int main(int argc, char *argv[])
{
char diagnosis[DRMAA_ERROR_STRING_BUFFER];
const char *all_jobids[NBULKS*JOB_CHUNK + JOB_CHUNK+1];
char jobid[100];
int drmaa_errno, i, pos = 0;
const char *job_path;
drmaa_job_template_t *jt;
if (argc<2) {
fprintf(stderr, "usage: example <path-to-job>\n");
return 1;
}
job_path = argv[1];
if (drmaa_init(NULL, diagnosis, sizeof(diagnosis)-1) != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_init() failed: %s\n", diagnosis);
return 1;
}
/*
* submit some bulk jobs
*/
if (!(jt = create_job_template(job_path, 5, 1))) {
fprintf(stderr, "create_job_template() failed\n");
return 1;
}
for (i=0; i<NBULKS; i++) {
drmaa_job_ids_t *jobids;
int j;
while ((drmaa_errno=drmaa_run_bulk_jobs(&jobids, jt, 1, JOB_CHUNK, 1, diagnosis,
sizeof(diagnosis)-1))==DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE) {
fprintf(stderr, "drmaa_run_bulk_jobs() failed - retry: %s %s\n", diagnosis, drmaa_strerror(drmaa_errno));
sleep(1);
}
if (drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_run_bulk_jobs() failed: %s %s\n", diagnosis, drmaa_strerror(drmaa_errno));
return 1;
}
printf("submitted bulk job with jobids:\n");
for (j=0; j<JOB_CHUNK; j++) {
drmaa_get_next_job_id(jobids, jobid, sizeof(jobid)-1);
all_jobids[pos++] = strdup(jobid);
printf("\t \"%s\"\n", jobid);
}
drmaa_release_job_ids(jobids);
}
drmaa_delete_job_template(jt, NULL, 0);
/*
* submit some sequential jobs
*/
if (!(jt = create_job_template(job_path, 5, 0))) {
fprintf(stderr, "create_sleeper_job_template() failed\n");
return 1;
}
for (i=0; i<JOB_CHUNK; i++) {
while ((drmaa_errno=drmaa_run_job(jobid, sizeof(jobid)-1, jt, diagnosis,
sizeof(diagnosis)-1)) == DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE) {
fprintf(stderr, "drmaa_run_job() failed - retry: %s\n", diagnosis);
sleep(1);
}
if (drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_run_job() failed: %s\n", diagnosis);
return 1;
}
printf("\t \"%s\"\n", jobid);
all_jobids[pos++] = strdup(jobid);
}
/* set string array end mark */
all_jobids[pos] = NULL;
drmaa_delete_job_template(jt, NULL, 0);
/*
* synchronize with all jobs
*/
drmaa_errno = drmaa_synchronize(all_jobids, DRMAA_TIMEOUT_WAIT_FOREVER, 0, diagnosis, sizeof(diagnosis)-1);
if (drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_synchronize(DRMAA_JOB_IDS_SESSION_ALL, dispose) failed: %s\n", diagnosis);
return 1;
}
printf("synchronized with all jobs\n");
/*
* wait all those jobs
*/
for (pos=0; pos<NBULKS*JOB_CHUNK + JOB_CHUNK; pos++) {
int stat;
int aborted, exited, exit_status, signaled;
drmaa_errno = drmaa_wait(all_jobids[pos], jobid, sizeof(jobid)-1,
&stat, DRMAA_TIMEOUT_WAIT_FOREVER, NULL, diagnosis, sizeof(diagnosis)-1);
if (drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_wait(%s) failed: %s\n", all_jobids[pos], diagnosis);
return 1;
}
/*
* report how job finished
*/
drmaa_wifaborted(&aborted, stat, NULL, 0);
if (aborted)
printf("job \"%s\" never ran\n", all_jobids[pos]);
else {
drmaa_wifexited(&exited, stat, NULL, 0);
if (exited) {
drmaa_wexitstatus(&exit_status, stat, NULL, 0);
printf("job \"%s\" finished regularly with exit status %d\n",
all_jobids[pos], exit_status);
} else {
drmaa_wifsignaled(&signaled, stat, NULL, 0);
if (signaled) {
char termsig[DRMAA_SIGNAL_BUFFER+1];
drmaa_wtermsig(termsig, DRMAA_SIGNAL_BUFFER, stat, NULL, 0);
printf("job \"%s\" finished due to signal %s\n",
all_jobids[pos], termsig);
} else
printf("job \"%s\" finished with unclear conditions\n",
all_jobids[pos]);
}
}
}
if (drmaa_exit(diagnosis, sizeof(diagnosis)-1) != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_exit() failed: %s\n", diagnosis);
return 1;
}
return 0;
}
int main(int argc, char **argv)
{
drmaa_job_template_t *jt_a = NULL;
drmaa_job_template_t *jt_b = NULL;
drmaa_job_ids_t *jobids_a = NULL;
drmaa_job_ids_t *jobids_b = NULL;
const char *base_name = "Job";
char name_a[512], name_b[512];
struct chunking_t {
int a;
int b;
} chunking[] = {
{ 1, 1 },
{ 1, 3 },
{ 1, 4 },
{ 1, 12 },
{ 3, 1 },
{ 3, 3 },
{ 3, 4 },
{ 3, 12 },
{ 4, 1 },
{ 4, 3 },
{ 4, 4 },
{ 4, 12 },
{ 12, 1 },
{ 12, 3 },
{ 12, 4 },
{ 12, 12 }
};
int samples, i, drmaa_errno = 0;
while (argc > 2) {
if (!strcmp(argv[1], "-ps")) {
fprintf(stderr, "use drmaa_job_ps() for progress monitoring\n");
do_ps = 1;
argc--;
argv++;
}
if (!strcmp(argv[1], "-1st")) {
fprintf(stderr, "first sample only\n");
do_1st = 1;
argc--;
argv++;
}
if (!strcmp(argv[1], "-exit_unknown")) {
fprintf(stderr, "first sample only\n");
do_exit = 0;
argc--;
argv++;
}
if (!strcmp(argv[1], "-nm")) {
argc--;
argv++;
base_name = argv[1];
argc--;
argv++;
fprintf(stderr, "use \"%s\" as base name\n", base_name);
}
}
if (argc != 2) {
printf("Usage: %s path_to_sleeper_script\n", argv[0]);
exit(1);
}
if (strstr(argv[1], "sleeper.sh") == NULL) {
printf("Usage: %s path_to_sleeper_script\n", argv[0]);
exit(1);
}
if (do_1st)
samples = 1;
else
samples = sizeof(chunking)/sizeof(void *);
while ((drmaa_errno=drmaa_init(NULL, errorbuf, sizeof(errorbuf)-1) == DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE))
sleep(1);
if (drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_init failed: %s\n", errorbuf);
exit(EXIT_FAILURE);
}
strcpy(name_a, base_name);
strcat(name_a, "A");
strcpy(name_b, base_name);
strcat(name_b, "B");
for (i=0; i< samples; i++) {
jt_a = create_job_template(argv[1], name_a, NULL, chunking[i].a, 1);
jt_b = create_job_template(argv[1], name_b, name_a, chunking[i].b, 0);
while ((drmaa_errno = drmaa_run_bulk_jobs(&jobids_a, jt_a, 1, BULK_SIZE, chunking[i].a,
errorbuf, sizeof(errorbuf)-1))==DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE)
sleep(1);
if(drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_run_bulk_jobs(%d-%d:%d) failed: %s\n", 1, BULK_SIZE, chunking[i].a, errorbuf);
return 1;
}
while ((drmaa_errno = drmaa_run_bulk_jobs(&jobids_b, jt_b, 1, BULK_SIZE, chunking[i].b,
errorbuf, sizeof(errorbuf)-1))==DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE)
sleep(1);
if(drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_run_bulk_jobs(%d-%d:%d) failed: %s\n", 1, BULK_SIZE, chunking[i].b, errorbuf);
return 1;
}
/* sleeper will sleep 60s by default, so this test should be okay */
validate_jobs(jobids_a, chunking[i].a, jobids_b, chunking[i].b);
drmaa_release_job_ids(jobids_a);
drmaa_release_job_ids(jobids_b);
drmaa_errno = drmaa_delete_job_template(jt_a, errorbuf, sizeof(errorbuf)-1);
if(drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_delete_job_template failed: %s\n", errorbuf);
return 1;
}
drmaa_errno = drmaa_delete_job_template(jt_b, errorbuf, sizeof(errorbuf)-1);
if(drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_delete_job_template failed: %s\n", errorbuf);
return 1;
}
}
while ((drmaa_errno = drmaa_exit(errorbuf, sizeof(errorbuf)-1)) == DRMAA_ERRNO_DRM_COMMUNICATION_FAILURE)
sleep(1);
if(drmaa_errno != DRMAA_ERRNO_SUCCESS) {
fprintf(stderr, "drmaa_exit failed: %s\n", errorbuf);
return 1;
}
printf ("OK\n");
return 0;
}
int launch (const char* script, const char* arg_vec[], int num_tasks) {
char error[DRMAA_ERROR_STRING_BUFFER];
int errnum = 0;
drmaa_job_template_t *jt = NULL;
errnum = drmaa_init (NULL, error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not initialize the DRMAA library: %s\n", error);
return 1;
}
errnum = drmaa_allocate_job_template (&jt, error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not create job template: %s\n", error);
}
else {
errnum = drmaa_set_attribute (jt, DRMAA_REMOTE_COMMAND, script,
error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not set attribute \"%s\": %s\n",
DRMAA_REMOTE_COMMAND, error);
}
else {
errnum = drmaa_set_vector_attribute (jt, DRMAA_V_ARGV, arg_vec, error,
DRMAA_ERROR_STRING_BUFFER);
}
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not set attribute \"%s\": %s\n",
DRMAA_REMOTE_COMMAND, error);
}else {
drmaa_job_ids_t *ids = NULL;
errnum = drmaa_run_bulk_jobs (&ids, jt, 1, num_tasks, 1, error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not submit job: %s\n", error);
}else {
char jobid[DRMAA_JOBNAME_BUFFER];
const char *jobids[2] = {DRMAA_JOB_IDS_SESSION_ALL, NULL};
while (drmaa_get_next_job_id (ids, jobid, DRMAA_JOBNAME_BUFFER) == DRMAA_ERRNO_SUCCESS) {
printf ("A job task has been submitted with id %s\n", jobid);
}
errnum = drmaa_synchronize (jobids, DRMAA_TIMEOUT_WAIT_FOREVER,
1, error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not wait for jobs: %s\n", error);
}else {
printf ("All job tasks have finished.\n");
}
}
drmaa_release_job_ids (ids);
} /* else */
errnum = drmaa_delete_job_template (jt, error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not delete job template: %s\n", error);
}
} /* else */
errnum = drmaa_exit (error, DRMAA_ERROR_STRING_BUFFER);
if (errnum != DRMAA_ERRNO_SUCCESS) {
fprintf (stderr, "Could not shut down the DRMAA library: %s\n", error);
return 1;
}
return 0;
}
