR.

export TMAX_PATHDIR = /home/tmax/path_new

When Tmax system starts, previous configuration file in the config directory is copied to PATHDIR of the configuration file. This will prevent the system from not terminating properly when the configuration file is modified after Tmax system starts. If you set the previous PATHDIR (the one that was referenced at startup) in the TMAX_PATHDIR, the previous configuration file will be created under this directory. This configuration file will not be modified even if the configuration file is modified after system startup.

Size: Up to 255 characters

Default: (TMAXDIR)/log/slog

The absolute path of directory where system messages are logged. The logged messages include those generated by TMM, TMS, CLL, and CLH, as well as those generated internally by the system. The slog.date file for each date is created in this directory, and messages are saved to a file according to the date they are created.

If no directory is specified in SLOGDIR, a log/slog folder will be created inside the directory configured in TMAXDIR.

Size: Up to 255 characters

Default: (TMAXDIR)/log/tlog

The absolute path where TXLOG file is created. The TXLOG file is used to log information about XA transactions. It is created as an unreadable binary file that is used to manage transaction information of Tmax engine. The file is created with a file system and TXLOG file logging is supported depending on the raw file.

If TLOGDIR is not defined, transaction information will be logged in the log/tlog folder inside the TMAXDIR directory.

During system boot, if a log file from the previous boot exists, the file is backed up as '.old', and a new transaction log file is created. The file cannot be larger than 16 MB, and the size can be modified by using the -l option of TLMOPT.

The size for a single xid in txlog is about 32 bytes. (txlog file size does not expand infinitely.)

Size: Up to 255 characters

Default: (TMAXDIR)/log/ulog

The absolute path where user messages are logged.

Users, or developers, can use the userlog() function to log debugging or various error and warning messages.

The messages generated by userlog() function are saved in the ULOGDIR directory. The ulog _date file for each date is created in this directory, and messages are saved to a file according to the date they are created. For more details about userlog(), refer to "Tmax Application Development Guide".

If no directory is set to ULOGDIR, messages will be logged in the log/ulog directory inside the TMAXDIR directory.

Size: Up to 15 characters

Since the system can consist of multiple domains, the domain that the node belongs to must be specified. The domain name must be configured in the DOMAIN element of the configuration file.

Range: 1 ~ MAX_INT

Maximum amount of time that a message can wait in a queue before being removed from the queue. This helps to prevent the messages from accumulating indefinitely in the queue due to problems with the host, network, etc. by notifying the client to take an appropriate action.

Size: Up to 255 characters

Configuration file that contains UNIX environment variables used to configure an appropriate environment for executing server application programs. The server application programs run in the environment configured with the settings in the ENVFILE.

Tmax system port number that clients can use to connect to servers. The port number is used as an alternative to TPORTNO.

A total of five ports can be used for connection if all ports from TPORTNO2 to TPORTNO5 are configured. Port numbers less than 1024 should not be used since they are reserved for system use.

TPORTNO and the next consecutive port number (TPORTNO+1) are reserved for communication between clients and Tmax or for inter-node communication. The port numbers must be unique, and the port number in TPORTNO should be different from those already configured.

Size: Up to 255 characters

Command options that can be sent to the TMM process when it starts. Options before '--' are used by Tmax system, and those that come after '--' can be used by general users.

The following are the most commonly used options. For more information, refer to Tmax. "3.2.2. _tmax_event_handler".

Size: Up to 255 characters

Command options that can be sent to the TLM process when it starts.

The following is the most commonly used option.

Command options that can be sent to the CLH process when it starts. Detailed options are same as that of TMMOPT.

(W)CLH3017 no such svc : svcname

The following conditions are required to transmit a message to a service designated by CLH.

1. The message must be a response to tpcall() or tpacall().

2. Data must be included whether it is a normal response or an error response.

3. The message may not be transmitted if CLH is terminated abnormally.

The name of the RDP server process, if using one.

There must be one real server per node. Clients can access all services of a node through the RDP server processes. The RSCPC value must be set to the number of channels that are used to receive results from other server processes. The number of CLH and RDP server processes must remain constant and the MIN and MAX values must be the same. In general, the ratio of the number of processes of RDP to CLH should be 2 to 1. The number of RDP processes must always be greater than that of CLH.

Range: 1 ~ 128

Number of channels used by RDP processes to communicate with other servers.

Range: 1 ~ 128

Default: 32

Maximum number of server groups per node.

Range: 1 ~ 128

Default: 32

Maximum number of TMSs that can be invoked on the node.

Range: 1 ~ 128

Default: 2

While MAXGW is the maximum number of gateways that can be used on the node, MAXGWSVR is the maximum number of gateway server processes that can be invoked on the node. The default value is 2, which means that there are two gateways and each gateway can have one gateway server.

Range: 1 ~ 128

Default: 2

Maximum number of RQ server processes that can be invoked on the node. The default value is 2, which means that there are two RQs and each RQ can have one RQ server.

Range: 1 ~ 128

Default: 8

Maximum number of connections between a gateway and a CLH.

CPC with a server is basically composed of a pair of channels (input/output). If CPC and the number of gateways are both set to 2, MAXGWCPC must be set to at least eight connections. The sum of MAXGWCPC must also be less than the value of MAXCPC.

To increase the MAXGWCPC value, the MAXCPC value must also be increased.

Default: Y

Option to automatically activate the backup server in the event of server failure.

Alternate port number for TPORTNO and TPORTNO[2-5]. It is recommended to use TMAXPORT instead of TPORTNO[2-5] when multiple port numbers are used.

Port numbers used by clients to communicate with Tmax system using the data compression function. If no port numbers are set, the data compression function is not used during communication. Note that the port numbers must be those that are managed by Tmax.

If TMAXPORT of the NODE element is specified, use one of the port numbers of TMAXPORT setting. If TMAXPORT is not specified, use the TPORTNO value of the DOMAIN element. Only the clients who are connected to the specified port can use the data compression function.

This feature is not recommended for use in a general network environment since overhead for data compression is greater than network transmission delay. Data compression is usually used when sending a relatively large volume of data under poor network conditions.

For example, data compression can be used when a client is exchanging tens of KB of data via a modem.

The following tables display the amount of time required for transmitting 1.5 MB of data using a 100 Mbps Local LAN (first two tables) and a 56 KB modem (third table). The results show that it is best to use the compression function when transmitting a large volume of text data via a modem.

<Transmitted data size>

Minimum size (Byte) of data that can be compressed when using the compression function.

Must be used in conjunction with COMPRESSPORT.

In the following example, Tmax system uses data compression when communicating with the client through port 9999. In the example, data compression will only be used if the data size is larger than 10000 bytes. If the data size is smaller than this value or the client is connected to port 8888, data will not be compressed.

TMAXPORT = "8888, 9999", CompressPort = "9999" CompressSize = 10000

Default: Y

Option to reload CLH, CAS, CLL, and TLM upon abnormal termination.

Settings for fault tolerance features. For details, refer to "3.8. Fault Tolerance Settings".

Tmax supports EXTERN server functions to use functions of Tmax configuration file from an external process, a non-Tmax process.

The server process TYPE must be set to EXTSVR. Two additional variables, EXTCLHPORT and EXTPORT, must be configured in the NODE element of the configuration file. Configure the listen port number of TMM in EXTPOT.

Listen port number of CLH. If not specified by the system administrator, the system will assign a value.

The following is an example of configuring EXTERN SERVER.

*NODE
tmaxh3        ... 
              EXTPORT = 9000, EXTCLHPORT = 9010

*SERVER
alinkjmapp    SVGNAME = svg1,  
              MIN = 1, MAX = 1, 
              CPC = 10, 
              SVRTYPE=EXTSVR

*SERVICE
JMAPPER       SVRNAME = alinkjmapp, 
              SVCTIME = 30

Default: N

Option to use SysMaster Trace function. If set to Y, trace function will be used. If set to N, the function will not be used. SysMaster Trace function uses the GID of a running service to trace system tasks.

The following describes the GID structure (12 bytes).

Range: 1024~MAX_INT

Default: 50000

Maximum number of SysMaster traces that can be stored per CLH. Only required if SMSUPPORT has been set to Y.

Default: N

Option to use the CLL Block. If set to Y, CLL will be blocked when a client connects to the Tmax system.

If there is a client request when Tmax service is not completely started up, the client is allowed to access the Tmax system if CLL is running, but it will receive the TPENOREADY error.

In order to prevent such problem, the administrator must have the client request blocked in CLL. This will prevent CLL from receiving service requests before the service is ready.

When the service is running and ready, you can cancel the CLL blocking function by using the tmadmin command.

Exception port numbers for the CLL Block. Select among the port numbers set in TMAXPORT.

Even if the CLL Block is set, you can specify exception ports for the CLL Block in the Tmax configuration file, and the clients that connect through these port numbers will not be blocked.

Default: 0

CLL Block time (seconds) for if CLLBLOCK=Y. If the block time expires, the CLL Block is released.

If this value is not set when the block time is less than 0 or CLLBLOCK=Y, the CLL Block must be released using tmadmin.

Default: RR

Method used by CLL to send a client connection to CLH.

If set to CLLCONNLB = LC, Least Connection method is used to make the next connection to the CLL with the least number of connections.

If CLLCONNLB is not set or is substituted by some other value, the default method (Round Robin) is used to make the next connection.

If this value is set in the NODE element, only the client data that were accessed through these ports are encrypted. This allows encryption at the session level which provides a more flexible encryption function.

Note:

A Back-up Node (TRB Node) that can process transaction recovery when a node is temporarily unavailable due to machine failure in a Multi-Node & Multi-Domain environment. This setting can help with the low availability problem caused by pending transactions.

As in the following figure, preconfigure TRB Node A' for the Node A.

When machine A fails, the TRB Node A' will start up and automatically recover any pending transaction.

[Figure 3.3] TRB

TRB item and Primary Node name must be configured in the Back-up Node (TRB Node).

TRB Node must have one server group. Database information and one TMS must be configured in the server group. To use a Gateway, the TRB item must also be configured in the GATEWAY element.

Note:

Startup/Termination of TRB Node

The –B option is used to start up and terminate TRB Node in a way that is different from other nodes.

$tmboot –B <nodename>

TRB is configured as follows:

*NODE
#Primary node
NODE1     TMAXDIR = /data/tmax

#TRB node
NODE2     TMAXDIR = /data2/tmax, TRB = NODE1

Range: 1 ~ 65535

Default: 1024

Maximum number of sessions that can be created in an HMS of the node. For details, refer to "3.7. HMS Settings".

Range: 32768~262143

Key value that refers to the shared memory segment used as Tmax Session Queue repository.

Range: 4 ~ 4000000 (Unit : KB)

Size of the shared memory segment used as SQ repository.

Range: 2 ~ MAX_INT-1 (Unit: KB)

Maximum number of SQs that can be created in the node. Must be an even number. If you use an odd number, it will be automatically decreased by 1.

Default: 1~ MAX_INT

Maximum number of keys that can be saved in each SQ.

Default: 1~MAX_INT (seconds)

Timeout for SQ data keeping. When timeout is reached, data will be automatically deleted.

Size: Up to 255 characters

IP address of the node.

Can be used if the IP address cannot be deduced from the host name when attempting to connect to another node.

SYSTEM_IPV6 must be set to Y to use an IPv6 address.

Used as IP address of an interface to bind to when creating a port for client connection (related setting: CLL_BIND_IP).

Default: N

Option to use the IPv6 protocol when creating a port for client connection.

Default: N

Option to use the IPv6 protocol when creating a port for inter-node communication in a multi-node environment.

Default: N

Option to use the IPv6 protocol when creating a port for connection to the extern server from Tmax system.

Default: Y

Option to use an interface of the IP address set in the NODE element when creating a port for client connection.

Size: Up to 15 characters

Logical name of the server group. Must be unique in the SVRGROUP element.

This variable will be referred to from the SVGNAME item of the SERVER element.

Size: Up to 15 characters

Node where the server group exists. Must be a node that is defined in the NODE element. If the node name includes a hyphen (-), it must be enclosed in double quotation marks.

Refer to DEFAULT of "3.2.2 NODE".

Default: TMAX

Server group type. One of TMAX, RQMGR, EXTSVG, and TXRQMGR. The default value is TMAX. Refer to "Tmax Programming Guide(MultipleRM) for detailed information about MTMAX and STMAX.

EXTSVG is same as EXTERN SERVER. It is selected when using the EXTERN TMS to support WAS(Web Application Server) and 2PC. When using EXTSVG, the OPENINFO item must also be configured. Even though its value is not significant, set the OPENINFO element to " ".

<Example of using EXTSVG Server Group>

*NODE
tmaxh3         ... 
               EXTPORT = 9000, EXTCLHPORT = 9010

*SVRGROUP
Extsvg         NODENAME = "tmaxh4", 
               SVGTYPE = "EXTSVG", TMSNAME = TMSWAS
               OPENINFO = " "

*SERVER
alinkjmapp     SVGNAME = Extsvg1, 
               MIN = 1, MAX = 1, CPC = 20, 
               SVRTYPE = EXTSVR

*SERVICE
JMAPPER        SVRNAME = alinkjmapp, 
               SVCTIME = 30

Size: Up to 8000 characters

Groups that can share a process or perform routing with. These sever groups can exist on a same node or on different nodes. Both COUSIN and BACKUP items can be configured, but they must be set to different server groups.

The server groups set in COUSIN are in charge of routing, and the server groups set in BACKUP handle failures. Gateway servers can also be included in COUSIN.

Size: Up to 8000 characters

Backup group name used to provide seamless services even in case of a failure.

LOAD = -1

Use Static Routing. Client service requests will only be routed to the server groups of the node to which they are connected. If LOAD is set to -1 (for local use only), the domain gateways in COUSIN are not called. Thus, LOAD cannot be set to -1 in the domain gateway's COUSIN.

This option cannot be used for transaction and non-transaction gateways.

[Figure 3.4] If LOAD is Set to -1

LOAD = -2

Configure each gateway in a multi node environment and create a separate client connection to each node. Then, by having the initial node, to which the client connects, process the requested service can help decrease the work load.

When you use the process local first function in a multi-node environment grouped by COUSIN, the server group of the node that the client is connected to will process the request one-on-one. The gateway of the node is used to make a request to another domain.

When you make a call to a certain server group using tpcallsvg and tpacallsvg, the request can be processed by all server groups even if the local node first function(LOAD=-2) is set. Moreover, if the server group is terminated, the service is not replaced by another server. Instead, the user receives a TPENOREADY error.

For a transaction gateway, XA server group must be used as general server group (dummy server group) for configuring COUSIN. LOAD must be set to -2 for all server groups and gateways grouped as COUSIN.

[Figure 3.5] If LOAD is Set to -2

LOAD = -3

Use when a backup is required between server groups in COUSIN.

The BACKUP item of the SVRGROUP element is not a backup configuration for server groups set in COUSIN but a backup configuration for the server group. For instance, let's assume that a server group A is configured with COUSIN="B,C" and BACKUP="D". If all the nodes where A, B, and C belong to are terminated, backup scheduling to D is attempted only when the node that A belongs to fails. LOAD=-3 can be used to failover to server group D when all servers that belong to A, B, and C server groups are disabled.

LOAD = 0

Use dynamic routing for load balancing. Tmax routes client requests to the server group with the required service and the lowest work load at the time.

LOAD > 0

Use Rule based routing. Client service requests are distributed on the target server group based on the specified LOAD ratio. Can only be used for a group specified in COUSIN. Load balancing for Tmax server group can be configured in the LOAD and ROUTING settings, and both must be set to the same value. If the service is configured in the ROUTING element, it can only be scheduled through ROUTING and the LOAD setting is ignored. Thus, it is impossible to perform scheduling by grouping the server groups of each ROUTING in the COUSIN setting.

[Figure 3.6] If LOAD > 0

Size: Up to 63 characters

Unique DB name required if TMS is configured.

Size: Up to 255 characters

Used to connect the server group to the database through TMS. Initializes the DB connection and must be defined by using the syntax required by each DB.

To encrypt some settings, use ***** (5 asterisks) in place of the setting. When using cfl, a password will be prompted for the encrypted parts.

Partial encryption

OPENINFO = "ORACLE_XA+Acc=P/scott/*****+SesTm=100"

Full encryption

OPENINFO = "*****"

Size: Up to 255 characters

Required if TMS, which is a group that integrates to the DB, is configured. Used to terminate the connection with the DB. Must be defined by using the syntax required by each DB. This item is not required by most DBs except for Informix.

Range: 1~16

Default: 2

Number of TMS processes that will be loaded in the server group when Tmax system starts up if TMS is configured.

Range:1~16

Default: 3

Maximum number of active TMS processes in the server group at any one time.

Size: Up to 15 characters

Name of the TMS process that will manage the database for the server group if TMS is configured.

Size: Up to 255 characters

Required to pass values using environment variables to servers that are part of the server group. It can also be used to connect to multiple homogeneous databases on the same node.

Only required for a limited number of DBMSs. For DBMSs like DB2 that only support dynamic registration, set it to 'Dynamic'.

If the DBMS is similar to Sybase or earlier versions of Informix and uses 'int' (not 'long') for the xaswitch field even though it has a 64 bit engine, set it to 'XASWITCH32'.

Range: 1 ~ 128

Number of channels with CLH processes. Used only when SVGTYPE is RQMGR. For details, refer to "3.6. Reliable Queue Settings".

Default: Y

Option to reload TMS processes if they are abnormally terminated.

Options to configure Tmax fault tolerance feature. For details, refer to "3.8. Fault Tolerance Settings".

Default: Y

Option to use transaction recovery. To use transaction recovery, set to Y. Otherwise, set to N.

Tmax v4.0 and later use the Transaction Recovery Function defined in X/OPEN DTP. When TMS is restarted, Tmax obtains an XID list of pending transactions from the RM and performs transaction recovery on the list. Transaction recovery function is supported in a single node or multi node environment, or a multi-domain system that uses Tmax Domain Gateways for communication.

Since recovery is performed when the entire group is started/terminated, recovery can be performed by each TMS group. A command option is added to tmboot/tmdown to start/terminate a specific TMS. For more details about the command, refer to "Tmax Reference Guide".

Permissions to query the Pending list must be given to RM to use this feature.

When different server groups of a single node share a database, dependencies can be established to prevent recovery from executing while TMS is running.

As in the following example, if three server groups share one DB, tms1 can be configured to perform recovery for both tms2 and tms3 to prevent duplicate recovery processing. For stable recovery, recovery is not performed while tm2 or tm3 is running.

svg1 TMSNAME = tms1, 
     TMSDEP = "tms2, tms3" 

svg2 TMSNAME = tms2, 
     TMSRECOVERY = N 

svg3 TMSNAME = tms3, 
     TMSRECOVERY = N

Default: DOMAIN

When server groups on different nodes share a single database, TMSRANGE can be set to NODE so that recovery is performed separately for each node. This will reduce inter-node traffic and increase stability of the recovery function.

Default: STD

TMS type used.

Number of working threads per TMS process.

Location and name of the TMS log file. Usage is same as TMMOPT and CLHOPT.

Unit: seconds

XA processing that takes longer than TMSXATIME is logged.

If XA processing, such as xa_prepare, xa_commit, xa_rollback, etc., takes unusually long time in TMS, this may indicate a DB failure. Logging is provided for cause analysis of the failure in the future.

Log format:

CLH2160 XA function name(xid) processing was delayed (elapsed time) by tms(server group no.) 
       : Return value of XA function
CLH2161 XA function name(xid) processing may be stopped due to closed tms(server group no.)

Default: N

Option to temporarily use a server group name as "__dummy".

If set to Y, TMS of the group is not started up when executing tmboot.

The DUMMY setting of the SVRGROUP element is not inherited by the SERVER element, and so DUMMY must be separately configured for each SERVER element.

Can be useful when a dummy server group setting, such as Domain Gateway Cousin, is needed. For details, refer to "B.2. Domain Gateway COUSIN Setting".

If not set, the following error occurs during tmboot.

(F) BOOT0014 exec error : /data1/tmaxkjh/tmax/appbin/tms_tbr [BOOT0029]: No such file or directory

Size: Up to 15 characters

Name of HMS process. To use HMS, SVGTYPE must be set to HMS.

For more details about the HMS setting, refer to "Tmax HMS User Guide".

Range: 0 ~ 65535

Unique value in the domain that is required for processing an HMS message.

Range: 0 ~ 65535

Number of threads that do not perform storage processing for HMS messages.

Range: 0 ~ 65535

Number of threads that perform storage processing for HMS messages.

Range: 1 ~ 65535

Number of threads that perform batch storage processing for HMS messages.

Range: 2 ~ 65535

Maximum number of HMS messages that can be simultaneously processed for batch storage processing.

Size: Up to 255 characters

Command options transmitted to an HMS process when HMS starts up.

Size: Up to 255 characters

Path and name of a configuration file where Durable Subscriber is defined. The durable subscriber is automatically registered when HMS starts up.

Range: 0 ~ 65535 (unit: hours)

An interval to delete persistent messages that HMS accumulates in a storage.

Port number used by HMS to communicate with other HMSs in a cluster.

Range: 0 ~ 65535 (seconds)

Heartbeat check interval. Clustered HMSs regularly exchange heartbeat messages to detect any network failure among them.

Range: 0 ~ 65535 (unit: milliseconds)

Interval for updating status information required for HMS queue clustering.

Must be set to use a clustered queue. Data is more evenly distributed in the queue if the interval is shorter, but network load is heavier. Set a value accordingly.

Size: Up to 255 characters

Paths of the directories that contain the executable files for server application programs.

APPDIR of the NODE element sets the paths of all applications running on the node, and it can be redefined in the SVRGROUP to set application file paths for the server group.

Size: Up to 255 characters

Paths of the directories where user messages are logged. Can be redefined for each server group. The method of logging user messages is same as ULOGDIR of the NODE element.

Log level of TMS.

Log level of the server.

Size: Up to 15 characters

Server name and also the executable name of the server application program. Each server name must be unique and must be defined only once in the SERVER element of the configuration file. In multi-domain systems, a server name must be unique in each domain.

Size: Up to 15 characters

Server group that the server belongs to.

The server group must already be defined in the SVRGROUP element. By mapping a server to a server group, the node it runs on and the resource manager (database) that it uses are also known. It can also be used to acquire parameters necessary for opening the resource manager.

When operating in XA mode, Tmax manages all database transactions and oversees 2PC processing. To facilitate this, the OPENINFO field must be properly configured in the SVRGROUP element as previously described.

When Tmax is operating in XA mode, users are not involved when the server connects to the database. However, if Tmax is not operating in XA mode, OPENINFO must not be included in the SVRGROUP element and the individual users must manage server-database connections.

Refer to DEFAULT of "3.2.2 NODE".

Size: Up to 255 characters

Command options that are passed to the server processes when they start up. Options that are defined before "--" are used by the system, and options that are defined after "--" can be used by users.

Major system options include:

MACRO Commands1

If the following macro is defined in the SERVER element, the file name is determined at server startup.

MACRO Commands2

If the following macro is defined in the SERVER element, the file name is determined immediately before tpsvrinit(), tpsvrdone() and the service functions are called for TCS server. For UCS server, It is determined immediately before tpsvrinit() and tpsvrdone() are called and tpschedule() is returned.

This MACRO is used to generate different types of log files (e.g., by date).

Example MACRO setting:

CLOPT="-o $(SVR).$(DATE).out –e $(SVR).$(DATE).err"

A log file will be created in the directory specified by the ULOGDIR field of [server name].03312001.out and [server name].03312001.err files. To modify a file name of standard error or standard output, use an absolute path name like '-e /usr/tmax/log/ulog/test.log'. The 'abc' option after '--' is passed to the tpsvrinit() function of the server program as argv[1] for the user.

Range: 1 ~ MAX_INT

Default: 1

Number of server processes that is created when Tmax starts up.

Unlike the conventional client/server model where one server process is loaded per client, Tmax manages and maintains a set number of active server processes. The CLH checks the status of server processes whenever a client issues a request and assigns the request to idle server processes to maximize system performance.

Since this value determines the default number of active server processes, it is important that the administrator sets MIN to a value appropriate for the system. If the number of server processes available for handling frequently requested services is too low, the wait time for processing the request will increase and system performance will decrease. On the other hand, if an excessive number of server processes are idle, system resources will be wasted and the performance of the system will decrease.

Range: 1 ~ MAX_INT

Default: 1

Maximum number of server processes that can be generated on the server.

While MIN specifies the number of server processes that are initially loaded, MAX sets the maximum number of server processes that can be loaded to handle requests. This value includes the total number of active server processes at any one time.

Initially as many server processes as MIN are loaded, and additional processes can be added up to the MAX value if load increases as client requests increase. Since this value has a strong impact on the system performance, care must be taken in setting the value. If a forked server process is used, set MAX to 0.

Default: N

Option to use a conversational communication method.

There are three types of server communication methods in Tmax: synchronous, asynchronous and conversational. Services that use conversational communication must be registered with a server that is set to conversational communication. If set to a value other than “Y”, the server can only facilitate synchronous or asynchronous communication.

If set to CONV=B, the server process becomes the scheduling process. B represents BOTH, which supports both conversational (Y) and non-conversational (N) types. It can also be used for inter-domain communication through the domain gateway.

Range: 1 ~ MAX_INT

Default: 0

Number of accumulated requests in the queue that automatically activates new server processes.

When the number of queued requests exceeds this value, additional server processes will be activated, up to the value set in MAX. If not set, additional server processes are not automatically activated.

If the number of server processes exceeds the MIN value, the status of the server processes are monitored and those that have been idle a given period of time are terminated to reduce the count. This option is not applied to POD type server processes.

Range: 1 ~ MAX_INT

Default: -1

If not set, the default value is used and MAXQCOUNT is not used.

Maximum number of requests that can be queued before additional requests are rejected by the server.

If the number of requests exceeds the value set in MAXQCOUNT, incoming requests will not be placed in the queue and an error message (tperrno = 26) is sent immediately to the requesting client. MAXQCOUNT is useful for systems that process large volumes of service requests at peak times, such as banks and public offices.

Default: STD

The following are the service types processed by the server.

Range: 1 ~ 128

Default: 1

Number of parallel communication channels between a UCS server and a CLH process. In situations where a single channel cannot handle the amount of requests, multiple communication channels can be used in parallel to increase processing speed.

Only applies to UCS, UCS_DYN, CUSTOM_GATEWAY, REALSVR, REALSVR_MT, STD_MT server types.

Required for STD_MT server type.

This value is related to the MAXCLH and MAXTHR items. MAXCLH x CPC value must be equal to or greater than the MAXTHR.

For instance, if MAXCLH is 1 and MAXTHR is 8, CPC must be at least 8, and if MAXCLH is 2 and MAXTHR is 8, CPC must be at least 4. This is because the maximum number of service requests that can be processed simultaneously by one server process is CPC x MAXCLH. Hence, MAXTHR does not need to be greater than CPC x MAXCLH.

If not set, the value, MAXTHR / MAXCLH, is used.

Range: 0 ~ MAXTHR

Default: 0

Minimum number of service threads that are initially created when a STD_MT server process starts. Required for STD_MT server.

If set to 0, a service request from CLH is processed after a thread is created.

Range: 1 ~ 1000

Default: 1

Maximum number of service threads managed by a thread pool of the STD_MT server process. Required for STD_MT server.

Creating and adding threads whenever there is a service request can degrade performance. Create threads only up to the maximum number, which must be set after careful consideration.

A total of MAXTHR + 1 threads including the main thread can be running. The main thread does not handle service requests but manages service threads and requests. If the value is 1, it is recommended to use a general server library since only one service thread is created.

The MAXTHR item is related to the CPC and MAXCLH items. Refer to the 'CPC' item for detailed information.

Range: 0 ~ MAX_INT

Default: 0 (Unit: KB)

Stack size of the service threads in the STD_MT server process. If not set, the default size of the OS is used.

Default: Y

Option to reload the server if it is terminated abnormally.

Fault tolerance related options. For more information, refer to "3.8. Fault Tolerance Settings".

A server name is a name of a server executable file, and TARGET is used to redefine a server process name. Server process refers to a process that executes the server executable file in APPDIR.

By setting TARGET as in the following, tmadmin can check for the actual server name and server process name by using ps.

Example) SVGNAME = svrgrp, MIN =1, MAX = 1, TARGET = svr2

/user1/starbj/tmax/sample/server> tmadmin   
--- Welcome to Tmax Admin (Type "quit" to leave) ---   
$$1 tmaxs1 (tmadm): st -p  
CLH 0:   
---------------------------------------------------
svr_name  svgname  spr_no  status  count  avg  svc
---------------------------------------------------
SVR      svrgrp     32     RDY      0   0 .00  -1  
 
/user2/starbj/tmax/sample/server> ps  
starbj 28702 1 0 11:13:03 pts/8 0:00 cll -b 28700  
starbj 28704 1 0 11:13:03 pts/8 0:00 svr2
 -b 28700 -S SVR -s SVR -d -1  
starbj 28701 1 0 11:13:03 pts/8 0:00 tmm -b 28700  
starbj 28703 1 0 11:13:03 pts/8 0:00 clh -b 28700 

Example configuration file for TARGET.

*SERVER
original        SVGNAME = svg0, 
copied1         SVGNAME = svg0, 
                TARGET  = "original", 
                CLOPT   = "-e $(SVR).err -o $(SVR).out -x ORIGINAL:COPIED1" 
copied2         SVGNAME = svg0, 
                TARGET  = "original", 
                CLOPT   = "-e $(SVR).err -o $(SVR).out -x ORIGINAL:COPIED2" 
*SERVICE 
ORIGINAL        SVRNAME = original 
COPIED1         SVRNAME = copied1 
COPIED2         SVRNAME = copied2 

On copied1 and copied2 servers, server processes are initiated with the name 'original'. In general, a server name and a server process name are the same in Tmax, but if TARGET is used, the names are different. Thus, when Tmax starts up, three processes named 'original' starts.

If a service request is sent to COPIED1 and COPIED 2, the target is changed to original by [-x] option in the CLOPT setting. A client can request a service to COPIED1 and COPIED2 separately. The server and service count are also incremented when checking in tmadmin.

TARGET is useful when checking a separate COUNT for each object that calls the tpcall within the same program.

Default: N

If a service is dynamically added to Tmax system using 'mksvr' while it is running, by default this value is set to 'N'.

The AUTORAN setting in the SERVER element will automatically load the transactions for services that have been added dynamically. This affects the setting of dynamically added service. Please refer to AUTOTRAN of the SERVICE element.

If AUTORAN is defined in both the SERVICE and SERVER elements, the SERVICE element setting is used.

Option B is used for compatibility with AUTOTRAN in old versions (~3.8.8). It works in the same way as Y, except that the transaction operates as a local transaction instead of a global transaction.

Default: FA

Processing order of transactions among server processes when the MIN value of the server is greater than 2.

Setting SCHEDULE

Default: IDLE_DOWN

Lifespan of the following additional server processes (more than MIN value).

Server processes started by POD (when MIN=0)

Server processes started through ASQCOUNT:

Default: N

Temporarily uses "__dummy" as the server name.

If DUMMY is set to Y in the SERVER element, the server process is not started during tmboot.

This item is useful when a dummy server setting, such as Domain Gateway Cousin, is required. For more details, refer to "B.2. Domain Gateway COUSIN Setting".

If not set, the following error occurs during tmboot.

(F) BOOT0014 exec error : /data1/tmaxkjh/tmax/appbin/tbrtest [BOOT0029]: No such file or directory

Default: N

Allow Unregistered Services

When a server is started after a service has been deleted from the configuration but not from the application (not rebuilt), the server is restarted as many times as MAXRSTART.

If set to AUS=Y in the SERVER element, the server is started ignoring this setting.

Default: Y

Only used for a UCS server. Other server types cannot use this option.

If a server process is terminated by tmdown -s, tmdown -s -i, or abnormal shutdown before tprelay() is called and after the UCS server calls tpsavectx() and saves the client information, the caller does not receive a response for the request. If set to 'Y', when the server process is being terminated an error message is sent to the client for requests that have not been handled by tprelay().

If the CTX_EREPLY option is set to default, when the server process is being terminated an error response is sent to the caller for requests that are being processed. This way, other processes on the server are not affected. This setting operates normally when the client information (CTX_T) created by tpsavectx() call is only managed internally in the server process.

But in an environment where multiple processes share the client information through IPC, when a specific process is abnormally terminated an error message may still be sent even if another process can execute tprelay(). Set to 'N' to prevent an error message from being sent.

Default: Y

If set to Y, a server process can concurrently receive requests from multiple CLHs when MINCLH >= 2. A server process handles concurrent service requests on a first-come-first-served basis. After the process completes the current request with the tpreturn call, it processes a request from another CLH.

If set to N, each server process can receive and handle service requests from a CLH. This prevents multiple CLHs from simultaneously requesting services to one server process. Each server process receives requests from an assigned CLH even though overload occurs on other CLHs which is inefficient in terms of performance.

Precautions if set to N

Default: Y

Option to use tmapm for each service.

As a user function, it is recommended not to use tmapm when using a user defined function (tpsetsvctimeout()) to readjust the timeout value.

Default: N

MAC(Master Alive Check) is used to terminate a server when TMM is terminated to prevent it from becoming a zombie process.

If set to MAC=Y, the server is terminated when TMM is terminated abnormally.

Default: N

ROC(Reconnect Outbound CPC) is used to reconnect to CPC when a server is disconnected from CLH.

If set to ROC=Y, reconnection is attempted when CLH is alive and the socket is disconnected.

Default: N

Number of times that the server calls a service before restarting the server.

If this setting triggers a server restart, TMM outputs "sever closed due to MAX_USE_COUNT or MEM_USAGE(CLOPT:-m)."

Log level of the server.

Path name of the directory where user messages are stored.

Users like Tmax application programmers can easily log messages for debugging and error and warning messages by using the userlog() function. The messages that are outputted by the userlog() function are stored in the ULOGDIR directory. Files called 'ulog_date' are created in the directory, and the messages are logged by date in each file.

If not set, messages are stored in the log/ulog directory under the directory specified in TMAXDIR. The system must exist in the specified path.

Size: Up to 15 characters

Function name (service routine name) of the service. The name must be unique within the SERVICE element. In a multi-domain environment, the Service Name must be unique across all domains in the system.

Size: Up to 15 characters

Name of the server that processes the service.

Use the executable file name of the server program that contains the service routine. The server must be defined in the SERVER element. If the server name includes any special characters, this must be wrapped in double quotes as in "Server Process Name". If multiple services are provided by a single server process, the same server process name can appear in the SERVICE element multiple times.

Refer to DEFAULT of "3.2.2 NODE".

File name of the module when the service name is different from the module name.

Range: 1 ~ 100

Default: 50

Priority level of the service. A service with a higher number is given higher priority. For example, a service with a PRIO value of 100 has the highest priority.

Data routing setting. For more information, refer to "3.5. Load Balancing Settings".

Range: 1 ~ MAX_INT

Default: 0

If not set, services wait indefinitely (default value).

Max time allowed for service processing. The service processing must be finished within this time period.

If the processing time exceeds SVCTIME, the server process will stop processing the service and will send an error message to the client. While BLOCKTIME in the DOMAIN element starts from the time that the client issues the requests (e.g., through tpcall() or tpacall()), SVCTIME starts from the time when the CLH passes the request to the server process and the server process begins to process the service. If SVCTIME is set to 0, time limit is infinite.

If service times out while using a database session, the next service cannot be processed since the session cannot be forced to disconnect from the database. Tmax resolves such problem in the following way. If service timeout occurs, a user can restart the server by using TPRETURN(TPEXIT, ...) in the tpsvctimeout routine which causes the MAXRSTART count in the SERVER element to be incremented.

It is also recommended to use the XA mode on Windows for application development. To use non-XA mode, set a sufficient number for RESTART count. SVCTIME only includes the processing time and excludes the time in the queue.

Default: Y

Option to process service requests from another domain in a multi domain environment.

If EXPORT is set to N and a request is sent from another domain via a gateway, a TPESECURITY error occurs.

Default: N

Option to automatically launch a global transaction for a service request that is received while in non-transaction mode.

It is different from AUTOTRAN in versions earlier than Tmax 3.8.8.

In the old versions, AUTOTRAN is set to Y by default, and a local transaction begins without executing tx_begin. In the new version, Tmax 3.8.8, AUTOTRAN is set to N by default, and a transaction is not automatically started.

If not set, tx_begin must be used to start a transaction. If set to Y, when the current service requests for another service they are grouped into a global transaction. If set to Y, and tx_begin is used to start a transaction, the TPESVCERR error occurs.

The following is an example of the configuration file and the tmadmin tool:

< config.m>

*SVRGROUP
svg2            NODENAME = "tmax1"

### tms for Oracle ###
svg3            NODENAME = "tmax1",    DBNAME = ORACLE,
                OPENINFO = "Oracle_XA+Acc=P/scott/tiger+SesTm=60",
                TMSNAME  = tms_ora

*SERVER
svr2            SVGNAME = svg2
### servers for Oracle sample program###
fdltest         SVGNAME = svg3

*SERVICE
### services for fdltest ###
FDLINS          SVRNAME = fdltest
FDLSEL          SVRNAME = fdltest
FDLUPT          SVRNAME = fdltest
FDLDEL          SVRNAME = fdltest
TOUPPER         SVRNAME = svr2

<tmadmin>

tmax@tmax1;/user/tmax/tmax/config>tmadmin
--- Welcome to Tmax Admin (Type "quit" to leave) --- 

$$1 tmax1 (tmadm): cfg -s
   ------------------------------------------------- 
   svc_name    prio(pr)  autotran  svctime(st)  routno  svrname      svgno 
   ------------------------------------------------- 
FDLINS          50      YES         0           -1      fdltest         2 
FDLUPT          50      YES         0           -1      fdltest         2 
FDLSEL          50      YES         0           -1      fdltest         2 
FDLDEL          50      YES         0           -1      fdltest         2
TOUPPER         50       NO         0           -1        svr2          0

$$2 tmax1 (tmadm): q
ADM quit for node (tmax1)

Range: 1 ~ MAX_INT

Default: 0

For more information, refer to TXTIME of "3.2.1 DOMAIN".

Size: Up to 15 characters

Logical name of the gateway. Must be unique within the GATEWAY element.

Size: Up to 255 characters

Name of the node where the gateway process runs on.

Node name must already be defined in the NODE element.

'listen' port number used by local gateway processes. Enables a remote gateway process to connect to a local gateway process. The port number must be configured in the RGWPORTNO item. Port numbers less than 1024 must not be used since they are reserved for system use.

Since port numbers are essential for communication between gateways, make sure not to use the port numbers that are used by other system functions.

Size: Up to 255 characters

Name or IP address of the node where the remote gateway process, to which the local gateway wants to connect, is running on.

If using the node name, it must be registered in the "/etc/hosts" file. The local gateway processes use both the IP address in RGWADDR and the port number in RGWPORTNO to connect to a remote gateway.

Port number through which the local gateway process connects to a remote gateway process.

This number is a listen port number of a remote gateway process. Port numbers less than 1024 must not be used since they are reserved for system use.

Type of the remote gateway to which the local gateway process is to connect.

Currently, the following nine types are supported.

Refer to DEFAULT of "3.2.2 NODE".

Range: 1 ~ 128

Default: 32

Number of channels that can be used concurrently in the gateway.

This is only used when TMAXNONTX is configured in GWTYPE. If not set, the gateway will only connect to and receive service requests from up to 32 domains.

Default: N

Option to use the IPv6 protocol when local gateway listens.

Default: N

Option to use the IPv6 protocol when a local gateway uses IPV6 to connect to a remote gateway.

Range: 1 ~ 128

Default: 1

Number of parallel communication channels between CLH process of both domains when using the gateway.

In situations where high volumes of data need to be passed through the gateway process, parallel communication using multiple channels can speed up the processing. For gateways set as TMAXNONTX, it is sufficient to set CPC to 2 or 3.

Range: 1 ~ MAX_INT

Default: 0

If not set, services wait indefinitely (default value).

Max time allowed for inter-domain tpcall / tpacall.

If set to the default value, time limit is infinite. If the delay is less than 30 seconds, TIMEOUT does not occur since the gateway checks timeout every 30 seconds which is different from a server.

Size: Up to 800 characters

Names of related gateways when a service needs to be distributed through multiple gateways (load balancing). Can use the names of gateways on the same node or different nodes. Both COUSIN and BACKUP items can be configured, but they must be set to different gateways.

The gateways set in COUSIN are in charge of routing, and the gateways set in BACKUP handle failures.

Can only use the gateway names registered in the GATEWAY element. Server groups cannot be used.

The following is an example of the environment file.

< config.m>

*SERVICE
GWSVC1       SVRNAME = gw1
GWSVC2       SVRNAME = gw1
GWSVC3       SVRNAME = gw1

*GATEWAY
gw1         NODENAME = node1, GWTYPE = TMAXNONTX, PORTNO = 10001,
            RGWADDR = "192.168.1.1",   RGWPORTNO = 10002,
            LOAD = 1, CPC = 5, CLOPT = "-i",
            COUSIN = "gw1"

gw2         NODENAME = node2, GWTYPE = TMAXNONTX, PORTNO = 10001,
            RGWADDR = "192.168.1.2",   RGWPORTNO = 10002,
            LOAD = 1, CPC = 5, CLOPT = "-i"

The gateways, gw1 and gw2, are set to COUSIN. If there is a GWSVC1 service request, it is distributed and processed according to the LOAD configuration.

Refer to "3.2.3. SVRGROUP Element" LOAD.

Size: Up to 7999 characters

Name of the backup gateway to be used in the event of main gateway failure. It must be a gateway set in the GATEWAY element of the Tmax configuration file.

Size: Up to 255 characters

IP address or node name of the backup remote gateway.

If using a node name, it must also be defined in the /etc/hosts file.

Port number of the backup remote gateway.

Used for listening. Port numbers less than 1024 must not be used since they are reserved for system use.

BACKUP, BACKUP_RGWADDR, and BACKUP_RGWPORTNO settings are all used for fault tolerance purposes.

Default: N

Option to use the IPv6 protocol when local gateway connects to a backup remote gateway.

Used when connecting to a remote gateway whose address is BACKUP_RGWADDR and the port is BACKUP_RGWPORTNO.

Size: Up to 255 characters

Used only for Tuxedo gateways (GWTYPE = TUXEDO | TUXEDO_ASYNC). Up to three backup Tuxedo gateways can be assigned by using BACKUP_RGWADDR2-BACKUP_RGWPORTNO2 and BACKUP_RGWPORT3-BACKUP_RGWPORTNO3.

Default: N

Option to use the IPv6 protocol to connect to second or third backup remote gateway. Used only for Tuxedo gateways (GWTYPE = TUXEDO | TUXEDO_ASYNC)

Used when connecting to a remote gateway whose address and port number are BACKUP_RGWADDR2 and BACKUP_RGWPORTNO2 or BACKUP_RGWADDR3 and BACKUP_RGWPORTNO3 respectively.

Default: BIDIR

Only used when GWTYPE is set to TMAX or TMAXNONTX.

Options are:

Size: Up to 255 characters

Gateway command options that can be sent to the gateway when it starts up. Options preceding '--' are used by the system, and options following '--' can be used by the users.

Common user options are:

Note

Default: Y

Option to restart the gateway if it is terminated abnormally.

These two items are used to configure fault tolerance. Refer to "3.8. Fault Tolerance Settings".

Range: 1 ~ MAX_INT/2

Default: -1

Maximum time that GW keeps the pending list. The maximum time must be less than or equal to the value of TIMEOUT divided by 2. (TIMEOUT >=PTIMEOUT)

Retransmit the pending transaction (prepare-done/commit/rollback) from the Domain Gateway to prevent any pending transactions even when a decision has not been received.

When Gateway1 and Gateway2 are connected through 2pc, if Gateway2 receives a prepare() and then does not receive a commit() or rollback(), Gateway2 will send prepare-done() to Gateway1. In this case, the PTIMEOUT and PTIMEINT settings can be used to prevent having a pending transaction when a decision is not received due to a network problem (or abnormal termination) between Gateways.

Range: 1 ~ MAX_INT/2

Default: -1

Specify the time interval for GW to retransmit a pending transaction (prepare-done, commit, rollback).

Maximum time must be less than or equal to the value of PTIMEOUT.

Range: -1 ~ MAX_INT

Default: -1

Interval for making a re-connect attempt to the main sever for failback while the gateway is connected to a backup.

Range: -1 ~ MAX_INT

Default: -1

Maximum waiting time for GW to connect to a remote gateway.

Range: -1 ~ MAX_INT

Default: -1

Interval for sending a message to check if the channel that connects a gateway to a remote gateway is available.

For details, refer to TRB of "3.2.2 NODE".

Size: Up to 63 characters

Name of the database to be used by the server group. Use standard DBMS names, such as ORACLE, INFORMIX, and SYBASE.

Size: Up to 255 characters

Database information related to server group.

If using the same resource manager, e.g., database, using one server group for management can raise resource efficiency and enable distributed transaction processing. The OPENINFO syntax for this item will vary depending on the resource manager vendor (Oracle, Informix, etc.). If you are unsure of what to enter, refer to the relevant database manual.

Size: Up to 255 characters

Default: NULL

Must include information required by the resource manager to close the database of the server group.

This item is usually omitted or set to NULL. Informix database requires that CLOSEINFO be set to a null character ("") and it may be omitted for Oracle databases.

<Server group that uses Oracle>

*SVRGROUP
svg1           NODENAME = tmax1,
               DBNAME = ORACLE,
               OPENINFO = "Oracle_XA+Acc = P/scott/tiger+SesTm= 60"

<Server group that uses Informix>

*SVRGROUP
svg2           NODENAME = tmax2,
               DBNAME = INFORMIX,
               OPENINFO = "test",
               CLOSEINFO = ''''

Size: Up to 15 characters

Name of the TMS that will manage database transactions for the server group.

If OPENINFO for the database has been configured, the TMSNAME setting must also be set. The TMS process is required for managing the database of the server group from the systems that use the database. In order for Tmax to manage a database, the database open/close information must be configured, and TMSNAME must be defined for each server group to start the TMS process. The TMS process is a transaction manager that processes transactions through XA linked to the database. It is created by linking the libtms.a file in the $TMAXDIR/lib directory to the database library of the DBNAME defined in the SVRGROUP element.

Range: 1 ~ 32

Default: 2

Number of TMS processes that can be started for the server group. This setting will influence transaction processing management more than database management. The default value is two TMS processes per server group, which is enough for only database management.

Range: 1 ~ 32

Default: 3

Number of additional TMS processes that can be loaded dynamically as needed. For more information, refer to "3.4. Distributed Transaction Settings".

Default: Y

Setting for the two-phase commit process (2PC) for transaction processing. 2PC transactions consist of two phases, preparation and commit/rollback. Tmax supports 2PC for distributed transaction implementation.

[Figure 3.9] Two-Phase Commit

There are two modes for 2PC. In the complete return mode, the service returns after completing all transactions using 2PC protocol and logging all rollbacks. In the return immediately mode, for faster processing the service returns immediately after issuing a commit/rollback command in Phase 2. Either mode can be selected according to the system environment.

CMTRET determines whether to do a complete return.

Range: 0 ~ MAX_INT

Default: 0

Timeout value for processing transactions, starting from when a transaction begins with tx_begin() to when the transaction ends with tx_commit() or tx_rollback(). If tx_commit() or tx_rollback() is not complete when the timeout expires, the transaction will be automatically rolled back.

Size: Up to 255 characters

Default: TMAXDIR

To handle a mid-transaction failure, all distributed transactions are logged. TLOGDIR is used to specify the file where transaction information will be stored.

This log file will be saved as an ordinary file since Tmax does not support raw files for logs. The general file defined in TLOGDIR with any name will only be used for logging transactions in situations with light transaction processing workload. To shorten response times in situations with heavy transaction processing workload, the specified raw file is used to set the device name to TLOGDIR without going through the UNIX file system. The device will only be used for logging.

Name of the database used by the server group.

Information required for opening the database.

Default: NULL

Information required for closing the database.

Name of the TMS process.

Minimum number of TMS processes for the server group.

This item is not important in terms of database management. However, it is very important in transaction processing. If there are too few TMS processes (one or two) when the transaction workload is high, commit/rollback procedures will generate a large resource overhead. Thus it is important to specify an appropriate number of TMSs to effectively manage resources and throughput.

Fault tolerance setting. For details, refer to "3.8. Fault Tolerance Settings".

Within the SVRGROUP element, this setting refers to the MAXRSTART of a TMS or RQ.

Maximum number of additional TMS processes that can be loaded within the server group. As the transaction workload increases, additional TMS processes can be loaded up to MAXTMS.

Size: Up to 255 characters

This is an item that specifies the name of server group when the Load Balancing function is used on 2 or more nodes.

Name of each server group must be registered in the SVRGROUP element. The Load Balancing function of Tmax system is processed in server group units. But when Load Balancing is processed across multiple nodes, each node name must be specified differently in each server group. The node name must already be configured in the NODE element of configuration file. When defining multiple server groups, separate them by using a comma(,). The required server program must exist in the APPDIR of each node. To use performance-based load balancing, a load value must be set for each server group.

Range: -1 ~ MAX_INT

Default: 1

Represents the server group's service performance level. A higher value represents higher performance level.

Set to a high value for a high-performing node and a low value for a low-performing node.

For system-based load balancing, set the COUSIN item to the server group and set an appropriate LOAD value for each server group. To disable distributed processing on a node, LOAD must be set to –1. If LOAD is not set or set to 0, load balancing will be automatically performed based on the system load.

Size: Up to 255 characters

DDR can only be used when multiple server groups provide the same service. Hence, the original SVRGROUP needs to be copied to a different node according to the COUSIN setting. For more information, refer to the description of COUSIN in "3.2.3. SVRGROUP".

Size: Up to 15 characters

Logical routing name.

This must be unique within the SERVICE element, and will be used in the ROUTING element, which needs to be defined for a service to use DDR (this is explained next). Additionally, the typed buffer name, and field used for routing and its value range must be set with the server group name.

Size: Up to 15 characters

Routing name defined in ROUTING item of the SERVICE element. Because DDR is implemented for each service, the ROUTING name must match the one defined in the SERVICE element.

Size: Up to 2000 characters

This item is configured according to the field data type, STRUCT,FIELD, STRING, or CARRAY type, of the services that use data dependent routing.

The following is the configuration syntax for each of the four data types.

Starting position of the input data that will be used for DDR. Both offset and length must be set to use only part of the data specified by the offset and length for DDR. For instance, STRING/0/4 represents offset of “0” and length of “4”. If the input data was “Tmaxsoft”, only "Tmax" which is 4 bytes in length starting from offset position of 0, will be used for comparison in performing DDR. This item is only used with CARRAY / STRING type DDR.

Offset and length must only be used for the range type DDR and not for the matching type because the offset and length items specify only the input data and are not related to the data specified in RANGES.

In the following example, only “abcd", 4 bytes counted from the offset (0), of the string “aaaaaaa” will be used as comparison criteria for DDR. Since “aaaa” (the comparison criteria for the input data) and “aaaaaaa” (specified in RANGES) do not match, the data will be processed by SVG2. Note that, in this case, all incoming data will be processed unconditionally by SVG2.

ROUTE_1   FIELD = "STRING/0/4",
          RANGES = "aaaaaaa":SVG1, *:SVG2

Offset and length are available when using the range type DDR.

In the following example (Range type DDR), if the input data is "aaaaaaa", only the "aaaa" part of the input data (4 bytes counted from the offset 0) will be used as comparison criteria for DDR. Since the string, "aaaa" is smaller than the range specified in SVG1, it will be processed by SVG2. If the input data is "bbbb", it falls in the range specified by SVG1, and hence it will be processed by SVG1.

ROUTE_1    FIELD = "STRING/0/4",
           RANGES = "aaaaaaa"-"ccccccc":SVG1, *:SVG2

As explained already, offset specifies the starting point of the element within the input data used for DDR criteria, and length specifies the length of element from the offset. Length must be used in conjunction with offset.

If length is not specified, Tmax will use the input data from offset until NULL, or the end of the data.

Size: Up to 2047 characters

Destination of client service requests based on the value of the field specified in FIELD.

Values specified must conform to the following.

When not using the MATCH option

*ROUTING
rout1    FIELD = "STRING, CARRAY, FIELD/INPUT,
         STRUCT/test/a",
         RANGES = "min-'100':svg1,'101'-'200':svg1_2,min-'100':svg1_3,*:svg1_4"

If the configuration file is configured like the previous example, in the previous version, a server in svg1 is first inspected to process data within the range min-100. If a node containing svg1 is not running normally, its backup is examined, and if backup is not running or does not exist, an error is sent to the client. However, in the new version, Tmax will operate in the following way by specifying MATCH in the ROUTING element.

When using MATCH

*ROUTING
rout1    FIELD = "STRING, CARRAY, FIELD/INPUT, 
         STRUCT/test/a",
         RANGES ="min-'100':svg1,'101'-'200':svg1_2, min-'100':svg1_3,*:svg1_4" , 
         MATCH=MULTIPLE

If the configuration file is configured as in the previous example, a server in svg1 is first inspected to process data within the range min-100. If a node containing svg1 is not running normally, its backup is examined. If the backup is not running or does not exist, Tmax searches for a next matching server group within the RANGES, and svg1_3 will be selected. If a node containing svg1_3 is alive, but if the server group svg1_3 is also not running, its backup is examined. If the backup does not exist or is not running normally, a next matching server group, svg1_4, will be selected.

Set "MATCH=FIRST" to use the function of the previous Tmax version.

Size: Up to 15 characters

Node where the RQ server group exists. Must be defined in the NODE element.

Range: 1 ~ 128

Default: 4

Number of parallel communication channels between the RQS and CLH processes. If the queue is used often, using multiple channels can increase the overall performance.

Server group type. To use RQ, this must be set to RQMGR.

Size: Up to 15 characters

Logical name of the RQ. Multiple RQs can be defined within the RQ section, but each RQ must have a unique name. Multiple RQs can be defined for a single node.

Size: Up to 15 characters

The server group that uses the RQ and must be a RQ SVRGROUP name in the SVRGROUP.

Range: 1 ~ 2047

Default: 16 (Unit: MB)

RQ space for storing enqueued data.

Since the RQ size will not automatically increase to account for increased demand, the size set here must be enough to avoid overflow errors. If there is no available RQ space to save a new RQ file, the file will be automatically deleted when Tmax shuts down.

Size: Up to 255 characters

Default: $(TMAXDIR)/path/rq-name

Location of the file used by the RQ. The full directory and file names must be specified to create the required RQ data file.

Default: COLD

How to treat service requests stored in the RQ after a reboot.

Refer to the following table.

Default: Y

Option to call fsync after saving to the disk.

Refer to the following table.

Default: Y

Option to cache RQ files in memory.

If set to N, RQ processing is slower but the amount of memory required is smaller.

Range: 1 ~ 65535

Default: 1024

Maximum number of sessions that can be created in HMS within the domain.

Range: 1 ~ 65535

Default: 1024

Maximum number of sessions that can be created in HMS within the node.

Size: Up to 15 characters

Logical name of the server group. Must be unique in the SVRGROUP element.

Size: Up to 15 characters

Node where the server group exists. Must be a node that is defined in the NODE element.

Server group type. To use HMS, this must be set to HMS.

Size: Up to 15 characters

Name of a user-built HMS process.

By default, HMS uses a database for storage. Specify OPENINFO of the DBMS to which HMS will connect.

Initialize the database connection, and define OPENINFO using the syntax required by the database.

Range: 0 ~ 65535

Used for processing an HMS message. Must be unique within the domain.

Range: 0 ~ 65535

Number of threads that will not be used for storage processing.

Must be set to a large value when sending and receiving only nonpersistent messages. For basic use of HMS, this item must be set above a certain value.

Range: 0 ~ 65535

Number of threads used for storage processing.

Must be set to a large value when there are many persistent messages. For basic use of HMS, this item must be set above a certain value.

Size: Up to 255 characters

Path and name of the configuration file where a durable subscriber is defined.

In the file, set LABEL to [TopicName], and then use a colon(:) as a delimiter to specify ClientName:Listener:Selector (in this order) on each line.

The durable subscriber defined in the configuration file is automatically registered when HMS is started.

Range: 0 ~ 65535 (Unit: hour)

HMS stores persistent messages in the storage. Specify a time interval for regularly deleting the accumulated messages from the storage.

Topic messages received by all subscribers and consumed queue messages are removed from the storage after the specified time. Messages that a durable subscriber has not received are not deleted from the storage even after the specified time.

For clustering multiple HMSs, each HMS creates a channel to communicate with others. Specify a port number to be used for this purpose.

If there is a destination that is set to GLOBAL, HMS will listen through the specified port number. Make sure to specify a port number that is not already in use.

Range: 0 ~ 65535 (seconds)

For clustering multiple HMSs, heartbeat messages are regularly exchanged to detect a network failure between HMSs. Specify a time interval to transmit this message. If a heartbeat message does not come from another HMS within the specified interval, it is regarded as a sign of failure and the connection to the failed HMS is terminated.

Range: 0 ~ 65535 (Unit: milliseconds)

To exchange messages between GLOBAL queues, state information of each queue must also be exchanged. Specify a time interval to exchange this information.

This item must be set to use clustered queues. The shorter the interval, more evenly the data will be distributed to the queues, but heavier network loads will be incurred. Thus, an appropriate value must be set.

Size: Up to 15 characters

Destination name. A producer and consumer are created using this name.

Size: Up to 15 characters

HMS server group name of the destination. The SVGNAME must be an HMS server group name specified in the SVRGROUP element.

Destination type. Currently, only TOPIC and QUEUE are supported.

Default: COLD

How to treat service requests stored in the RQ after a reboot.

Like in RQ, set either to WARM or COLD.

Refer to the following table.

Default: N

Option to use destinations with the same name in clustered HMSs logically as a single destination.

For GLOBAL setting, HMSPORT must be set in HMS of the SVRGROUP element.

For GLOBAL queue setting, the following items are also required.

Range: 0 ~ 65535

Coefficient to determine the buffer size maintained by a clustered queue.

A buffer size is the number of consumers X GQTHR value. If the length of the current queue is shorter than the buffer size, the queue takes messages from another clustered queue. If the queue contains less messages than the buffer size, it does not distribute its messages upon a distribution request.

Range: 1 ~ 65535

Number of messages that can be requested at once to another queue. If the specified value is large, the queue will be filled with messages very quickly. If the value is too large, there may be unnecessary message transmissions between nodes. Thus, an appropriate value must be set.

Range: 0 ~ 65535

Number of messages that can be accumulated in a queue. If more than the specified number of messages is accumulated in a queue, any subsequent messages are evenly distributed to the clustered queues on another node.

Size: Up to 255 characters

Backup server group.

Multiple backup server groups can be specified by using a comma (,) as a separator. If the original server group fails, the first specified backup will be activated, and if this backup server group also fails, the next server group will be activated, and so on.

All of the specified backup server groups must be registered within the SVGROUP element. The backup server groups must be configured to comply with the settings of the original server group. To change the configuration options, relevant items can be redefined for each server group or server. However, service settings must be the same for the original and backup server groups. These rules are the same as those for duplicate server groups created for load balancing.

Unlike the duplicate server groups created for load balancing, backup nodes are not initially active when Tmax is launched. Instead, they are activated to replace the original server group when the original server group fails.

Default: Y

Option to restart CLH, CAS, CLL, and TLM upon abnormal termination.

If set to Y, the server processes will be restarted when they shut down unexpectedly due to a system failure.

Range: -1 ~ MAX_INT

Default: -1

Maximum number of times a CHL or CAS process can be restarted.

MAXRSTART is used in combination with the RESTART and GPERIOD items. If RESTART is set to 'Y', a process will be restarted MAXRSTART times during the GPERIOD. If set to 'N', Tmax will ignore this setting. If set to '-1', there will be no limit to the number of times that a process can be restarted.

Range: 1 ~ MAX_INT

Default: 86400 (24 hours) (Unit: second)

Time period for which MAXRSTART is valid.

If a server process is unexpectedly terminated, it will be restarted MAXRSTART times during the GPERIOD. Once the GPERIOD expires, this procedure will be repeated.

If the SVGTYPE is RQMGR, these items configure the restart setting of the RQ. If the SVGTYPE is XA svg, they configure the restart setting of the TMS.

If a server process is abnormally terminated, the process automatically restarts to provide services.

For more information of each item, refer to the previous "NODE" section.

These items configure the restart setting of the server processes.

If a gateway process is abnormally terminated, the process automatically restarts to provide services.

For more information of each item, refer to the previous "NODE" section.

These items configure the restart setting of the gateway.

If the gateway process is abnormally terminated, the process automatically restarts to provide services.

For more information of each item, refer to the previous "NODE" section.

Default: "NO_SECURITY"

Refer to the following table.

Utilities such as mkgrp, mkpw and mkacl must be used to generate each of these files. Both ACL and MANDATORY are set to level 3 security. The following describes their differences.

Size: Up to 15 characters

Password for the dompwd of TPSTART_T. This password will be used to log into the Tmax system. If no password or an invalid one is provided, users will not be able to access the Tmax system. The user and password defined here can be created with the mkpw utility.

After configuring the security, a password file must be created with the mkpw utility. This file includes passwords of each user, and is encrypted. For Tmax security system to work properly, the password file must be created before configuring the Tmax system.

For DOMAIN_SEC, if the specified password is different from the dompwd value of the TPSTART_T structure in tpstart(), the client program cannot make a service request. The tperrno is set to TPESECURITY(25). For USER_AUTH, the usrpwd and usrname fields in the TPSTART_T struct must be the same as the account and password registered in the password file created with the mkpw utility.

Since level 2 (user authentication) security includes level 1 (system access control), the specified dompwd must be the same as the set domain password.

IPv4

192.168.1.43
192.168.1.48

IPv6

fe80::213:77ff:fe4d:c57d
fe80::213:77ff:fe4d:c570

IPv4

192.168.1.35
192.168.1.45

IPv6

fe80::213:77ff:fe4d:c50d
fe80::213:77ff:fe4d:c500

Search the tmax.allow file first, and if there are clients that are in the ACL, allow them to access the system.

Search the tmax.deny file, and if there are clients that are in the ACL, deny them from accessing the system. If they attempt to access the system (TPSTART), the TPECLOSE error occurs.

Allow access to clients who exist in neither tmax.allow nor tmax.deny files.

If the first character is '#', it is processed as a comment.

Allow only IP or NETWORK/NETMASK type.

Example)

(IPv4)(IP_V4)192.168.1.1 or 192.168.1.0/24, 
(IPv6)fe80::213:77ff:fe4d:c57d or fe80::213:77ff:fe4d:c57d/64

One ACCESS CONTROL LIST is allowed per line, and a blank or tab character is not allowed.

ALL is a reserved word meaning all IPs.