static void _process_terminated(List job_list, char *f[], int lc,
int show_full, int len)
{
filetxt_job_rec_t *job = NULL;
filetxt_job_rec_t *temp = NULL;
_parse_line(f, (void **)&temp, len);
if (temp == NULL) {
error("Unknown proccess terminated");
return;
}
job = _find_job_record(job_list, temp->header, JOB_TERMINATED);
if (!job) { /* fake it for now */
job = _create_filetxt_job_rec(temp->header);
job->jobname = xstrdup("(unknown)");
debug("Note: JOB_TERMINATED record for job "
"%u preceded "
"other job records at line %d\n",
temp->header.jobnum, lc);
} else if (job->job_terminated_seen) {
if (temp->status == JOB_NODE_FAIL) {
/* multiple node failures - extra TERMINATED records */
debug("Note: Duplicate JOB_TERMINATED "
"record (nf) for job %u at "
"line %d\n",
temp->header.jobnum, lc);
/* JOB_TERMINATED/NF records may be preceded
* by a JOB_TERMINATED/CA record; NF is much
* more interesting.
*/
job->status = temp->status;
goto finished;
}
fprintf(stderr,
"Conflicting JOB_TERMINATED record (%s) for "
"job %u at line %d -- ignoring it\n",
job_state_string(temp->status),
job->header.jobnum, lc);
goto finished;
}
job->job_terminated_seen = 1;
job->elapsed = temp->elapsed;
job->end = temp->header.timestamp;
job->status = temp->status;
job->requid = temp->requid;
job->exitcode = temp->exitcode;
if (list_count(job->steps) > 1)
job->track_steps = 1;
job->show_full = show_full;
finished:
_destroy_filetxt_job_rec(temp);
}
static void _process_suspend(List job_list, char *f[], int lc,
int show_full, int len)
{
filetxt_job_rec_t *job = NULL;
filetxt_job_rec_t *temp = NULL;
_parse_line(f, (void **)&temp, len);
job = _find_job_record(job_list, temp->header, JOB_SUSPEND);
if (!job) { /* fake it for now */
job = _create_filetxt_job_rec(temp->header);
job->jobname = xstrdup("(unknown)");
}
job->show_full = show_full;
if (job->status == JOB_SUSPENDED)
job->elapsed -= temp->elapsed;
//job->header.timestamp = temp->header.timestamp;
job->status = temp->status;
_destroy_filetxt_job_rec(temp);
}
static void _process_step(List job_list, char *f[], int lc,
int show_full, int len)
{
filetxt_job_rec_t *job = NULL;
filetxt_step_rec_t *step = NULL;
filetxt_step_rec_t *temp = NULL;
_parse_line(f, (void **)&temp, len);
job = _find_job_record(job_list, temp->header, JOB_STEP);
if (temp->stepnum == -2) {
_destroy_filetxt_step_rec(temp);
return;
}
if (!job) { /* fake it for now */
job = _create_filetxt_job_rec(temp->header);
job->jobname = xstrdup("(unknown)");
debug2("Note: JOB_STEP record %u.%u preceded "
"JOB_START record at line %d\n",
temp->header.jobnum, temp->stepnum, lc);
}
job->show_full = show_full;
if ((step = _find_step_record(job, temp->stepnum))) {
if (temp->status == JOB_RUNNING) {
_destroy_filetxt_step_rec(temp);
return;/* if "R" record preceded by F or CD; unusual */
}
if (step->status != JOB_RUNNING) { /* if not JOB_RUNNING */
fprintf(stderr,
"Conflicting JOB_STEP record for "
"jobstep %u.%u at line %d "
"-- ignoring it\n",
step->header.jobnum,
step->stepnum, lc);
_destroy_filetxt_step_rec(temp);
return;
}
step->status = temp->status;
step->exitcode = temp->exitcode;
step->ntasks = temp->ntasks;
step->ncpus = temp->ncpus;
step->elapsed = temp->elapsed;
step->tot_cpu_sec = temp->tot_cpu_sec;
step->tot_cpu_usec = temp->tot_cpu_usec;
job->requid = temp->requid;
step->requid = temp->requid;
memcpy(&step->rusage, &temp->rusage, sizeof(struct rusage));
memcpy(&step->stats, &temp->stats, sizeof(slurmdb_stats_t));
xfree(step->stepname);
step->stepname = xstrdup(temp->stepname);
step->end = temp->header.timestamp;
_destroy_filetxt_step_rec(temp);
goto got_step;
}
step = temp;
temp = NULL;
list_append(job->steps, step);
if (!job->track_steps) {
/* If we don't have track_steps we want to see
if we have multiple steps. If we only have
1 step check the job name against the step
name in most all cases it will be
different. If it is different print out
the step separate.
*/
if (list_count(job->steps) > 1)
job->track_steps = 1;
else if (step && step->stepname && job->jobname) {
if (strcmp(step->stepname, job->jobname))
job->track_steps = 1;
}
}
if (job->header.timestamp == 0)
job->header.timestamp = step->header.timestamp;
job->job_step_seen = 1;
job->ntasks += step->ntasks;
if (!job->nodes || !strcmp(job->nodes, "(unknown)")) {
xfree(job->nodes);
job->nodes = xstrdup(step->nodes);
}
got_step:
if (job->job_terminated_seen == 0) { /* If the job is still running,
this is the most recent
status */
if ( job->exitcode == 0 )
job->exitcode = step->exitcode;
job->status = JOB_RUNNING;
job->elapsed = step->header.timestamp - job->header.timestamp;
}
}
static int _parse_line(char *f[], void **data, int len)
{
int i = atoi(f[F_RECTYPE]);
filetxt_job_rec_t **job = (filetxt_job_rec_t **)data;
filetxt_step_rec_t **step = (filetxt_step_rec_t **)data;
filetxt_header_t header;
_parse_header(f, &header);
switch(i) {
case JOB_START:
*job = _create_filetxt_job_rec(header);
(*job)->jobname = xstrdup(f[F_JOBNAME]);
(*job)->track_steps = atoi(f[F_TRACK_STEPS]);
(*job)->priority = atoi(f[F_PRIORITY]);
(*job)->ncpus = atoi(f[F_NCPUS]);
(*job)->nodes = xstrdup(f[F_NODES]);
for (i=0; (*job)->nodes[i]; i++) { /* discard trailing <CR> */
if (isspace((*job)->nodes[i]))
(*job)->nodes[i] = '\0';
}
if (!strcmp((*job)->nodes, "(null)")) {
xfree((*job)->nodes);
(*job)->nodes = xstrdup("(unknown)");
}
if (len > F_JOB_ACCOUNT) {
(*job)->account = xstrdup(f[F_JOB_ACCOUNT]);
for (i=0; (*job)->account[i]; i++) {
/* discard trailing <CR> */
if (isspace((*job)->account[i]))
(*job)->account[i] = '\0';
}
}
break;
case JOB_STEP:
*step = _create_filetxt_step_rec(header);
(*step)->stepnum = atoi(f[F_JOBSTEP]);
(*step)->status = atoi(f[F_STATUS]);
(*step)->exitcode = atoi(f[F_EXITCODE]);
(*step)->ntasks = atoi(f[F_NTASKS]);
(*step)->ncpus = atoi(f[F_STEPNCPUS]);
(*step)->elapsed = atoi(f[F_ELAPSED]);
(*step)->tot_cpu_sec = atoi(f[F_CPU_SEC]);
(*step)->tot_cpu_usec = atoi(f[F_CPU_USEC]);
(*step)->rusage.ru_utime.tv_sec = atoi(f[F_USER_SEC]);
(*step)->rusage.ru_utime.tv_usec = atoi(f[F_USER_USEC]);
(*step)->rusage.ru_stime.tv_sec = atoi(f[F_SYS_SEC]);
(*step)->rusage.ru_stime.tv_usec = atoi(f[F_SYS_USEC]);
(*step)->rusage.ru_maxrss = atoi(f[F_RSS]);
(*step)->rusage.ru_ixrss = atoi(f[F_IXRSS]);
(*step)->rusage.ru_idrss = atoi(f[F_IDRSS]);
(*step)->rusage.ru_isrss = atoi(f[F_ISRSS]);
(*step)->rusage.ru_minflt = atoi(f[F_MINFLT]);
(*step)->rusage.ru_majflt = atoi(f[F_MAJFLT]);
(*step)->rusage.ru_nswap = atoi(f[F_NSWAP]);
(*step)->rusage.ru_inblock = atoi(f[F_INBLOCKS]);
(*step)->rusage.ru_oublock = atoi(f[F_OUBLOCKS]);
(*step)->rusage.ru_msgsnd = atoi(f[F_MSGSND]);
(*step)->rusage.ru_msgrcv = atoi(f[F_MSGRCV]);
(*step)->rusage.ru_nsignals = atoi(f[F_NSIGNALS]);
(*step)->rusage.ru_nvcsw = atoi(f[F_NVCSW]);
(*step)->rusage.ru_nivcsw = atoi(f[F_NIVCSW]);
(*step)->stats.vsize_max = atoi(f[F_MAX_VSIZE]);
if (len > F_STEPNODES) {
(*step)->stats.vsize_max_taskid =
atoi(f[F_MAX_VSIZE_TASK]);
(*step)->stats.vsize_ave = atof(f[F_AVE_VSIZE]);
(*step)->stats.rss_max = atoi(f[F_MAX_RSS]);
(*step)->stats.rss_max_taskid =
atoi(f[F_MAX_RSS_TASK]);
(*step)->stats.rss_ave = atof(f[F_AVE_RSS]);
(*step)->stats.pages_max = atoi(f[F_MAX_PAGES]);
(*step)->stats.pages_max_taskid =
atoi(f[F_MAX_PAGES_TASK]);
(*step)->stats.pages_ave = atof(f[F_AVE_PAGES]);
(*step)->stats.cpu_min = atoi(f[F_MIN_CPU]);
(*step)->stats.cpu_min_taskid =
atoi(f[F_MIN_CPU_TASK]);
(*step)->stats.cpu_ave = atof(f[F_AVE_CPU]);
(*step)->stepname = xstrdup(f[F_STEPNAME]);
(*step)->nodes = xstrdup(f[F_STEPNODES]);
} else {
(*step)->stats.vsize_max_taskid = (uint16_t)NO_VAL;
(*step)->stats.vsize_ave = (float)NO_VAL;
(*step)->stats.rss_max = NO_VAL;
(*step)->stats.rss_max_taskid = (uint16_t)NO_VAL;
(*step)->stats.rss_ave = (float)NO_VAL;
(*step)->stats.pages_max = NO_VAL;
(*step)->stats.pages_max_taskid = (uint16_t)NO_VAL;
(*step)->stats.pages_ave = (float)NO_VAL;
(*step)->stats.cpu_min = NO_VAL;
(*step)->stats.cpu_min_taskid = (uint16_t)NO_VAL;
(*step)->stats.cpu_ave = (float)NO_VAL;
(*step)->stepname = NULL;
(*step)->nodes = NULL;
}
if (len > F_MIN_CPU_NODE) {
(*step)->stats.vsize_max_nodeid =
atoi(f[F_MAX_VSIZE_NODE]);
(*step)->stats.rss_max_nodeid =
atoi(f[F_MAX_RSS_NODE]);
(*step)->stats.pages_max_nodeid =
atoi(f[F_MAX_PAGES_NODE]);
(*step)->stats.cpu_min_nodeid =
atoi(f[F_MIN_CPU_NODE]);
} else {
(*step)->stats.vsize_max_nodeid = NO_VAL;
(*step)->stats.rss_max_nodeid = NO_VAL;
(*step)->stats.pages_max_nodeid = NO_VAL;
(*step)->stats.cpu_min_nodeid = NO_VAL;
}
if (len > F_STEP_ACCOUNT)
(*step)->account = xstrdup(f[F_STEP_ACCOUNT]);
if (len > F_STEP_REQUID)
(*step)->requid = atoi(f[F_STEP_REQUID]);
break;
case JOB_SUSPEND:
case JOB_TERMINATED:
*job = _create_filetxt_job_rec(header);
(*job)->elapsed = atoi(f[F_TOT_ELAPSED]);
(*job)->status = atoi(f[F_STATUS]);
if (len > F_JOB_REQUID)
(*job)->requid = atoi(f[F_JOB_REQUID]);
if (len > F_JOB_EXITCODE)
(*job)->exitcode = atoi(f[F_JOB_EXITCODE]);
break;
default:
error("UNKNOWN TYPE %d",i);
break;
}
return SLURM_SUCCESS;
}
