diff --git a/doc/man/man1/scancel.1 b/doc/man/man1/scancel.1
index 277c4c7d6b22fca4d91d62894f460d08c05ffcb6..058d17013481b8257a54cd68749eaeb08055ec4d 100644
--- a/doc/man/man1/scancel.1
+++ b/doc/man/man1/scancel.1
@@ -1,4 +1,4 @@
-.TH SCANCEL "1" "April 2009" "scancel 2.0" "Slurm components"
+.TH SCANCEL "1" "January 2011" "scancel 2.3" "Slurm components"
.SH "NAME"
scancel \- Used to signal jobs or job steps that are under the control of Slurm.
@@ -33,6 +33,8 @@ for details.
\fB-\-ctld\fR
Send the job signal request to the slurmctld daemon rather than directly to the
slurmd daemons. This increases overhead, but offers better fault tolerance.
+This is the default behavior on architectures using front end nodes (e.g.
+BlueGene and Cray computers) or when the \fB\-\-clusters\fR option is used.
.TP
\fB\-\-help\fR
@@ -215,7 +217,7 @@ scancel \-\-state=PENDING \-\-user=bob \-\-partition=debug
.SH "COPYING"
Copyright (C) 2002-2007 The Regents of the University of California.
-Copyright (C) 2008-2009 Lawrence Livermore National Security.
+Copyright (C) 2008-2011 Lawrence Livermore National Security.
Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
CODE\-OCEC\-09\-009. All rights reserved.
.LP
diff --git a/src/common/read_config.c b/src/common/read_config.c
index 927b402c02f6ba9d76810fa6efe5d9d6fb6fc40b..67e032eaa9530541d779e206f5e963773722a72a 100644
--- a/src/common/read_config.c
+++ b/src/common/read_config.c
@@ -472,6 +472,7 @@ static int _parse_frontend(void **dest, slurm_parser_enum_t type,
*dest = (void *)n;
+ s_p_hashtbl_destroy(tbl);
return 1;
}
diff --git a/src/scontrol/update_node.c b/src/scontrol/update_node.c
index f453e76a613e36284f76497eac17c5efe42901fa..ddd55497d62e88e1b2faf6b01239fd46efb4f705 100644
--- a/src/scontrol/update_node.c
+++ b/src/scontrol/update_node.c
@@ -309,6 +309,13 @@ scontrol_update_front_end (int argc, char *argv[])
}
}
+ if ((front_end_msg.node_state == NODE_STATE_DOWN) &&
+ ((front_end_msg.reason == NULL) ||
+ (strlen(front_end_msg.reason) == 0))) {
+ fprintf (stderr, "You must specify a reason when DOWNING a "
+ "frontend node\nRequest aborted\n");
+ goto done;
+ }
if ((front_end_msg.node_state == NODE_STATE_DRAIN) &&
((front_end_msg.reason == NULL) ||
(strlen(front_end_msg.reason) == 0))) {
diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index b8dddb16e0973385b974514c7ee7f2fe26d7c1c3..3553f070c73089b9dde194d8ad126606962e6602 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -1973,6 +1973,7 @@ extern int kill_job_by_part_name(char *part_name)
* resource for its jobs and kill them.
* IN node_name - name of a front end node
* RET number of jobs associated with this front end node
+ * NOTE: Patterned after kill_running_job_by_node_name()
*/
extern int kill_job_by_front_end_name(char *node_name)
{
@@ -7736,7 +7737,8 @@ abort_job_on_node(uint32_t job_id, struct job_record *job_ptr,
#ifdef HAVE_FRONT_END
xassert(job_ptr->batch_host);
agent_info->hostlist = hostlist_create(job_ptr->batch_host);
- debug("Aborting job %u on node %s", job_id, job_ptr->batch_host);
+ debug("Aborting job %u on front end node %s", job_id,
+ job_ptr->batch_host);
#else
agent_info->hostlist = hostlist_create(node_ptr->name);
debug("Aborting job %u on node %s", job_id, node_ptr->name);
@@ -7781,7 +7783,8 @@ kill_job_on_node(uint32_t job_id, struct job_record *job_ptr,
#ifdef HAVE_FRONT_END
xassert(job_ptr->batch_host);
agent_info->hostlist = hostlist_create(job_ptr->batch_host);
- debug("Killing job %u on node %s", job_id, job_ptr->batch_host);
+ debug("Killing job %u on front end node %s", job_id,
+ job_ptr->batch_host);
#else
agent_info->hostlist = hostlist_create(node_ptr->name);
debug("Killing job %u on node %s", job_id, node_ptr->name);
@@ -7966,7 +7969,6 @@ _xmit_new_end_time(struct job_record *job_ptr)
hostlist_push(agent_args->hostlist, job_ptr->batch_host);
agent_args->node_count = 1;
#else
- agent_args->hostlist = hostlist_create("");
for (i = 0; i < node_record_count; i++) {
if (bit_test(job_ptr->node_bitmap, i) == 0)
continue;
diff --git a/src/slurmctld/proc_req.c b/src/slurmctld/proc_req.c
index c3816de90046f21f7ed66ea5b5d69700fb1fc123..76702a95b0ab6ccba8e195c8694a96fdf12e20e0 100644
--- a/src/slurmctld/proc_req.c
+++ b/src/slurmctld/proc_req.c
@@ -1118,15 +1118,15 @@ static void _slurm_rpc_dump_nodes(slurm_msg_t * msg)
/* Locks: Read config, write node (reset allocated CPU count in some
* select plugins) */
slurmctld_lock_t node_write_lock = {
- READ_LOCK, NO_LOCK, NO_LOCK, WRITE_LOCK };
+ READ_LOCK, NO_LOCK, WRITE_LOCK, NO_LOCK };
uid_t uid = g_slurm_auth_get_uid(msg->auth_cred, NULL);
START_TIMER;
debug2("Processing RPC: REQUEST_NODE_INFO from uid=%d", uid);
lock_slurmctld(node_write_lock);
- if ((slurmctld_conf.private_data & PRIVATE_DATA_NODES)
- && (!validate_operator(uid))) {
+ if ((slurmctld_conf.private_data & PRIVATE_DATA_NODES) &&
+ (!validate_operator(uid))) {
unlock_slurmctld(node_write_lock);
error("Security violation, REQUEST_NODE_INFO RPC from uid=%d",
uid);
@@ -1435,8 +1435,8 @@ static void _slurm_rpc_complete_batch_script(slurm_msg_t * msg)
lock_slurmctld(job_write_lock);
/* Send batch step info to accounting */
- if (association_based_accounting
- && (job_ptr = find_job_record(comp_msg->job_id))) {
+ if (association_based_accounting &&
+ (job_ptr = find_job_record(comp_msg->job_id))) {
struct step_record batch_step;
memset(&batch_step, 0, sizeof(struct step_record));
batch_step.job_ptr = job_ptr;