diff --git a/NEWS b/NEWS
index ae7a3cbdcb2ea547341b25de34dd02ee10482536..83b93b6fb7efd4395b76f6677b075d5615f677ad 100644
--- a/NEWS
+++ b/NEWS
@@ -192,6 +192,9 @@ documents those changes that are of interest to users and admins.
-- Reword the explanation of cputime and cputimeraw in the sacct man page.
-- JobCompType allows "jobcomp/mysql" as valid name but the code used
"job_comp/mysql" setting an incorrect default database.
+ -- Try to load libslurm.so only when necessary.
+ -- When nodes scheduled for reboot, set state to DOWN rather than FUTURE so
+ they are still visible to sinfo. State set to IDLE after reboot completes.
* Changes in Slurm 14.03.6
==========================
diff --git a/contribs/pam/pam_slurm.c b/contribs/pam/pam_slurm.c
index ffa0620c2618f7135b0e33a4dd6578a323d9bb07..e01c928c723a5604b86b24d41790a2d6df27a6fd 100644
--- a/contribs/pam/pam_slurm.c
+++ b/contribs/pam/pam_slurm.c
@@ -422,7 +422,9 @@ extern void libpam_slurm_init (void)
SLURM_API_REVISION, SLURM_API_AGE) >=
sizeof(libslurmname) ) {
_log_msg (LOG_ERR, "Unable to write libslurmname\n");
- } else if (!(slurm_h = dlopen(libslurmname, RTLD_NOW|RTLD_GLOBAL))) {
+ } else if ((slurm_h = dlopen(libslurmname, RTLD_NOW|RTLD_GLOBAL))) {
+ return;
+ } else {
_log_msg (LOG_INFO, "Unable to dlopen %s: %s\n",
libslurmname, dlerror ());
}
@@ -430,7 +432,9 @@ extern void libpam_slurm_init (void)
if (snprintf(libslurmname, sizeof(libslurmname), "libslurm.so.%d",
SLURM_API_CURRENT) >= sizeof(libslurmname) ) {
_log_msg (LOG_ERR, "Unable to write libslurmname\n");
- } else if (!(slurm_h = dlopen(libslurmname, RTLD_NOW|RTLD_GLOBAL))) {
+ } else if ((slurm_h = dlopen(libslurmname, RTLD_NOW|RTLD_GLOBAL))) {
+ return;
+ } else {
_log_msg (LOG_INFO, "Unable to dlopen %s: %s\n",
libslurmname, dlerror ());
}
diff --git a/src/slurmctld/controller.c b/src/slurmctld/controller.c
index 91abde3cf63dcd32f8c8c479cf85246d6d24f532..a51c90509502a4319c6d2822b9f2c8d2346fdc8b 100644
--- a/src/slurmctld/controller.c
+++ b/src/slurmctld/controller.c
@@ -1324,9 +1324,10 @@ static void _queue_reboot_msg(void)
}
hostlist_push(reboot_agent_args->hostlist, node_ptr->name);
reboot_agent_args->node_count++;
- node_ptr->node_state = NODE_STATE_FUTURE |
- (node_ptr->node_state & NODE_STATE_FLAGS);
node_ptr->node_state &= ~NODE_STATE_MAINT;
+ node_ptr->node_state &= NODE_STATE_FLAGS;
+ node_ptr->node_state |= NODE_STATE_DOWN;
+ node_ptr->reason = xstrdup("Scheduled reboot");
bit_clear(avail_node_bitmap, i);
bit_clear(idle_node_bitmap, i);
node_ptr->last_response = now;
diff --git a/src/slurmctld/node_mgr.c b/src/slurmctld/node_mgr.c
index 8b6d59ee0a10c8f2e65348d122add4b3cc176343..29749c58c53e3bada450f0943886437b6c13e978 100644
--- a/src/slurmctld/node_mgr.c
+++ b/src/slurmctld/node_mgr.c
@@ -742,8 +742,7 @@ extern void pack_all_node (char **buffer_ptr, int *buffer_size,
if (((show_flags & SHOW_ALL) == 0) && (uid != 0) &&
(_node_is_hidden(node_ptr)))
hidden = true;
- else if (IS_NODE_FUTURE(node_ptr) &&
- !IS_NODE_MAINT(node_ptr)) /* reboot req sent */
+ else if (IS_NODE_FUTURE(node_ptr))
hidden = true;
else if (IS_NODE_CLOUD(node_ptr) &&
IS_NODE_POWER_SAVE(node_ptr))
@@ -830,8 +829,7 @@ extern void pack_one_node (char **buffer_ptr, int *buffer_size,
if (((show_flags & SHOW_ALL) == 0) && (uid != 0) &&
(_node_is_hidden(node_ptr)))
hidden = true;
- else if (IS_NODE_FUTURE(node_ptr) &&
- !IS_NODE_MAINT(node_ptr)) /* reboot req sent */
+ else if (IS_NODE_FUTURE(node_ptr))
hidden = true;
else if (IS_NODE_CLOUD(node_ptr) &&
IS_NODE_POWER_SAVE(node_ptr))
@@ -2203,10 +2201,9 @@ extern int validate_node_specs(slurm_node_registration_status_msg_t *reg_msg,
}
} else if (IS_NODE_DOWN(node_ptr) &&
((slurmctld_conf.ret2service == 2) ||
+ !xstrcmp(node_ptr->reason, "Scheduled reboot") ||
((slurmctld_conf.ret2service == 1) &&
- (node_ptr->reason != NULL) &&
- (strncmp(node_ptr->reason, "Not responding", 14)
- == 0)))) {
+ !xstrcmp(node_ptr->reason, "Not responding")))) {
if (reg_msg->job_count) {
node_ptr->node_state = NODE_STATE_ALLOCATED |
node_flags;
@@ -2564,10 +2561,11 @@ extern int validate_nodes_via_front_end(
}
} else if (IS_NODE_DOWN(node_ptr) &&
((slurmctld_conf.ret2service == 2) ||
+ !xstrcmp(node_ptr->reason,
+ "Scheduled reboot") ||
((slurmctld_conf.ret2service == 1) &&
- (node_ptr->reason != NULL) &&
- (strncmp(node_ptr->reason,
- "Not responding", 14) == 0)))) {
+ !xstrcmp(node_ptr->reason,
+ "Not responding")))) {
update_node_state = true;
*newly_up = true;
if (node_ptr->run_job_cnt) {
@@ -2679,9 +2677,10 @@ static void _node_did_resp(front_end_record_t *fe_ptr)
fe_ptr->node_state = NODE_STATE_IDLE | node_flags;
}
if (IS_NODE_DOWN(fe_ptr) &&
- (slurmctld_conf.ret2service == 1) &&
- (fe_ptr->reason != NULL) &&
- (strncmp(fe_ptr->reason, "Not responding", 14) == 0)) {
+ ((slurmctld_conf.ret2service == 2) ||
+ !xstrcmp(fe_ptr->reason, "Scheduled reboot") ||
+ ((slurmctld_conf.ret2service == 1) &&
+ !xstrcmp(fe_ptr->reason, "Not responding")))) {
last_front_end_update = now;
fe_ptr->node_state = NODE_STATE_IDLE | node_flags;
info("node_did_resp: node %s returned to service",
@@ -2730,9 +2729,10 @@ static void _node_did_resp(struct node_record *node_ptr)
}
}
if (IS_NODE_DOWN(node_ptr) &&
- (slurmctld_conf.ret2service == 1) &&
- (node_ptr->reason != NULL) &&
- (strncmp(node_ptr->reason, "Not responding", 14) == 0)) {
+ ((slurmctld_conf.ret2service == 2) ||
+ !xstrcmp(node_ptr->reason, "Scheduled reboot") ||
+ ((slurmctld_conf.ret2service == 1) &&
+ !xstrcmp(node_ptr->reason, "Not responding")))) {
node_ptr->last_idle = now;
node_ptr->node_state = NODE_STATE_IDLE | node_flags;
info("node_did_resp: node %s returned to service",