diff --git a/NEWS b/NEWS
index 8df683795ac19dcb46f017ece4732cdba7f27f13..b3da22b7e71913bc7e6546251392ef64ddf08330 100644
--- a/NEWS
+++ b/NEWS
@@ -30,6 +30,8 @@ documents those changes that are of interest to users and admins.
exit with a return code of zero and others are killed with SIGKILL. Only an
exit code of zero did this.
-- task/cgroup make TaskAffinity=hard bind to threads instead of cores.
+ -- Add a new pending reason WaitingForScheduling which indicates a job has
+ not been processed by the scheduler yet.
* Changes in Slurm 14.03.0
==========================
diff --git a/doc/man/man1/scontrol.1 b/doc/man/man1/scontrol.1
index 864f9f9a274e91425cdb2772ddcd771447706aa7..7d1f93f3360d1b05fd02e27ae3436f5917bd842b 100644
--- a/doc/man/man1/scontrol.1
+++ b/doc/man/man1/scontrol.1
@@ -1541,7 +1541,7 @@ JobId=71701 Name=hostname
.br
Priority=66264 Account=none QOS=normal WCKey=*123
.br
- JobState=COMPLETED Reason=None Dependency=(null)
+ JobState=COMPLETED Reason=WaitingForScheduling Dependency=(null)
.br
TimeLimit=UNLIMITED Requeue=1 Restarts=0 BatchFlag=0 ExitCode=0:0
.br
diff --git a/doc/man/man1/squeue.1 b/doc/man/man1/squeue.1
index 027152e358a24101045726b4ee146e9afbd634cc..479f582ae6c2d79103dba636b0f21851c1b340ba 100644
--- a/doc/man/man1/squeue.1
+++ b/doc/man/man1/squeue.1
@@ -481,8 +481,8 @@ one of those reasons is displayed.
\fBDependency\fR
This job is waiting for a dependent job to complete.
.TP
-\fBNone\fR
-No reason is set for this job.
+\fBWaitingForScheduling\fR
+The job is waiting for the next scheduling session.
.TP
\fBPartitionDown\fR
The partition required by this job is in a DOWN state.
diff --git a/slurm/slurm.h.in b/slurm/slurm.h.in
index b893f1fef22af88f03644c945c84030a9796f0dd..0ec53bf9920fa1b2b613d7761ff68552226ff888 100644
--- a/slurm/slurm.h.in
+++ b/slurm/slurm.h.in
@@ -295,7 +295,7 @@ enum job_states {
* system efficiency */
enum job_state_reason {
/* Reasons for job to be pending */
- WAIT_NO_REASON = 0, /* not set or job not pending */
+ WAIT_FOR_SCHED = 0, /* job is waiting for scheduling */
WAIT_PRIORITY, /* higher priority jobs exist */
WAIT_DEPENDENCY, /* dependent job has not completed */
WAIT_RESOURCES, /* required resources not available */
diff --git a/src/common/slurm_protocol_defs.c b/src/common/slurm_protocol_defs.c
index db6946343f2d86aa531e4fe0ada525a16adce3c1..c853a039abde36a6b66d7c492a6e4d9275cec003 100644
--- a/src/common/slurm_protocol_defs.c
+++ b/src/common/slurm_protocol_defs.c
@@ -984,8 +984,8 @@ extern void slurm_free_spank_env_responce_msg(spank_env_responce_msg_t *msg)
extern char *job_reason_string(enum job_state_reason inx)
{
switch (inx) {
- case WAIT_NO_REASON:
- return "None";
+ case WAIT_FOR_SCHED:
+ return "WaitingForScheduling";
case WAIT_PROLOG:
return "Prolog";
case WAIT_PRIORITY:
diff --git a/src/plugins/sched/wiki/start_job.c b/src/plugins/sched/wiki/start_job.c
index eb232e4f09dd060891b9161dfdac676c369e36bd..5d2b8d9c2bfeb30836df0b6519ef26a0093d5dcd 100644
--- a/src/plugins/sched/wiki/start_job.c
+++ b/src/plugins/sched/wiki/start_job.c
@@ -251,7 +251,7 @@ static int _start_job(uint32_t jobid, int task_cnt, char *hostlist,
old_task_cnt = job_ptr->details->min_cpus;
job_ptr->details->min_cpus = MAX(task_cnt, old_task_cnt);
job_ptr->priority = 100000000;
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
fini: unlock_slurmctld(job_write_lock);
if (rc)
diff --git a/src/plugins/sched/wiki2/start_job.c b/src/plugins/sched/wiki2/start_job.c
index 3693c8f2578a1f8dcc9005689a9b0e2b8dc20111..0fb0c1097d269aa6041e82e3f3e4488b4565a02c 100644
--- a/src/plugins/sched/wiki2/start_job.c
+++ b/src/plugins/sched/wiki2/start_job.c
@@ -306,7 +306,7 @@ static int _start_job(uint32_t jobid, int task_cnt, char *hostlist,
old_task_cnt = job_ptr->details->min_cpus;
job_ptr->details->min_cpus = MAX(task_cnt, old_task_cnt);
job_ptr->priority = 100000000;
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
fini: unlock_slurmctld(job_write_lock);
if (rc)
diff --git a/src/slurmctld/acct_policy.c b/src/slurmctld/acct_policy.c
index 8bb8785cee8ed0b812acb9b745f601521d3eed78..eb5d50bd8a26eb687b53954d1b4620cd5b4c8840 100644
--- a/src/slurmctld/acct_policy.c
+++ b/src/slurmctld/acct_policy.c
@@ -1059,7 +1059,7 @@ extern bool acct_policy_job_runnable_pre_select(struct job_record *job_ptr)
/* clear old state reason */
if (!acct_policy_job_runnable_state(job_ptr))
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
assoc_mgr_lock(&locks);
qos_ptr = job_ptr->qos_ptr;
@@ -1341,7 +1341,7 @@ extern bool acct_policy_job_runnable_post_select(
/* clear old state reason */
if (!acct_policy_job_runnable_state(job_ptr))
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
job_cpu_time_limit = (uint64_t)job_ptr->time_limit * (uint64_t)cpu_cnt;
diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index 52b7448ccab233d07c3bfa16ea19510a77669a97..fda1e2d089bdb554d4dec81b320096de58eb2f44 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -3203,7 +3203,7 @@ static int _select_nodes_parts(struct job_record *job_ptr, bool test_only,
job_ptr->part_ptr = part_ptr;
debug2("Try job %u on next partition %s",
job_ptr->job_id, part_ptr->name);
- if (job_limits_check(&job_ptr, false) != WAIT_NO_REASON)
+ if (job_limits_check(&job_ptr, false) != WAIT_FOR_SCHED)
continue;
rc = select_nodes(job_ptr, test_only,
select_node_bitmap);
@@ -3213,7 +3213,7 @@ static int _select_nodes_parts(struct job_record *job_ptr, bool test_only,
}
list_iterator_destroy(iter);
} else {
- if (job_limits_check(&job_ptr, false) != WAIT_NO_REASON)
+ if (job_limits_check(&job_ptr, false) != WAIT_FOR_SCHED)
test_only = true;
rc = select_nodes(job_ptr, test_only, select_node_bitmap);
}
@@ -3688,7 +3688,7 @@ extern int prolog_complete(uint32_t job_id, bool requeue,
error("Prolog launch failure, JobId=%u", job_ptr->job_id);
}
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
return SLURM_SUCCESS;
}
@@ -4250,7 +4250,7 @@ _valid_job_part_qos(struct part_record *part_ptr, slurmdb_qos_rec_t *qos_ptr)
* IN job_ptr - pointer to job table entry.
* IN check_min_time - if true test job's minimum time limit,
* otherwise test maximum time limit
- * RET WAIT_NO_REASON on success, fail status otherwise.
+ * RET WAIT_FOR_SCHED on success, fail status otherwise.
*/
extern int job_limits_check(struct job_record **job_pptr, bool check_min_time)
{
@@ -4276,7 +4276,7 @@ extern int job_limits_check(struct job_record **job_pptr, bool check_min_time)
assoc_ptr = job_ptr->assoc_ptr;
if (!detail_ptr) { /* To prevent CLANG error */
fatal("job %u has NULL details_ptr", job_ptr->job_id);
- return WAIT_NO_REASON;
+ return WAIT_FOR_SCHED;
}
#ifdef HAVE_BG
@@ -4291,7 +4291,7 @@ extern int job_limits_check(struct job_record **job_pptr, bool check_min_time)
part_max_nodes = part_ptr->max_nodes;
#endif
- fail_reason = WAIT_NO_REASON;
+ fail_reason = WAIT_FOR_SCHED;
if (check_min_time && job_ptr->time_min)
time_check = job_ptr->time_min;
@@ -7499,7 +7499,7 @@ static bool _top_priority(struct job_record *job_ptr)
job_ptr->state_reason = WAIT_HELD;
xfree(job_ptr->state_desc);
}
- } else if (job_ptr->state_reason == WAIT_NO_REASON) {
+ } else if (job_ptr->state_reason == WAIT_FOR_SCHED) {
job_ptr->state_reason = WAIT_PRIORITY;
xfree(job_ptr->state_desc);
}
@@ -8393,7 +8393,7 @@ int update_job(job_desc_msg_t * job_specs, uid_t uid)
set_job_prio(job_ptr);
info("sched: update_job: releasing user hold "
"for job_id %u", job_specs->job_id);
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
job_ptr->job_state &= ~JOB_SPECIAL_EXIT;
xfree(job_ptr->state_desc);
} else if (authorized ||
@@ -8421,7 +8421,7 @@ int update_job(job_desc_msg_t * job_specs, uid_t uid)
xfree(job_ptr->state_desc);
} else if ((job_ptr->state_reason == WAIT_HELD) ||
(job_ptr->state_reason == WAIT_HELD_USER)) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
job_ptr->job_state &= ~JOB_SPECIAL_EXIT;
xfree(job_ptr->state_desc);
}
@@ -8925,7 +8925,7 @@ int update_job(job_desc_msg_t * job_specs, uid_t uid)
goto fini;
fail_reason = job_limits_check(&job_ptr, false);
- if (fail_reason != WAIT_NO_REASON) {
+ if (fail_reason != WAIT_FOR_SCHED) {
if (fail_reason == WAIT_QOS_THRES)
error_code = ESLURM_QOS_THRES;
else if ((fail_reason == WAIT_PART_TIME_LIMIT) ||
@@ -8940,7 +8940,7 @@ int update_job(job_desc_msg_t * job_specs, uid_t uid)
return error_code;
} else if ((job_ptr->state_reason != WAIT_HELD) &&
(job_ptr->state_reason != WAIT_HELD_USER)) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
}
#ifdef HAVE_BG
@@ -10149,14 +10149,14 @@ extern bool job_independent(struct job_record *job_ptr, int will_run)
/* Job is eligible to start now */
if (job_ptr->state_reason == WAIT_DEPENDENCY) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
}
if ((detail_ptr && (detail_ptr->begin_time == 0) &&
(job_ptr->priority != 0))) {
detail_ptr->begin_time = now;
} else if (job_ptr->state_reason == WAIT_TIME) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
}
return true;
diff --git a/src/slurmctld/job_scheduler.c b/src/slurmctld/job_scheduler.c
index 9d4a18020d8daa29e0b5db79e6169402a10f444d..9980e592ee09346c93171b5ab68d0e16ac2ebd60 100644
--- a/src/slurmctld/job_scheduler.c
+++ b/src/slurmctld/job_scheduler.c
@@ -187,7 +187,7 @@ static bool _job_runnable_test1(struct job_record *job_ptr, bool clear_start)
#ifdef HAVE_FRONT_END
/* At least one front-end node up at this point */
if (job_ptr->state_reason == WAIT_FRONT_END) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
last_job_update = time(NULL);
}
@@ -236,12 +236,12 @@ static bool _job_runnable_test2(struct job_record *job_ptr, bool check_min_time)
reason = job_limits_check(&job_ptr, check_min_time);
if ((reason != job_ptr->state_reason) &&
- ((reason != WAIT_NO_REASON) ||
+ ((reason != WAIT_FOR_SCHED) ||
(!part_policy_job_runnable_state(job_ptr)))) {
job_ptr->state_reason = reason;
xfree(job_ptr->state_desc);
}
- if (reason != WAIT_NO_REASON)
+ if (reason != WAIT_FOR_SCHED)
return false;
return true;
}
@@ -275,7 +275,7 @@ extern List build_job_queue(bool clear_start, bool backfill)
list_next(part_iterator))) {
job_ptr->part_ptr = part_ptr;
reason = job_limits_check(&job_ptr, backfill);
- if ((reason != WAIT_NO_REASON) &&
+ if ((reason != WAIT_FOR_SCHED) &&
(reason != job_ptr->state_reason) &&
(!part_policy_job_runnable_state(job_ptr))){
job_ptr->state_reason = reason;
@@ -284,7 +284,7 @@ extern List build_job_queue(bool clear_start, bool backfill)
/* priority_array index matches part_ptr_list
* position: increment inx*/
inx++;
- if (reason != WAIT_NO_REASON)
+ if (reason != WAIT_FOR_SCHED)
continue;
if (job_ptr->priority_array) {
_job_queue_append(job_queue, job_ptr,
@@ -534,7 +534,7 @@ next_part: part_ptr = (struct part_record *)
continue;
}
}
- if (job_limits_check(&job_ptr, false) != WAIT_NO_REASON)
+ if (job_limits_check(&job_ptr, false) != WAIT_FOR_SCHED)
continue;
/* Test for valid account, QOS and required nodes on each pass */
@@ -550,7 +550,7 @@ next_part: part_ptr = (struct part_record *)
accounting_enforce,
(slurmdb_association_rec_t **)
&job_ptr->assoc_ptr)) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
job_ptr->assoc_id = assoc_rec.id;
last_job_update = now;
@@ -573,7 +573,7 @@ next_part: part_ptr = (struct part_record *)
continue;
} else if (job_ptr->state_reason == FAIL_QOS) {
xfree(job_ptr->state_desc);
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
last_job_update = now;
}
}
@@ -581,7 +581,7 @@ next_part: part_ptr = (struct part_record *)
if ((job_ptr->state_reason == WAIT_QOS_JOB_LIMIT) ||
(job_ptr->state_reason == WAIT_QOS_RESOURCE_LIMIT) ||
(job_ptr->state_reason == WAIT_QOS_TIME_LIMIT)) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
last_job_update = now;
}
@@ -833,7 +833,7 @@ extern int schedule(uint32_t job_limit)
list_next(job_iterator))) {
if (!IS_JOB_PENDING(job_ptr))
continue;
- if ((job_ptr->state_reason != WAIT_NO_REASON) &&
+ if ((job_ptr->state_reason != WAIT_FOR_SCHED) &&
(job_ptr->state_reason != WAIT_RESOURCES) &&
(job_ptr->state_reason != WAIT_NODE_NOT_AVAIL))
continue;
@@ -928,7 +928,7 @@ next_part: part_ptr = (struct part_record *)
if (part_ptr) {
job_ptr->part_ptr = part_ptr;
if (job_limits_check(&job_ptr, false) !=
- WAIT_NO_REASON)
+ WAIT_FOR_SCHED)
continue;
} else {
list_iterator_destroy(part_iterator);
@@ -997,7 +997,7 @@ next_part: part_ptr = (struct part_record *)
}
}
if (found_resv) {
- if (job_ptr->state_reason == WAIT_NO_REASON) {
+ if (job_ptr->state_reason == WAIT_FOR_SCHED) {
job_ptr->state_reason = WAIT_PRIORITY;
xfree(job_ptr->state_desc);
}
@@ -1012,7 +1012,7 @@ next_part: part_ptr = (struct part_record *)
} else if (_failed_partition(job_ptr->part_ptr, failed_parts,
failed_part_cnt)) {
if ((job_ptr->state_reason == WAIT_NODE_NOT_AVAIL)
- || (job_ptr->state_reason == WAIT_NO_REASON)) {
+ || (job_ptr->state_reason == WAIT_FOR_SCHED)) {
job_ptr->state_reason = WAIT_PRIORITY;
xfree(job_ptr->state_desc);
last_job_update = now;
@@ -1039,7 +1039,7 @@ next_part: part_ptr = (struct part_record *)
accounting_enforce,
(slurmdb_association_rec_t **)
&job_ptr->assoc_ptr)) {
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
job_ptr->assoc_id = assoc_rec.id;
last_job_update = now;
@@ -1062,7 +1062,7 @@ next_part: part_ptr = (struct part_record *)
continue;
} else if (job_ptr->state_reason == FAIL_QOS) {
xfree(job_ptr->state_desc);
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
last_job_update = now;
}
}
diff --git a/src/slurmctld/node_scheduler.c b/src/slurmctld/node_scheduler.c
index e404728d1fe91e88303c90f69e6bd47793b1a969..3857980f717af25045787a51bc01530ecac0e188 100644
--- a/src/slurmctld/node_scheduler.c
+++ b/src/slurmctld/node_scheduler.c
@@ -1746,7 +1746,7 @@ extern int select_nodes(struct job_record *job_ptr, bool test_only,
}
/* assign the nodes and stage_in the job */
- job_ptr->state_reason = WAIT_NO_REASON;
+ job_ptr->state_reason = WAIT_FOR_SCHED;
xfree(job_ptr->state_desc);
if (job_ptr->job_resrcs && job_ptr->job_resrcs->nodes)
diff --git a/src/slurmctld/slurmctld.h b/src/slurmctld/slurmctld.h
index bd05aacf982e3f8f79dd5ea238b3106c74dcf18d..2d51510ac1f81424e5bd89708bf3e4ff74392fac 100644
--- a/src/slurmctld/slurmctld.h
+++ b/src/slurmctld/slurmctld.h
@@ -807,7 +807,7 @@ extern struct part_record *create_part_record (void);
* IN job_ptr - pointer to job table entry.
* IN check_min_time - if true test job's minimum time limit,
* otherwise test maximum time limit
- * RET WAIT_NO_REASON on success, fail status otherwise.
+ * RET WAIT_FOR_SCHED on success, fail status otherwise.
*/
extern int job_limits_check(struct job_record **job_pptr, bool check_min_time);