diff --git a/doc/man/man1/scontrol.1 b/doc/man/man1/scontrol.1
index 68b63570efcbc9c57f4942c29cd95e0551419662..eda3bdf783f514b22f026a4c56469ca0c0ec7be4 100644
--- a/doc/man/man1/scontrol.1
+++ b/doc/man/man1/scontrol.1
@@ -1107,9 +1107,11 @@ not possible to also explicitly specify allowed accounts.
.TP
\fICoreCnt\fP=<num>
-Identify number of cores to be reserved. This should only be used for
-reservations that are less than one node in size. Otherwise use the
-\fINodeCnt\fP option described below.
+This option is only suported when SelectType=select/cons_res. Identify number of
+cores to be reserved. If NodeCnt is used, this is the total number of cores to
+reserve where cores per node is CoreCnt/NodeCnt. If a nodelist is used, this
+should be an array of core numbers by node: Nodes=node[1-5] CoreCnt=2,2,3,3,4
+
.TP
\fILicenses\fP=<license>
diff --git a/slurm/slurm.h.in b/slurm/slurm.h.in
index 7795fc833f4597bfd8540fb4c1a5a8635c2fdcf1..1d192d624dbfb3c236ce05a4d7c41ccc3dc1018d 100644
--- a/slurm/slurm.h.in
+++ b/slurm/slurm.h.in
@@ -1790,7 +1790,7 @@ typedef struct resv_desc_msg {
uint16_t flags; /* see RESERVE_FLAG_* above */
char *licenses; /* names of licenses to be reserved */
char *name; /* name of reservation (optional on create) */
- uint32_t core_cnt; /* Count of cores required */
+ uint32_t *core_cnt; /* Count of cores required */
uint32_t *node_cnt; /* Count of nodes required. Specify set of job
* sizes with trailing zero to optimize layout
* for those jobs just specify their total size
diff --git a/src/common/node_select.c b/src/common/node_select.c
index c6cdf4d0dda673142fb542cf43c0bb7b45617317..01beca48821afe20d8d3ec330319796d6bcc5f9b 100644
--- a/src/common/node_select.c
+++ b/src/common/node_select.c
@@ -1244,7 +1244,7 @@ extern int select_g_reconfigure (void)
* RET - nodes selected for use by the reservation
*/
extern bitstr_t * select_g_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
- uint32_t core_cnt, bitstr_t **core_bitmap)
+ uint32_t *core_cnt, bitstr_t **core_bitmap)
{
if (slurm_select_init(0) < 0)
return NULL;
diff --git a/src/common/node_select.h b/src/common/node_select.h
index 608410b3dde861317c1176607447a746481ead2b..ef8990b09e1890350f8094bfea12244f28ea52fb 100644
--- a/src/common/node_select.h
+++ b/src/common/node_select.h
@@ -675,7 +675,7 @@ extern int select_g_step_finish(struct step_record *step_ptr);
* RET - nodes selected for use by the reservation
*/
extern bitstr_t * select_g_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
- uint32_t core_cnt,
+ uint32_t *core_cnt,
bitstr_t **core_bitmap);
/*****************************\
diff --git a/src/common/slurm_protocol_pack.c b/src/common/slurm_protocol_pack.c
index bdb88803b35b246e29ebc431f19f387cc8db7091..ad697fac1f914de0f15523a7f349f799e989c610 100644
--- a/src/common/slurm_protocol_pack.c
+++ b/src/common/slurm_protocol_pack.c
@@ -3215,7 +3215,15 @@ _pack_update_resv_msg(resv_desc_msg_t * msg, Buf buffer,
} else
array_len = 0;
pack32_array(msg->node_cnt, array_len, buffer);
- pack32(msg->core_cnt, buffer);
+ if (msg->core_cnt) {
+ for (array_len = 0; msg->core_cnt[array_len];
+ array_len++) {
+ /* determine array length */
+ }
+ array_len++; /* Include trailing zero */
+ } else
+ array_len = 0;
+ pack32_array(msg->core_cnt, array_len, buffer);
packstr(msg->node_list, buffer);
packstr(msg->features, buffer);
packstr(msg->licenses, buffer);
@@ -3297,7 +3305,15 @@ _unpack_update_resv_msg(resv_desc_msg_t ** msg, Buf buffer,
/* This avoids a pointer to a zero length buffer */
xfree(tmp_ptr->node_cnt);
}
- safe_unpack32(&tmp_ptr->core_cnt, buffer);
+ safe_unpack32_array(&tmp_ptr->core_cnt, &uint32_tmp, buffer);
+ if (uint32_tmp > 0) {
+ /* Must be zero terminated */
+ if (tmp_ptr->core_cnt[uint32_tmp-1] != 0)
+ goto unpack_error;
+ } else {
+ /* This avoids a pointer to a zero length buffer */
+ xfree(tmp_ptr->core_cnt);
+ }
safe_unpackstr_xmalloc(&tmp_ptr->node_list,
&uint32_tmp, buffer);
safe_unpackstr_xmalloc(&tmp_ptr->features,
diff --git a/src/plugins/select/cons_res/select_cons_res.c b/src/plugins/select/cons_res/select_cons_res.c
index 18ace3d46c2a23c9942ecf85e6df44d885770e65..abdc27c015b84168dba11ee408ded86c2e3f74f0 100644
--- a/src/plugins/select/cons_res/select_cons_res.c
+++ b/src/plugins/select/cons_res/select_cons_res.c
@@ -2493,25 +2493,52 @@ bitstr_t *_make_core_bitmap_filtered(bitstr_t *node_map, int filter)
return core_map;
}
-/* Once here, if core_cnt=0, avail_bitmap has nodes not used by any job or
+/* Once here, if core_cnt is NULL, avail_bitmap has nodes not used by any job or
* reservation */
bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
- uint32_t core_cnt, bitstr_t **core_bitmap)
+ uint32_t *core_cnt, bitstr_t **core_bitmap)
{
bitstr_t *sp_avail_bitmap;
char str[300];
- /* Just allowing symetric requests today */
- uint32_t cores_per_node = core_cnt / MAX(node_cnt, 1);
+ uint32_t cores_per_node = 0;
bitstr_t *tmpcore;
+ int total_core_cnt = 0;
+
+ /* We have these cases here:
+ * 1) Reservation requests using just number of nodes
+ * - core_cnt is null
+ * 2) Reservations request using number of nodes + number of cores
+ * 3) Reservations request using node list
+ * - node_cnt is 0
+ * - core_cnt is null
+ * 4) Reservation request using node list + number of cores list
+ * - node_cnt is 0
+ */
+
+ if ((node_cnt) && (core_cnt)) {
+ debug2("reserving %u cores per node in %d nodes",
+ cores_per_node, node_cnt);
+ total_core_cnt = core_cnt[0];
+ cores_per_node = core_cnt[0] / MAX(node_cnt, 1);
+ }
+ if ((!node_cnt) && (core_cnt)) {
+ int num_nodes = bit_set_count(avail_bitmap);
+ int i;
+ bit_fmt(str, (sizeof(str) - 1), avail_bitmap);
+ debug2("Reserving cores from nodes: %s", str);
+ for (i=0; i < num_nodes; i++)
+ total_core_cnt += core_cnt[i];
+ }
- debug2("reserving %u cores per node in %d nodes",
- cores_per_node, node_cnt);
+ debug2("Reservations requires %d cores", total_core_cnt);
sp_avail_bitmap = bit_alloc(bit_size(avail_bitmap));
bit_fmt(str, (sizeof(str) - 1), avail_bitmap);
bit_fmt(str, (sizeof(str) - 1), sp_avail_bitmap);
if (core_cnt) { /* Reservation is using partial nodes */
+ int node_list_inx = 0;
+
debug2("Reservation is using partial nodes");
/* if not NULL = Cores used by other core based reservations
@@ -2527,12 +2554,15 @@ bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
debug2("tmpcore contains just current free cores: %s", str);
bit_and(*core_bitmap, tmpcore); /* clear core_bitmap */
- while (core_cnt) {
+ while (total_core_cnt) {
int inx, coff, coff2;
int i;
int cores_in_node;
int local_cores;
+ if (node_cnt == 0)
+ cores_per_node = core_cnt[node_list_inx];
+
inx = bit_ffs(avail_bitmap);
if (inx < 0) {
info("reservation request can not be satisfied");
@@ -2543,8 +2573,6 @@ bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
debug2("Using node %d", inx);
coff = cr_get_coremap_offset(inx);
- /* TODO: is next right for the last possible node at
- * avail_bitmap? */
coff2 = cr_get_coremap_offset(inx + 1);
local_cores = coff2 - coff;
@@ -2564,14 +2592,17 @@ bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
if (cores_in_node < cores_per_node)
continue;
+ debug2("Using node %d (avail: %d, needed: %d)",
+ inx, cores_in_node, cores_per_node);
+
cores_in_node = 0;
for (i = 0; i < local_cores; i++) {
if (bit_test(tmpcore, coff + i)) {
bit_set(*core_bitmap, coff + i);
- core_cnt--;
+ total_core_cnt--;
cores_in_node++;
if ((cores_in_node == cores_per_node) ||
- (core_cnt == 0))
+ (total_core_cnt == 0))
break;
}
}
@@ -2584,6 +2615,7 @@ bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
} else {
debug2("Reservation NOT using node %d", inx);
}
+ node_list_inx++;
}
FREE_NULL_BITMAP(tmpcore);
@@ -2591,7 +2623,7 @@ bitstr_t *sequential_pick(bitstr_t *avail_bitmap, uint32_t node_cnt,
bit_fmt(str, (sizeof(str) - 1), *core_bitmap);
info("sequential pick using coremap: %s", str);
- if (core_cnt) {
+ if (total_core_cnt) {
info("reservation request can not be satisfied");
FREE_NULL_BITMAP(sp_avail_bitmap);
return NULL;
@@ -2661,7 +2693,7 @@ static int _get_avail_core_in_node(bitstr_t *core_bitmap, int node)
* RET - nodes selected for use by the reservation
*/
extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
- uint32_t core_cnt, bitstr_t **core_bitmap)
+ uint32_t *core_cnt, bitstr_t **core_bitmap)
{
bitstr_t **switches_bitmap; /* nodes on this switch */
bitstr_t **switches_core_bitmap; /* cores on this switch */
@@ -2681,7 +2713,8 @@ extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
xassert(avail_bitmap);
- if (!switch_record_cnt || !switch_record_table) {
+ /* When reservation includes a nodelist we use sequential_pick code */
+ if (!switch_record_cnt || !switch_record_table || !node_cnt) {
return sequential_pick(avail_bitmap, node_cnt, core_cnt,
core_bitmap);
}
@@ -2694,10 +2727,11 @@ extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
*core_bitmap = _make_core_bitmap_filtered(avail_bitmap, 0);
rem_nodes = node_cnt;
- rem_cores = core_cnt;
+ rem_cores = core_cnt[0];
- /* TODO: allowing asymmetric cluster */
- cores_per_node = core_cnt / MAX(node_cnt, 1);
+ /* Assuming symmetric cluster */
+ if(core_cnt)
+ cores_per_node = core_cnt[0] / MAX(node_cnt, 1);
/* Construct a set of switch array entries,
* use the same indexes as switch_record_table in slurmctld */
@@ -2766,7 +2800,7 @@ extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
best_fit_inx = -1;
for (j=0; j<switch_record_cnt; j++) {
if ((switches_node_cnt[j] < rem_nodes) ||
- (core_cnt && (switches_cpu_cnt[j] < core_cnt)))
+ (core_cnt && (switches_cpu_cnt[j] < core_cnt[0])))
continue;
if ((best_fit_inx == -1) ||
(switch_record_table[j].level <
@@ -2800,8 +2834,12 @@ extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
for (j=0; j<switch_record_cnt; j++) {
if (switches_node_cnt[j] == 0)
continue;
- sufficient = (switches_node_cnt[j] >= rem_nodes) &&
- (switches_cpu_cnt[j] >= core_cnt);
+ if(core_cnt)
+ sufficient =
+ (switches_node_cnt[j] >= rem_nodes) &&
+ (switches_cpu_cnt[j] >= core_cnt[0]);
+ else
+ sufficient = switches_node_cnt[j] >= rem_nodes;
/* If first possibility OR */
/* first set large enough for request OR */
/* tightest fit (less resource waste) OR */
@@ -2886,9 +2924,9 @@ fini: for (i=0; i<switch_record_cnt; i++) {
*core_bitmap = bit_alloc(bit_size(exc_core_bitmap));
}
- cores_per_node = core_cnt / MAX(node_cnt, 1);
+ cores_per_node = core_cnt[0] / MAX(node_cnt, 1);
- while (core_cnt) {
+ while (core_cnt[0]) {
uint32_t inx, coff;
int i;
int avail_cores_in_node;
@@ -2897,8 +2935,8 @@ fini: for (i=0; i<switch_record_cnt; i++) {
if ((inx < 0) || (inx > bit_size(avail_bitmap)))
break;
- debug2("Using node inx %d cores_per_node: %d "
- "core_cnt: %d", inx, cores_per_node, core_cnt);
+ debug2("Using node inx %d cores_per_node %d "
+ "core_cnt %d", inx, cores_per_node, core_cnt[0]);
coff = cr_get_coremap_offset(inx);
/* Clear this node from the initial available bitmap */
@@ -2924,12 +2962,12 @@ fini: for (i=0; i<switch_record_cnt; i++) {
for (i = 0; i < cr_node_num_cores[inx]; i++) {
if (!bit_test(exc_core_bitmap, coff + i)) {
bit_set(*core_bitmap, coff + i);
- core_cnt--;
+ core_cnt[0]--;
avail_cores_in_node++;
}
if ((avail_cores_in_node == cores_per_node) ||
- (core_cnt == 0))
+ (core_cnt[0] == 0))
break;
}
@@ -2943,7 +2981,7 @@ fini: for (i=0; i<switch_record_cnt; i++) {
//bit_fmt(str, (sizeof(str) - 1), *core_bitmap);
//info("sequential pick using coremap: %s", str);
- if (core_cnt) {
+ if (core_cnt[0]) {
info("reservation request can not be satisfied");
FREE_NULL_BITMAP(sp_avail_bitmap);
return NULL;
diff --git a/src/plugins/select/cray/select_cray.c b/src/plugins/select/cray/select_cray.c
index 9ef87b10d1b3b6f5eeb713a0948c0105c26f85cd..a97dd67ad275b7c63fc30e69f33efd30c728ee06 100644
--- a/src/plugins/select/cray/select_cray.c
+++ b/src/plugins/select/cray/select_cray.c
@@ -848,7 +848,7 @@ extern int select_p_reconfigure(void)
}
extern bitstr_t * select_p_resv_test(bitstr_t *avail_bitmap, uint32_t node_cnt,
- uint32_t core_cnt, bitstr_t **core_bitmap)
+ uint32_t *core_cnt, bitstr_t **core_bitmap)
{
return other_resv_test(avail_bitmap, node_cnt, core_cnt, core_bitmap);
}
diff --git a/src/scontrol/create_res.c b/src/scontrol/create_res.c
index 78007f517fe59b0a08f7e5aa8c563be8c28a7877..326eb794d688c35b5c232a77c997bb531fc986dd 100644
--- a/src/scontrol/create_res.c
+++ b/src/scontrol/create_res.c
@@ -295,8 +295,29 @@ scontrol_parse_res_options(int argc, char *argv[], const char *msg,
strncasecmp(tag, "CoreCount", MAX(taglen,5)) == 0 ||
strncasecmp(tag, "CPUCnt", MAX(taglen,5)) == 0 ||
strncasecmp(tag, "CPUCount", MAX(taglen,5)) == 0) {
- char *endptr = NULL;
- resv_msg_ptr->core_cnt = strtol(val, &endptr, 10);
+
+ char *endptr = NULL, *core_cnt, *tok, *ptrptr = NULL;
+ int node_inx = 0;
+
+ core_cnt = xstrdup(val);
+ tok = strtok_r(core_cnt, ",", &ptrptr);
+ while (tok) {
+ xrealloc(resv_msg_ptr->core_cnt,
+ sizeof(uint32_t) * (node_inx + 2));
+ resv_msg_ptr->core_cnt[node_inx] =
+ strtol(tok, &endptr, 10);
+ if ((endptr == NULL) ||
+ (endptr[0] != '\0') ||
+ (tok[0] == '\0')) {
+ exit_code = 1;
+ error("Invalid core count %s. %s", argv[i], msg);
+ xfree(core_cnt);
+ return -1;
+ }
+ node_inx++;
+ tok = strtok_r(NULL, ",", &ptrptr);
+ }
+ xfree(core_cnt);
} else if (strncasecmp(tag, "Nodes", MAX(taglen, 5)) == 0) {
resv_msg_ptr->node_list = val;
diff --git a/src/slurmctld/reservation.c b/src/slurmctld/reservation.c
index 9ea669e2f38971c0edcc4003bd4f0315b67a81cf..78bf2f4655cc13e9777e28955137d990e1b9eac8 100644
--- a/src/slurmctld/reservation.c
+++ b/src/slurmctld/reservation.c
@@ -1406,6 +1406,8 @@ static bool _resv_overlap(time_t start_time, time_t end_time,
continue; /* no specific nodes in reservation */
if (!bit_overlap(resv_ptr->node_bitmap, node_bitmap))
continue; /* no overlap */
+ if (!resv_ptr->full_nodes)
+ continue;
for (i=0; ((i<7) && (!rc)); i++) { /* look forward one week */
s_time1 = start_time;
@@ -1579,9 +1581,6 @@ extern int create_resv(resv_desc_msg_t *resv_desc_ptr)
}
}
- if (resv_desc_ptr->core_cnt == NO_VAL)
- resv_desc_ptr->core_cnt = 0;
-
#ifdef HAVE_BG
if (!cnodes_per_bp) {
select_g_alter_node_cnt(SELECT_GET_NODE_SCALING,
@@ -1648,18 +1647,37 @@ extern int create_resv(resv_desc_msg_t *resv_desc_ptr)
}
total_node_cnt = bit_set_count(node_bitmap);
if (!(resv_desc_ptr->flags & RESERVE_FLAG_IGN_JOBS) &&
+ !resv_desc_ptr->core_cnt &&
_job_overlap(resv_desc_ptr->start_time,
resv_desc_ptr->flags, node_bitmap)) {
info("Reservation request overlaps jobs");
rc = ESLURM_NODES_BUSY;
goto bad_parse;
}
- /* We do no allow to request cores with nodelist */
- info("Reservation CoreCnt cleared due to Nodes specification");
- resv_desc_ptr->core_cnt = 0;
+ /* We do allow to request cores with nodelist */
+ if (resv_desc_ptr->core_cnt) {
+ int nodecnt = bit_set_count(node_bitmap);
+ int nodeinx = 0;
+ while (nodeinx < nodecnt) {
+ if (!resv_desc_ptr->core_cnt[nodeinx]) {
+ info("Core count for reservation node \
+ list is not consistent!");
+ rc = ESLURM_INVALID_NODE_NAME;
+ goto bad_parse;
+ }
+ debug2("Requesting %d cores for node_list %d",
+ resv_desc_ptr->core_cnt[nodeinx],
+ nodeinx);
+ nodeinx++;
+ }
+ if ((rc = _select_nodes(resv_desc_ptr, &part_ptr, &node_bitmap,
+ &core_bitmap)) != SLURM_SUCCESS) {
+ goto bad_parse;
+ }
+ }
} else if (((resv_desc_ptr->node_cnt == NULL) ||
(resv_desc_ptr->node_cnt[0] == 0)) &&
- (resv_desc_ptr->core_cnt == 0) &&
+ (!resv_desc_ptr->core_cnt) &&
((resv_desc_ptr->flags & RESERVE_FLAG_LIC_ONLY) == 0)) {
info("Reservation request lacks node specification");
rc = ESLURM_INVALID_NODE_NAME;
@@ -1731,14 +1749,14 @@ extern int create_resv(resv_desc_msg_t *resv_desc_ptr)
resv_ptr->user_not = user_not;
resv_desc_ptr->users = NULL; /* Nothing left to free */
- if (resv_desc_ptr->core_cnt == 0) {
+ if (!resv_desc_ptr->core_cnt) {
debug2("reservation using full nodes");
_set_cpu_cnt(resv_ptr);
resv_ptr->full_nodes = 1;
} else {
debug2("reservation using partial nodes: core count %u",
- resv_desc_ptr->core_cnt);
- resv_ptr->cpu_cnt = resv_desc_ptr->core_cnt;
+ bit_set_count(resv_ptr->core_bitmap));
+ resv_ptr->cpu_cnt = bit_set_count(resv_ptr->core_bitmap);
resv_ptr->full_nodes = 0;
}
@@ -2836,7 +2854,10 @@ static int _select_nodes(resv_desc_msg_t *resv_desc_ptr,
}
/* Start with all nodes in the partition */
- node_bitmap = bit_copy((*part_ptr)->node_bitmap);
+ if (*resv_bitmap)
+ node_bitmap = bit_copy(*resv_bitmap);
+ else
+ node_bitmap = bit_copy((*part_ptr)->node_bitmap);
/* Don't use node already reserved */
if (!(resv_desc_ptr->flags & RESERVE_FLAG_OVERLAP)) {
@@ -3135,7 +3156,7 @@ static bitstr_t *_pick_idle_node_cnt(bitstr_t *avail_bitmap,
if (job_ptr->end_time < resv_desc_ptr->start_time)
continue;
- if (resv_desc_ptr->core_cnt == 0) {
+ if (!resv_desc_ptr->core_cnt) {
bit_not(job_ptr->node_bitmap);
bit_and(avail_bitmap, job_ptr->node_bitmap);
bit_not(job_ptr->node_bitmap);
diff --git a/testsuite/expect/inc3.11.1 b/testsuite/expect/inc3.11.1
index b2698db8c73210aba1b6f40eb85f3b3e941fab48..f34e73bc92756118a17efe56bcfbbee699d6a21d 100644
--- a/testsuite/expect/inc3.11.1
+++ b/testsuite/expect/inc3.11.1
@@ -33,7 +33,7 @@
proc inc3_11_1 {} {
global def_node user_name def_partition exit_code cluster_cpus res_name
- global cons_res_actived
+ global cons_res_actived def_node_name
set num_nodes [available_nodes $def_partition]
@@ -60,6 +60,7 @@ proc inc3_11_1 {} {
{StartTime=now Duration=5 Nodes=$def_node User=$user_name Flags=badtype,ignore_jobs}
{StartTime=now+10minutes EndTime=now Nodes=$def_node User=$user_name Flags=ignore_jobs}
{StartTime=now Duration=5 Nodes=$def_node User=$user_name Licenses=DUMMY_FOR_TESTING Flags=ignore_jobs}
+ {StartTime=now Duration=5 Nodes=$def_node_name\[1\-2\] CoreCnt=1 User=$user_name}
"
#{StartTime=now Duration=5 Nodes=$def_node Account=badaccountname}
if {$cons_res_actived == 1} {
diff --git a/testsuite/expect/inc3.11.9 b/testsuite/expect/inc3.11.9
new file mode 100644
index 0000000000000000000000000000000000000000..113abac4ea8575e86da38295baa10901b32ef3fb
--- /dev/null
+++ b/testsuite/expect/inc3.11.9
@@ -0,0 +1,560 @@
+############################################################################
+# Purpose: Test of SLURM functionality
+# to be called from test3.11
+# Several cases for core based reservations using nodelists
+# Pluging select/cons_res needed
+#
+############################################################################
+# Copyright (C) 2009 Lawrence Livermore National Security
+# Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
+# Written by Dave Bremer <dbremer@llnl.gov>
+# CODE-OCEC-09-009. All rights reserved.
+#
+#
+# This file is part of SLURM, a resource management program.
+# For details, see <http://www.schedmd.com/slurmdocs/>.
+# Please also read the included file: DISCLAIMER.
+#
+# SLURM is free software; you can redistribute it and/or modify it under
+# the terms of the GNU General Public License as published by the Free
+# Software Foundation; either version 2 of the License, or (at your option)
+# any later version.
+#
+# SLURM is distributed in the hope that it will be useful, but WITHOUT ANY
+# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
+# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
+# details.
+#
+# You should have received a copy of the GNU General Public License along
+# with SLURM; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+############################################################################
+
+proc inc3_11_9 {} {
+ global user_name exit_code res_name res_nodes res_nodecnt res_corecnt
+ global bin_rm file_in bin_sleep sbatch number scontrol
+ global alpha_numeric_under scancel
+ global cluster_cpus cores_per_node def_partition
+ global res_nodes res_thread_cnt node_count
+ global def_node_name
+
+ send_user "\n+++++ STARTING TEST 9 +++++\n"
+
+ # Make the job script
+ exec $bin_rm -f $file_in
+ make_bash_script $file_in "$bin_sleep 100"
+
+ # Make a reservation, just to get node size infomation
+ set ret_code [create_res "StartTime=now Duration=1 NodeCnt=1 User=$user_name" 0]
+ if {$ret_code != 0} {
+ send_user "\n\033\[31mFAILURE: Unable to create a valid reservation\033\[m\n"
+ exit $ret_code
+ }
+ # Delete the reservation
+ set ret_code [delete_res $res_name]
+ if {$ret_code != 0} {
+ exit $ret_code
+ }
+
+ set num_nodes [available_nodes $def_partition]
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set thread_res_num [ expr $core_res_num * $res_thread_cnt ]
+
+ # Submit a batch job using half the threads on the nodes
+ set sbatch_pid [spawn $sbatch --nodes=1-$num_nodes --time=30:00 --ntasks-per-node=$thread_res_num --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ exit 1
+ }
+
+ if {[wait_for_job $job_id "RUNNING"] != 0} {
+ send_user "\nFAILURE: job failed to start\n"
+ cancel_job $job_id
+ exit 1
+ }
+ send_user "\nJOB is running as expected\n"
+
+ # Make a reservation using 1 core per node in first 5 nodes
+
+ if {$num_nodes < 5} {
+ send_user "\nWARNING: test can not work with current nodes\n";
+ cancel_job $job_id
+ exit 1
+
+ }
+
+ set ret_code [create_res "StartTime=now Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,1 User=$user_name" 0]
+ if {$ret_code != 0} {
+ send_user "\n\033\[31mFAILURE: Unable to create a valid reservation\033\[m\n"
+ exit $ret_code
+ }
+
+ if {$res_nodecnt != 5} {
+ send_user "\n\033\[31mFAILURE: reservation created with $res_nodecnt nodes when 5 were requested\033\[m\n"
+ set exit_code 1
+ exit $ret_code
+ }
+
+ if {$res_corecnt != 5} {
+ send_user "\n\033\[31mFAILURE: reservation created with $res_corecnt cores when just 5 was requested\033\[m\n"
+ set exit_code 1
+ exit $ret_code
+ }
+
+ send_user "\n\033\[32mSUCCESS: reservation was created as expected\033\[m\n"
+
+ # Delete the reservation
+ set ret_code [delete_res $res_name]
+ if {$ret_code != 0} {
+ cancel_job $job_id
+ exit $ret_code
+ }
+
+ set core_res_num [expr $core_res_num + 1]
+ # Make the reservation using more cores then free in a node
+ set ret_code [create_res "StartTime=now Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,$core_res_num User=$user_name" 1]
+ if {$ret_code != 0} {
+ send_user "\n\033\[32mSUCCESS: Reservation can not be created as expected\033\[m\n"
+ } else {
+ send_user "\n\033\[31mFAILURE: reservation was created when it should have not\033\[m\n"
+ set exit_code 1
+
+ # Delete the reservation
+ set ret_code [delete_res $res_name]
+ }
+
+ # Make the reservation using more cores than free in a node (now)
+ # but those cores being free at reservation start time
+ set ret_code [create_res "StartTime=now+3600 Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,$core_res_num User=$user_name" 0]
+ if {$ret_code != 0} {
+ send_user "\n\033\[31mFAILURE: Reservation can not be created when it should\033\[m\n"
+ set exit_code 1
+ } else {
+ send_user "\n\033\[32mSUCCESS: Reservation can be created as expected\033\[m\n"
+ # Delete the reservation
+ set ret_code [delete_res $res_name]
+ }
+
+ # Make the reservation using more cores than free at reservation start time
+ set ret_code [create_res "StartTime=now+300 Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,$core_res_num User=$user_name" 1]
+ if {$ret_code != 0} {
+ send_user "\n\033\[32mSUCCESS: Reservation can not be created as expected\033\[m\n"
+ } else {
+ send_user "\n\033\[31mFAILURE: Reservation can be created when it should not\033\[m\n"
+ set exit_code 1
+
+ # Delete the reservation
+ set ret_code [delete_res $res_name]
+ }
+
+ cancel_job $job_id
+
+
+ send_user "\n\nLet's check overlapping reservations\n\n";
+
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set total_core_res [ expr $core_res_num * $node_count ]
+ # Make a reservation for all nodes using just half the processor in each node
+ set ret_code [create_res "StartTime=now Duration=60 Nodecnt=$node_count CoreCnt=$total_core_res User=$user_name" 1]
+ if {$ret_code != 0} {
+ send_user "\n\033\[31mFAILURE: Unable to create a valid reservation\033\[m\n"
+ exit $ret_code
+ }
+
+ send_user "\n\033\[32mSUCCESS: reservation was created as expected\033\[m\n"
+
+
+ if {$core_res_num < 2} {
+ send_user "\n\033\[32mWARNING: not enough cores for testing\033\[m\n"
+ set ret_code [delete_res $res_name]
+ }
+
+ set res_name1 "$res_name"
+
+ set total_core_res [ expr $core_res_num + 1 ]
+ # Now creating a reservation using first 5 nodes and more cores per node than available
+ set ret_code [create_res "StartTime=now Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,$total_core_res User=$user_name" 0]
+ if {$ret_code == 0} {
+ send_user "\n\033\[31mFAILURE: reservation was created when it should not\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ exit $ret_code
+ }
+
+ send_user "\n\033\[32mSUCCESS: reservation was not created as expected\033\[m\n"
+
+ # Now creating a reservation using first 5 nodes and just 1 core per node
+ set ret_code [create_res "StartTime=now Duration=60 Nodes=$def_node_name\[1\-5\] CoreCnt=1,1,1,1,1 User=$user_name" 1]
+ if {$ret_code != 0} {
+ send_user "\n\033\[31mFAILURE: Unable to create a valid reservation\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ exit $ret_code
+ }
+
+ send_user "\n\033\[32mSUCCESS: reservation was created as expected\033\[m\n"
+
+ # Submit a batch job: a job using cores availbale in first 5 nodes
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set core_res_num [ expr $core_res_num - 1 ]
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=$core_res_num --nodelist=$def_node_name\[1\-5\] --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is PENDING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ send_user "\n\033\[32mJOB is running as expected\033\[m\n"
+
+ cancel_job $job_id
+
+ # Submit a batch job: a job using more cores than availbale in first 5 nodes
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=$core_res_num --nodelist=$def_node_name\[1\-5\] --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[32m Job is PENDING as expected\033\[m\n"
+ exp_continue
+ }
+ -re "JobState=RUNNING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is RUNNING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ cancel_job $job_id
+
+ # Submit a batch job: a job using cores reserved in first reservation
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=$core_res_num --nodelist=$def_node_name\[1\-5\] --reservation=$res_name1 --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=RUNNING" {
+ send_user "\n\033\[32m Job is RUNNING as expected\033\[m\n"
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is PENDING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ cancel_job $job_id
+
+ # Submit a batch job: a job using more cores than reserved in first reservation
+ set core_res_num [ expr $cores_per_node / 2 ]
+ set core_res_num [ expr $core_res_num * 5 ]
+ set core_res_num [ expr $core_res_num + 1 ]
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=$core_res_num --nodelist=$def_node_name\[1\-5\] --reservation=$res_name1 --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[32m Job is PENDING as expected\033\[m\n"
+ exp_continue
+ }
+ -re "JobState=RUNNING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is RUNNING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ cancel_job $job_id
+
+ # Submit a batch job: a job using cores reserved in second reservation
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=1 --nodelist=$def_node_name\[1\-5\] --reservation=$res_name --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=RUNNING" {
+ send_user "\n\033\[32m Job is RUNNING as expected\033\[m\n"
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is PENDING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ cancel_job $job_id
+
+ # Submit a batch job: a job using more cores than reserved in second reservation
+ set sbatch_pid [spawn $sbatch --ntasks-per-node=2 --nodelist=$def_node_name\[1\-5\] --reservation=$res_name --output=/dev/null $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id $expect_out(1,string)
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: sbatch not responding\033\[m\n"
+ slow_kill $sbatch_pid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if {$job_id == 0} {
+ send_user "\n\033\[31mFAILURE: batch submit failure\033\[m\n"
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ sleep 1
+ # Show the job, make sure reservation tag is right
+ spawn $scontrol show job $job_id
+ expect {
+ -re "Invalid job id specified" {
+ send_user "\n\033\[31mFAILURE: Job $job_id not found\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "JobState=PENDING" {
+ send_user "\n\033\[32m Job is PENDING as expected\033\[m\n"
+ exp_continue
+ }
+ -re "JobState=RUNNING" {
+ send_user "\n\033\[31mFAILURE: Job $job_id is RUNNING\033\[m\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\n\033\[31mFAILURE: scontrol not responding\033\[m\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+ if { $exit_code == 1 } {
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+ exit 1
+ }
+
+ cancel_job $job_id
+
+ set ret_code [delete_res $res_name1]
+ set ret_code [delete_res $res_name]
+
+}
diff --git a/testsuite/expect/test3.11 b/testsuite/expect/test3.11
index 824c93675bf939566c31025434e2288d850d2713..354d0b232fc5df9ec540aa46a1b09a8b63e23cd3 100755
--- a/testsuite/expect/test3.11
+++ b/testsuite/expect/test3.11
@@ -40,6 +40,7 @@ source ./inc3.11.5
source ./inc3.11.6
source ./inc3.11.7
source ./inc3.11.8
+source ./inc3.11.9
@@ -53,6 +54,7 @@ set res_thread_cnt 0
set user_name ""
set def_partition ""
set def_node ""
+set def_node_name ""
set ii 0
print_header $test_id
@@ -276,6 +278,16 @@ spawn $sinfo -h -o "=%N=" -p $def_partition
expect {
-re "=(.+)=" {
set def_node $expect_out(1,string)
+ send_user "\nDefault node $def_node\n";
+ }
+ eof {
+ wait
+ }
+}
+spawn $sinfo -h -o "=%N=" -p $def_partition
+expect {
+ -re "=($alpha_numeric)(\[)($alpha_numeric)-($alpha_numeric)(\])=" {
+ set def_node_name $expect_out(1,string)
exp_continue
}
eof {
@@ -355,6 +367,7 @@ inc3_11_6
if {$cons_res_actived == 1} {
inc3_11_7
inc3_11_8
+ inc3_11_9
}
if {$exit_code == 0} {