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);
}
void
doSkipSection_conf(struct lsConf *conf, int *LineNum, char *lsfile, char *sectionName)
{
char *word;
char *cp;
if (conf == NULL)
return;
while ((cp = getNextLineC_conf(conf, LineNum, TRUE)) != NULL) {
word = getNextWord_(&cp);
if (strcasecmp(word, "end") == 0) {
word = getNextWord_(&cp);
if (! word) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5419,
"%s(%d): Section ended without section name, ignored"), lsfile, *LineNum); /* catgets 5419 */
} else {
if (strcasecmp(word, sectionName) != 0)
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5420,
"%s(%d): Section %s ended with wrong section name: %s, ignored"), lsfile, *LineNum, sectionName, word); /* catgets 5420 */
}
return;
}
}
ls_syslog(LOG_ERR, I18N_PREMATURE_EOF,
"doSkipSection_conf", lsfile, *LineNum, sectionName);
}
static int
replyHdrWithRC(int rc, int chfd, int jobId)
{
XDR xdrs2;
char reply_buf[MSGSIZE];
static char fname[] = "replyHdrWithRC";
struct LSFHeader replyHdr;
xdrmem_create(&xdrs2, reply_buf, MSGSIZE, XDR_ENCODE);
replyHdr.opCode = rc;
replyHdr.length = 0;
if (!xdr_LSFHeader(&xdrs2, &replyHdr)) {
ls_syslog(LOG_ERR, "%s: xdr_LSFHeader() failed for job <%d>", fname,
jobId);
xdr_destroy(&xdrs2);
return -1;
}
if (chanWrite_(chfd, reply_buf, XDR_GETPOS(&xdrs2)) <= 0) {
ls_syslog(LOG_ERR, "%s: chanWrite_(%d) failed for job <%d>: %m",
fname, XDR_GETPOS(&xdrs2), jobId);
xdr_destroy(&xdrs2);
return -1;
}
xdr_destroy(&xdrs2);
return 0;
} /* replyHdrWithRC */
static char *
readPIMBuf(char *pfile)
{
char *fname="readPIMBuf";
struct stat bstat;
FILE *fp;
FREEUP(pimInfoBuf);
pimInfoLen = 0;
if (stat(pfile,&bstat) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "stat", pfile);
return(NULL);
}
pimInfoLen = bstat.st_size;
if ((pimInfoBuf = (char *)malloc(pimInfoLen+1)) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_M, fname, "malloc");
return(NULL);
}
if ((fp = openPIMFile(pfile)) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "openPIMFile", pfile);
return (FALSE);
}
if (fread(pimInfoBuf,sizeof(char),pimInfoLen,fp) <= 0) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_M, fname, "fread");
FREEUP(pimInfoBuf);
return(NULL);
}
fclose(fp);
pimInfoBuf[pimInfoLen] = '\0';
return(pimInfoBuf);
}
static void
initConn2NIOS (void)
{
static char fname[] = "initConn2NIOS";
conn2NIOS.task_duped =
(int *) calloc (sysconf (_SC_OPEN_MAX), sizeof (int));
if (!conn2NIOS.task_duped)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "calloc");
resExit_ (-1);
}
conn2NIOS.sock.rbuf = (RelayBuf *) malloc (sizeof (RelayBuf));
if (!conn2NIOS.sock.rbuf)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "calloc");
resExit_ (-1);
}
conn2NIOS.sock.wbuf = (RelayLineBuf *) malloc (sizeof (RelayLineBuf));
if (!conn2NIOS.sock.wbuf)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "calloc");
resExit_ (-1);
}
conn2NIOS.sock.fd = -1;
conn2NIOS.wtag = conn2NIOS.rtag = -1;
conn2NIOS.num_duped = 0;
conn2NIOS.sock.rcount = conn2NIOS.sock.wcount = 0;
conn2NIOS.sock.rbuf->bcount = conn2NIOS.sock.wbuf->bcount = 0;
}
void
pingReq(XDR *xdrs, struct sockaddr_in *from, struct LSFHeader *reqHdr)
{
static char fname[] = "pingReq()";
char buf[MSGSIZE/4];
XDR xdrs2;
enum limReplyCode limReplyCode;
struct LSFHeader replyHdr;
limReplyCode = LIME_NO_ERR;
replyHdr.opCode = (short) limReplyCode;
replyHdr.refCode = reqHdr->refCode;
xdrmem_create(&xdrs2, buf, MSGSIZE/4, XDR_ENCODE);
if (!xdr_LSFHeader(&xdrs2, &replyHdr) ||
!xdr_string(&xdrs2, &myHostPtr->hostName, MAXHOSTNAMELEN)) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_string");
xdr_destroy(&xdrs2);
return;
}
if (chanSendDgram_(limSock, buf, XDR_GETPOS(&xdrs2), from) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "chanSendDgram_",
sockAdd2Str_(from));
xdr_destroy(&xdrs2);
return;
}
xdr_destroy(&xdrs2);
return;
}
static char *
getLSFAdmin(void)
{
static char admin[MAXLSFNAMELEN];
static char fname[] = "getLSFAdmin";
char *mycluster;
struct clusterInfo *clusterInfo;
struct passwd *pw;
char *lsfUserName;
if (admin[0] != '\0')
return admin;
if ((mycluster = ls_getclustername()) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_MM, fname, "ls_getclustername");
return NULL;
}
if ((clusterInfo = ls_clusterinfo(NULL, NULL, NULL, 0, 0)) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_MM, fname, "ls_clusterinfo");
return NULL;
}
lsfUserName = (clusterInfo->nAdmins == 0 ? clusterInfo->managerName :
clusterInfo->admins[0]);
if ((pw = getpwnam(lsfUserName)) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M,
fname, "getpwnam", lsfUserName);
return NULL;
}
strcpy(admin, lsfUserName);
return admin;
}
static int
readPIMFile(char *pfile)
{
char *fname = "readPIMFile";
struct lsPidInfo *tmp;
char *buffer,*tmpbuf;
char pimString[MAXLINELEN];
if ((buffer=readPIMBuf(pfile))==NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "readPIMBuf");
return(FALSE);
}
FREEUP(pinfoList);
npinfoList = 0;
pinfoList = (struct lsPidInfo *)malloc(sizeof(struct lsPidInfo) * MAX_NUM_PID);
if (pinfoList == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_D_FAIL_M, fname, "malloc",
sizeof(struct lsPidInfo) * MAX_NUM_PID);
return(FALSE);
}
tmpbuf = getNextString(buffer,pimString);
if (tmpbuf == NULL) {
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 5908,
"%s format error"), "pim.info"); /* catgets 5908 */
return(FALSE);
}
buffer = tmpbuf;
while ((tmpbuf = getNextString(buffer,pimString))!=NULL) {
buffer = tmpbuf;
if (logclass & LC_PIM)
ls_syslog(LOG_DEBUG3,"pim info string is %s",pimString);
sscanf(pimString, "%d %d %d %d %d %d %d %d %d %d %d %d",
&pinfoList[npinfoList].pid, &pinfoList[npinfoList].ppid,
&pinfoList[npinfoList].pgid, &pinfoList[npinfoList].jobid,
&pinfoList[npinfoList].utime, &pinfoList[npinfoList].stime,
&pinfoList[npinfoList].cutime, &pinfoList[npinfoList].cstime,
&pinfoList[npinfoList].proc_size, &pinfoList[npinfoList].resident_size,
&pinfoList[npinfoList].stack_size,
(int *)&pinfoList[npinfoList].status);
npinfoList ++;
if (npinfoList % MAX_NUM_PID == 0) {
tmp = (struct lsPidInfo *)realloc(pinfoList,
sizeof(struct lsPidInfo) * (npinfoList + MAX_NUM_PID));
if (tmp == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_D_FAIL_M, fname, "realloc",
sizeof(struct lsPidInfo)
* (npinfoList + MAX_NUM_PID));
return(FALSE);
}
pinfoList = tmp;
}
}
return(TRUE);
}
int
equivalentXferFile(lsRcpXfer *lsXfer, char *szLocalFile, char *szRemoteFile,
struct stat *psLstat, struct stat *psRstat, char *szRhost)
{
char *pszH;
char szHost1[MAXHOSTNAMELEN], szHost2[MAXHOSTNAMELEN];
char *hostlist[1];
struct hostInfo *hostinfo;
char * szFileName1, * szFileName2;
if (logclass & (LC_FILE))
ls_syslog(LOG_DEBUG,"equivalentXferFile(), ls_getmnthost() for '%s'",
szLocalFile);
hostlist[0] = szRhost;
hostinfo = ls_gethostinfo((char *)NULL, (int *)NULL, (char **)hostlist, 1, 0);
if ( hostinfo == (struct hostInfo *)NULL ) {
return -1;
} else {
if ( strcmp(hostinfo->hostType, "NTX86") == 0
|| strcmp(hostinfo->hostType, "NTALPHA") == 0) {
return (1);
}
}
if ((pszH = ls_getmnthost(szLocalFile)) == NULL) {
return -1;
}
strcpy(szHost1, pszH);
if (logclass & (LC_FILE))
ls_syslog(LOG_DEBUG,
"equivalentXferFile(),ls_rgetmnthost() for '%s' on '%s'"
, szLocalFile, szRhost);
if ((pszH = ls_rgetmnthost(szRhost, szRemoteFile)) == NULL) {
return -1;
}
strcpy(szHost2, pszH);
szFileName1 = strrchr(lsXfer->ppszHostFnames[0],'/');
szFileName2 = strrchr(lsXfer->ppszDestFnames[0],'/');
if (szFileName1 == NULL)
szFileName1 = lsXfer->ppszHostFnames[0];
else szFileName1 ++;
if (szFileName2 == NULL)
szFileName2 = lsXfer->ppszDestFnames[0];
else szFileName2 ++;
if ( psLstat->st_ino == psRstat->st_ino
&& (strcmp(szFileName1,szFileName2) == 0)
&& equalHost_(szHost1,szHost2)) {
return 0;
}
return(1);
}
int
putInLists (char *word, struct admins *admins, int *numAds, char *forWhat)
{
static char fname[] = "putInLists";
struct passwd *pw;
char **tempNames;
int i, *tempIds, *tempGids;
if ((pw = getpwlsfuser_ (word)) == NULL)
{
if (logclass & LC_TRACE)
{
ls_syslog (LOG_DEBUG, _i18n_msg_get (ls_catd, NL_SETN, 5410, "%s: <%s> is not a valid user on this host"), fname, word); /* catgets 5410 */
}
}
if (isInlist (admins->adminNames, word, admins->nAdmins))
{
ls_syslog (LOG_WARNING, _i18n_msg_get (ls_catd, NL_SETN, 5411, "%s: Duplicate user name <%s> %s; ignored"), fname, word, forWhat); /* catgets 5411 */
return (0);
}
admins->adminIds[admins->nAdmins] = (pw == NULL) ? -1 : pw->pw_uid;
admins->adminGIds[admins->nAdmins] = (pw == NULL) ? -1 : pw->pw_gid;
admins->adminNames[admins->nAdmins] = putstr_ (word);
admins->nAdmins += 1;
if (admins->nAdmins >= *numAds)
{
*numAds = *numAds * 2;
tempIds = (int *) realloc (admins->adminIds, *numAds * sizeof (int));
tempGids = (int *) realloc (admins->adminGIds, *numAds * sizeof (int));
tempNames =
(char **) realloc (admins->adminNames, *numAds * sizeof (char *));
if (tempIds == NULL || tempGids == NULL || tempNames == NULL)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "realloc");
FREEUP (tempIds);
FREEUP (tempGids);
FREEUP (tempNames);
FREEUP (admins->adminIds);
FREEUP (admins->adminGIds);
for (i = 0; i < admins->nAdmins; i++)
FREEUP (admins->adminNames[i]);
FREEUP (admins->adminNames);
admins->nAdmins = 0;
lserrno = LSE_MALLOC;
return (-1);
}
else
{
admins->adminIds = tempIds;
admins->adminGIds = tempGids;
admins->adminNames = tempNames;
}
}
return (0);
}
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);
}
bool_t
xdr_sbdPackage1 (XDR *xdrs, struct sbdPackage *sbdPackage, struct LSFHeader *hdr)
{
static char fname[] = "xdr_sbdPackage1";
int i;
if (!(xdr_int(xdrs, &sbdPackage->uJobLimit))) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_int", "uJobLimit");
return(FALSE);
}
if (!(xdr_int(xdrs, &sbdPackage->rusageUpdateRate) &&
xdr_int(xdrs, &sbdPackage->rusageUpdatePercent)))
return(FALSE);
if (!xdr_int(xdrs, &sbdPackage->jobTerminateInterval)) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL, fname,
"xdr_int", "jobTerminateInterval");
return (FALSE);
}
if (xdrs->x_op == XDR_ENCODE || (xdrs->x_op != XDR_FREE)) {
if (!xdr_int(xdrs, &sbdPackage->nAdmins))
return FALSE;
if (xdrs->x_op == XDR_DECODE && sbdPackage->nAdmins > 0) {
sbdPackage->admins
= (char **) calloc (sbdPackage->nAdmins, sizeof (char *));
if (sbdPackage->admins == NULL) {
sbdPackage->nAdmins = 0;
return (FALSE);
}
}
for (i = 0; i < sbdPackage->nAdmins; i++) {
if (!xdr_var_string (xdrs, &sbdPackage->admins[i]))
return (FALSE);
}
}
if (xdrs->x_op == XDR_FREE) {
for (i = 0; i < sbdPackage->nAdmins; i++) {
FREEUP(sbdPackage->admins[i]);
}
FREEUP(sbdPackage->admins);
}
return(TRUE);
}
bool_t
xdr_jobSetup (XDR *xdrs, struct jobSetup *jsetup, struct LSFHeader *hdr)
{
static char fname[]="xdr_jobSetup";
char *sp1, *sp2, *sp3;
int jobArrId, jobArrElemId;
if (xdrs->x_op == XDR_ENCODE) {
jobId64To32(jsetup->jobId, &jobArrId, &jobArrElemId);
}
if (!(xdr_int(xdrs, &jobArrId) &&
xdr_int(xdrs, &jsetup->jStatus) &&
xdr_float(xdrs, &jsetup->cpuTime) &&
xdr_int(xdrs, &jsetup->w_status) &&
xdr_int(xdrs, &jsetup->reason) &&
xdr_int(xdrs, &jsetup->jobPid) &&
xdr_int(xdrs, &jsetup->jobPGid) &&
xdr_int(xdrs, &jsetup->execGid) &&
xdr_int(xdrs, &jsetup->execUid) &&
xdr_int(xdrs, &jsetup->execJobFlag))) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_int");
return (FALSE);
}
if (!xdr_arrayElement(xdrs, (char *) &jsetup->lsfRusage, hdr,
xdr_lsfRusage))
return (FALSE);
sp1 = jsetup->execUsername;
sp2 = jsetup->execHome;
sp3 = jsetup->execCwd;
if (xdrs->x_op == XDR_DECODE) {
sp1[0] = '\0';
sp2[0] = '\0';
sp3[0] = '\0';
}
if (!(xdr_string(xdrs, &sp1, MAX_LSB_NAME_LEN) &&
xdr_string(xdrs, &sp2, MAXFILENAMELEN) &&
xdr_string(xdrs, &sp3, MAXFILENAMELEN))) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_string");
return (FALSE);
}
if (!xdr_int(xdrs, &jobArrElemId)) {
return (FALSE);
}
if (xdrs->x_op == XDR_DECODE) {
jobId32To64(&jsetup->jobId,jobArrId,jobArrElemId);
}
return (TRUE);
}
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
fcp (char *file1, char *file2, struct hostent *hp)
{
static char fname[] = "fcp()";
struct stat sbuf;
int fd1, fd2;
char buf[BSIZE];
int cc;
fd1 = myopen_ (file1, O_RDONLY, 0, hp);
if (fd1 < 0)
return -1;
if (fstat (fd1, &sbuf) < 0)
{
ls_syslog (LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "fstat", file1);
close (fd1);
return -1;
}
fd2 = myopen_ (file2, O_CREAT | O_TRUNC | O_WRONLY, (int) sbuf.st_mode, hp);
if (fd2 < 0)
{
ls_syslog (LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "creat", file1);
close (fd1);
return -1;
}
for (;;)
{
cc = read (fd1, buf, BSIZE);
if (cc == 0)
break;
if (cc < 0)
{
close (fd1);
close (fd2);
return -1;
}
if (write (fd2, buf, cc) != cc)
{
close (fd1);
close (fd2);
return -1;
}
}
close (fd1);
close (fd2);
return (0);
}
char *
runEGroup_(char *type, char *gname)
{
static char fname[] = "runEGroup";
struct lenData ed;
char lsfUserName[MAXLSFNAMELEN];
char egroupPath[MAXFILENAMELEN];
char *argv[4];
char *managerIdStr;
int uid;
struct stat sbuf;
sprintf(egroupPath, "%s/%s", genParams_[LSF_SERVERDIR].paramValue,
EGROUPNAME);
argv[0] = egroupPath;
argv[1] = type;
argv[2] = gname;
argv[3] = NULL;
uid = getuid();
if (uid == 0 && (managerIdStr = getenv("LSB_MANAGERID")) != NULL) {
uid = atoi(managerIdStr);
if (getUserByUid(uid, lsfUserName, sizeof(lsfUserName)) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_D_FAIL_MM, fname, "getUserByUid", uid);
return NULL;
}
} else {
if (getUser(lsfUserName, sizeof(lsfUserName)) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_MM, fname, "getUser");
return NULL;
}
}
if (stat(egroupPath, &sbuf) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "stat", egroupPath);
return NULL;
}
if (getEData(&ed, argv, lsfUserName) < 0) {
ls_syslog(LOG_ERR, I18N_FUNC_S_FAIL_M, fname, "getEData", egroupPath);
return NULL;
}
if (ed.len > 0) {
ed.data[ed.len] = '\0';
return ed.data;
}
return NULL;
}
int
getElock(void)
{
static char fname[] = "getElock";
int force = 0;
int retry;
char *myhostnm, *mastername;
struct stat statbuf;
time_t lastmodtime;
char first = TRUE;
char buf[MAXHOSTNAMELEN+8];
int lock_fd = -1;
if (lsb_CheckMode)
return 0;
if ((myhostnm = ls_getmyhostname()) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_MM, fname, "ls_getmyhostname");
return MASTER_FATAL;
}
sprintf(lockfile, "%s/logdir/%s",
daemonParams[LSB_SHAREDIR].paramValue,
EVENT_LOCK_FILE);
access:
chuser(managerId);
if (force)
lock_fd = open(lockfile, O_RDWR | O_CREAT | O_TRUNC, 0644);
else
lock_fd = open(lockfile, O_RDWR | O_CREAT | O_EXCL, 0644);
if (lock_fd >= 0) {
sprintf(buf, "%s:%d", myhostnm, (int)getpid());
write(lock_fd, buf, strlen(buf));
close(lock_fd);
chuser(batchId);
ls_syslog(LOG_INFO, "%s: Got lock file", fname);
gotLock = TRUE;
return 0;
} else if (errno == EEXIST) {
int fd,cc,i,pid;
fd = open(lockfile, O_RDONLY, 0444);
chuser(batchId);
if (fd < 0) {
ls_syslog(LOG_ERR, "\
%s: Can't open existing lock file <%s>: %m", fname, lockfile);
return MASTER_FATAL;
}
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;
}
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;
}
int
ackReturnCode_(int s)
{
struct LSFHeader repHdr;
int rc;
char hostname[MAXHOSTNAMELEN];
static char fname[] = "ackReturnCode_";
if (logclass & (LC_TRACE))
ls_syslog(LOG_DEBUG, "%s: Entering this routine...", fname);
gethostbysock_(s, hostname);
currentSN = _getcurseqno_(hostname);
rc = expectReturnCode_(s, currentSN, &repHdr);
if (rc < 0)
return rc;
lsf_res_version = (int)repHdr.version;
rc = resRC2LSErr_(repHdr.opCode);
if (rc == 0)
return 0;
else
return -1;
}
static int
chuserId (uid_t uid)
{
if (seteuid(uid) < 0) {
ls_syslog(LOG_ERR, "%s: Failed to change to uid <%d>: %m", __func__, uid);
if (lsb_CheckMode) {
lsb_CheckError = FATAL_ERR;
return -1;
} else {
die(MASTER_FATAL);
}
}
if (uid == 0) {
if (setreuid(0, 0) < 0) {
ls_syslog(LOG_ERR, "\
%s: Failed to change to uid <%d>: %m", __func__, uid);
if (lsb_CheckMode) {
lsb_CheckError = FATAL_ERR;
return -1;
} else {
if (masterme)
die(MASTER_FATAL);
else
die(SLAVE_FATAL);
}
}
}
static int
intoPidList(struct lsPidInfo *pinfo)
{
pidList[npidList].pid = pinfo->pid;
pidList[npidList].ppid = pinfo->ppid;
pidList[npidList].pgid = pinfo->pgid;
pidList[npidList].jobid = pinfo->jobid;
npidList++;
if (npidList % PID_LIST_SIZE == 0) {
struct pidInfo *tmpPtr;
tmpPtr = (struct pidInfo *) realloc((char *) pidList,
(npidList + PID_LIST_SIZE) *
sizeof(struct pidInfo));
if (tmpPtr == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_D_FAIL_M, "intoPidList", "realloc",
(npidList + PID_LIST_SIZE) * sizeof(struct pidInfo));
return (-1);
}
pidList = tmpPtr;
}
return (0);
}
int
getAuth_(struct lsfAuth *auth, char *host)
{
auth->uid = getuid();
if (getUser(auth->lsfUserName, sizeof(auth->lsfUserName)) < 0) {
ls_syslog(LOG_DEBUG, I18N_FUNC_FAIL_MM, "getAuth", "getUser");
lserrno = LSE_BADUSER;
return -1;
}
auth->gid = getgid();
if (!genParams_[LSF_AUTH].paramValue)
auth->kind = CLIENT_SETUID;
else if (!strcmp(genParams_[LSF_AUTH].paramValue, AUTH_IDENT))
auth->kind = CLIENT_IDENT;
else if (!strcmp(genParams_[LSF_AUTH].paramValue, AUTH_PARAM_EAUTH)) {
auth->kind = CLIENT_EAUTH;
return getEAuth(&auth->k.eauth, host);
} else
auth->kind = CLIENT_SETUID;
return 0;
}
void
do_jobSyslog(XDR * xdrs, int chfd, struct LSFHeader * reqHdr)
{
static char fname[] = "do_jobSyslog()";
struct jobSyslog sysMsg;
if (logclass & LC_TRACE)
ls_syslog(LOG_DEBUG, "%s: Entering ...", fname);
if (!xdr_jobSyslog(xdrs, &sysMsg, reqHdr)) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "xdr_syslog");
return;
}
if (replyHdrWithRC(LSBE_NO_ERROR, chfd, -1) < 0)
ls_syslog(LOG_ERR, I18N_FUNC_FAIL, fname, "replyHdrWithRC");
ls_syslog(sysMsg.logLevel, sysMsg.msg);
}
/* prm_init()
*/
int
prm_init(LIST_T *qList)
{
if (logclass & LC_PREEMPT)
ls_syslog(LOG_INFO, "%s: plugin initialized", __func__);
return 0;
}
void
lsb_merr(char *s)
{
char fname[] = "lsb_merr";
FILE *mail;
char *myhostnm;
if (lsb_CheckMode)
return;
if ((myhostnm = ls_getmyhostname()) == NULL) {
ls_syslog(LOG_ERR, I18N_FUNC_FAIL_MM, fname, "ls_getmyhostname");
if (masterme)
die(MASTER_FATAL);
else
die (SLAVE_FATAL);
}
if (lsbManager == NULL || (getpwnam(lsbManager)) == NULL) {
if (lsbManager == NULL)
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 8601,
"%s: LSF administrator name is NULL"), /* catgets 8601 */
fname);
else
ls_syslog(LOG_ERR, _i18n_msg_get(ls_catd , NL_SETN, 8602,
"%s: Bad LSF administrator name <%s>"), /* catgets 8602 */
fname,
lsbManager);
if (masterme)
die (MASTER_FATAL);
else
die (SLAVE_FATAL);
}
mail = smail(lsbManager, myhostnm);
if (masterme)
fprintf(mail, _i18n_msg_get(ls_catd, NL_SETN, 3201,
"Subject: mbatchd on %s: %s\n"), /* catgets 3201 */
myhostnm, s);
else
fprintf(mail, _i18n_msg_get(ls_catd, NL_SETN, 3202,
"Subject: sbatchd on %s: %s\n"), /* catgets 3202 */
myhostnm, s);
mclose(mail);
}
int
ctrlQueue (struct controlReq *qcReq, struct lsfAuth *auth)
{
struct qData *qp;
if ((qp = getQueueData (qcReq->name)) == NULL)
return (LSBE_BAD_QUEUE);
if (!isAuthManager (auth) && auth->uid != 0 && !isAuthQueAd (qp, auth))
{
ls_syslog (LOG_CRIT, I18N (7511, "ctrlQueue: uid <%d> not allowed to perform control operation"), auth->uid); /* catgets 7511 */
return (LSBE_PERMISSION);
}
if ((qcReq->opCode < QUEUE_OPEN) || (qcReq->opCode > QUEUE_INACTIVATE))
return (LSBE_LSBLIB);
if (qcReq->opCode == QUEUE_ACTIVATE && !(qp->qStatus & QUEUE_STAT_RUN))
{
qp->qStatus |= QUEUE_STAT_ACTIVE;
log_queuestatus (qp, qcReq->opCode, auth->uid, auth->lsfUserName);
return (LSBE_QUEUE_WINDOW);
}
if (((qp->qStatus & QUEUE_STAT_OPEN) &&
(qcReq->opCode == QUEUE_OPEN))
|| (!(qp->qStatus & QUEUE_STAT_OPEN) &&
(qcReq->opCode == QUEUE_CLOSED))
|| ((qp->qStatus & QUEUE_STAT_ACTIVE) &&
(qcReq->opCode == QUEUE_ACTIVATE))
|| (!(qp->qStatus & QUEUE_STAT_ACTIVE) &&
(qcReq->opCode == QUEUE_INACTIVATE)))
return (LSBE_NO_ERROR);
if (qcReq->opCode == QUEUE_OPEN)
{
qp->qStatus |= QUEUE_STAT_OPEN;
log_queuestatus (qp, qcReq->opCode, auth->uid, auth->lsfUserName);
}
if (qcReq->opCode == QUEUE_CLOSED)
{
qp->qStatus &= ~QUEUE_STAT_OPEN;
log_queuestatus (qp, qcReq->opCode, auth->uid, auth->lsfUserName);
}
if (qcReq->opCode == QUEUE_ACTIVATE)
{
qp->qStatus |= QUEUE_STAT_ACTIVE;
log_queuestatus (qp, qcReq->opCode, auth->uid, auth->lsfUserName);
}
if (qcReq->opCode == QUEUE_INACTIVATE)
{
qp->qStatus &= ~QUEUE_STAT_ACTIVE;
log_queuestatus (qp, qcReq->opCode, auth->uid, auth->lsfUserName);
}
return (LSBE_NO_ERROR);
}
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 void
initChildRes (char *envdir)
{
static char fname[] = "initChildRes";
int i, maxfds;
getLogClass_ (resParams[LSF_DEBUG_RES].paramValue,
resParams[LSF_TIME_RES].paramValue);
openChildLog ("res", resParams[LSF_LOGDIR].paramValue,
(debug > 1), &(resParams[LSF_LOG_MASK].paramValue));
if ((Myhost = ls_getmyhostname ()) == NULL)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "ls_getmyhostname");
resExit_ (-1);
}
client_cnt = child_cnt = 0;
for (i = 0; i < MAXCLIENTS_HIGHWATER_MARK + 1; i++)
{
clients[i] = NULL;
}
children = calloc (sysconf (_SC_OPEN_MAX), sizeof (struct children *));
if (!children)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "calloc");
resExit_ (-1);
}
maxfds = sysconf (_SC_OPEN_MAX);
for (i = 0; i < maxfds; i++)
{
children[i] = NULL;
}
initConn2NIOS ();
resNotifyList = listCreate ("resNotifyList");
if (!resNotifyList)
{
ls_syslog (LOG_ERR, I18N_FUNC_FAIL_M, fname, "listCreate");
resExit_ (-1);
}
}
static int
callLimTcp_(char *reqbuf,
char **rep_buf,
int req_size,
struct LSFHeader *replyHdr,
int options)
{
static char fname[]="callLimTcp_";
char retried = FALSE;
int cc;
XDR xdrs;
struct Buffer sndbuf;
struct Buffer rcvbuf;
if (logclass & (LC_COMM | LC_TRACE))
ls_syslog(LOG_DEBUG2,"%s: Entering...req_size=%d",fname, req_size);
*rep_buf = NULL;
if (!sockIds_[TCP].sin_addr.s_addr) {
if (ls_getmastername() == NULL)
return(-1);
}
contact:
if (limchans_[TCP] < 0) {
limchans_[TCP] = chanClientSocket_(AF_INET, SOCK_STREAM, 0);
if (limchans_[TCP] < 0 )
return(-1);
cc = chanConnect_(limchans_[TCP],
&sockIds_[TCP],
conntimeout_ * 1000,
0);
if (cc < 0) {
ls_syslog(LOG_DEBUG,"\
%s: failed in connecting to limChans_[TCP]=<%d> <%s>",
fname, limchans_[TCP], sockAdd2Str_(&sockIds_[TCP]));
if (errno == ECONNREFUSED || errno == ENETUNREACH) {
if (errno == ECONNREFUSED) {
lserrno = LSE_LIM_DOWN;
}
if (! retried) {
if (ls_getmastername() != NULL) {
retried = 1;
CLOSECD(limchans_[TCP]);
goto contact;
}
}
}
sockIds_[TCP].sin_addr.s_addr = 0;
sockIds_[TCP].sin_port = 0;
CLOSECD(limchans_[TCP]);
return(-1);
}
