diff --git a/NEWS b/NEWS
index 28f663da962372594f45adc4be5adcf1aa61daf4..e0cacbb89b995aed9cbc1ee19acaaa55857510d5 100644
--- a/NEWS
+++ b/NEWS
@@ -4,6 +4,7 @@ documents those changes that are of interest to users and admins.
* Changes in SLURM 0.5.0-pre10
==============================
-- squeue and sinfo print timestamp in interate mode by default.
+ -- Latest accounting patches from Riebs/HP, retry communications.
* Changes in SLURM 0.5.0-pre9
=============================
diff --git a/src/plugins/jobacct/log/jobacct_log.c b/src/plugins/jobacct/log/jobacct_log.c
index 3df8af1d0cdb67b3649a2f109db8376e3dd6f59d..781fe36d8993ec6c67b1779a699e1825acd027bb 100644
--- a/src/plugins/jobacct/log/jobacct_log.c
+++ b/src/plugins/jobacct/log/jobacct_log.c
@@ -102,7 +102,11 @@ const uint32_t plugin_version = 100;
#define HOST_NAME_MAX 256
#endif
#endif
+#define MAX_BUFFER_SIZE 1024
#define MAX_MSG_SIZE 1024
+#define MAX_SEND_RETRIES 3
+#define MAX_SEND_RETRY_DELAY 5
+#define NOT_FOUND "NOT_FOUND"
typedef enum _stats_msg_type {
TO_CONTROLLER=100,
@@ -204,7 +208,7 @@ static _prec_t *_get_process_data(void);
static int _get_process_data_line(FILE *in, _prec_t *prec);
static void _get_slurmctld_syms(void);
static int _pack_jobrec(_jrec_t *outrec, _jrec_t *inrec);
-static void _print_record(struct job_record *job_ptr, char *data);
+static int _print_record(struct job_record *job_ptr, char *data);
static void _process_mynode_msg(_mynode_msg_t *msg);
static void _process_mynode_msg_launch(_jrec_t *jrec);
static void _process_mynode_msg_taskdata(_jrec_t *jrec);
@@ -236,7 +240,6 @@ int slurm_jobacct_process_message(struct slurm_msg *msg)
_stats_msg_type_t msgtype;
_stats_msg_t *stats;
- debug2("jobacct(%d): in slurm_jobacct_process_message", getpid());
jmsg = msg->data;
stats = (_stats_msg_t *)jmsg->data;
msgtype = ntohl(stats->msg_type);
@@ -265,7 +268,7 @@ int slurm_jobacct_process_message(struct slurm_msg *msg)
getpid(), jobid, jmsg->data);
return(SLURM_ERROR);
}
- _print_record(job_ptr, stats->data);
+ rc = _print_record(job_ptr, stats->data);
break;
/*
@@ -303,9 +306,11 @@ int slurm_jobacct_process_message(struct slurm_msg *msg)
static int _send_msg_to_slurmctld(_stats_msg_t *stats) {
- slurm_msg_t *msg;
+ slurm_msg_t *msg,
+ *retmsg=NULL;
jobacct_msg_t *jmsg;
- int rc = SLURM_SUCCESS;
+ int rc = SLURM_SUCCESS,
+ retry;
debug2("jobacct(%d): _send_msg_to_slurmctld, msgtype=%d",
getpid(), ntohl(stats->msg_type));
@@ -313,14 +318,25 @@ static int _send_msg_to_slurmctld(_stats_msg_t *stats) {
jmsg->len = sizeof(_stats_msg_t);
jmsg->data = (char *)stats;
msg = xmalloc(sizeof(slurm_msg_t));
+ retmsg = xmalloc(sizeof(slurm_msg_t));
msg->msg_type = MESSAGE_JOBACCT_DATA;
msg->data = jmsg;
msg->data_size = sizeof(jobacct_msg_t);
- if (slurm_send_only_controller_msg(msg) < 0) {
- error("Unable to send message to slurmctld: %m");
- rc = SLURM_ERROR;
+ for (retry=0; retry<MAX_SEND_RETRIES; retry++) {
+ if ((rc=slurm_send_recv_controller_msg(msg, retmsg)) >= 0)
+ break;
+ sleep(1+(rand()/(RAND_MAX/MAX_SEND_RETRY_DELAY)));
+ /* pause 1-MAX_SEND_RETRY_DELAY seconds */
}
+ if (rc<0)
+ error("jobacct(%d): _send_data_to_slurmctld(msg, retmsg)"
+ " says %d (%m) after %d tries",
+ getpid(), rc, retry);
+ else
+ debug3("jobacct(%d): slurm_send_recv_controller_msg says %d",
+ rc);
+ slurm_free_cred(retmsg->cred);
xfree(jmsg);
xfree(msg);
debug2("jobacct(%d): leaving _send_msg_to_slurmctld, rc=%d",
@@ -375,7 +391,7 @@ int slurmctld_jobacct_job_complete(struct job_record *job_ptr)
{
int rc = SLURM_SUCCESS,
tmp;
- _stats_msg_t *stats;
+ char *buf;
struct tm ts; /* timestamp decoder */
debug2("slurmctld_jobacct_job_complete() called");
@@ -384,24 +400,21 @@ int slurmctld_jobacct_job_complete(struct job_record *job_ptr)
return rc;
}
gmtime_r(&job_ptr->end_time, &ts);
- stats = xmalloc(sizeof(_stats_msg_t));
- stats->msg_type = htonl(TO_CONTROLLER);
- stats->jobid = htonl(job_ptr->job_id);
- stats->stepid = 0;
- tmp = snprintf(stats->data, MAX_MSG_SIZE,
+ buf = xmalloc(MAX_BUFFER_SIZE);
+ tmp = snprintf(buf, MAX_MSG_SIZE,
"JOB_TERMINATED 1 12 %u %04d%02d%02d%02d%02d%02d %s",
(int) (job_ptr->end_time - job_ptr->start_time),
1900+(ts.tm_year), 1+(ts.tm_mon), ts.tm_mday,
- ts.tm_hour, ts.tm_min, ts.tm_sec,
- job_state_string_compact((job_ptr->job_state) & ~JOB_COMPLETING));
+ ts.tm_hour, ts.tm_min, ts.tm_sec,
+ job_state_string_compact(
+ (job_ptr->job_state) & ~JOB_COMPLETING));
if (tmp >= MAX_MSG_SIZE) {
error("slurmctld_jobacct_job_complete buffer overflow");
rc = SLURM_ERROR;
} else {
- stats->datalen = htonl(tmp+1);
- rc = _send_msg_to_slurmctld(stats);
+ rc = _print_record(job_ptr, buf);
}
- xfree(stats);
+ xfree(buf);
return rc;
}
@@ -409,11 +422,11 @@ int slurmctld_jobacct_job_start(struct job_record *job_ptr)
{
int i,
ncpus=0,
- rc,
+ rc=SLURM_SUCCESS,
tmp;
- char *jname;
+ char *buf,
+ *jname;
long priority;
- _stats_msg_t *stats;
debug2("slurmctld_jobacct_job_start() called");
for (i=0; i < job_ptr->num_cpu_groups; i++)
@@ -434,11 +447,8 @@ int slurmctld_jobacct_job_start(struct job_record *job_ptr)
jname = (char*) xmalloc(10);
strncpy(jname, "(noname)", 10);
}
- stats = xmalloc(sizeof(_stats_msg_t));
- stats->msg_type = htonl(TO_CONTROLLER);
- stats->jobid = htonl(job_ptr->job_id);
- stats->stepid = 0;
- tmp = snprintf(stats->data, MAX_MSG_SIZE,
+ buf = xmalloc(MAX_BUFFER_SIZE);
+ tmp = snprintf(buf, MAX_MSG_SIZE,
"JOB_START 1 16 %d %d %s %u %ld %u %s",
job_ptr->user_id, job_ptr->group_id, jname,
job_ptr->batch_flag, priority, ncpus,
@@ -447,10 +457,9 @@ int slurmctld_jobacct_job_start(struct job_record *job_ptr)
error("slurmctld_jobacct_job_start buffer overflow");
rc = SLURM_ERROR;
} else {
- stats->datalen = htonl(tmp+1);
- rc = _send_msg_to_slurmctld(stats);
+ rc = _print_record(job_ptr, buf);
}
- xfree(stats);
+ xfree(buf);
xfree(jname);
return rc;
}
@@ -481,26 +490,29 @@ static void _get_slurmctld_syms(void)
* Print the record to the log file.
*/
-static void _print_record(struct job_record *job_ptr, char *data)
+static int _print_record(struct job_record *job_ptr, char *data)
{
struct tm *ts; /* timestamp decoder */
- static int chk=0;
+ static int chk=0,
+ rc=SLURM_SUCCESS;
ts = xmalloc(sizeof(struct tm));
gmtime_r(&job_ptr->start_time, ts);
+ debug2("jobacct:_print_record, chk=%d, rec starts \"%20s", chk++, data);
slurm_mutex_lock( &logfile_lock );
- debug2("jobacct:_print_record, chk=%d, rec starts \"%20s",
- chk++, data);
- fprintf(LOGFILE,
- "%u %s %04d%02d%02d%02d%02d%02d %u %d.%d - %s\n",
- job_ptr->job_id, job_ptr->partition,
- 1900+(ts->tm_year), 1+(ts->tm_mon), ts->tm_mday,
- ts->tm_hour, ts->tm_min, ts->tm_sec,
- (int) job_ptr->start_time,
- job_ptr->user_id, job_ptr->group_id, data);
+ if (fprintf(LOGFILE,
+ "%u %s %04d%02d%02d%02d%02d%02d %u %d.%d - %s\n",
+ job_ptr->job_id, job_ptr->partition,
+ 1900+(ts->tm_year), 1+(ts->tm_mon), ts->tm_mday,
+ ts->tm_hour, ts->tm_min, ts->tm_sec,
+ (int) job_ptr->start_time,
+ job_ptr->user_id, job_ptr->group_id, data)
+ < 0)
+ rc=SLURM_ERROR;
fdatasync(LOGFILE_FD);
slurm_mutex_unlock( &logfile_lock );
xfree(ts);
+ return rc;
}
@@ -582,27 +594,30 @@ int slurmd_jobacct_init(char *job_acct_parameters)
int slurmd_jobacct_jobstep_launched(slurmd_job_t *job)
{
_jrec_t *jrec;
-
+ int rc=SLURM_SUCCESS;
+
+ debug3("slurmd_jobacct_jobstep_launched() called");
jrec = _alloc_jrec(job);
jrec->nodeid = job->nodeid;
debug2("jobacct(%d): in slurmd_jobacct_jobstep_launched with %d cpus,"
" node0,1=%s,%s, this is node %d of %d",
getpid(), job->cpus, jrec->node0, jrec->node1,
job->nodeid, job->nnodes);
- _send_data_to_mynode(LAUNCH, jrec);
+ rc = _send_data_to_mynode(LAUNCH, jrec);
xfree(jrec);
- return SLURM_SUCCESS;
+ return rc;
}
int slurmd_jobacct_jobstep_terminated(slurmd_job_t *job)
{
_jrec_t *jrec;
+ int rc=SLURM_SUCCESS;
jrec = _alloc_jrec(job);
jrec->status = job->smgr_status;
- _send_data_to_mynode(TERMINATE, jrec);
+ rc = _send_data_to_mynode(TERMINATE, jrec);
xfree(jrec);
- return SLURM_SUCCESS;
+ return rc;
}
/*
@@ -632,6 +647,7 @@ int slurmd_jobacct_smgr(void)
int slurmd_jobacct_task_exit(slurmd_job_t *job, pid_t pid, int status, struct rusage *rusage)
{
_jrec_t *jrec;
+ int rc=SLURM_SUCCESS;
debug2("slurmd_jobacct_task_exit for job %u.%u,"
" node %d, status=%d",
@@ -647,9 +663,9 @@ int slurmd_jobacct_task_exit(slurmd_job_t *job, pid_t pid, int status, struct ru
jrec->max_psize = max_psize;
jrec->max_vsize = max_vsize;
}
- _send_data_to_mynode(TASKDATA, jrec);
+ rc = _send_data_to_mynode(TASKDATA, jrec);
xfree(jrec);
- return SLURM_SUCCESS;
+ return rc;
}
/*
@@ -780,9 +796,8 @@ static void _get_node_01_names( char *node0, char *node1, char **env ) {
int i=0,
slen;
const char *env_var = "SLURM_NODELIST=";
-
slen = strlen(env_var);
- while (*(hname=env[i++])) {
+ while (env && env[i+1] && *(hname=env[i++])) {
if (strncmp(env_var, hname, slen)==0) {
hname += slen;
hl = hostlist_create(hname);
@@ -794,7 +809,10 @@ static void _get_node_01_names( char *node0, char *node1, char **env ) {
hostlist_destroy(hl);
goto done;
}
- strncpy(node0, "NOTFOUND", HOST_NAME_MAX);}
+ /* Either user cleared SLURM_NODELIST or
+ * never set (as is case with POE on AIX) */
+ strncpy(node0, NOT_FOUND, HOST_NAME_MAX);
+ }
done:
debug2("jobacct(%d): node0 is \"%s\"\n", getpid(), node0);
return;
@@ -1120,8 +1138,8 @@ static void _process_node0_msg(_jrec_t *nrec) {
hrec->rusage.ru_utime.tv_usec);
jrec = _get_jrec_by_jobstep(hrec->jobid, hrec->stepid);
if (jrec == NULL) {
- error("jobacct(%i): in process_node0_msg, "
- "did not find %i.%i record",
+ error("jobacct(%d): in process_node0_msg, "
+ "did not find %u.%u record",
getpid(), hrec->jobid, hrec->stepid);
xfree(hrec);
return;
@@ -1133,7 +1151,7 @@ static void _process_node0_msg(_jrec_t *nrec) {
}
xfree(hrec);
--jrec->not_reported;
- debug2("jobacct(%i): not_reported=%i after node0 message, "
+ debug2("jobacct(%d): not_reported=%d after node0 message, "
"cum. utime=%d.%06d",
getpid(), jrec->not_reported,
jrec->rusage.ru_utime.tv_sec,
@@ -1159,8 +1177,8 @@ static void _remove_jrec_from_list(uint32_t jobid, uint32_t stepid)
while ((jrec = list_next(i))) {
if ( (jrec->jobid == jobid )
&& (jrec->stepid == stepid)) {
- debug2("jobacct(%i): in _remove_jrec_from_list, "
- "found %i.%i record",
+ debug2("jobacct(%d): in _remove_jrec_from_list, "
+ "found %u.%u record",
getpid(), jobid, stepid);
break;
}
@@ -1178,14 +1196,16 @@ static void _remove_jrec_from_list(uint32_t jobid, uint32_t stepid)
*/
static int _send_data_to_mynode(_mynode_msg_type_t msgtype, _jrec_t *jrec) {
- slurm_msg_t msg;
+ slurm_msg_t *msg,
+ *retmsg;
jobacct_msg_t jmsg;
_mynode_msg_t *node_msg;
_stats_msg_t stats;
int rc,
+ retry,
s_port;
- debug2("jobacct(%i): in _send_data_to_mynode(msgtype %i, job %i)",
+ debug2("jobacct(%d): in _send_data_to_mynode(msgtype %d, job %u)",
getpid(), msgtype, jrec->jobid);
stats.msg_type = htonl(TO_MYNODE);
stats.jobid = htonl(jrec->jobid);
@@ -1202,23 +1222,33 @@ static int _send_data_to_mynode(_mynode_msg_type_t msgtype, _jrec_t *jrec) {
jmsg.len = sizeof(_mynode_msg_t);
jmsg.data = (char*)&stats;
/* finally, set up the slurm_msg */
+ msg = xmalloc(sizeof(slurm_msg_t));
+ retmsg = xmalloc(sizeof(slurm_msg_t));
s_port=slurm_get_slurmd_port();
- slurm_set_addr(&msg.address, s_port, "localhost");
- msg.msg_type = MESSAGE_JOBACCT_DATA;
- msg.data = &jmsg;
- msg.data_size = sizeof(jobacct_msg_t);
- debug2("jobacct(%i): attempting send_only_node_msg(msg, %i, localhost)"
- " for job %i.%i",
+ slurm_set_addr(&msg->address, s_port, "localhost");
+ msg->msg_type = MESSAGE_JOBACCT_DATA;
+ msg->data = &jmsg;
+ msg->data_size = sizeof(jobacct_msg_t);
+ debug2("jobacct(%d): attempting send_recv_node_msg(msg, %d, localhost)"
+ " for job %u.%u",
getpid(), s_port, jrec->jobid, jrec->stepid);
- rc = slurm_send_only_node_msg(&msg);
+ for (retry=0; retry<MAX_SEND_RETRIES; retry++) {
+ if (( rc = slurm_send_recv_node_msg(msg, retmsg, 0)) >=0 )
+ break;
+ sleep(1+(rand()/(RAND_MAX/MAX_SEND_RETRY_DELAY)));
+ /* pause 1-MAX_SEND_RETRY_DELAY seconds */
+ }
+ slurm_free_cred(retmsg->cred);
if (rc<0)
- error("jobacct(%i): _send_data_to_mynode(msg, %i, localhost)"
- " says %i (%m)",
- getpid(), s_port, rc);
+ error("jobacct(%d): _send_data_to_mynode(msg, %d, localhost)"
+ " says %d (%m) after %d tries",
+ getpid(), s_port, rc, retry);
else
- debug2("jobacct(%i): _send_data_to_mynode(msg, %i, localhost) "
+ debug2("jobacct(%d): _send_data_to_mynode(msg, %d, localhost) "
"succeeded",
getpid(), s_port);
+ xfree(msg);
+ xfree(retmsg);
return rc;
}
@@ -1229,17 +1259,25 @@ static int _send_data_to_mynode(_mynode_msg_type_t msgtype, _jrec_t *jrec) {
*/
static int _send_data_to_node_0(_jrec_t *jrec) {
- slurm_msg_t msg;
+ slurm_msg_t *msg,
+ *retmsg;
jobacct_msg_t jmsg;
_stats_msg_t stats;
int rc,
+ retry,
s_port;
- debug2("jobacct(%i): in _send_data_to_node_0(job %i), nodes0,1=%s,%s"
+ if (!strcmp(jrec->node0, NOT_FOUND)) {
+ error("jobacct(%d): job %d has no node0");
+ return SLURM_SUCCESS; /* can't do anything here */
+ }
+
+ debug2("jobacct(%d): in _send_data_to_node_0(job %u), nodes0,1=%s,%s"
", utime=%d.%06d",
getpid(), jrec->jobid, jrec->node0, jrec->node1,
jrec->rusage.ru_utime.tv_sec,
jrec->rusage.ru_utime.tv_usec);
+
/* make a stats_msg */
stats.msg_type = htonl(TO_NODE0);
stats.jobid = htonl(jrec->jobid);
@@ -1251,22 +1289,32 @@ static int _send_data_to_node_0(_jrec_t *jrec) {
jmsg.len = sizeof(_stats_msg_t);
jmsg.data = (char*)&stats;
/* finally, set up the slurm_msg */
+ msg = xmalloc(sizeof(slurm_msg_t));
+ retmsg = xmalloc(sizeof(slurm_msg_t));
s_port=slurm_get_slurmd_port();
- slurm_set_addr(&msg.address, s_port, jrec->node0);
- msg.msg_type = MESSAGE_JOBACCT_DATA;
- msg.data = &jmsg;
- msg.data_size = sizeof(jobacct_msg_t);
- debug2("jobacct(%i): attempting send_only_node_msg(msg, %i, %s)",
+ slurm_set_addr(&msg->address, s_port, jrec->node0);
+ msg->msg_type = MESSAGE_JOBACCT_DATA;
+ msg->data = &jmsg;
+ msg->data_size = sizeof(jobacct_msg_t);
+ debug2("jobacct(%d): attempting send_recv_node_msg(msg, %d, %s)",
getpid(), s_port, jrec->node0);
- rc = slurm_send_only_node_msg(&msg);
+ for (retry=0; retry<MAX_SEND_RETRIES; retry++) {
+ if ((rc = slurm_send_recv_node_msg(msg, retmsg, 0))>=0 )
+ break;
+ sleep(1+(rand()/(RAND_MAX/MAX_SEND_RETRY_DELAY)));
+ /* pause 1-MAX_SEND_RETRY_DELAY seconds */
+ }
+ slurm_free_cred(retmsg->cred);
if (rc<0)
- error("jobacct(%i): _send_data_to_node_0(msg, %i, %s)"
- " says %i (%m)",
- getpid(), s_port, jrec->node0);
+ error("jobacct(%d): _send_data_to_node_0(msg, %d, %s)"
+ " says %d (%m) after %d tries",
+ getpid(), s_port, jrec->node0, rc, retry);
else
- debug2("jobacct(%i): _send_data_to_node_0(msg, %i, %s)"
+ debug2("jobacct(%d): _send_data_to_node_0(msg, %d, %s)"
" succeeded",
getpid(), s_port, jrec->node0);
+ xfree(msg);
+ xfree(retmsg);
return rc;
}
@@ -1286,7 +1334,7 @@ static int _send_data_to_slurmctld(_jrec_t *jrec) {
struct tm ts; /* timestamp decoder */
time_t now;
- debug2("jobacct(%i): in _send_data_to_slurmctld", getpid());
+ debug2("jobacct(%d): in _send_data_to_slurmctld", getpid());
if (jrec->status)
comp_status = "F";
else
diff --git a/src/sacct/sacct.c b/src/sacct/sacct.c
index 85ee30a51743995e48dcda243a3f1a434e5bd58b..e5bd29bdea9674bd76138b02e84d087a70c32342 100644
--- a/src/sacct/sacct.c
+++ b/src/sacct/sacct.c
@@ -25,15 +25,8 @@
/*
* HISTORY
* $Log$
- * Revision 1.4 2005/06/03 23:33:48 jette
- * Add info on new slurm.conf parameters for job accounting.
- * Sacct now reports right away if accounting is disabled.
- *
- * Revision 1.3 2005/06/03 20:35:52 jette
- * Fix several sacct bugs.
- *
- * Revision 1.2 2005/06/01 18:35:11 jette
- * Fix a couple of minor issues for clean build on AIX with new HP code.
+ * Revision 1.5 2005/06/11 00:49:43 jette
+ * Get all the latest accounting software patches.
*
* Revision 1.1 2005/06/01 17:26:11 jette
* Extensive mods checked it for HP work, see NEWS for details.
@@ -184,60 +177,60 @@
/* Map field names to positions */
/* Fields common to all records */
-#define F_JOB 0
-#define F_PARTITION 1
-#define F_SUBMITTED 2
-#define F_STARTTIME 3
-#define F_UIDGID 4
-#define F_RESERVED1 5
-#define F_RECTYPE 6
-#define F_RECVERSION 7
-#define F_NUMFIELDS 8
+#define F_JOB 0
+#define F_PARTITION 1
+#define F_SUBMITTED 2
+#define F_STARTTIME 3
+#define F_UIDGID 4
+#define F_RESERVED1 5
+#define F_RECTYPE 6
+#define F_RECVERSION 7
+#define F_NUMFIELDS 8
/* JOB_START fields */
-#define F_UID 9
-#define F_GID 10
-#define F_JOBNAME 11
-#define F_BATCH 12
-#define F_PRIORITY 13
-#define F_NCPUS 14
-#define F_NODES 15
+#define F_UID 9
+#define F_GID 10
+#define F_JOBNAME 11
+#define F_BATCH 12
+#define F_PRIORITY 13
+#define F_NCPUS 14
+#define F_NODES 15
/* JOB_STEP fields */
-#define F_JOBSTEP 9
-#define F_FINISHED 10
-#define F_CSTATUS 11
-#define F_ERROR 12
-#define F_NPROCS 13
+#define F_JOBSTEP 9
+#define F_FINISHED 10
+#define F_CSTATUS 11
+#define F_ERROR 12
+#define F_NPROCS 13
/*define F_NCPUS 14 (defined above) */
-#define F_ELAPSED 15
-#define F_CPU_SEC 16
-#define F_CPU_USEC 17
-#define F_USER_SEC 18
-#define F_USER_USEC 19
-#define F_SYS_SEC 20
-#define F_SYS_USEC 21
-#define F_RSS 22
-#define F_IXRSS 23
-#define F_IDRSS 24
-#define F_ISRSS 25
-#define F_MINFLT 26
-#define F_MAJFLT 27
-#define F_NSWAP 28
-#define F_INBLOCKS 29
-#define F_OUBLOCKS 30
-#define F_MSGSND 31
-#define F_MSGRCV 32
-#define F_NSIGNALS 33
-#define F_NVCSW 34
-#define F_NIVCSW 35
-#define F_VSIZE 36
-#define F_PSIZE 37
+#define F_ELAPSED 15
+#define F_CPU_SEC 16
+#define F_CPU_USEC 17
+#define F_USER_SEC 18
+#define F_USER_USEC 19
+#define F_SYS_SEC 20
+#define F_SYS_USEC 21
+#define F_RSS 22
+#define F_IXRSS 23
+#define F_IDRSS 24
+#define F_ISRSS 25
+#define F_MINFLT 26
+#define F_MAJFLT 27
+#define F_NSWAP 28
+#define F_INBLOCKS 29
+#define F_OUBLOCKS 30
+#define F_MSGSND 31
+#define F_MSGRCV 32
+#define F_NSIGNALS 33
+#define F_NVCSW 34
+#define F_NIVCSW 35
+#define F_VSIZE 36
+#define F_PSIZE 37
/* JOB_COMPLETION fields */
-#define F_TOT_ELAPSED 9
-/*define F_FINISHED 10 */
-/*define F_CSTATUS 11 */
+#define F_TOT_ELAPSED 9
+/*define F_FINISHED 10 */
+/*define F_CSTATUS 11 */
/* On output, use fields 12-37 from JOB_STEP */
@@ -1532,7 +1525,7 @@ void getOptions(int argc, char **argv)
&& (stat(opt_filein, &stat_buf) != 0)) {
fprintf(stderr, "SLURM accounting is disabled\n");
exit(1);
- }
+ }
free(acct_type);
/* specific partitions requested? */
@@ -2630,6 +2623,7 @@ char *prefix_filename(char *path, char *prefix) {
i++;
*out = 0;
strncpy(out, path, i);
+ out[i] = 0;
strcat(out, prefix);
strcat(out, path+i);
return(out);
diff --git a/src/slurmctld/proc_req.c b/src/slurmctld/proc_req.c
index 2029d942980e44e359dd752dee827270a71fdeed..a3f9422280ec217a28c950305c963d468a877d4d 100644
--- a/src/slurmctld/proc_req.c
+++ b/src/slurmctld/proc_req.c
@@ -243,8 +243,16 @@ void slurmctld_req (slurm_msg_t * msg)
/* Note: No data to free */
break;
case MESSAGE_JOBACCT_DATA:
- g_slurm_jobacct_process_message(msg);
- slurm_free_jobacct_msg(msg->data);
+ {
+ int rc=SLURM_SUCCESS;
+ debug3("proc_req received jobacct message");
+ slurm_send_rc_msg(msg,rc); /* ACK the message */
+ debug3("proc_req sent rc=%d to jobacct", rc);
+ rc=g_slurm_jobacct_process_message(msg);
+ debug3("proc_req slurm_jobacct_process_message rc=%d",
+ rc);
+ slurm_free_jobacct_msg(msg->data);
+ }
break;
default:
error("invalid RPC msg_type=%d", msg->msg_type);
@@ -1792,11 +1800,11 @@ static void _slurm_rpc_job_ready(slurm_msg_t * msg)
END_TIMER;
if (error_code) {
- debug2("_slurm_rpc_job_ready: %s",
+ debug("_slurm_rpc_job_ready: %s",
slurm_strerror(error_code));
slurm_send_rc_msg(msg, error_code);
} else {
- debug2("_slurm_rpc_job_ready(%u)=%d %s", id_msg->job_id,
+ debug("_slurm_rpc_job_ready(%u)=%d %s", id_msg->job_id,
result, TIME_STR);
rc_msg.return_code = result;
response_msg.address = msg->address;