17.2.285. MPI_Publish_name

MPI_Publish_name - Publishes a service name associated with a port

17.2.285.1. SYNTAX

17.2.285.1.1. C Syntax

#include <mpi.h>

int MPI_Publish_name(const char *service_name, MPI_Info info,
     const char *port_name)

17.2.285.1.2. Fortran Syntax

USE MPI
! or the older form: INCLUDE 'mpif.h'
MPI_PUBLISH_NAME(SERVICE_NAME, INFO, PORT_NAME, IERROR)
     CHARACTER*(*)   SERVICE_NAME, PORT_NAME
     INTEGER         INFO, IERROR

17.2.285.1.3. Fortran 2008 Syntax

USE mpi_f08
MPI_Publish_name(service_name, info, port_name, ierror)
     TYPE(MPI_Info), INTENT(IN) :: info
     CHARACTER(LEN=*), INTENT(IN) :: service_name, port_name
     INTEGER, OPTIONAL, INTENT(OUT) :: ierror

17.2.285.2. INPUT PARAMETERS

  • service_name: A service name (string).

  • info: Options to the name service functions (handle).

  • port_name: A port name (string).

17.2.285.3. OUTPUT PARAMETER

  • IERROR: Fortran only: Error status (integer).

17.2.285.4. DESCRIPTION

This routine publishes the pair (service_name, port_name) so that an application may retrieve port_name by calling MPI_Lookup_name with service_name as an argument. It is an error to publish the same service_name twice, or to use a port_name argument that was not previously opened by the calling process via a call to MPI_Open_port.

17.2.285.5. INFO ARGUMENTS

The following keys for info are recognized:

Key                   Type      Description
---                   ----      -----------

ompi_global_scope     bool      If set to true, publish the name in
                                the global scope.  Publish in the local
                                scope otherwise.  See the NAME SCOPE
                                section for more details.

ompi_unique           bool      If set to true, return an error if the
                                specified service_name already exists.
                                Default to overwriting any pre-existing
                                value.

bool info keys are actually strings but are evaluated as follows: if the string value is a number, it is converted to an integer and cast to a boolean (meaning that zero integers are false and non-zero values are true). If the string value is (case-insensitive) “yes” or “true”, the boolean is true. If the string value is (case-insensitive) “no” or “false”, the boolean is false. All other string values are unrecognized, and therefore false.

If no info key is provided, the function will first check to see if a global server has been specified and is available. If so, then the publish function will default to global scope first, followed by local. Otherwise, the data will default to publish with local scope.

17.2.285.6. NAME SCOPE

Open MPI supports two name scopes: global and local. Local scope will place the specified service/port pair in a data store located on the mpirun of the calling process’ job. Thus, data published with local scope will only be accessible to processes in jobs spawned by that mpirun - e.g., processes in the calling process’ job, or in jobs spawned via MPI_Comm_spawn.

Global scope places the specified service/port pair in a data store located on a central server that is accessible to all jobs running in the cluster or environment. Thus, data published with global scope can be accessed by multiple mpiruns and used for MPI_Comm_connect and MPI_Comm_accept between jobs.

Note that global scope operations require both the presence of the central server and that the calling process be able to communicate to that server. MPI_Publish_name will return an error if global scope is specified and a global server is either not specified or cannot be found.

Open MPI provides a server called ompi-server to support global scope operations. Please refer to its manual page for a more detailed description of data store/lookup operations.

As an example of the impact of these scoping rules, consider the case where a job has been started with mpirun - call this job “job1”. A process in job1 creates and publishes a service/port pair using a local scope. Open MPI will store this data in the data store within mpirun.

A process in job1 (perhaps the same as did the publish, or perhaps some other process in the job) subsequently calls MPI_Comm_spawn to start another job (call it “job2”) under this mpirun. Since the two jobs share a common mpirun, both jobs have access to local scope data. Hence, a process in job2 can perform an MPI_Lookup_name with a local scope to retrieve the information.

However, assume another user starts a job using mpirun - call this job “job3”. Because the service/port data published by job1 specified local scope, processes in job3 cannot access that data. In contrast, if the data had been published using global scope, then any process in job3 could access the data, provided that mpirun was given knowledge of how to contact the central server and the process could establish communication with it.

17.2.285.7. ERRORS

Almost all MPI routines return an error value; C routines as the return result of the function and Fortran routines in the last argument.

Before the error value is returned, the current MPI error handler associated with the communication object (e.g., communicator, window, file) is called. If no communication object is associated with the MPI call, then the call is considered attached to MPI_COMM_SELF and will call the associated MPI error handler. When MPI_COMM_SELF is not initialized (i.e., before MPI_INIT / MPI_INIT_THREAD, after MPI_FINALIZE, or when using the Sessions Model exclusively) the error raises the initial error handler. The initial error handler can be changed by calling MPI_COMM_SET_ERRHANDLER on MPI_COMM_SELF when using the World model, or the mpi_initial_errhandler CLI argument to mpiexec or info key to MPI_COMM_SPAWN[_MULTIPLE]. If no other appropriate error handler has been set, then the MPI_ERRORS_RETURN error handler is called for MPI I/O functions and the MPI_ERRORS_ABORT error handler is called for all other MPI functions.

In the sessions model, the error handler can be set during MPI_Session_init.

Open MPI includes three predefined error handlers that can be used:

MPI_ERRORS_ARE_FATAL: Causes the program to abort all connected MPI processes.
MPI_ERRORS_ABORT:     An error handler that can be invoked on a communicator,
                      window, file, or session. When called on a communicator, it
                      acts as if MPI_ABORT was called on that communicator. If
                      called on a window or file, acts as if MPI_ABORT was called
                      on a communicator containing the group of processes in the
                      corresponding window or file. If called on a session,
                      aborts only the local process.
MPI_ERRORS_RETURN:    Returns an error code to the application.

MPI applications can also implement their own error handlers.

Custom MPI error handlers can be created by calling: MPI_Comm_create_errhandler(3) MPI_File_create_errhandler(3) MPI_Session_create_errhandler(3) MPI_Win_create_errhandler(3)

Predefined and custom error handlers can be set by calling: MPI_Comm_set_errhandler(3) MPI_File_set_errhandler(3) MPI_Session_set_errhandler(3) MPI_Win_set_errhandler(3)

Note that MPI does not guarantee that an MPI program can continue past an error.

See the MPI man page for a full list of MPI error codes.

See the Error Handling section of the MPI-3 standard for more information.