static void
operateHost (char *host, int opCode, int confirm)
{
char msg1[MAXLINELEN];
char msg[MAXLINELEN];
if (opCode == LIM_CMD_REBOOT)
sprintf(msg1, I18N(256, "Restart LIM on <%s>"), host); /* catgets 256 */
else
sprintf(msg1, I18N(257, "Shut down LIM on <%s>"), host); /* catgets 257 */
if (confirm) {
sprintf(msg, "%s ? [y/n] ", msg1);
if (!getConfirm(msg))
return;
}
fprintf(stderr, "%s ...... ", msg1);
fflush(stderr);
if (ls_limcontrol(host, opCode) == -1) {
ls_perror ("ls_limcontrol");
exitrc = -1;
} else {
char *delay = getenv("LSF_RESTART_DELAY");
int delay_time;
if (delay == 0)
delay_time = 500;
else
delay_time = atoi(delay) * 1000;
millisleep_(delay_time);
fprintf (stderr, "%s\n", I18N_done);
}
fflush(stderr);
}
int
nb_write_fix(int s, char *buf, int len)
{
int cc;
int length;
struct timeval start, now;
struct timezone junk;
gettimeofday(&start, &junk);
for (length = len; len > 0 ; ) {
if ((cc = write(s, buf, len)) > 0) {
len -= cc;
buf += cc;
} else if (cc < 0 && BAD_IO_ERR(errno)) {
if (errno == EPIPE)
lserrno = LSE_LOSTCON;
return (-1);
}
if (len > 0)
{
gettimeofday(&now, &junk);
if (US_DIFF(now, start) > IO_TIMEOUT * 1000) {
errno = ETIMEDOUT;
return(-1);
}
millisleep_(IO_TIMEOUT / 20);
}
}
return (length);
}
static FILE *
openPIMFile(char *pfile)
{
static char fname[] = "openPIMFile";
FILE *fp;
if ((fp = fopen(pfile, "r")) == NULL) {
millisleep_(1000);
if ((fp = fopen(pfile, "r")) == NULL) {
if (logclass & LC_PIM)
ls_syslog(LOG_DEBUG, "%s: fopen(%s) failed: %m", fname, pfile);
lserrno = LSE_FILE_SYS;
return (NULL);
}
}
return (fp);
}
int
nb_read_fix (int s, char *buf, int len)
{
int cc;
int length;
struct timeval start, now;
struct timezone junk;
if (logclass & LC_TRACE)
ls_syslog (LOG_DEBUG, "nb_read_fix: Entering this routine...");
gettimeofday (&start, &junk);
for (length = len; len > 0;)
{
if ((cc = read (s, buf, len)) > 0)
{
len -= cc;
buf += cc;
}
else if (cc == 0 || BAD_IO_ERR (errno))
{
if (cc == 0)
errno = ECONNRESET;
return (-1);
}
if (len > 0)
{
gettimeofday (&now, &junk);
if (US_DIFF (now, start) > IO_TIMEOUT * 1000)
{
errno = ETIMEDOUT;
return (-1);
}
millisleep_ (IO_TIMEOUT / 20);
}
}
return (length);
}
int
main(int argc, char **argv)
{
static char fname[] = "res/main";
int nready;
int maxfd;
int i;
char *sp;
char *pathname = NULL;
int didSomething = 0;
char exbuf;
time_t thisPeriodic, lastPeriodic = 0, timeDiff;
fd_set rm, wm, em;
int sbdPty = FALSE;
char *sbdClHost = NULL;
ushort sbdClPort = 0;
char **sbdArgv = NULL;
int selectError = 0;
_i18n_init(I18N_CAT_RES);
saveDaemonDir_(argv[0]);
for (i=1; i<argc; i++) {
if (strcmp(argv[i], "-d") == 0 && argv[i+1] != NULL) {
pathname = argv[i+1];
putEnv("LSF_ENVDIR",pathname);
break;
}
}
if (pathname == NULL) {
if ((pathname = getenv("LSF_ENVDIR")) == NULL)
pathname = LSETCDIR;
}
if (argc > 1) {
if (!strcmp(argv[1],"-V")) {
fputs(_LS_VERSION_, stderr);
exit(0);
}
}
if ((ls_readconfenv(resConfParams, NULL) < 0) ||
(initenv_(resParams, pathname) < 0) ) {
if ((sp = getenv("LSF_LOGDIR")) != NULL)
resParams[LSF_LOGDIR].paramValue = sp;
ls_openlog("res", resParams[LSF_LOGDIR].paramValue, (debug > 1),
resParams[LSF_LOG_MASK].paramValue);
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_MM, fname, "initenv_",
pathname);
ls_syslog(LOG_ERR, I18N_Exiting);
exit(-1);
}
restart_argc = argc;
restart_argv = argv;
for (i=1; i<argc; i++) {
if (strcmp(argv[i], "-d") == 0 && argv[i+1] != NULL) {
pathname = argv[i+1];
i++;
continue;
}
if (strcmp(argv[i], "-1") == 0) {
debug = 1;
continue;
}
if (strcmp(argv[i], "-2") == 0) {
debug = 2;
continue;
}
if (strcmp(argv[i], "-PTY_FIX") == 0) {
printf("PTY_FIX");
exit(0);
}
if ( (strcmp(argv[i], "-j") == 0) && (argv[i+1] != NULL) ) {
lsbJobStarter = argv[++i];
continue;
}
if (strcmp(argv[i], "-P") == 0) {
sbdPty = TRUE;
continue;
}
if (strcmp(argv[i], "-i") == 0) {
sbdFlags |= SBD_FLAG_STDIN;
continue;
}
if (strcmp(argv[i], "-o") == 0) {
sbdFlags |= SBD_FLAG_STDOUT;
continue;
}
if (strcmp(argv[i], "-e") == 0) {
sbdFlags |= SBD_FLAG_STDERR;
continue;
}
if (strcmp(argv[i], "-m") == 0 && argv[i+1] != NULL) {
sbdClHost = argv[i+1];
i++;
sbdMode = TRUE;
continue;
}
if (strcmp(argv[i], "-p") == 0 && argv[i+1] != NULL) {
sbdClPort = atoi(argv[i+1]);
i++;
sbdMode = TRUE;
continue;
}
if (argv[i][0] != '-') {
sbdMode = TRUE;
sbdArgv = argv + i;
break;
}
usage(argv[0]);
}
if (sbdMode) {
if (sbdClHost == NULL || sbdArgv == NULL) {
usage(argv[0]);
exit(-1);
}
if (sbdClPort) {
sbdFlags |= SBD_FLAG_TERM;
} else {
sbdFlags |= SBD_FLAG_STDIN | SBD_FLAG_STDOUT | SBD_FLAG_STDERR;
}
} else {
if (debug < 2)
for (i = sysconf(_SC_OPEN_MAX) ; i >= 0 ; i--)
close(i);
}
if (resParams[LSF_SERVERDIR].paramValue == NULL) {
ls_openlog("res", resParams[LSF_LOGDIR].paramValue, (debug > 1),
resParams[LSF_LOG_MASK].paramValue);
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5001,
"LSF_SERVERDIR not defined in %s/lsf.conf: %M; res exiting"), /* catgets 5001 */
pathname);
resExit_(-1);
}
if (! debug && resParams[LSF_RES_DEBUG].paramValue != NULL) {
debug = atoi(resParams[LSF_RES_DEBUG].paramValue);
if (debug <= 0)
debug = 1;
}
getLogClass_(resParams[LSF_DEBUG_RES].paramValue,
resParams[LSF_TIME_RES].paramValue);
if (getuid() == 0 && debug) {
if (sbdMode) {
debug = 0;
} else {
ls_openlog("res", resParams[LSF_LOGDIR].paramValue, FALSE,
resParams[LSF_LOG_MASK].paramValue);
ls_syslog(LOG_ERR, I18N(5005,"Root cannot run RES in debug mode ... exiting."));/*catgets 5005 */
exit(-1);
}
}
if (debug > 1)
ls_openlog("res", resParams[LSF_LOGDIR].paramValue, TRUE, "LOG_DEBUG");
else {
ls_openlog("res", resParams[LSF_LOGDIR].paramValue, FALSE,
resParams[LSF_LOG_MASK].paramValue);
}
if (logclass & (LC_TRACE | LC_HANG))
ls_syslog(LOG_DEBUG, "%s: logclass=%x", fname, logclass);
ls_syslog(LOG_DEBUG, "%s: LSF_SERVERDIR=%s", fname, resParams[LSF_SERVERDIR].paramValue);
init_res();
initSignals();
periodic(0);
if (sbdMode) {
lsbJobStart(sbdArgv, sbdClPort, sbdClHost, sbdPty);
}
maxfd = FD_SETSIZE;
for (;;) {
struct timeval *timep;
struct timeval timeout;
loop:
didSomething = 0;
for (i = 0; i < child_cnt; i++) {
if (children[i]->backClnPtr == NULL
&& !FD_IS_VALID(conn2NIOS.sock.fd)
&& children[i]->running == 0) {
delete_child (children[i]);
}
}
if (logclass & LC_TRACE) {
ls_syslog(LOG_DEBUG,"\
%s: %s Res child_res=<%d> child_go=<%d> child_cnt=<%d> client_cnt=<%d>",
fname, ((child_res) ? "Application" : "Root") ,
child_res, child_go, child_cnt, client_cnt);
if (child_cnt == 1 && children != NULL && children[0] != NULL) {
dumpChild(children[0], 1, "in main()");
}
}
if (child_res && child_go && child_cnt == 0 && client_cnt == 0) {
if (debug > 1)
printf (" \n Child <%d> Retired! \n", (int)getpid());
if (logclass & LC_TRACE) {
ls_syslog(LOG_DEBUG,"\
%s: Application Res is exiting.....", fname);
}
millisleep_(5000);
if (sbdMode) {
close(1);
close(2);
exit(lastChildExitStatus);
}
resExit_(EXIT_NO_ERROR);
}
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;
}
}
}
}
