From 5569138a04b2446dcaf8452b181d8e8cc7b5396f Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Fri, 30 May 2003 18:07:50 +0000
Subject: [PATCH] Fix bug in node filtering by configuration if fast_schedule
flag is set. Previously slurmctld was not filtering out nodes failing to
satisfy the job's mincpus, mem, or tmp constraints.
---
src/slurmctld/node_scheduler.c | 15 ++++++++++-----
1 file changed, 10 insertions(+), 5 deletions(-)
diff --git a/src/slurmctld/node_scheduler.c b/src/slurmctld/node_scheduler.c
index 6f7fb52f5c8..471bfec42c9 100644
--- a/src/slurmctld/node_scheduler.c
+++ b/src/slurmctld/node_scheduler.c
@@ -818,7 +818,7 @@ static int _build_node_list(struct job_record *job_ptr,
struct config_record *config_ptr;
struct part_record *part_ptr = job_ptr->part_ptr;
ListIterator config_record_iterator;
- int tmp_feature, check_node_config;
+ int tmp_feature, check_node_config, config_filter = 0;
struct job_details *detail_ptr = job_ptr->details;
bitstr_t *exc_node_mask = NULL;
@@ -842,14 +842,19 @@ static int _build_node_list(struct job_record *job_ptr,
if (tmp_feature == 0)
continue;
+ if ((detail_ptr->min_procs > config_ptr->cpus ) ||
+ (detail_ptr->min_memory > config_ptr->real_memory) ||
+ (detail_ptr->min_tmp_disk > config_ptr->tmp_disk))
+ config_filter = 1;
+
/* since nodes can register with more resources than defined */
/* in the configuration, we want to use those higher values */
/* for scheduling, but only as needed (slower) */
- if (slurmctld_conf.fast_schedule)
+ if (slurmctld_conf.fast_schedule) {
+ if (config_filter)
+ continue;
check_node_config = 0;
- else if ((detail_ptr->min_procs > config_ptr->cpus ) ||
- (detail_ptr->min_memory > config_ptr->real_memory) ||
- (detail_ptr->min_tmp_disk > config_ptr->tmp_disk)) {
+ } else if (config_filter) {
check_node_config = 1;
} else
check_node_config = 0;
--
GitLab