Fix mpi_abort, minimize error output.
This commit was SVN r26266.
Этот коммит содержится в:
родитель
d3dfba3872
Коммит
5d14fa7546
@ -347,12 +347,6 @@ static void report_sync(int status, orte_process_name_t* sender,
|
||||
{
|
||||
/* flag as complete */
|
||||
sync_recvd = true;
|
||||
|
||||
/* (not really necessary, but good practice) */
|
||||
orte_proc_info_finalize();
|
||||
|
||||
/* Now Exit */
|
||||
exit(status);
|
||||
}
|
||||
|
||||
void orte_ess_base_app_abort(int status, bool report)
|
||||
@ -386,14 +380,13 @@ void orte_ess_base_app_abort(int status, bool report)
|
||||
* process exiting
|
||||
*/
|
||||
sync_recvd = false;
|
||||
if (ORTE_SUCCESS == orte_rml.recv_buffer_nb(ORTE_NAME_WILDCARD, ORTE_RML_TAG_ABORT,
|
||||
ORTE_RML_NON_PERSISTENT, report_sync, NULL)) {
|
||||
return;
|
||||
if (ORTE_SUCCESS != orte_rml.recv_buffer_nb(ORTE_NAME_WILDCARD, ORTE_RML_TAG_ABORT,
|
||||
ORTE_RML_NON_PERSISTENT, report_sync, NULL)) {
|
||||
exit(status);
|
||||
}
|
||||
while (!sync_recvd) {
|
||||
opal_progress();
|
||||
}
|
||||
return;
|
||||
}
|
||||
|
||||
/* - Clean out the global structures
|
||||
|
@ -377,10 +377,6 @@ static void dump_aborted_procs(void)
|
||||
orte_show_help("help-orterun.txt", "orterun:proc-sensor-exceeded", true);
|
||||
break;
|
||||
}
|
||||
} else if (ORTE_JOB_STATE_CALLED_ABORT == job->state) {
|
||||
orte_show_help("help-orterun.txt", "orterun:proc-called-abort", true,
|
||||
orte_basename,
|
||||
(0 == strncmp("orte", orte_basename, 4)) ? "orte" : "MPI");
|
||||
} else if (ORTE_JOB_STATE_HEARTBEAT_FAILED == job->state) {
|
||||
orte_show_help("help-orterun.txt", "orterun:proc-heartbeat-failed", true,
|
||||
orte_basename, ORTE_NAME_PRINT(&proc->name), node->name);
|
||||
|
@ -607,10 +607,6 @@ failed to show the required activity.
|
||||
One or more processes have exceeded a specified sensor limit, but
|
||||
no further info is available.
|
||||
#
|
||||
[orterun:proc-called-abort]
|
||||
%s detected that one or more processes called %s_abort, thus causing
|
||||
the job to be terminated.
|
||||
#
|
||||
[orterun:proc-heartbeat-failed]
|
||||
%s failed to receive scheduled heartbeat communications from a remote
|
||||
process:
|
||||
|
Загрузка…
Ссылка в новой задаче
Block a user