Unix and Linux Discussions Tagged with hang |
|
Thread / Thread Starter |
Last Post |
Replies |
Views |
Forum |
|
|
|
7 |
3,243 |
Solaris |
|
|
|
5 |
7,028 |
Solaris |
|
|
|
0 |
3,149 |
Shell Programming and Scripting |
|
|
|
6 |
11,188 |
Programming |
|
|
|
0 |
5,334 |
AIX |
|
|
|
1 |
4,549 |
AIX |
|
|
|
1 |
6,419 |
UNIX for Dummies Questions & Answers |
|
|
|
0 |
1,184 |
Solaris BigAdmin RSS |
|
|
|
3 |
14,063 |
AIX |
|
|
|
4 |
2,386 |
UNIX for Dummies Questions & Answers |
|
|
|
0 |
4,230 |
AIX |
|
|
|
14 |
7,838 |
UNIX for Dummies Questions & Answers |
|
|
|
2 |
5,044 |
Programming |
|
|
|
1 |
5,631 |
UNIX for Advanced & Expert Users |
|
|
|
4 |
4,587 |
UNIX for Dummies Questions & Answers |
|
|
|
2 |
2,433 |
UNIX for Dummies Questions & Answers |
|
|
|
2 |
4,621 |
UNIX for Dummies Questions & Answers |
|
|
|
2 |
2,414 |
UNIX for Dummies Questions & Answers |
|
|
|
9 |
5,413 |
UNIX for Dummies Questions & Answers |
|
|
|
3 |
3,668 |
UNIX for Dummies Questions & Answers |
|
|
|
1 |
5,383 |
UNIX for Dummies Questions & Answers |
MPI_Abort(3) LAM/MPI MPI_Abort(3)
NAME
MPI_Abort - Terminates MPI execution environment
SYNOPSIS
#include <mpi.h>
int MPI_Abort(MPI_Comm comm, int errcode)
INPUT PARAMETERS
comm - communicator of tasks to abort
errcode
- error code to return to invoking environment
NOTES
Makes a "best attempt" to terminate all MPI processes associated with the communicator comm ; for non-IMPI jobs, this will terminate all
processes. MPI_Abort will hang, however, if the LAM daemon on the node on which a rank was running has died. For example, if a node gets
powered off in the middle of an MPI run and MPI_Abort is invoked, it may hang trying to contact the LAM daemon on the downed node.
In the case of an IMPI job, in addition to terminating all local IMPI procs, this function will also send an abort message to the local
IMPI host, which will, in turn send "closing" messages to all the other IMPI hosts. Their behavior upon receipt of these "closing" mes-
sages is implementation dependant.
If the LAM IMPI host receives a premature "closing" message from a remote IMPI host, it will attempt to continue processing. Any new mes-
sages to that host will likely hang, however.
'ERRCODE' IS SPLIT INTO THREE SECTIONS
lower 8 bits: error class
- . next 8 bits: function
upper 16 bits: error code
- If the error class is nonzero, the error code is the POSIX return value for the program (i.e., exit(error_code) is eventually
called to terminate the program). If the error class is zero, errcode is returned as the POSIX return value for the program.
This is mainly because MPI_Abort is used internally in LAM/MPI to abort MPI upon error; LAM packs all three pieces of data into the
POSIX return code. Users who wish to return a specific value from the
PROGRAM SHOULD USE SOMETHING SIMILAR TO
return ((return_code << 16) + 1);
NOTES FOR FORTRAN
All MPI routines in Fortran (except for MPI_WTIME and MPI_WTICK ) have an additional argument ierr at the end of the argument list. ierr
is an integer and has the same meaning as the return value of the routine in C. In Fortran, MPI routines are subroutines, and are invoked
with the call statement.
All MPI objects (e.g., MPI_Datatype , MPI_Comm ) are of type INTEGER in Fortran.
MORE INFORMATION
For more information, please see the official MPI Forum web site, which contains the text of both the MPI-1 and MPI-2 standards. These
documents contain detailed information about each MPI function (most of which is not duplicated in these man pages).
http://www.mpi-forum.org/
ACKNOWLEDGEMENTS
The LAM Team would like the thank the MPICH Team for the handy program to generate man pages ("doctext" from ftp://ftp.mcs.anl.gov/pub/sow-
ing/sowing.tar.gz ), the initial formatting, and some initial text for most of the MPI-1 man pages.
LOCATION
abort.c
LAM
/MPI 6.5.8 11/10/2002 MPI_Abort(3)