From 78aa6a508b9dbf321da3f0e597a6061b8c337cbc Mon Sep 17 00:00:00 2001
From: Danny Auble <da@llnl.gov>
Date: Tue, 16 Jun 2009 18:12:32 +0000
Subject: [PATCH] svn merge -r17853:17864
https://eris.llnl.gov/svn/slurm/branches/slurm-2.0
---
NEWS | 7 ++++--
src/plugins/sched/backfill/backfill.c | 24 +++++++++----------
.../select/bluegene/plugin/bg_job_place.c | 3 ---
src/scontrol/scontrol.c | 12 ++++++++--
src/slurmctld/reservation.c | 2 +-
src/slurmctld/reservation.h | 4 ++--
6 files changed, 30 insertions(+), 22 deletions(-)
diff --git a/NEWS b/NEWS
index aadfb3c97b0..19c4c0dd401 100644
--- a/NEWS
+++ b/NEWS
@@ -66,7 +66,7 @@ documents those changes that are of interest to users and admins.
-- Fix select/linear to properly set a job's count of allocated processors
(all processors on the allocated nodes).
-- Fix select/cons_res to allocate proper CPU count when --ntasks-per-node
- option is used.
+ option is used without a task count in the job request.
-- Insure that no node is allocated to a job for which the CPU count is less
than --ntasks-per-node * --cpus-per-task.
-- Correct AllocProcs reported by "scontrol show node" when ThreadsPerCore
@@ -86,9 +86,12 @@ documents those changes that are of interest to users and admins.
-- Make sched/backfill properly schedule jobs with constraints having node
counts. NOTE: Backfill of jobs with constraings having exclusive OR
operators are not fully supported.
- -- if srun is cancelled by SIGINT that job is marked as cancelled.
+ -- If srun is cancelled by SIGINT, set the job state to cancelled, not
+ failed.
-- BLUEGENE - Fix, for if you are setting an subbp into an error mode
where the subbp stated isn't the first ionode in a nodecard.
+ -- fix for backfill to not core when checking shared nodes.
+ -- fix for scontrol to not core when hitting just return in interactive mode.
* Changes in SLURM 2.0.1
========================
diff --git a/src/plugins/sched/backfill/backfill.c b/src/plugins/sched/backfill/backfill.c
index 6f15b8b6356..f05a4586ff3 100644
--- a/src/plugins/sched/backfill/backfill.c
+++ b/src/plugins/sched/backfill/backfill.c
@@ -122,7 +122,7 @@ static void _diff_tv_str(struct timeval *tv1,struct timeval *tv2,
static bool _more_work(void);
static int _num_feature_count(struct job_record *job_ptr);
static int _start_job(struct job_record *job_ptr, bitstr_t *avail_bitmap);
-static int _try_sched(struct job_record *job_ptr, bitstr_t *avail_bitmap,
+static int _try_sched(struct job_record *job_ptr, bitstr_t **avail_bitmap,
uint32_t min_nodes, uint32_t max_nodes,
uint32_t req_nodes);
@@ -191,7 +191,7 @@ static int _num_feature_count(struct job_record *job_ptr)
* IN/OUT avail_bitmap - nodes available/selected to use
* RET SLURM_SUCCESS on success, otherwise an error code
*/
-static int _try_sched(struct job_record *job_ptr, bitstr_t *avail_bitmap,
+static int _try_sched(struct job_record *job_ptr, bitstr_t **avail_bitmap,
uint32_t min_nodes, uint32_t max_nodes,
uint32_t req_nodes)
{
@@ -225,12 +225,12 @@ static int _try_sched(struct job_record *job_ptr, bitstr_t *avail_bitmap,
}
list_iterator_destroy(feat_iter);
- if ((job_req_node_filter(job_ptr, avail_bitmap) !=
+ if ((job_req_node_filter(job_ptr, *avail_bitmap) !=
SLURM_SUCCESS) ||
- (bit_set_count(avail_bitmap) < high_cnt)) {
+ (bit_set_count(*avail_bitmap) < high_cnt)) {
rc = ESLURM_NODES_BUSY;
} else {
- rc = select_g_job_test(job_ptr, avail_bitmap,
+ rc = select_g_job_test(job_ptr, *avail_bitmap,
high_cnt, max_nodes, req_nodes,
SELECT_MODE_WILL_RUN);
}
@@ -250,15 +250,15 @@ static int _try_sched(struct job_record *job_ptr, bitstr_t *avail_bitmap,
uint16_t orig_shared;
orig_shared = job_ptr->details->shared;
job_ptr->details->shared = 0;
- tmp_bitmap = bit_copy(avail_bitmap);
- rc = select_g_job_test(job_ptr, avail_bitmap, min_nodes,
+ tmp_bitmap = bit_copy(*avail_bitmap);
+ rc = select_g_job_test(job_ptr, *avail_bitmap, min_nodes,
max_nodes, req_nodes,
SELECT_MODE_WILL_RUN);
job_ptr->details->shared = orig_shared;
if ((rc != SLURM_SUCCESS) && (orig_shared != 0)) {
- FREE_NULL_BITMAP(avail_bitmap);
- avail_bitmap= tmp_bitmap;
- rc = select_g_job_test(job_ptr, avail_bitmap,
+ FREE_NULL_BITMAP(*avail_bitmap);
+ *avail_bitmap= tmp_bitmap;
+ rc = select_g_job_test(job_ptr, *avail_bitmap,
min_nodes, max_nodes, req_nodes,
SELECT_MODE_WILL_RUN);
} else
@@ -462,11 +462,11 @@ static void _attempt_backfill(void)
if (job_req_node_filter(job_ptr, avail_bitmap))
continue; /* nodes lack features */
- j = _try_sched(job_ptr, avail_bitmap,
+ j = _try_sched(job_ptr, &avail_bitmap,
min_nodes, max_nodes, req_nodes);
if (j != SLURM_SUCCESS)
continue; /* not runable */
-
+
job_ptr->start_time = MAX(job_ptr->start_time, start_res);
if (job_ptr->start_time <= now) {
int rc = _start_job(job_ptr, resv_bitmap);
diff --git a/src/plugins/select/bluegene/plugin/bg_job_place.c b/src/plugins/select/bluegene/plugin/bg_job_place.c
index 4acd56f0053..215966fbc78 100644
--- a/src/plugins/select/bluegene/plugin/bg_job_place.c
+++ b/src/plugins/select/bluegene/plugin/bg_job_place.c
@@ -97,7 +97,6 @@ static int _check_for_booted_overlapping_blocks(
bool test_only);
static int _dynamically_request(List block_list, int *blocks_added,
ba_request_t *request,
- bitstr_t* slurm_block_bitmap,
char *user_req_nodes);
static int _find_best_block_match(List block_list, int *blocks_added,
struct job_record* job_ptr,
@@ -709,7 +708,6 @@ static int _check_for_booted_overlapping_blocks(
static int _dynamically_request(List block_list, int *blocks_added,
ba_request_t *request,
- bitstr_t* slurm_block_bitmap,
char *user_req_nodes)
{
List list_of_lists = NULL;
@@ -1092,7 +1090,6 @@ static int _find_best_block_match(List block_list,
if((rc = _dynamically_request(block_list, blocks_added,
&request,
- slurm_block_bitmap,
job_ptr->details->req_nodes))
== SLURM_SUCCESS) {
create_try = 1;
diff --git a/src/scontrol/scontrol.c b/src/scontrol/scontrol.c
index 8b514cc7075..2d1e2749b1a 100644
--- a/src/scontrol/scontrol.c
+++ b/src/scontrol/scontrol.c
@@ -461,14 +461,22 @@ _process_command (int argc, char *argv[])
{
int error_code = 0;
char *tag = argv[0];
- int taglen = strlen(tag);
+ int taglen = 0;
if (argc < 1) {
exit_code = 1;
if (quiet_flag == -1)
fprintf(stderr, "no input");
+ return 0;
+ } else if(tag)
+ taglen = strlen(tag);
+ else {
+ if (quiet_flag == -1)
+ fprintf(stderr, "input problem");
+ return 0;
}
- else if (strncasecmp (tag, "abort", MAX(taglen, 5)) == 0) {
+
+if (strncasecmp (tag, "abort", MAX(taglen, 5)) == 0) {
/* require full command name */
if (argc > 2) {
exit_code = 1;
diff --git a/src/slurmctld/reservation.c b/src/slurmctld/reservation.c
index 8b178d7b0bd..95856585267 100644
--- a/src/slurmctld/reservation.c
+++ b/src/slurmctld/reservation.c
@@ -2337,7 +2337,7 @@ extern int job_test_resv(struct job_record *job_ptr, time_t *when,
if (rc == SLURM_SUCCESS)
break;
-
+ /* rc == ESLURM_NODES_BUSY here from above break */
if (move_time && (i<10)) { /* Retry for later start time */
bit_nset(*node_bitmap, 0, (node_record_count - 1));
rc = SLURM_SUCCESS;
diff --git a/src/slurmctld/reservation.h b/src/slurmctld/reservation.h
index f019aba2955..d05ee8ee080 100644
--- a/src/slurmctld/reservation.h
+++ b/src/slurmctld/reservation.h
@@ -50,10 +50,10 @@ extern time_t last_resv_update;
/* Create a resource reservation */
extern int create_resv(resv_desc_msg_t *resv_desc_ptr);
-/* Update an exiting resource reservation */
+/* Update an existing resource reservation */
extern int update_resv(resv_desc_msg_t *resv_desc_ptr);
-/* Delete an exiting resource reservation */
+/* Delete an existing resource reservation */
extern int delete_resv(reservation_name_msg_t *resv_desc_ptr);
/* Dump the reservation records to a buffer */
--
GitLab