From 585e81908f31e646ef4a28361312453b59d9cfe7 Mon Sep 17 00:00:00 2001
From: Mehdi Dogguy <mehdi@debian.org>
Date: Mon, 8 Sep 2014 22:34:34 +0200
Subject: [PATCH] Imported Upstream version 2.6.9
---
META | 4 +--
NEWS | 7 +++++
src/common/slurm_protocol_api.c | 8 ++---
src/plugins/sched/backfill/backfill.c | 3 +-
src/sacctmgr/user_functions.c | 7 +++++
src/sinfo/print.c | 10 -------
src/sinfo/sinfo.c | 42 +++++++++++++++++++++++++++
src/slurmctld/proc_req.c | 2 +-
8 files changed, 64 insertions(+), 19 deletions(-)
diff --git a/META b/META
index 61dc5fc5d..5fe143bd8 100644
--- a/META
+++ b/META
@@ -9,8 +9,8 @@
Name: slurm
Major: 2
Minor: 6
- Micro: 8
- Version: 2.6.8
+ Micro: 9
+ Version: 2.6.9
Release: 1
##
diff --git a/NEWS b/NEWS
index 51e2c712d..5dc1fcc3d 100644
--- a/NEWS
+++ b/NEWS
@@ -1,6 +1,13 @@
This file describes changes in recent versions of Slurm. It primarily
documents those changes that are of interest to users and admins.
+* Changes in Slurm 2.6.9
+========================
+ -- Fix sinfo to work correctly with draining/mixed nodes as well as filtering
+ on Mixed state.
+ -- Fix sacctmgr update user with no "where" condition.
+ -- Fix logic bugs for SchedulerParameters option of max_rpc_cnt.
+
* Changes in Slurm 2.6.8
========================
-- Add support for Torque/PBS job array options and environment variables.
diff --git a/src/common/slurm_protocol_api.c b/src/common/slurm_protocol_api.c
index 294ed10fa..fdf353b65 100644
--- a/src/common/slurm_protocol_api.c
+++ b/src/common/slurm_protocol_api.c
@@ -2199,10 +2199,9 @@ slurm_fd_t slurm_open_controller_conn(slurm_addr_t *addr)
myproto->primary_controller.sin_port;
}
- msg_timeout *= 10; /* Do iteration every 0.1 secs */
for (retry = 0; retry < msg_timeout; retry++) {
if (retry)
- usleep(100000);
+ sleep(1);
if (working_cluster_rec) {
if (working_cluster_rec->control_addr.sin_port == 0) {
slurm_set_addr(
@@ -3683,12 +3682,11 @@ List slurm_send_addr_recv_msgs(slurm_msg_t *msg, char *name, int timeout)
if (conn_timeout == (uint16_t) NO_VAL)
conn_timeout = MIN(slurm_get_msg_timeout(), 10);
- conn_timeout *= 10; /* Do iteration every 0.1 secs */
/* This connect retry logic permits Slurm hierarchical communications
* to better survive slurmd restarts */
for (i = 0; i <= conn_timeout; i++) {
- if (i > 0)
- usleep(100000);
+ if (i)
+ sleep(1);
fd = slurm_open_msg_conn(&msg->address);
if ((fd >= 0) || (errno != ECONNREFUSED))
break;
diff --git a/src/plugins/sched/backfill/backfill.c b/src/plugins/sched/backfill/backfill.c
index a989df161..12836ecd2 100644
--- a/src/plugins/sched/backfill/backfill.c
+++ b/src/plugins/sched/backfill/backfill.c
@@ -244,7 +244,8 @@ static bool _job_is_completing(void)
static bool _many_pending_rpcs(void)
{
//info("thread_count = %u", slurmctld_config.server_thread_count);
- if (slurmctld_config.server_thread_count >= defer_rpc_cnt)
+ if ((defer_rpc_cnt > 0) &&
+ (slurmctld_config.server_thread_count >= defer_rpc_cnt))
return true;
return false;
}
diff --git a/src/sacctmgr/user_functions.c b/src/sacctmgr/user_functions.c
index 198e3b4bf..55ea73bef 100644
--- a/src/sacctmgr/user_functions.c
+++ b/src/sacctmgr/user_functions.c
@@ -1697,6 +1697,13 @@ extern int sacctmgr_modify_user(int argc, char *argv[])
user_cond->assoc_cond = xmalloc(sizeof(slurmdb_association_cond_t));
user_cond->assoc_cond->cluster_list = list_create(slurm_destroy_char);
+ /* We need this to make sure we only change users, not
+ * accounts if this list didn't exist it would change
+ * accounts. Having it blank is fine, it just needs to
+ * exist. This also happens in _set_cond, but that doesn't
+ * always happen.
+ */
+ user_cond->assoc_cond->user_list = list_create(slurm_destroy_char);
for (i=0; i<argc; i++) {
int command_len = strlen(argv[i]);
diff --git a/src/sinfo/print.c b/src/sinfo/print.c
index 2b31371b4..dda081392 100644
--- a/src/sinfo/print.c
+++ b/src/sinfo/print.c
@@ -995,16 +995,6 @@ int _print_state_compact(sinfo_data_t * sinfo_data, int width,
if (sinfo_data && sinfo_data->nodes_total) {
my_state = sinfo_data->node_state;
- if (IS_NODE_DRAIN(sinfo_data)) {
- /* don't worry about mixed since the
- * whole node is being drained. */
- } else if ((sinfo_data->cpus_alloc && sinfo_data->cpus_other)
- || (sinfo_data->cpus_idle
- && (sinfo_data->cpus_idle
- != sinfo_data->cpus_total))) {
- my_state &= NODE_STATE_FLAGS;
- my_state |= NODE_STATE_MIXED;
- }
upper_state = node_state_string_compact(my_state);
lower_state = _str_tolower(upper_state);
diff --git a/src/sinfo/sinfo.c b/src/sinfo/sinfo.c
index a19716ed0..3aa5c7a8a 100644
--- a/src/sinfo/sinfo.c
+++ b/src/sinfo/sinfo.c
@@ -245,6 +245,8 @@ _query_server(partition_info_msg_t ** part_pptr,
static reserve_info_msg_t *old_resv_ptr = NULL, *new_resv_ptr;
int error_code;
uint16_t show_flags = 0;
+ int cc;
+ node_info_t *node_ptr;
if (params.all_flag)
show_flags |= SHOW_ALL;
@@ -304,6 +306,46 @@ _query_server(partition_info_msg_t ** part_pptr,
old_node_ptr = new_node_ptr;
*node_pptr = new_node_ptr;
+ /* Set the node state as NODE_STATE_MIXED. */
+ for (cc = 0; cc < new_node_ptr->record_count; cc++) {
+ node_ptr = &(new_node_ptr->node_array[cc]);
+ if (IS_NODE_DRAIN(node_ptr)) {
+ /* don't worry about mixed since the
+ * whole node is being drained. */
+ } else {
+ uint16_t alloc_cpus = 0, err_cpus = 0, idle_cpus;
+ int single_node_cpus =
+ (node_ptr->cpus / g_node_scaling);
+
+ select_g_select_nodeinfo_get(node_ptr->select_nodeinfo,
+ SELECT_NODEDATA_SUBCNT,
+ NODE_STATE_ALLOCATED,
+ &alloc_cpus);
+ if (params.cluster_flags & CLUSTER_FLAG_BG) {
+ if (!alloc_cpus &&
+ (IS_NODE_ALLOCATED(node_ptr) ||
+ IS_NODE_COMPLETING(node_ptr)))
+ alloc_cpus = node_ptr->cpus;
+ else
+ alloc_cpus *= single_node_cpus;
+ }
+ idle_cpus = node_ptr->cpus - alloc_cpus;
+ select_g_select_nodeinfo_get(node_ptr->select_nodeinfo,
+ SELECT_NODEDATA_SUBCNT,
+ NODE_STATE_ERROR,
+ &err_cpus);
+ if (params.cluster_flags & CLUSTER_FLAG_BG)
+ err_cpus *= single_node_cpus;
+ idle_cpus -= err_cpus;
+
+ if ((alloc_cpus && err_cpus) ||
+ (idle_cpus && (idle_cpus != node_ptr->cpus))) {
+ node_ptr->node_state &= NODE_STATE_FLAGS;
+ node_ptr->node_state |= NODE_STATE_MIXED;
+ }
+ }
+ }
+
if (old_resv_ptr) {
if (clear_old)
old_resv_ptr->last_update = 0;
diff --git a/src/slurmctld/proc_req.c b/src/slurmctld/proc_req.c
index c598437ae..a2b2a71be 100644
--- a/src/slurmctld/proc_req.c
+++ b/src/slurmctld/proc_req.c
@@ -2819,7 +2819,7 @@ static void _slurm_rpc_submit_batch_job(slurm_msg_t * msg)
if (error_code == SLURM_SUCCESS) {
error_code = validate_job_create_req(job_desc_msg);
if (job_desc_msg->array_bitmap)
- schedule_cnt = 0; /* Do full schedule cycle */
+ schedule_cnt = 100;
}
dump_job_desc(job_desc_msg);
if (error_code == SLURM_SUCCESS) {
--
GitLab