From 87d03dd8ccaccc6775155e9f9f23c439c05b1ac1 Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Tue, 14 Sep 2010 21:22:41 +0000
Subject: [PATCH] Add support for SLURM_CLUSTERS environment variable in the
sbatch command.
---
NEWS | 1 +
doc/man/man1/sbatch.1 | 23 +++++++++++++----------
src/sbatch/mult_cluster.c | 20 +++++++++++---------
src/sbatch/opt.c | 8 +++++---
src/sbatch/sbatch.c | 6 +++---
src/slurmctld/controller.c | 5 +++--
6 files changed, 36 insertions(+), 27 deletions(-)
diff --git a/NEWS b/NEWS
index 12e353ad20e..e4fc8550bd2 100644
--- a/NEWS
+++ b/NEWS
@@ -8,6 +8,7 @@ documents those changes that are of interest to users and admins.
partitions and use lua metatables to reference the job and partition fields.
-- Add support for serveral new trigger types: SlurmDBD failure/restart,
Database failure/restart, Slurmctld failure/restart.
+ -- Add support for SLURM_CLUSTERS environment variable in the sbatch command.
* Changes in SLURM 2.2.0.pre10
==============================
diff --git a/doc/man/man1/sbatch.1 b/doc/man/man1/sbatch.1
index a90b979001c..8f098d8eb51 100644
--- a/doc/man/man1/sbatch.1
+++ b/doc/man/man1/sbatch.1
@@ -1,4 +1,4 @@
-.TH "sbatch" "1" "SLURM 2.2" "April 2010" "SLURM Commands"
+.TH "sbatch" "1" "SLURM 2.2" "September 2010" "SLURM Commands"
.SH "NAME"
sbatch \- Submit a batch script to SLURM.
@@ -126,15 +126,6 @@ The default value is the current working directory.
Checkpoint files will be of the form "<job_id>.ckpt" for jobs
and "<job_id>.<step_id>.ckpt" for job steps.
-.TP
-\fB\-\-clusters\fR=<\fIstring\fR>
-Clusters to issue commands to. Multiple cluster names may be comma separated.
-The job will be submitted to the one cluster providing the earliest expected
-job initiation time. The default value is the current cluster. A value of
-of '\fIall\fR' will query to run on all clusters. Note the
-\fB\-\-export\fR option to control environment variables exported
-between clusters.
-
.TP
\fB\-\-comment\fR=<\fIstring\fR>
An arbitrary comment.
@@ -502,6 +493,15 @@ License names can be followed by an asterisk and count
Multiple license names should be comma separated (e.g.
"\-\-licenses=foo*4,bar").
+.TP
+\fB\-M\fR, \fB\-\-clusters\fR=<\fIstring\fR>
+Clusters to issue commands to. Multiple cluster names may be comma separated.
+The job will be submitted to the one cluster providing the earliest expected
+job initiation time. The default value is the current cluster. A value of
+of '\fIall\fR' will query to run on all clusters. Note the
+\fB\-\-export\fR option to control environment variables exported
+between clusters.
+
.TP
\fB\-m\fR, \fB\-\-distribution\fR=
<\fIblock\fR|\fIcyclic\fR|\fIarbitrary\fR|\fIplane=<options>\fR>
@@ -1070,6 +1070,9 @@ Same as \fB\-\-checkpoint\fR
\fBSLURM_CHECKPOINT_DIR\fR
Same as \fB\-\-checkpoint\-dir\fR
.TP
+\fBSBATCH_CLUSTERS\fR or \fBSLURM_CLUSTERS\fR
+Same as \fB\-\-clusters\fR
+.TP
\fBSBATCH_CONN_TYPE\fR
Same as \fB\-\-conn\-type\fR
.TP
diff --git a/src/sbatch/mult_cluster.c b/src/sbatch/mult_cluster.c
index 2b203317f0c..7e52f366398 100644
--- a/src/sbatch/mult_cluster.c
+++ b/src/sbatch/mult_cluster.c
@@ -108,7 +108,7 @@ local_cluster_rec_t *_job_will_run (job_desc_msg_t *req)
slurm_seterrno(rc);
break;
case RESPONSE_JOB_WILL_RUN:
- if(working_cluster_rec->flags & CLUSTER_FLAG_BG)
+ if (working_cluster_rec->flags & CLUSTER_FLAG_BG)
type = "cnodes";
will_run_resp = (will_run_response_msg_t *) resp_msg.data;
slurm_make_time_str(&will_run_resp->start_time,
@@ -123,7 +123,7 @@ local_cluster_rec_t *_job_will_run (job_desc_msg_t *req)
if (will_run_resp->preemptee_job_id) {
local_cluster->preempt_cnt =
list_count(will_run_resp->preemptee_job_id);
- if(opt.verbose >= LOG_LEVEL_DEBUG) {
+ if (opt.verbose >= LOG_LEVEL_DEBUG) {
ListIterator itr;
uint32_t *job_id_ptr;
char *job_list = NULL, *sep = "";
@@ -160,9 +160,9 @@ extern int sbatch_set_first_avail_cluster(job_desc_msg_t *req)
List ret_list = NULL;
/* return if we only have 1 or less clusters here */
- if(!opt.clusters || !list_count(opt.clusters)) {
+ if (!opt.clusters || !list_count(opt.clusters)) {
return rc;
- } else if(list_count(opt.clusters) == 1) {
+ } else if (list_count(opt.clusters) == 1) {
working_cluster_rec = list_peek(opt.clusters);
return rc;
}
@@ -174,12 +174,14 @@ extern int sbatch_set_first_avail_cluster(job_desc_msg_t *req)
}
ret_list = list_create(_destroy_local_cluster_rec);
+ if (ret_list == NULL)
+ fatal("list_create malloc failure");
itr = list_iterator_create(opt.clusters);
- while((working_cluster_rec = list_next(itr))) {
- if((local_cluster = _job_will_run(req))) {
- if(!ret_list)
+ while ((working_cluster_rec = list_next(itr))) {
+ if ((local_cluster = _job_will_run(req))) {
+ if (!ret_list)
ret_list = list_create(
- _destroy_local_cluster_rec);
+ _destroy_local_cluster_rec);
list_append(ret_list, local_cluster);
} else
error("Problem with submit to cluster %s: %m",
@@ -190,7 +192,7 @@ extern int sbatch_set_first_avail_cluster(job_desc_msg_t *req)
if (host_set)
req->alloc_node = NULL;
- if(!ret_list) {
+ if (!ret_list) {
error("Can't run on any of the clusters given");
return SLURM_ERROR;
}
diff --git a/src/sbatch/opt.c b/src/sbatch/opt.c
index 9622febec05..6b725d8b97e 100644
--- a/src/sbatch/opt.c
+++ b/src/sbatch/opt.c
@@ -438,6 +438,8 @@ env_vars_t env_vars[] = {
{"SBATCH_BLRTS_IMAGE", OPT_STRING, &opt.blrtsimage, NULL },
{"SBATCH_CHECKPOINT", OPT_STRING, &opt.ckpt_interval_str, NULL },
{"SBATCH_CHECKPOINT_DIR",OPT_STRING, &opt.ckpt_dir, NULL },
+ {"SBATCH_CLUSTERS", OPT_STRING, &opt.clusters, NULL },
+ {"SLURM_CLUSTERS", OPT_STRING, &opt.clusters, NULL },
{"SBATCH_CNLOAD_IMAGE", OPT_STRING, &opt.linuximage, NULL },
{"SBATCH_CONN_TYPE", OPT_CONN_TYPE, NULL, NULL },
{"SBATCH_CPU_BIND", OPT_CPU_BIND, NULL, NULL },
@@ -1210,10 +1212,10 @@ static void _set_options(int argc, char **argv)
}
break;
case 'M':
- if(opt.clusters)
+ if (opt.clusters)
list_destroy(opt.clusters);
- if(!(opt.clusters =
- slurmdb_get_info_cluster(optarg))) {
+ if (!(opt.clusters =
+ slurmdb_get_info_cluster(optarg))) {
error("'%s' invalid entry for --clusters",
optarg);
exit(1);
diff --git a/src/sbatch/sbatch.c b/src/sbatch/sbatch.c
index 94344b6865d..7f0d08dfb79 100644
--- a/src/sbatch/sbatch.c
+++ b/src/sbatch/sbatch.c
@@ -148,8 +148,8 @@ int main(int argc, char *argv[])
desc.script = (char *)script_body;
/* If can run on multiple clusters find the earliest run time
- and run it there */
- if(sbatch_set_first_avail_cluster(&desc) != SLURM_SUCCESS)
+ * and run it there */
+ if (sbatch_set_first_avail_cluster(&desc) != SLURM_SUCCESS)
exit(error_exit);
while (slurm_submit_batch_job(&desc, &resp) < 0) {
@@ -180,7 +180,7 @@ int main(int argc, char *argv[])
}
printf("Submitted batch job %u", resp->job_id);
- if(working_cluster_rec)
+ if (working_cluster_rec)
printf(" on cluster %s", working_cluster_rec->name);
printf("\n");
diff --git a/src/slurmctld/controller.c b/src/slurmctld/controller.c
index c6f316e7900..4167d60ac2f 100644
--- a/src/slurmctld/controller.c
+++ b/src/slurmctld/controller.c
@@ -360,10 +360,11 @@ int main(int argc, char *argv[])
assoc_init_arg.remove_assoc_notify = _remove_assoc;
assoc_init_arg.remove_qos_notify = _remove_qos;
assoc_init_arg.cache_level = ASSOC_MGR_CACHE_ASSOC |
- ASSOC_MGR_CACHE_USER | ASSOC_MGR_CACHE_QOS;
+ ASSOC_MGR_CACHE_USER |
+ ASSOC_MGR_CACHE_QOS;
if (assoc_mgr_init(acct_db_conn, &assoc_init_arg)) {
- if(accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS)
+ if (accounting_enforce & ACCOUNTING_ENFORCE_ASSOCS)
error("Association database appears down, "
"reading from state file.");
else
--
GitLab