static int
rmJobBufFilesPid(struct jobCard *jp)
{
static char fname[] = "rmJobBufFilesPid()";
int pid;
if ((pid = fork()) < 0) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_M, fname,
lsb_jobid2str(jp->jobSpecs.jobId), "fork");
return (pid);
}
if (pid)
return (pid);
closeBatchSocket();
putEnv(LS_EXEC_T, "END");
if (postJobSetup(jp) < 0) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S,
fname,
lsb_jobid2str(jp->jobSpecs.jobId),
"postSetupUser");
exit(-1);
}
rmJobBufFiles(jp);
exit(0);
}
static
void prtSignaled (int signalValue, LS_LONG_INT jobId)
{
char *op;
switch (signalValue) {
case SIGCHK:
op = (_i18n_msg_get(ls_catd,NL_SETN,473, "checkpointed")); /* catgets 473 */
break;
case SIGSTOP:
op = (_i18n_msg_get(ls_catd,NL_SETN,474, "stopped")); /* catgets 474 */
break;
case SIGCONT:
op = (_i18n_msg_get(ls_catd,NL_SETN,475, "resumed")); /* catgets 475 */
break;
case SIGKILL:
case SIGFORCE:
op = (_i18n_msg_get(ls_catd,NL_SETN,476, "terminated")); /* catgets 476 */
break;
case SIGDEL:
op = (_i18n_msg_get(ls_catd,NL_SETN,477, "deleted")); /* catgets 477 */
break;
default:
op = (_i18n_msg_get(ls_catd,NL_SETN,478, "signaled")); /* catgets 478 */
break;
}
if (signalValue == SIGDEL && runCount != 0)
printf((_i18n_msg_get(ls_catd,NL_SETN,479, "Job <%s> will be deleted after running next %d times\n")), /* catgets 479 */
lsb_jobid2str(jobId), runCount);
else
printf((_i18n_msg_get(ls_catd,NL_SETN,480, "Job <%s> is being %s\n")), lsb_jobid2str(jobId), op); /* catgets 480 */
}
static int
xdr_thresholds(XDR *xdrs, struct jobSpecs *jobSpecs)
{
static char fname[] = "xdr_thresholds";
int i, j;
if (xdrs->x_op == XDR_DECODE) {
jobSpecs->thresholds.loadSched = NULL;
jobSpecs->thresholds.loadStop = NULL;
}
if (xdrs->x_op == XDR_FREE) {
for(i=0; i < jobSpecs->thresholds.nThresholds; i++) {
FREEUP(jobSpecs->thresholds.loadSched[i]);
FREEUP(jobSpecs->thresholds.loadStop[i]);
}
FREEUP(jobSpecs->thresholds.loadSched);
FREEUP(jobSpecs->thresholds.loadStop);
return(TRUE);
}
if (!(xdr_int(xdrs, &jobSpecs->thresholds.nIdx) &&
xdr_int(xdrs, &jobSpecs->thresholds.nThresholds))) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(jobSpecs->jobId),
"xdr_int", "nIdx/nThresholds");
return(FALSE);
}
if (xdrs->x_op == XDR_DECODE) {
jobSpecs->thresholds.loadSched = (float **)
my_calloc (jobSpecs->thresholds.nThresholds, sizeof(float *), fname);
jobSpecs->thresholds.loadStop = (float **)
my_calloc (jobSpecs->thresholds.nThresholds, sizeof(float *), fname);
for (i = 0; i < jobSpecs->thresholds.nThresholds; i++) {
jobSpecs->thresholds.loadSched[i] = (float *)
my_calloc (jobSpecs->thresholds.nIdx, sizeof(float), fname);
jobSpecs->thresholds.loadStop[i] = (float *)
my_calloc (jobSpecs->thresholds.nIdx, sizeof(float), fname);
}
}
for (j = 0; j < jobSpecs->thresholds.nThresholds; j++) {
for (i = 0; i < jobSpecs->thresholds.nIdx; i++) {
if (!(xdr_float(xdrs, &jobSpecs->thresholds.loadStop[j][i]) &&
xdr_float(xdrs, &jobSpecs->thresholds.loadSched[j][i]))) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(jobSpecs->jobId),
"xdr_float", "loadStop/loadSched");
return(FALSE);
}
}
}
return (TRUE);
}
static int
jobResumeAction (struct jobCard *jp, int sigValue, int suspReason)
{
static char fname[] = "jobResumeAction";
if (jp->jobSpecs.reasons & SUSP_MBD_LOCK) {
return -1;
};
if (jp->jobSpecs.actPid)
return 0;
if (!(jp->jobSpecs.reasons & suspReason))
return -1;
if (logclass & (LC_TRACE | LC_SCHED | LC_EXEC))
ls_syslog(LOG_DEBUG1, "%s: Try to resume job %s with the current reason %d and the triggered reason %d;",
fname, lsb_jobid2str(jp->jobSpecs.jobId), jp->jobSpecs.reasons, suspReason);
if (jobSigStart(jp, sigValue, 0, 0, SIGLOG) < 0)
if (jobsig(jp, 0, FALSE) < 0) {
SBD_SET_STATE(jp, JOB_STAT_EXIT);
return -1;
}
sbdlog_newstatus(jp);
return 0;
}
void
prtJobSubmit(struct jobInfoEnt *job, int prt_q, int tFormat)
{
char prline[MAXLINELEN];
char *timestr;
timestr = putstr_(_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &job->submitTime));
if (tFormat) {
sprintf(prline, (_i18n_msg_get(ls_catd,NL_SETN,569, "%s: Job <%s> submitted from host <%s>")), /* catgets 569 */
timestr, lsb_jobid2str(job->jobId), job->fromHost);
} else {
sprintf(prline, (_i18n_msg_get(ls_catd,NL_SETN,570, "%s: Submitted from host <%s>")), /* catgets 570 */
timestr, job->fromHost);
}
FREEUP(timestr);
prtLine(prline);
if (job->submit.options2 & SUB2_HOLD) {
sprintf(prline, (_i18n_msg_get(ls_catd,NL_SETN,570, " with hold")));
/* catgets 570 */
prtLine(prline);
}
if (prt_q) {
sprintf(prline, (_i18n_msg_get(ls_catd,NL_SETN,571, ", to Queue <%s>")), job->submit.queue); /* catgets 571 */
prtLine(prline);
}
TIMEIT(2, prtBTTime(job), "prtBTTime");
}
void
prtJobSubmit(struct jobInfoEnt *job, int prt_q, int tFormat)
{
char prline[MAXLINELEN];
char timeBuf[128];
if (tFormat) {
sprintf(timeBuf, " Job <%s> s", lsb_jobid2str(job->jobId));
} else {
sprintf(timeBuf, " S");
}
sprintf(prline, "%-12.19s:%submitted from host <%s>",
ctime(&job->submitTime), timeBuf, job->fromHost);
prtLineWUF(prline);
if (job->submit.options2 & SUB2_HOLD) {
sprintf(prline, " on hold");
prtLineWUF(prline);
}
if (prt_q) {
sprintf(prline, " to Queue <%s>", job->submit.queue);
prtLineWUF(prline);
}
TIMEIT(2, prtBTTime(job), "prtBTTime");
}
int
job_resume (struct jobCard *jp)
{
static char fname[] = "job_resume";
int rep;
if (jp->jobSpecs.actPid)
return 0;
if (jobsig(jp, SIGCONT, FALSE) < 0)
return -1;
SBD_SET_STATE(jp, JOB_STAT_RUN);
jp->jobSpecs.reasons = 0;
jp->jobSpecs.subreasons = 0;
rep = status_job (BATCH_STATUS_JOB, jp, jp->jobSpecs.jStatus,
ERR_NO_ERROR);
if (rep < 0)
jp->notReported++;
else {
if (jp->notReported > 0)
jp->notReported = 0;
}
if (logclass & (LC_TRACE | LC_SCHED | LC_EXEC))
ls_syslog(LOG_DEBUG1, "%s: Resume job %s",
fname, lsb_jobid2str(jp->jobSpecs.jobId));
return 0;
}
static int
cleanupMigJob(struct jobCard *jp)
{
static char fname[] = "cleanupMigJob()";
int pid;
unlockHosts (jp, jp->jobSpecs.numToHosts);
if (!jp->jobSpecs.postCmd || jp->jobSpecs.postCmd[0] == '\0')
return 0;
if ((pid = fork()) < 0) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_M, fname,
lsb_jobid2str(jp->jobSpecs.jobId), "fork");
lsb_merr2(_i18n_msg_get(ls_catd , NL_SETN, 700,
"Unable to fork a child to run the queue's post-exec command for job <%s>. Please run <%s> manually if necessary.\n"), /* catgets 700 */
lsb_jobid2str(jp->jobSpecs.jobId),
jp->jobSpecs.postCmd);
return (pid);
}
if (pid)
return (pid);
closeBatchSocket();
putEnv(LS_EXEC_T, "END");
if (postJobSetup(jp) == -1) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S, fname,
lsb_jobid2str(jp->jobSpecs.jobId), "postJobSetup");
lsb_merr2(_i18n_msg_get(ls_catd , NL_SETN, 701,
"Unable to setup the environment for job <%s> to run the queue's post exec. Please run <%s> manually if necessary.\n"), /* catgets 701 */
lsb_jobid2str(jp->jobSpecs.jobId),
jp->jobSpecs.postCmd);
exit(-1);
}
runQPost(jp);
exit(0);
}
static int
signalJobs (LS_LONG_INT *jobIds, int numJobs)
{
int failsignal = FALSE, signaled = FALSE;
int i, cc;
char msg[80];
for (i = 0; i < numJobs; i++) {
if (sigValue == SIGCHK)
cc = lsb_chkpntjob(jobIds[i], chkPeriod, chkOptions);
else if (sigValue == SIGDEL)
cc = lsb_deletejob(jobIds[i], runCount, 0);
else if (sigValue == SIGFORCE)
cc = lsb_forcekilljob(jobIds[i]);
else
cc = lsb_signaljob(jobIds[i], sigValue);
if (cc < 0) {
if (sigValue == SIGCHK && lsberrno == LSBE_NOT_STARTED &&
chkPeriod != LSB_CHKPERIOD_NOCHNG) {
if (chkPeriod)
printf((_i18n_msg_get(ls_catd,NL_SETN,470, "Job <%s>: Checkpoint period is now %d min.\n")), /* catgets 470 */
lsb_jobid2str(jobIds[i]),
(int) (chkPeriod / 60));
else
printf((_i18n_msg_get(ls_catd,NL_SETN,471, "Job <%s>: Periodic checkpointing is disabled\n")), /* catgets 471 */
lsb_jobid2str(jobIds[i]));
signaled = TRUE;
} else {
failsignal = TRUE;
sprintf (msg, "%s <%s>", I18N_Job, lsb_jobid2str(jobIds[i]));
lsb_perror (msg);
}
} else {
signaled = TRUE;
prtSignaled (sigValue, jobIds[i]);
}
}
return (signaled ? !failsignal : FALSE);
}
int
main(int argc, char **argv)
{
int cc;
char *msg;
LS_LONG_INT *jobIDs;
if (lsb_init(argv[0]) < 0) {
lsb_perror("lsb_init");
return -1;
}
while ((cc = getopt(argc, argv, "Vhd:")) != EOF) {
switch (cc) {
case 'V':
fputs(_LS_VERSION_, stderr);
return -1;
case 'h':
usage();
return -1;
case 'd':
msg = optarg;
break;
default:
usage();
return -1;
}
}
if (strlen(msg) > LSB_MAX_MSGSIZE) {
fprintf(stderr, "bpost: message bigger than %d\n", LSB_MAX_MSGSIZE);
return -1;
}
getJobIds(argc,
argv,
NULL,
NULL,
NULL,
NULL,
&jobIDs,
0);
cc = lsb_postjobmsg(jobIDs[0], msg);
if (cc < 0) {
lsb_perror("lsb_jobmsg()");
return -1;
}
printf("Message to job %s posted all right.\n", lsb_jobid2str(jobIDs[0]));
return 0;
}
void
jobSuspendAction(struct jobCard *jp,
int sigValue,
int suspReasons,
int suspSubReasons)
{
static char fname[] = "jobSuspendAction";
if (logclass & (LC_TRACE | LC_SCHED | LC_EXEC))
ls_syslog(LOG_DEBUG1, "%s: Suspend job %s; reasons=%x, subresons=%d, sigValue=%d, status=%x",
fname, lsb_jobid2str(jp->jobSpecs.jobId),
jp->jobSpecs.reasons,
jp->jobSpecs.subreasons, sigValue, jp->jobSpecs.jStatus);
jp->actReasons = suspReasons;
jp->actSubReasons = suspSubReasons;
if (!JOB_RUNNING(jp))
return;
if( jp->postJobStarted ) {
return;
}
if (IS_SUSP (jp->jobSpecs.jStatus)) {
if (jp->jobSpecs.reasons & suspReasons)
return;
else if (jp->jobSpecs.sigMap[-sigValue] == 0)
return;
}
if ((jp->jobSpecs.actPid)
&& ((jp->jobSpecs.actValue == sigValue)
|| (jp->jobSpecs.actValue == (sigValue + jp->jobSpecs.sigMap[-sigValue]))))
return;
if (logclass & (LC_TRACE | LC_SCHED | LC_EXEC))
ls_syslog(LOG_DEBUG1, "%s: Call jobSigStart(sigValue =%d) to suspend job", fname, sigValue + jp->jobSpecs.sigMap[-(sigValue)]);
jobSigStart(jp,
sigValue + jp->jobSpecs.sigMap[-(sigValue)],
0,
0,
SIGLOG);
sbdlog_newstatus(jp);
}
int
main(int argc, char **argv)
{
struct submit req;
struct submitReply reply;
char *job;
LS_LONG_INT jobId = -1, *jobIdList = NULL;
int numJobIds;
time_t beginTime, terminTime;
if (lsb_init(argv[0]) < 0) {
sub_perror("lsb_init");
fprintf(stderr, ". Job not modified.\n");
exit (-1);
}
if (fillReq (argc, argv, CMD_BMODIFY, &req) < 0) {
fprintf(stderr, ". Job not modified.\n");
exit (-1);
}
job = req.command;
beginTime = req.beginTime;
terminTime = req.termTime;
if ((numJobIds = getJobIdList(job, &jobIdList)) < 0) {
exit(-1);
}
jobId = jobIdList[0];
if ((jobId = lsb_modify(&req, &reply, jobId)) < 0) {
if (lsberrno == LSBE_JOB_ARRAY) {
fprintf(stderr, "Options -q and -O cannot be applied on job array");
} else {
prtErrMsg (&req, &reply);
}
fprintf(stderr, ". Job not modified.\n");
if (req.nxf)
free(req.xf);
exit (-1);
}
printf("Parameters of job <%s> are being changed\n", lsb_jobid2str(jobId));
if (beginTime > 0 || terminTime > 0)
prtBETime_(&req);
if (req.nxf)
free(req.xf);
return 0;
}
void
shout_err (struct jobCard *jobPtr, char *msg)
{
char buf[MSGSIZE];
sprintf(buf, \
"We are unable to run your job %s:<%s>. The error is:\n%s.",
lsb_jobid2str(jobPtr->jobSpecs.jobId),
jobPtr->jobSpecs.command, msg);
if (jobPtr->jobSpecs.options & SUB_MAIL_USER) {
merr_user(jobPtr->jobSpecs.mailUser, jobPtr->jobSpecs.fromHost,
buf, I18N_error);
} else {
merr_user(jobPtr->jobSpecs.userName, jobPtr->jobSpecs.fromHost,
buf, I18N_error);
}
}
void
suspendActEnd (struct jobCard *jobCard, int w_status)
{
int sbdStartStop = 0;
if (logclass & (LC_TRACE | LC_SIGNAL))
ls_syslog(LOG_DEBUG, "suspendActEnd: Suspend job %s; reasons=%x, subresons=%d",
lsb_jobid2str(jobCard->jobSpecs.jobId),
jobCard->actReasons,
jobCard->actReasons);
sbdStartStop = (jobCard->actReasons & SUSP_SBD_STARTUP);
jobCard->jobSpecs.lastSSuspTime = now;
jobCard->jobSpecs.reasons |= jobCard->actReasons & (~SUSP_SBD_STARTUP);
jobCard->jobSpecs.subreasons = jobCard->actSubReasons;
if ((jobCard->jobSpecs.actValue == SIG_SUSP_USER) ||
(jobCard->jobSpecs.actValue == SIG_TERM_USER))
SET_STATE(jobCard->jobSpecs.jStatus, JOB_STAT_USUSP);
else
SET_STATE(jobCard->jobSpecs.jStatus, JOB_STAT_SSUSP);
if (w_status == 0)
jobCard->actStatus = ACT_DONE;
else
jobCard->actStatus = ACT_FAIL;
if (sbdStartStop)
jobCard->actStatus = ACT_NO;
if (jobSigLog(jobCard, w_status) == 0) {
jobCard->jobSpecs.actValue = SIG_NULL;
jobCard->jobSpecs.actPid = 0;
}
}
void
prtHeader(struct jobInfoEnt *job, int prt_q, int tFormat)
{
char prline[MAXLINELEN];
if (!tFormat) {
sprintf(prline, "\nJob%s <%s>,", uf_format?"":" Id", lsb_jobid2str(job->jobId));
prtLineWUF(prline);
if (job->submit.options & SUB_JOB_NAME) {
char *jobName, *pos;
jobName = job->submit.jobName;
if ((pos = strchr(jobName, '[')) && LSB_ARRAY_IDX(job->jobId)) {
*pos = '\0';
sprintf(jobName, "%s[%d]", jobName, LSB_ARRAY_IDX(job->jobId));
}
sprintf(prline, " Job Name <%s>,", jobName);
prtLineWUF(prline);
}
}
if (tFormat) {
sprintf(prline, ",");
prtLine(prline);
}
sprintf(prline, " User <%s>,", job->user);
prtLineWUF(prline);
if (lsbMode_ & LSB_MODE_BATCH) {
sprintf(prline, " Project <%s>,", job->submit.projectName);
prtLineWUF(prline);
}
if (job->submit.userGroup && job->submit.userGroup[0] != '\0') {
sprintf(prline, " User Group <%s>,", job->submit.userGroup);
prtLineWUF(prline);
}
if (job->submit.options & SUB_MAIL_USER) {
sprintf(prline, " Mail <%s>,", job->submit.mailUser);
prtLineWUF(prline);
}
if (prt_q) {
sprintf(prline, " Status <%s>, Queue <%s>,",
get_status(job),
job->submit.queue);
prtLineWUF(prline);
}
/* Interactive job */
if (job->submit.options & SUB_INTERACTIVE) {
sprintf(prline, " Interactive");
if (job->submit.options & SUB_PTY) {
strcat(prline, " pseudo-terminal");
if (job->submit.options & (SUB_PTY_SHELL))
strcat(prline, " shell");
}
strcat(prline, " mode,");
prtLineWUF(prline);
}
if (job->jobPriority > 0) {
sprintf(prline, " Job Priority <%d>,", job->jobPriority);
prtLineWUF(prline);
}
if (job->submit.options2 & (SUB2_JOB_CMD_SPOOL)) {
if (tFormat)
sprintf(prline, " Command(Spooled) <%s>", job->submit.command);
else
sprintf(prline, " Command(Spooled) <%s>", job->submit.command);
} else {
if (tFormat)
sprintf(prline, " Command <%s>", job->submit.command);
else
sprintf(prline, " Command <%s>", job->submit.command);
}
prtLineWUF(prline);
if (job->submit.options2 & SUB2_JOB_GROUP) {
sprintf(prline, ", Job Group <%s>", job->submit.job_group);
prtLineWUF(prline);
}
sprintf(prline, "\n");
prtLineWUF(prline);
}
void
child_handler(int sig)
{
int pid;
LS_WAIT_T status;
struct rusage rusage;
register float cpuTime;
struct lsfRusage lsfRusage;
struct jobCard *jobCard;
static short lastMbdExitVal = MASTER_NULL;
static int sbd_finish_sleep = -1;
cleanRusage (&rusage);
now = time(0);
while ((pid=wait3(&status, WNOHANG, &rusage)) > 0) {
if (pid == mbdPid) {
int sig = WTERMSIG(status);
if (mbdExitCnt > 150)
mbdExitCnt = 150;
mbdExitVal = WIFSIGNALED(status);
if (mbdExitVal) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5600,
"mbatchd died with signal <%d> termination"), /* catgets 5600 */
sig);
if (WCOREDUMP(status))
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5601,
"mbatchd core dumped")); /* catgets 5601 */
mbdExitVal = sig;
if (mbdExitVal == lastMbdExitVal)
mbdExitCnt++;
else {
mbdExitCnt = 0;
lastMbdExitVal = mbdExitVal;
}
continue;
} else {
mbdExitVal = WEXITSTATUS(status);
if (mbdExitVal == lastMbdExitVal)
mbdExitCnt++;
else {
mbdExitCnt = 0;
lastMbdExitVal = mbdExitVal;
}
if (mbdExitVal == MASTER_RECONFIG) {
ls_syslog(LOG_NOTICE, _i18n_msg_get(ls_catd , NL_SETN, 5602,
"mbatchd resigned for reconfiguration")); /* catgets 5602 */
start_master();
} else
ls_syslog(LOG_NOTICE, _i18n_msg_get(ls_catd , NL_SETN, 5603,
"mbatchd exited with value <%d>"), /* catgets 5603 */
mbdExitVal);
continue;
}
}
ls_ruunix2lsf (&rusage, &lsfRusage);
cpuTime = lsfRusage.ru_utime + lsfRusage.ru_stime;
for (jobCard = jobQueHead->forw; (jobCard != jobQueHead);
jobCard = jobCard->forw) {
if (jobCard->exitPid == pid) {
jobCard->w_status = LS_STATUS(status);
jobCard->exitPid = -1;
if (logclass & LC_EXEC) {
ls_syslog(LOG_DEBUG, I18N(5604,
"child_handler: Job <%s> exitPid <%d> status <%d> exitcode <%d>"),/*catgets 5604*/
lsb_jobid2str(jobCard->jobSpecs.jobId),
pid, jobCard->w_status,
WEXITSTATUS(status));
}
}
if (jobCard->jobSpecs.jobPid == pid) {
jobCard->collectedChild = TRUE;
jobCard->cpuTime = cpuTime;
jobCard->w_status = LS_STATUS(status);
jobCard->exitPid = -1;
memcpy ((char *) &jobCard->lsfRusage, (char *) &lsfRusage,
sizeof (struct lsfRusage));
jobCard->notReported++;
if (sbd_finish_sleep < 0) {
if (daemonParams[LSB_SBD_FINISH_SLEEP].paramValue) {
errno = 0;
sbd_finish_sleep = atoi(daemonParams[LSB_SBD_FINISH_SLEEP].paramValue);
if (errno)
sbd_finish_sleep = 0;
} else {
sbd_finish_sleep=0;
}
}
if (sbd_finish_sleep > 0) {
millisleep_(sbd_finish_sleep);
}
if (logclass & LC_EXEC) {
ls_syslog(LOG_DEBUG, I18N(5605,
"child_handler: Job <%s> Pid <%d> status <%d> exitcode <%d>"), /*catgets 5605*/
lsb_jobid2str(jobCard->jobSpecs.jobId), pid,
jobCard->w_status, WEXITSTATUS(status));
}
need_checkfinish = TRUE;
break;
}
}
}
}
int
main (int argc, char **argv, char **environ)
{
char *queue = NULL, *host = NULL, *jobName = NULL, *user = NULL;
LS_LONG_INT jobId;
int options;
struct jobInfoEnt *jInfo;
char *outFile;
char fflag = FALSE;
int cc;
int rc;
rc = _i18n_init (I18N_CAT_MIN);
if (lsb_init (argv[0]) < 0)
{
lsb_perror ("lsb_init");
exit (-1);
}
while ((cc = getopt (argc, argv, "Vhfq:m:J:")) != EOF)
{
switch (cc)
{
case 'q':
if (queue || host || jobName)
oneOf (argv[0]);
queue = optarg;
break;
case 'm':
if (queue || host || jobName)
oneOf (argv[0]);
host = optarg;
break;
case 'J':
if (queue || host || jobName)
oneOf (argv[0]);
jobName = optarg;
break;
case 'V':
fputs (_LS_VERSION_, stderr);
exit (0);
case 'f':
fflag = TRUE;
break;
case 'h':
default:
usage (argv[0]);
}
}
jobId = 0;
options = LAST_JOB;
if (argc >= optind + 1)
{
if (queue || host || jobName)
{
oneOf (argv[0]);
}
else if ((argc > 2 && !fflag) || (argc > 3 && fflag))
usage (argv[0]);
if (getOneJobId (argv[optind], &jobId, 0))
{
usage (argv[0]);
}
options = 0;
}
if (lsb_openjobinfo (jobId, jobName, NULL, queue, host, options) < 0
|| (jInfo = lsb_readjobinfo (NULL)) == NULL)
{
if (jobId != 0 || jobName != NULL)
{
user = ALL_USERS;
if (lsb_openjobinfo (jobId, jobName, user, queue, host, options) < 0
|| (jInfo = lsb_readjobinfo (NULL)) == NULL)
{
jobInfoErr (jobId, jobName, NULL, queue, host, options);
exit (-1);
}
}
else
{
jobInfoErr (jobId, jobName, NULL, queue, host, options);
exit (-1);
}
}
lsb_closejobinfo ();
if (jobId && jInfo->jobId != jobId)
{
lsberrno = LSBE_JOB_ARRAY;
lsb_perror ("bpeek");
exit (-1);
}
if ((jInfo->submit.options & SUB_INTERACTIVE) &&
!(jInfo->submit.options & (SUB_OUT_FILE | SUB_ERR_FILE)))
{
fprintf (stderr, _i18n_msg_get (ls_catd, NL_SETN, 2456, "Job <%s> : Cannot bpeek an interactive job.\n"), /* catgets 2456 */
lsb_jobid2str (jInfo->jobId));
exit (-1);
}
if (IS_PEND (jInfo->status) || jInfo->execUsername[0] == '\0')
{
fprintf (stderr, _i18n_msg_get (ls_catd, NL_SETN, 2454, "Job <%s> : Not yet started.\n"), /* catgets 2454 */
lsb_jobid2str (jInfo->jobId));
exit (-1);
}
if (IS_FINISH (jInfo->status))
{
fprintf (stderr, _i18n_msg_get (ls_catd, NL_SETN, 2455, "Job <%s> : Already finished.\n"), /* catgets 2455 */
lsb_jobid2str (jInfo->jobId));
exit (-1);
}
if ((outFile = lsb_peekjob (jInfo->jobId)) == NULL)
{
char msg[50];
sprintf (msg, "%s <%s>", I18N_Job, lsb_jobid2str (jInfo->jobId));
lsb_perror (msg);
exit (-1);
}
displayOutput (outFile, jInfo, fflag, environ);
_i18n_end (ls_catd);
exit (0);
}
void
bmove (int argc, char **argv, int opCode)
{
int position, reqPos;
LS_LONG_INT jobId = 0;
int achar;
if (lsb_init(argv[0]) < 0) {
lsb_perror("lsb_init");
exit(-1);
}
opterr = 0;
while((achar = getopt(argc, argv, "hV")) != EOF) {
switch(achar) {
case 'V':
fputs(_LS_VERSION_, stderr);
exit(0);
case 'h':
default:
usage(argv[0]);
}
}
if (argc == optind) {
fprintf(stderr, "%s.\n",
(_i18n_msg_get(ls_catd,NL_SETN,852, "Job ID must be specified"))); /* catgets 852 */
usage(argv[0]);
}
if (optind < argc-2) {
fprintf(stderr, "%s.\n",
(_i18n_msg_get(ls_catd,NL_SETN,853, "Command syntax error: too many arguments"))); /* catgets 853 */
usage(argv[0]);
}
if (getOneJobId (argv[optind], &jobId, 0)) {
usage(argv[0]);
}
position = 1;
if (optind == argc - 2) {
if (!isint_(argv[++optind]) || atoi(argv[optind]) <= 0) {
fprintf(stderr, "%s: %s.\n", argv[optind],
I18N(854, "Position value must be a positive integer")); /* catgets854*/
usage(argv[0]);
}
position = atoi(argv[optind]);
}
reqPos = position;
if (lsb_movejob(jobId, &position, opCode) <0) {
lsb_perror(lsb_jobid2str (jobId));
exit(-1);
}
if (position != reqPos)
fprintf(stderr, (_i18n_msg_get(ls_catd,NL_SETN,855, "Warning: position value <%d> is beyond movable range.\n")), /* catgets 855 */
reqPos);
if (opCode == TO_TOP)
fprintf(stderr, (_i18n_msg_get(ls_catd,NL_SETN,856, "Job <%s> has been moved to position %d from top.\n")), /* catgets 856 */
lsb_jobid2str (jobId), position);
else
fprintf(stderr, (_i18n_msg_get(ls_catd,NL_SETN,857, "Job <%s> has been moved to position %d from bottom.\n")), /* catgets 857 */
lsb_jobid2str (jobId), position);
exit(0);
}
void
do_sigjob(XDR * xdrs, int chfd, struct LSFHeader * reqHdr)
{
static char fname[] = "do_sigjob()";
char reply_buf[MSGSIZE];
XDR xdrs2;
struct jobSig jobSig;
sbdReplyType reply;
struct jobReply jobReply;
struct LSFHeader replyHdr;
char *replyStruct;
struct jobCard *jp = NULL;
char found = FALSE;
int cc;
int sigValue;
int savedActReasons;
int savedActSubReasons;
struct lsfAuth *auth = NULL;
memset(&jobReply, 0, sizeof(struct jobReply));
if (!xdr_jobSig(xdrs, &jobSig, reqHdr)) {
reply = ERR_BAD_REQ;
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_jobSig");
goto Reply1;
}
jobSig.sigValue = sig_decode(jobSig.sigValue);
sigValue = jobSig.sigValue;
if (logclass & LC_SIGNAL)
ls_syslog(LOG_DEBUG, "do_sigJob: sigValue =%d", sigValue);
for (jp = jobQueHead->forw; (jp != jobQueHead); jp = jp->forw) {
if (jp->jobSpecs.jobId != jobSig.jobId)
continue;
found = TRUE;
break;
}
if (found == FALSE) {
reply = ERR_NO_JOB;
jp = NULL;
goto Reply1;
}
if (jobSig.reasons & SUSP_MBD_LOCK) {
jp->jobSpecs.reasons = jobSig.reasons;
jp->jobSpecs.subreasons = jobSig.subReasons;
savedActReasons = jp->actReasons;
savedActSubReasons = jp->actSubReasons;
jp->actReasons = jobSig.reasons;
jp->actSubReasons = jobSig.subReasons;
}
if (jp->postJobStarted) {
reply = ERR_NO_ERROR;
goto Reply1;
}
if (IS_FINISH(jp->jobSpecs.jStatus)) {
reply = ERR_NO_ERROR;
goto Reply1;
}
if (jp->jobSpecs.jobPGid == -1) {
SBD_SET_STATE(jp, JOB_STAT_EXIT);
reply = ERR_NO_ERROR;
goto Reply;
}
if (!JOB_STARTED(jp)) {
if (isSigTerm(sigValue) == TRUE) {
if ((cc = jobSigStart (jp, sigValue, jobSig.actFlags, jobSig.chkPeriod, NO_SIGLOG)) < 0)
reply = ERR_SIG_RETRY;
else
reply = ERR_NO_ERROR;
goto Reply;
}
reply = ERR_SIG_RETRY;
if (logclass & LC_EXEC)
ls_syslog(LOG_DEBUG, "%s: Retry signal %s for job <%s>",
fname, getLsbSigSymbol(sigValue),
lsb_jobid2str(jp->jobSpecs.jobId));
goto Reply1;
}
if (IS_PEND(jp->jobSpecs.jStatus)) {
reply = ERR_SIG_RETRY;
goto Reply1;
}
if (jp->jobSpecs.actPid || (jp->jobSpecs.jStatus & JOB_STAT_MIG)) {
if ((cc = jobSigStart(jp,
sigValue,
jobSig.actFlags,
jobSig.chkPeriod,
NO_SIGLOG)) < 0)
reply = ERR_SIG_RETRY;
else {
jp->jobSpecs.jStatus &= ~JOB_STAT_MIG;
reply = ERR_NO_ERROR;
}
goto Reply;
}
if ((cc = jobSigStart(jp,
sigValue,
jobSig.actFlags,
jobSig.chkPeriod,
NO_SIGLOG)) < 0)
reply = ERR_SIG_RETRY;
else
reply = ERR_NO_ERROR;
Reply:
sbdlog_newstatus(jp);
Reply1:
xdrmem_create(&xdrs2, reply_buf, MSGSIZE, XDR_ENCODE);
initLSFHeader_(&replyHdr);
replyHdr.opCode = reply;
if (reply == ERR_NO_ERROR) {
jobReply.jobPid = jp->jobSpecs.jobPid;
jobReply.actPid = jp->jobSpecs.actPid;
jobReply.jobId = jp->jobSpecs.jobId;
jobReply.jobPGid = jp->jobSpecs.jobPGid;
jobReply.jStatus = jp->jobSpecs.jStatus;
jobReply.reasons = jp->jobSpecs.reasons;
jobReply.actStatus = jp->actStatus;
replyStruct = (char *) &jobReply;
} else {
if (reply != ERR_NO_JOB)
if ((jp != NULL) && (jobSig.reasons & SUSP_MBD_LOCK)) {
jp->actReasons = savedActReasons;
jp->actSubReasons = savedActSubReasons;
}
replyStruct = (char *) 0;
}
if (!xdr_encodeMsg(&xdrs2, replyStruct, &replyHdr, xdr_jobReply, 0, auth)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_M, fname,
lsb_jobid2str(jp->jobSpecs.jobId), "xdr_jobReply");
relife();
}
if (chanWrite_(chfd, reply_buf, XDR_GETPOS(&xdrs2)) <= 0) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5821,
"%s: Sending jobReply (len=%d) to master failed for job <%s>: %m"), fname, XDR_GETPOS(&xdrs2), lsb_jobid2str(jobSig.jobId)); /* catgets 5821 */
}
if (jp != NULL)
jp->actStatus = ACT_NO;
xdr_destroy(&xdrs2);
return;
}
void
do_jobSetup(XDR * xdrs, int chfd, struct LSFHeader * reqHdr)
{
static char fname[] = "do_jobSetup()";
struct jobSetup jsetup;
struct jobCard *jp = NULL;
char found = FALSE;
struct jobCard savejp;
if (logclass & LC_EXEC)
ls_syslog(LOG_DEBUG, "%s: Entering ...", fname);
if (!xdr_jobSetup(xdrs, &jsetup, reqHdr)) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_jobSetup");
return;
}
for (jp = jobQueHead->forw; (jp != jobQueHead); jp = jp->forw) {
if (jp->jobSpecs.jobId != jsetup.jobId)
continue;
found = TRUE;
break;
}
if (found == FALSE) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5838,
"%s: Job <%s> is not found"), /* catgets 5838 */
fname, lsb_jobid2str(jsetup.jobId));
replyHdrWithRC(LSBE_NO_JOB, chfd, jsetup.jobId);
return;
}
if (jp->jobSpecs.actPid)
return;
memcpy((char *) &savejp, (char *) jp, sizeof(savejp));
jp->execJobFlag |= JOB_EXEC_QPRE_KNOWN;
if (jsetup.execJobFlag & JOB_EXEC_QPRE_OK)
jp->execJobFlag |= JOB_EXEC_QPRE_OK;
jp->jobSpecs.jobPid = jsetup.jobPid;
jp->jobSpecs.jobPGid = jsetup.jobPGid;
jp->jobSpecs.execUid = jsetup.execUid;
strcpy(jp->jobSpecs.execUsername, jsetup.execUsername);
jp->execGid = jsetup.execGid;
strcpy(jp->execUsername, jsetup.execUsername);
strcpy(jp->jobSpecs.execCwd, jsetup.execCwd);
strcpy(jp->jobSpecs.execHome, jsetup.execHome);
if (jsetup.jStatus & JOB_STAT_RUN) {
if (!(jsetup.jStatus & JOB_STAT_PRE_EXEC))
jp->jobSpecs.jStatus &= ~JOB_STAT_PRE_EXEC;
if (status_job(BATCH_STATUS_JOB, jp, jp->jobSpecs.jStatus,
ERR_NO_ERROR) < 0) {
memcpy((char *) jp, (char *) &savejp, sizeof(savejp));
return;
}
jp->execJobFlag |= JOB_EXEC_STARTED;
} else {
jp->jobSpecs.reasons = jsetup.reason;
jp->collectedChild = TRUE;
jp->notReported = 0;
jp->exitPid = -1;
jp->needReportRU = FALSE;
jp->jobSpecs.jStatus = jsetup.jStatus;
jp->w_status = jsetup.w_status;
jp->lsfRusage = jsetup.lsfRusage;
jp->cpuTime = jsetup.cpuTime;
if (job_finish(jp, TRUE) < 0) {
memcpy((char *) jp, (char *) &savejp, sizeof(savejp));
return;
}
}
if (replyHdrWithRC(LSBE_NO_ERROR, chfd, jsetup.jobId) < 0) {
ls_syslog(LOG_DEBUG, "%s: Reply header failed for job <%s>",
fname, lsb_jobid2str(jsetup.jobId));
}
if (logclass & LC_EXEC)
ls_syslog(LOG_DEBUG1, "%s: JobId %s jstatus %d reason %x jobPid %d jobPGid %d execUid %d execGid <%d> execUser <%s> execHome <%s> execCwd <%s> execJobFlag %x cpuTime %f w_status %d",
fname, lsb_jobid2str(jsetup.jobId), jsetup.jStatus,
jsetup.reason, jsetup.jobPid,
jsetup.jobPGid, jsetup.execUid, jsetup.execGid,
jsetup.execUsername, jsetup.execHome, jsetup.execCwd,
jsetup.execJobFlag, jsetup.cpuTime, jsetup.w_status);
}
int
main(int argc, char** argv)
{
char* hosts = NULL;
struct runJobRequest runJobRequest;
int cc;
int c;
bool_t fFlag = FALSE;
bool_t bFlag = FALSE;
int rc;
rc = _i18n_init ( I18N_CAT_MIN );
if (lsb_init(argv[0]) < 0) {
lsb_perror("lsb_init");
exit (-1);
}
while((c = getopt(argc, argv, "m:fbhV")) != EOF) {
switch(c) {
case 'm':
hosts = putstr_(optarg);
if (hosts == NULL) {
perror("putstr_");
exit(-1);
}
break;
case 'f':
fFlag = TRUE;
break;
case 'b':
bFlag = TRUE;
break;
case 'V':
fputs(_LS_VERSION_, stderr);
return (0);
case 'h':
usage(argv[0]);
exit(-1);
}
}
if (argc <= optind) {
usage(argv[0]);
exit(-1);
}
memset((struct runJobRequest* )&runJobRequest, 0,
sizeof(struct runJobRequest));
if (getOneJobId (argv[argc - 1], &(runJobRequest.jobId), 0)) {
usage(argv[0]);
exit(-1);
}
runJobRequest.numHosts = countHosts(hosts);
if (runJobRequest.numHosts > 1) {
int i;
runJobRequest.hostname = (char **)calloc(runJobRequest.numHosts,
sizeof(char *));
if (runJobRequest.hostname == NULL) {
perror("calloc");
exit(-1);
}
for (i = 0; i < runJobRequest.numHosts; i++) {
while (isspace(*hosts)) hosts++;
runJobRequest.hostname[i] = hosts;
hosts += strlen(hosts) + 1;
}
} else
runJobRequest.hostname = &hosts;
runJobRequest.options = (fFlag == TRUE) ?
RUNJOB_OPT_NOSTOP : RUNJOB_OPT_NORMAL;
if (bFlag) {
runJobRequest.options |= RUNJOB_OPT_FROM_BEGIN;
}
cc = lsb_runjob(&runJobRequest);
if (cc < 0) {
lsb_perror((_i18n_msg_get(ls_catd,NL_SETN,2755, "Failed to run the job"))); /* catgets 2755 */
exit(-1);
}
printf((_i18n_msg_get(ls_catd,NL_SETN,2756, "Job <%s> is being forced to run.\n")), /* catgets 2756 */
lsb_jobid2str(runJobRequest.jobId));
_i18n_end ( ls_catd );
return (0);
}
static int
shouldResume (struct hostLoad *loadV, struct jobCard *jp, int num)
{
static char fname[] = "shouldResume";
int i, j, numHosts = -1;
int resume = TRUE, found;
int lastReasons = jp->jobSpecs.reasons;
int lastSubreasons = jp->jobSpecs.subreasons;
struct hostLoad *loads = NULL;
struct tclHostData *tclHostData = NULL;
if (logclass & (LC_SCHED | LC_EXEC))
ls_syslog(LOG_DEBUG3, "%s: job=%s; jStatus=%d; reasons=%x, subreasons=%d, numHosts=%d", fname, lsb_jobid2str(jp->jobSpecs.jobId), jp->jobSpecs.jStatus, jp->jobSpecs.reasons, jp->jobSpecs.subreasons, num);
if (num <= 0)
return FALSE;
if (!(jp->jobSpecs.jStatus & JOB_STAT_SSUSP))
return FALSE;
if ((jp->jobSpecs.reasons & SUSP_QUEUE_WINDOW)
|| (jp->jobSpecs.reasons & SUSP_USER_STOP)
|| (jp->jobSpecs.reasons & SUSP_MBD_LOCK))
return FALSE;
loads = (struct hostLoad *)
my_malloc (num * sizeof (struct hostLoad), fname);
if (jp->resumeCondVal != NULL) {
tclHostData = (struct tclHostData *)
my_malloc (num * sizeof (struct tclHostData), fname);
for (i = 0; i < num; i++) {
initTclHostData (&tclHostData[i]);
}
} else {
tclHostData = NULL;
}
for (j = 0; j <jp->jobSpecs.numToHosts; j++) {
if (j > 0 && !strcmp (jp->jobSpecs.toHosts[j],
jp->jobSpecs.toHosts[j-1]))
continue;
numHosts++;
found = FALSE;
for (i = 0; i < num; i++) {
if (equalHost_(jp->jobSpecs.toHosts[j], loadV[i].hostName)) {
loads[numHosts] = loadV[i];
if (tclHostData != NULL) {
if (getTclHostData (&loadV[i],
&tclHostData[numHosts], FALSE) < 0) {
break;
}
}
found = TRUE;
break;
}
}
if (found != TRUE) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5706,
"%s: Can not find load information for host <%s> to check resume condiftions for job <%s>"), fname, jp->jobSpecs.toHosts[j], lsb_jobid2str(jp->jobSpecs.jobId)); /* catgets 5706 */
loads[numHosts].li = NULL;
continue;
}
}
if (numHosts >= 0) {
numHosts++;
resume = checkResumeByLoad (jp->jobSpecs.jobId, numHosts,
jp->jobSpecs.thresholds, loads, &jp->jobSpecs.reasons,
&jp->jobSpecs.subreasons,
jp->jobSpecs.jAttrib, jp->resumeCondVal, tclHostData);
FREEUP (loads);
if (tclHostData != NULL) {
for (i = 0; i < numHosts; i++) {
FREEUP (tclHostData[i].resBitMaps);
FREEUP (tclHostData[i].loadIndex);
}
FREEUP (tclHostData);
}
} else {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5707,
"%s: No valid load information is found for job <%s>"), fname, lsb_jobid2str(jp->jobSpecs.jobId)); /* catgets 5707 */
}
if ((logclass & (LC_SCHED | LC_EXEC)) && !resume)
ls_syslog(LOG_DEBUG2, "%s: Can't resume job %s; reason=%x, subreasons=%d", fname, lsb_jobid2str(jp->jobSpecs.jobId), jp->jobSpecs.reasons, jp->jobSpecs.subreasons);
if (!resume) {
if ((jp->jobSpecs.reasons != lastReasons ||
(jp->jobSpecs.reasons == lastReasons &&
jp->jobSpecs.subreasons != lastSubreasons)) &&
(now - jp->lastStatusMbdTime > rusageUpdateRate * sbdSleepTime))
jp->notReported++;
}
return (resume);
}
bool_t
xdr_jobSpecs (XDR *xdrs, struct jobSpecs *jobSpecs, struct LSFHeader *hdr)
{
static char fname[] = "xdr_jobSpecs";
char *sp[15];
char *pTemp;
int i, nLimits;
int jobArrId, jobArrElemId;
LS_LONG_INT tmpJobId;
if (xdrs->x_op == XDR_DECODE) {
jobSpecs->numToHosts = 0;
jobSpecs->toHosts = NULL;
jobSpecs->nxf = 0;
jobSpecs->xf = NULL;
jobSpecs->numEnv = 0;
jobSpecs->env = NULL;
jobSpecs->eexec.len = 0;
jobSpecs->eexec.data = NULL;
jobSpecs->loginShell = NULL;
jobSpecs->schedHostType= NULL;
jobSpecs->execHosts = NULL;
}
if (xdrs->x_op == XDR_FREE) {
for(i=0; i < jobSpecs->numToHosts; i++) {
FREEUP(jobSpecs->toHosts[i]);
}
FREEUP(jobSpecs->toHosts);
for(i=0; i < jobSpecs->numEnv; i++)
FREEUP(jobSpecs->env[i]);
FREEUP(jobSpecs->env);
FREEUP(jobSpecs->xf);
FREEUP(jobSpecs->loginShell);
FREEUP(jobSpecs->schedHostType);
FREEUP(jobSpecs->execHosts);
if (!xdr_thresholds(xdrs, jobSpecs) ||
!xdr_lenData(xdrs, &jobSpecs->eexec))
return(FALSE);
return(TRUE);
}
if (xdrs->x_op == XDR_ENCODE) {
jobId64To32(jobSpecs->jobId, &jobArrId, &jobArrElemId);
}
if (!xdr_int(xdrs, &jobArrId)) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname, "xdr_int", "jobId");
return(FALSE);
}
if (!(xdr_int(xdrs, &jobSpecs->userId) &&
xdr_int(xdrs, &jobSpecs->options) &&
xdr_short(xdrs, &jobSpecs->nice) &&
xdr_int(xdrs, &jobSpecs->priority) &&
xdr_int(xdrs, &jobSpecs->chkSig) &&
xdr_int(xdrs, &jobSpecs->actPid) &&
xdr_time_t(xdrs, &jobSpecs->chkPeriod) &&
xdr_time_t(xdrs, &jobSpecs->migThresh) &&
xdr_time_t(xdrs, &jobSpecs->lastSSuspTime) &&
xdr_float(xdrs, &jobSpecs->lastCpuTime))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_int", "userId");
return(FALSE);
}
nLimits = LSF_RLIM_NLIMITS;
tmpJobId = jobArrId;
if (!xdr_int(xdrs, &nLimits)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_int", "nLimits");
return(FALSE);
}
for (i = 0; i < nLimits && i < LSF_RLIM_NLIMITS; i++) {
if(!xdr_lsfLimit(xdrs, &jobSpecs->lsfLimits[i], hdr)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_lsfLimit");
return(FALSE);
}
}
if (nLimits > LSF_RLIM_NLIMITS) {
for (i=LSF_RLIM_NLIMITS; i<nLimits; i++) {
struct lsfLimit lsfLimit;
if (!xdr_lsfLimit(xdrs, &lsfLimit, hdr)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_lsfLimit");
return(FALSE);
}
}
}
if (!(xdr_int(xdrs, &jobSpecs->jStatus) &&
xdr_int(xdrs, &jobSpecs->reasons) &&
xdr_int(xdrs, &jobSpecs->subreasons) &&
xdr_time_t(xdrs, &jobSpecs->termTime) &&
xdr_time_t(xdrs, &jobSpecs->startTime) &&
xdr_int(xdrs, &jobSpecs->runTime) &&
xdr_time_t(xdrs, &jobSpecs->submitTime) &&
xdr_int(xdrs, &jobSpecs->jobPid) &&
xdr_int(xdrs, &jobSpecs->jobPGid) &&
xdr_int(xdrs, &jobSpecs->restartPid) &&
xdr_int(xdrs, &jobSpecs->sigValue) &&
xdr_int(xdrs, &jobSpecs->umask) &&
xdr_int(xdrs, &jobSpecs->jAttrib))) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname, lsb_jobid2str(tmpJobId),
"xdr_int", "jStatus");
return(FALSE);
}
sp[0] = jobSpecs->jobFile;
sp[1] = jobSpecs->inFile;
sp[2] = jobSpecs->outFile;
sp[3] = jobSpecs->errFile;
sp[4] = jobSpecs->chkpntDir;
sp[5] = jobSpecs->cwd;
sp[6] = jobSpecs->subHomeDir;
sp[7] = jobSpecs->command;
sp[8] = jobSpecs->jobName;
sp[9] = jobSpecs->preExecCmd;
sp[10] = jobSpecs->fromHost;
sp[11] = jobSpecs->resReq;
if (xdrs->x_op == XDR_DECODE)
for (i = 0; i < 11; i++)
sp[i][0] = '\0';
if (!(xdr_string(xdrs, &sp[0], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[1], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[2], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[3], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[4], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[5], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[6], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[7], MAXLINELEN) &&
xdr_string(xdrs, &sp[8], MAXLINELEN) &&
xdr_string(xdrs, &sp[9], MAXLINELEN) &&
xdr_string(xdrs, &sp[10], MAXHOSTNAMELEN))) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_int", "jobFile");
return(FALSE);
}
if (xdrs->x_op == XDR_DECODE)
sp[11][0] = '\0';
if (!xdr_string(xdrs, &sp[11], MAXLINELEN)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_int", "jobFile");
return(FALSE);
}
sp[12] = jobSpecs->queue;
sp[13] = jobSpecs->windows;
sp[14] = jobSpecs->userName;
if (xdrs->x_op == XDR_DECODE)
for (i = 12; i < 15; i++)
sp[i][0] = '\0';
if (!(xdr_string(xdrs, &sp[12], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[13], MAXLINELEN) &&
xdr_string(xdrs, &sp[14], MAX_LSB_NAME_LEN))) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname, lsb_jobid2str(tmpJobId),
"xdr_int", "jobFile");
return(FALSE);
}
if (!xdr_int(xdrs, &jobSpecs->numToHosts)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname, lsb_jobid2str(tmpJobId),
"xdr_int", "numToHosts");
return(FALSE);
}
if (xdrs->x_op == XDR_DECODE && jobSpecs->numToHosts) {
jobSpecs->toHosts = (char **) my_calloc(jobSpecs->numToHosts,
sizeof (char *), fname);
}
for (i = 0; i < jobSpecs->numToHosts; i++) {
if (!xdr_var_string(xdrs, &jobSpecs->toHosts[i])) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_var_string", "toHosts");
return(FALSE);
}
}
if (!xdr_thresholds(xdrs, jobSpecs))
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_thresholds");
if (!xdr_int(xdrs, &jobSpecs->nxf)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_int", "nxf");
return(FALSE);
}
if (xdrs->x_op == XDR_DECODE && jobSpecs->nxf > 0) {
jobSpecs->xf = (struct xFile *) my_calloc(jobSpecs->nxf,
sizeof(struct xFile), fname);
}
for (i = 0; i < jobSpecs->nxf; i++) {
if (!xdr_arrayElement(xdrs, (char *) &(jobSpecs->xf[i]),
hdr, xdr_xFile)) {
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_S, fname,
lsb_jobid2str(tmpJobId),
"xdr_arrayElement", "xf");
return(FALSE);
}
}
sp[0] = jobSpecs->mailUser;
sp[1] = jobSpecs->clusterName;
sp[2] = jobSpecs->projectName;
sp[3] = jobSpecs->preCmd;
sp[4] = jobSpecs->postCmd;
sp[5] = jobSpecs->execCwd;
sp[6] = jobSpecs->execHome;
sp[7] = jobSpecs->requeueEValues;
if (xdrs->x_op == XDR_DECODE) {
for(i=0; i < 8; i++)
sp[i][0]= '\0';
}
if (!(xdr_string(xdrs, &sp[0], MAXLINELEN) &&
xdr_string(xdrs, &sp[1], MAX_LSB_NAME_LEN) &&
xdr_string(xdrs, &sp[2], MAX_LSB_NAME_LEN) &&
xdr_string(xdrs, &sp[3], MAXLINELEN) &&
xdr_string(xdrs, &sp[4], MAXLINELEN) &&
xdr_string(xdrs, &sp[5], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[6], MAXFILENAMELEN) &&
xdr_string(xdrs, &sp[7], MAXLINELEN) &&
xdr_int(xdrs, &jobSpecs->execUid) &&
xdr_int(xdrs, &jobSpecs->maxNumProcessors) &&
xdr_int(xdrs, &jobSpecs->numEnv)))
return(FALSE);
if (xdrs->x_op == XDR_DECODE && jobSpecs->numEnv) {
jobSpecs->env = (char **) my_calloc(jobSpecs->numEnv,
sizeof (char *), fname);
}
for (i = 0; i < jobSpecs->numEnv; i++) {
if (!xdr_var_string(xdrs, &jobSpecs->env[i]))
return(FALSE);
}
if (!xdr_lenData(xdrs, &jobSpecs->eexec))
return (FALSE);
if (!xdr_int(xdrs, &jobSpecs->niosPort))
return (FALSE);
sp[0] = jobSpecs->resumeCond;
sp[1] = jobSpecs->stopCond;
sp[2] = jobSpecs->suspendActCmd;
sp[3] = jobSpecs->resumeActCmd;
sp[4] = jobSpecs->terminateActCmd;
if (xdrs->x_op == XDR_DECODE) {
sp[0][0] = '\0';
sp[1][0] = '\0';
}
if (!(xdr_string(xdrs, &sp[0], MAXLINELEN)) ||
!(xdr_string(xdrs, &sp[1], MAXLINELEN)))
return (FALSE);
if (xdrs->x_op == XDR_DECODE) {
for ( i = 2; i < 5; i++)
sp[i][0] = '\0';
}
for ( i = 2; i < 5; i++)
if (!(xdr_string(xdrs, &sp[i], MAXLINELEN)))
return(FALSE);
for (i = 0; i <LSB_SIG_NUM; i++)
if (!(xdr_int(xdrs, &jobSpecs->sigMap[i])))
return(FALSE);
if (!(xdr_int(xdrs, &jobSpecs->actValue)))
return (FALSE);
if (!xdr_var_string(xdrs, &jobSpecs->loginShell))
return (FALSE);
if (!xdr_var_string(xdrs, &jobSpecs->schedHostType))
return (FALSE);
if (!xdr_var_string(xdrs, &jobSpecs->execHosts))
return (FALSE);
if (!xdr_int(xdrs, &jobSpecs->options2)) {
return(FALSE);
}
pTemp = jobSpecs->jobSpoolDir;
if(!( xdr_string(xdrs, &pTemp, MAXPATHLEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_string", "jobSpoolDir");
return(FALSE);
}
if (!(xdr_int(xdrs, &jobArrElemId))) {
return (FALSE);
}
if (xdrs->x_op == XDR_DECODE) {
jobId32To64(&jobSpecs->jobId,jobArrId,jobArrElemId);
}
sp[0] = jobSpecs->inFileSpool;
sp[1] = jobSpecs->commandSpool;
if (!(xdr_string(xdrs, &sp[0], MAXFILENAMELEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_string", "inFileSpool");
return (FALSE);
}
if (!(xdr_string(xdrs, &sp[1], MAXFILENAMELEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_string", "commandSpool");
return (FALSE);
}
if (!(xdr_int(xdrs, &jobSpecs->userPriority))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname, "xdr_int",
"userPriority");
return (FALSE);
}
sp[0] = jobSpecs->execUsername;
if (!(xdr_string(xdrs, &sp[0], MAX_LSB_NAME_LEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_string", "execUsername");
return (FALSE);
}
sp[0] = jobSpecs->prepostUsername;
if (!(xdr_string(xdrs, &sp[0], MAX_LSB_NAME_LEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_string", "prepostUsername");
return (FALSE);
}
return(TRUE);
}
static void
chkpntEnd (struct jobCard *jobCard, int w_status, bool_t *freed)
{
static char fname[] = "chkpntEnd()";
int savePid, saveStatus;
if (IS_SUSP(jobCard->jobSpecs.jStatus)
&& !(jobCard->jobSpecs.jStatus & JOB_STAT_MIG))
jobsig(jobCard, SIGSTOP, TRUE);
saveStatus = jobCard->jobSpecs.jStatus;
if (jobCard->jobSpecs.jStatus & JOB_STAT_MIG) {
if (w_status == 0) {
if (!jobCard->missing) {
jobCard->missing = TRUE;
need_checkfinish = TRUE;
return;
} else if (jobCard->notReported == 0)
return;
if (jobCard->cleanupPid == 0) {
if ((jobCard->cleanupPid = rmJobBufFilesPid(jobCard)) > 0)
return;
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5709,
"%s: Unable to cleanup migrating job <%s>"), /* catgets 5709 */
fname, lsb_jobid2str(jobCard->jobSpecs.jobId));
}
SBD_SET_STATE(jobCard, JOB_STAT_PEND);
} else {
jobCard->jobSpecs.jStatus &= ~JOB_STAT_MIG;
}
}
savePid = jobCard->jobSpecs.actPid;
if (status_job (BATCH_STATUS_JOB, jobCard, jobCard->jobSpecs.jStatus,
w_status == 0 ? ERR_NO_ERROR :
ERR_SYSACT_FAIL) < 0) {
jobCard->jobSpecs.actPid = savePid;
jobCard->jobSpecs.jStatus = saveStatus;
} else {
jobCard->lastChkpntTime = now;
jobCard->jobSpecs.actPid = 0;
jobCard->actStatus = ACT_NO;
jobCard->jobSpecs.actValue = SIG_NULL;
if (w_status == 0) {
jobCard->migCnt = 1;
}
if (saveStatus & JOB_STAT_MIG) {
if (w_status == 0) {
cleanupMigJob(jobCard);
deallocJobCard(jobCard);
*freed = TRUE;
} else
jobCard->migCnt *= 2;
}
}
}
int
status_job (mbdReqType reqType,
struct jobCard *jp, int newStatus, sbdReplyType err)
{
static char fname[] = "status_job()";
static int seq = 1;
static char lastHost[MAXHOSTNAMELEN];
int reply;
char *request_buf;
char *reply_buf = NULL;
XDR xdrs;
struct LSFHeader hdr;
int cc;
struct statusReq statusReq;
int flags;
int i;
int len;
struct lsfAuth *auth = NULL;
if ((logclass & LC_TRACE) && (logclass & LC_SIGNAL))
ls_syslog (LOG_DEBUG, "%s: Entering ... regType %d jobId %s",
fname, reqType, lsb_jobid2str (jp->jobSpecs.jobId));
if (newStatus == JOB_STAT_EXIT)
{
jp->userJobSucc = FALSE;
}
if (MASK_STATUS (newStatus) == JOB_STAT_DONE)
{
jp->userJobSucc = TRUE;
}
if (IS_POST_FINISH (newStatus))
{
if (jp->userJobSucc != TRUE)
{
return 0;
}
}
if (masterHost == NULL)
return -1;
if (jp->notReported < 0)
{
jp->notReported = -INFINIT_INT;
return (0);
}
statusReq.jobId = jp->jobSpecs.jobId;
statusReq.actPid = jp->jobSpecs.actPid;
statusReq.jobPid = jp->jobSpecs.jobPid;
statusReq.jobPGid = jp->jobSpecs.jobPGid;
statusReq.newStatus = newStatus;
statusReq.reason = jp->jobSpecs.reasons;
statusReq.subreasons = jp->jobSpecs.subreasons;
statusReq.sbdReply = err;
statusReq.lsfRusage = jp->lsfRusage;
statusReq.execUid = jp->jobSpecs.execUid;
statusReq.numExecHosts = 0;
statusReq.execHosts = NULL;
statusReq.exitStatus = jp->w_status;
statusReq.execCwd = jp->jobSpecs.execCwd;
statusReq.execHome = jp->jobSpecs.execHome;
statusReq.execUsername = jp->execUsername;
statusReq.queuePostCmd = "";
statusReq.queuePreCmd = "";
statusReq.msgId = jp->delieveredMsgId;
if (IS_FINISH (newStatus))
{
if (jp->maxRusage.mem > jp->runRusage.mem)
jp->runRusage.mem = jp->maxRusage.mem;
if (jp->maxRusage.swap > jp->runRusage.swap)
jp->runRusage.swap = jp->maxRusage.swap;
if (jp->maxRusage.stime > jp->runRusage.stime)
jp->runRusage.stime = jp->maxRusage.stime;
if (jp->maxRusage.utime > jp->runRusage.utime)
jp->runRusage.utime = jp->maxRusage.utime;
}
statusReq.runRusage.mem = jp->runRusage.mem;
statusReq.runRusage.swap = jp->runRusage.swap;
statusReq.runRusage.utime = jp->runRusage.utime;
statusReq.runRusage.stime = jp->runRusage.stime;
statusReq.runRusage.npids = jp->runRusage.npids;
statusReq.runRusage.pidInfo = jp->runRusage.pidInfo;
statusReq.runRusage.npgids = jp->runRusage.npgids;
statusReq.runRusage.pgid = jp->runRusage.pgid;
statusReq.actStatus = jp->actStatus;
statusReq.sigValue = jp->jobSpecs.actValue;
statusReq.seq = seq;
seq++;
if (seq >= MAX_SEQ_NUM)
seq = 1;
len = 1024 + ALIGNWORD_ (sizeof (struct statusReq));
len += ALIGNWORD_ (strlen (statusReq.execHome)) + 4 +
ALIGNWORD_ (strlen (statusReq.execCwd)) + 4 +
ALIGNWORD_ (strlen (statusReq.execUsername)) + 4;
for (i = 0; i < statusReq.runRusage.npids; i++)
len += ALIGNWORD_ (sizeof (struct pidInfo)) + 4;
for (i = 0; i < statusReq.runRusage.npgids; i++)
len += ALIGNWORD_ (sizeof (int)) + 4;
if (logclass & (LC_TRACE | LC_COMM))
ls_syslog (LOG_DEBUG, "%s: The length of the job message is: <%d>", fname,
len);
if ((request_buf = malloc (len)) == NULL)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "malloc");
return (-1);
}
xdrmem_create (&xdrs, request_buf, len, XDR_ENCODE);
initLSFHeader_ (&hdr);
hdr.opCode = reqType;
if (!xdr_encodeMsg (&xdrs, (char *) &statusReq, &hdr, xdr_statusReq, 0,
auth))
{
ls_syslog (LOG_ERR, I18N_JOB_FAIL_S_M,
fname, lsb_jobid2str (jp->jobSpecs.jobId), "xdr_statusReq");
lsb_merr2 (I18N_FUNC_FAIL, fname, "xdr_statusReq");
xdr_destroy (&xdrs);
FREEUP (request_buf);
relife ();
}
flags = CALL_SERVER_NO_HANDSHAKE;
if (statusChan >= 0)
flags |= CALL_SERVER_USE_SOCKET;
if (reqType == BATCH_RUSAGE_JOB)
flags |= CALL_SERVER_NO_WAIT_REPLY;
if (logclass & LC_COMM)
ls_syslog (LOG_DEBUG1, "%s: before call_server statusChan=%d flags=%d",
fname, statusChan, flags);
cc = call_server (masterHost,
mbd_port,
request_buf,
XDR_GETPOS (&xdrs),
&reply_buf,
&hdr,
connTimeout, readTimeout, &statusChan, NULL, NULL, flags);
if (cc < 0)
{
statusChan = -1;
if (!equalHost_ (masterHost, lastHost))
{
if (errno != EINTR)
ls_syslog (LOG_DEBUG,
"%s: Failed to reach mbatchd on host <%s> for job <%s>: %s",
fname, masterHost, lsb_jobid2str (jp->jobSpecs.jobId),
lsb_sysmsg ());
strcpy (lastHost, masterHost);
}
xdr_destroy (&xdrs);
FREEUP (request_buf);
failcnt++;
return (-1);
}
else if (cc == 0)
{
}
failcnt = 0;
lastHost[0] = '\0';
xdr_destroy (&xdrs);
FREEUP (request_buf);
if (cc)
free (reply_buf);
if (flags & CALL_SERVER_NO_WAIT_REPLY)
{
struct timeval timeval;
timeval.tv_sec = 0;
timeval.tv_usec = 0;
if (rd_select_ (chanSock_ (statusChan), &timeval) == 0)
{
jp->needReportRU = FALSE;
jp->lastStatusMbdTime = now;
return 0;
}
CLOSECD (statusChan);
if (logclass & LC_COMM)
ls_syslog (LOG_DEBUG1,
"%s: Job <%s> rd_select() failed, assume connection broken",
fname, lsb_jobid2str (jp->jobSpecs.jobId));
return (-1);
}
reply = hdr.opCode;
switch (reply)
{
case LSBE_NO_ERROR:
case LSBE_LOCK_JOB:
jp->needReportRU = FALSE;
jp->lastStatusMbdTime = now;
if (reply == LSBE_LOCK_JOB)
{
if (IS_SUSP (jp->jobSpecs.jStatus))
jp->jobSpecs.reasons |= SUSP_MBD_LOCK;
else
ls_syslog (LOG_ERR, _i18n_msg_get (ls_catd, NL_SETN, 5204, "%s: Job <%s> is in status <%x> and mbatchd wants to lock it, ignored."), /* catgets 5204 */
fname,
lsb_jobid2str (jp->jobSpecs.jobId),
jp->jobSpecs.jStatus);
}
return (0);
case LSBE_NO_JOB:
if (!IS_POST_FINISH (jp->jobSpecs.jStatus))
{
ls_syslog (LOG_ERR, _i18n_msg_get (ls_catd, NL_SETN, 5205, "%s: Job <%s> is forgotten by mbatchd on host <%s>, ignored."), fname, lsb_jobid2str (jp->jobSpecs.jobId), masterHost); /* catgets 5205 */
}
jp->notReported = -INFINIT_INT;
return (0);
case LSBE_STOP_JOB:
if (jobsig (jp, SIGSTOP, TRUE) < 0)
SET_STATE (jp->jobSpecs.jStatus, JOB_STAT_EXIT);
else
{
SET_STATE (jp->jobSpecs.jStatus, JOB_STAT_USUSP);
jp->jobSpecs.reasons |= SUSP_USER_STOP;
}
return (-1);
case LSBE_SBATCHD:
ls_syslog (LOG_ERR, _i18n_msg_get (ls_catd, NL_SETN, 5206, "%s: mbatchd on host <%s> doesn't think I'm configured as a batch server when I report the status for job <%s>"), /* catgets 5206 */
fname, masterHost, lsb_jobid2str (jp->jobSpecs.jobId));
return (-1);
default:
ls_syslog (LOG_ERR, _i18n_msg_get (ls_catd, NL_SETN, 5207, "%s: Illegal reply code <%d> from mbatchd on host <%s> for job <%s>"), /* catgets 5207 */
fname,
reply, masterHost, lsb_jobid2str (jp->jobSpecs.jobId));
return (-1);
}
}
static void
sigActEnd (struct jobCard *jobCard)
{
int w_status;
struct stat st;
bool_t freed = FALSE;
char exitFile[MAXFILENAMELEN];
if (jobCard->jobSpecs.actValue < 0) {
sprintf(exitFile, "%s/.%s.%s.%s",
LSTMPDIR, jobCard->jobSpecs.jobFile,
lsb_jobidinstr(jobCard->jobSpecs.jobId),
exitFileSuffix(jobCard->jobSpecs.actValue));
w_status = stat(exitFile, &st);
if (w_status == 0)
jobCard->actStatus = ACT_DONE;
else {
jobCard->actStatus = ACT_FAIL;
}
}
jobCard->jobSpecs.jStatus &= ~JOB_STAT_SIGNAL;
switch (jobCard->jobSpecs.actValue) {
case SIG_CHKPNT:
case SIG_CHKPNT_COPY:
chkpntEnd (jobCard, w_status, &freed);
break;
case SIG_SUSP_USER:
case SIG_SUSP_LOAD:
case SIG_SUSP_WINDOW:
case SIG_SUSP_OTHER:
suspendActEnd (jobCard, w_status);
break;
case SIG_RESUME_USER:
case SIG_RESUME_LOAD:
case SIG_RESUME_WINDOW:
case SIG_RESUME_OTHER:
resumeActEnd (jobCard, w_status);
break;
case SIG_TERM_USER:
case SIG_KILL_REQUEUE:
case SIG_TERM_OTHER:
case SIG_TERM_FORCE:
if (jobSigLog (jobCard, w_status) == 0) {
jobCard->jobSpecs.actPid = 0;
jobCard->jobSpecs.actValue = SIG_NULL;
}
break;
case SIG_TERM_LOAD:
case SIG_TERM_WINDOW:
case SIG_TERM_RUNLIMIT:
case SIG_TERM_DEADLINE:
case SIG_TERM_PROCESSLIMIT:
case SIG_TERM_CPULIMIT:
case SIG_TERM_MEMLIMIT:
suspendActEnd (jobCard, w_status);
break;
default:
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5708,
"sigActEnd: unknown sigValue <%d> for job <%s> at the job status <%d> with actPid <%d>"), /* catgets 5708 */
jobCard->jobSpecs.actValue,
lsb_jobid2str(jobCard->jobSpecs.jobId),
jobCard->jobSpecs.jStatus,
jobCard->jobSpecs.actPid);
jobCard->jobSpecs.actPid = 0;
return;
}
if (!freed) {
sbdlog_newstatus (jobCard);
}
}
static void
tryResume (struct hostLoad *myload)
{
char fname[] = "tryResume";
struct jobCard *jobCard, *next;
static int errCount = 0, lastTryResumeTime = 0;
if (now - lastTryResumeTime < sbdSleepTime) {
return;
}
lastTryResumeTime = now;
for (jobCard = jobQueHead->back; jobCard != jobQueHead; jobCard = next) {
next = jobCard->back;
if (!(jobCard->jobSpecs.jStatus & JOB_STAT_SSUSP) ||
jobCard->jobSpecs.actPid)
continue;
if (jobCard->jobSpecs.numToHosts == 1) {
if (shouldResume (myload, jobCard, 1)) {
if (jobResumeAction(jobCard, SIG_RESUME_LOAD, LOAD_REASONS) < 0)
continue;
else
return;
}
} else {
int numh;
struct hostLoad *load;
struct nameList *hostList;
numh = jobCard->jobSpecs.numToHosts;
hostList = lsb_compressStrList(jobCard->jobSpecs.toHosts, numh);
numh = hostList->listSize;
load = ls_loadofhosts ("-", &numh, EFFECTIVE, 0,
hostList->names, hostList->listSize);
if (load == NULL) {
if (errCount < 3)
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_M, fname,
lsb_jobid2str(jobCard->jobSpecs.jobId),
"ls_loadofhosts");
errCount++;
if (lserrno == LSE_LIM_BADHOST)
relife();
if (lserrno == LSE_BAD_XDR)
relife();
if (lserrno == LSE_LIM_DOWN || lserrno == LSE_TIME_OUT)
myStatus |= NO_LIM;
continue;
} else {
myStatus = 0;
errCount = 0;
}
if (!shouldResume (load, jobCard, numh))
continue;
if (jobResumeAction(jobCard, SIG_RESUME_LOAD, LOAD_REASONS) < 0)
continue;
else
return;
}
}
return;
}
static void
tryStop (char *myhostnm, struct hostLoad *myload)
{
static char fname[] = "tryStop";
struct jobCard *jobCard, *next;
int reasons, subreasons, stopmore = FALSE;
static int errCount = 0, lastTryStopTime = 0;
if (now - lastTryStopTime < sbdSleepTime) {
return;
}
lastTryStopTime = now;
for (jobCard = jobQueHead->forw; jobCard != jobQueHead; jobCard = next) {
next = jobCard->forw;
if (jobCard->jobSpecs.numToHosts == 1) {
if ((jobCard->jobSpecs.jStatus & JOB_STAT_RUN)
&& (now >= jobCard->jobSpecs.startTime + sbdSleepTime)
&& shouldStop (myload, jobCard, &reasons, &subreasons, 1, &stopmore)) {
jobSuspendAction (jobCard, SIG_SUSP_LOAD, reasons, subreasons);
if (stopmore)
continue;
else
return;
}
} else {
struct hostLoad *load;
int numh;
struct nameList *hostList;
numh = jobCard->jobSpecs.numToHosts;
hostList = lsb_compressStrList(jobCard->jobSpecs.toHosts, numh);
numh = hostList->listSize;
if (hostList->listSize == 1) {
load = myload;
} else {
load = ls_loadofhosts ("-", &numh, EFFECTIVE, 0,
hostList->names, hostList->listSize);
}
if (load == NULL) {
if (errCount < 3)
ls_syslog(LOG_ERR, I18N_JOB_FAIL_S_MM, fname,
lsb_jobid2str(jobCard->jobSpecs.jobId),
"ls_loadofhosts");
errCount++;
if (lserrno == LSE_LIM_BADHOST)
relife();
if (lserrno == LSE_BAD_XDR)
relife();
if (lserrno == LSE_LIM_DOWN || lserrno == LSE_TIME_OUT)
myStatus |= NO_LIM;
continue;
} else {
errCount = 0;
myStatus = 0;
}
if ((jobCard->jobSpecs.jStatus & JOB_STAT_RUN)
&& now >= jobCard->jobSpecs.startTime + sbdSleepTime) {
if (shouldStop (load, jobCard, &reasons, &subreasons, numh, &stopmore)) {
jobSuspendAction (jobCard, SIG_SUSP_LOAD, reasons, subreasons);
if (stopmore)
break;
else
return;
}
}
}
}
return;
}
void
prtJobStart(struct jobInfoEnt *job, int prtFlag, int jobPid, int tFormat)
{
char prline[MAXLINELEN], tBuff[20];
time_t startTime;
int i = 0;
struct nameList *hostList = NULL;
if (lsbParams[LSB_SHORT_HOSTLIST].paramValue && job->numExHosts > 1
&& strcmp(lsbParams[LSB_SHORT_HOSTLIST].paramValue, "1") == 0) {
hostList = lsb_compressStrList(job->exHosts, job->numExHosts);
if (!hostList) {
exit(99);
}
}
if (tFormat) {
sprintf (tBuff, "%s <%s>", I18N_Job, lsb_jobid2str(job->jobId));
}
else if (LSB_ARRAY_IDX(job->jobId) > 0 )
sprintf (tBuff, " [%d]", LSB_ARRAY_IDX(job->jobId));
else
tBuff[0] = '\0';
if (job->startTime && job->numExHosts) {
if (job->startTime < job->submitTime)
startTime = job->submitTime;
else
startTime = job->startTime;
if ((job->submit.options & SUB_PRE_EXEC)
&& (prtFlag != BJOBS_PRINT)) {
if (prtFlag == BHIST_PRINT_PRE_EXEC)
{
if (tBuff[0] == '\0')
sprintf(prline, "%s: %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T,
&startTime),
I18N(604, "The pre-exec command is started on")); /* catgets 604 */
else
sprintf(prline, "%s:%s, %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T,
&startTime),
tBuff,
I18N(605, "the pre-exec command is started on")); /* catgets 605 */
}
else
{
if (tBuff[0] == '\0')
sprintf(prline, "%s: %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
I18N(606, "The batch job command is started on")); /*catgets 606 */
else
sprintf(prline, "%s:%s, %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
tBuff,
I18N(607, "the batch job command is started on")); /*catgets 607 */
}
} else {
if (jobPid > 0)
{
if (tBuff[0] == '\0')
sprintf(prline, "%s: %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
I18N(608, "Started on")); /* catgets 608 */
else
sprintf(prline, "%s:%s %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
tBuff,
I18N(609, "started on")); /* catgets 609 */
}
else
{
if (tBuff[0] == '\0')
sprintf(prline, "%s: %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
I18N(610, "Dispatched to")); /* catgets 610 */
else
sprintf(prline, "%s: %s %s",
_i18n_ctime(ls_catd, CTIME_FORMAT_a_b_d_T, &startTime),
tBuff,
I18N(611, "dispatched to")); /* catgets 611 */
}
}
prtLineWUF(prline);
if (job->numExHosts > 1) {
sprintf(prline, " %d %s",
job->numExHosts,
I18N(612, "Hosts/Processors")); /* catgets 612 */
prtLineWUF(prline);
}
if (lsbParams[LSB_SHORT_HOSTLIST].paramValue && job->numExHosts > 1
&& strcmp(lsbParams[LSB_SHORT_HOSTLIST].paramValue, "1") == 0) {
for (i = 0; i < hostList->listSize; i++) {
sprintf(prline, " <%d*%s>", hostList->counter[i],
hostList->names[i]);
prtLineWUF(prline);
}
} else {
for (i = 0; i < job->numExHosts; i++) {
sprintf(prline, " <%s>", job->exHosts[i]);
prtLineWUF(prline);
}
}
if (job->execHome && strcmp (job->execHome, "")) {
sprintf(prline, ", %s <%s>",
I18N(615, "Execution Home"), /* catgets 615 */
job->execHome);
prtLineWUF(prline);
}
if (job->execCwd && strcmp (job->execCwd, "")) {
sprintf(prline, ", %s <%s>",
I18N(616, "Execution CWD"), /* catgets 616 */
job->execCwd);
prtLineWUF(prline);
}
if (job->execUsername && strcmp(job->execUsername, "") &&
strcmp(job->user, job->execUsername)) {
sprintf(prline, ", %s <%s>",
I18N(617, "Execution user name"), /* catgets 617 */
job->execUsername);
prtLineWUF(prline);
}
sprintf(prline, ";\n");
prtLineWUF(prline);
}
}
static int
shouldStop (struct hostLoad *loadV,
struct jobCard *jobCard, int *reasons, int *subreasons, int num, int *stopmore)
{
static char fname[] = "shouldStop";
int i, numLoad = -1, j;
struct hostLoad *load = NULL;
static struct tclHostData tclHostData;
static int first = TRUE;
*reasons = 0;
*subreasons = 0;
if( jobCard->postJobStarted ) {
return false;
}
if (jobCard->jobSpecs.jAttrib & JOB_URGENT_NOSTOP)
return false;
if (now - jobCard->windWarnTime < sbdSleepTime)
return FALSE;
if (!JOB_STARTED(jobCard))
return FALSE;
if (LS_ISUNAVAIL(loadV->status))
return FALSE;
if (num <= 0)
return FALSE;
for (i = 0; i <jobCard->jobSpecs.numToHosts && (*reasons) == 0; i++) {
if (i > 0 && !strcmp (jobCard->jobSpecs.toHosts[i],
jobCard->jobSpecs.toHosts[i-1]))
continue;
numLoad++;
load = NULL;
for (j = 0; j < num; j ++) {
if (equalHost_(jobCard->jobSpecs.toHosts[i], loadV[j].hostName)) {
load = &(loadV[j]);
break;
}
}
if (load == NULL) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5705,
"%s: Can not find load information for host <%s>"), fname, jobCard->jobSpecs.toHosts[i]); /* catgets 5705 */
return FALSE;
}
if (LS_ISLOCKEDU(load->status)
&& !(jobCard->jobSpecs.jAttrib & Q_ATTRIB_EXCLUSIVE)) {
*reasons = SUSP_HOST_LOCK;
*stopmore = TRUE;
}
else if (LS_ISLOCKEDM(load->status)) {
*reasons = SUSP_HOST_LOCK_MASTER;
*stopmore = TRUE;
}
else if (load->li[IT] <= jobCard->jobSpecs.thresholds.loadStop[numLoad][IT]
&& load->li[IT] != -INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][IT] != -INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = IT;
*stopmore = TRUE;
}
else if (load->li[LS] >=
jobCard->jobSpecs.thresholds.loadStop[numLoad][LS]
&& load->li[LS] != INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][LS]
!= INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = LS;
*stopmore = TRUE;
}
else if (load->li[UT] >=
jobCard->jobSpecs.thresholds.loadStop[numLoad][UT]
&& load->li[UT] != INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][UT] !=
INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = UT;
}
else if(load->li[PG] >=
jobCard->jobSpecs.thresholds.loadStop[numLoad][PG]
&& load->li[PG] != INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][PG]
!= INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = PG;
}
else if(load->li[IO] >=
jobCard->jobSpecs.thresholds.loadStop[numLoad][IO]
&& load->li[IO] != INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][IO]
!= INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = IO;
}
else if(load->li[MEM]
<= jobCard->jobSpecs.thresholds.loadStop[numLoad][MEM]
&& load->li[MEM] != -INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][MEM]
!= -INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = MEM;
}
else if(load->li[SWP]
<= jobCard->jobSpecs.thresholds.loadStop[numLoad][SWP]
&& load->li[SWP] != -INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][SWP]
!= -INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = SWP;
}
else if(load->li[TMP]
<= jobCard->jobSpecs.thresholds.loadStop[numLoad][TMP]
&& load->li[TMP] != -INFINIT_LOAD
&& jobCard->jobSpecs.thresholds.loadStop[numLoad][TMP]
!= -INFINIT_LOAD) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = TMP;
}
for (j = R15S; !(*reasons) && j <= R15M; j++)
if ((load->li[j] != INFINIT_LOAD)
&& (jobCard->jobSpecs.thresholds.loadStop[numLoad][j]
!= INFINIT_LOAD)
&& (load->li[j]
>= jobCard->jobSpecs.thresholds.loadStop[numLoad][j])) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = j;
break;
}
for (j = MEM + 1; !(*reasons) &&
j < MIN(allLsInfo->numIndx, jobCard->jobSpecs.thresholds.nIdx);
j++) {
if (load->li[j] >= INFINIT_LOAD || load->li[j] <= -INFINIT_LOAD
|| jobCard->jobSpecs.thresholds.loadStop[numLoad][j]
>= INFINIT_LOAD
|| jobCard->jobSpecs.thresholds.loadStop[numLoad][j]
<= -INFINIT_LOAD) {
continue;
}
if (allLsInfo->resTable[j].orderType == INCR) {
if (load->li[j]
>= jobCard->jobSpecs.thresholds.loadStop[numLoad][j]) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = j;
break;
}
} else {
if (load->li[j]
<= jobCard->jobSpecs.thresholds.loadStop[numLoad][j]) {
*reasons |= SUSP_LOAD_REASON;
*subreasons = j;
break;
}
}
}
if (!(*reasons) && jobCard->stopCondVal != NULL) {
int returnCode;
if (first == TRUE) {
initTclHostData (&tclHostData);
returnCode = getTclHostData (load, &tclHostData, FALSE);
first = FALSE;
} else {
returnCode = getTclHostData (load, &tclHostData, TRUE);
}
if (returnCode >= 0
&& evalResReq (jobCard->stopCondVal->selectStr,
&tclHostData, DFT_FROMTYPE) == 1) {
*reasons |= SUSP_QUE_STOP_COND;
break;
}
}
}
if (! (*reasons))
return FALSE;
if (LS_ISLOCKEDU(load->status) || LS_ISLOCKEDM(load->status)) {
return TRUE;
} else if (shouldStop1 (load)) {
if (logclass & (LC_SCHED | LC_EXEC))
ls_syslog (LOG_DEBUG2,
"%s: Should stop job %s; reason=%x, subreasons=%d",
fname, lsb_jobid2str(jobCard->jobSpecs.jobId),
*reasons, *subreasons);
return TRUE;
}
return FALSE;
}