/*****************************************************************************\
* job_scheduler.c - manage the scheduling of pending jobs in priority order
* Note there is a global job list (job_list)
*****************************************************************************
* Copyright (C) 2002-2007 The Regents of the University of California.
* Copyright (C) 2008-2010 Lawrence Livermore National Security.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
* Written by Morris Jette <jette1@llnl.gov>
* CODE-OCEC-09-009. All rights reserved.
*
* This file is part of SLURM, a resource management program.
* For details, see <http://www.schedmd.com/slurmdocs/>.
* Please also read the included file: DISCLAIMER.
*
* SLURM is free software; you can redistribute it and/or modify it under
* the terms of the GNU General Public License as published by the Free
* Software Foundation; either version 2 of the License, or (at your option)
* any later version.
*
* In addition, as a special exception, the copyright holders give permission
* to link the code of portions of this program with the OpenSSL library under
* certain conditions as described in each individual source file, and
* distribute linked combinations including the two. You must obey the GNU
* General Public License in all respects for all of the code used other than
* OpenSSL. If you modify file(s) with this exception, you may extend this
* exception to your version of the file(s), but you are not obligated to do
* so. If you do not wish to do so, delete this exception statement from your
* version. If you delete this exception statement from all source files in
* the program, then also delete it here.
*
* SLURM is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
* FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
* details.
*
* You should have received a copy of the GNU General Public License along
* with SLURM; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
\*****************************************************************************/
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "src/common/assoc_mgr.h"
#include "src/common/env.h"
#include "src/common/gres.h"
#include "src/common/list.h"
#include "src/common/macros.h"
#include "src/common/node_select.h"
#include "src/common/slurm_accounting_storage.h"
#include "src/common/timers.h"
#include "src/common/uid.h"
#include "src/common/xassert.h"
#include "src/common/xstring.h"
#include "src/slurmctld/acct_policy.h"
#include "src/slurmctld/agent.h"
#include "src/slurmctld/front_end.h"
#include "src/slurmctld/job_scheduler.h"
#include "src/slurmctld/licenses.h"
#include "src/slurmctld/locks.h"
#include "src/slurmctld/node_scheduler.h"
#include "src/slurmctld/preempt.h"
#include "src/slurmctld/proc_req.h"
#include "src/slurmctld/reservation.h"
#include "src/slurmctld/slurmctld.h"
#include "src/slurmctld/srun_comm.h"
#define _DEBUG 0
#define MAX_RETRIES 10
static char ** _build_env(struct job_record *job_ptr);
static void _depend_list_del(void *dep_ptr);
static void _feature_list_delete(void *x);
static void _job_queue_append(List job_queue, struct job_record *job_ptr,
struct part_record *part_ptr);
static void _job_queue_rec_del(void *x);
static void * _run_epilog(void *arg);
static void * _run_prolog(void *arg);
static bool _scan_depend(List dependency_list, uint32_t job_id);
static int _valid_feature_list(uint32_t job_id, List feature_list);
static int _valid_node_feature(char *feature);
static int save_last_part_update = 0;
/*
* _build_user_job_list - build list of jobs for a given user
* and an optional job name
* IN user_id - user id
* IN job_name - job name constraint
* RET the job queue
* NOTE: the caller must call list_destroy() on RET value to free memory
*/
static List _build_user_job_list(uint32_t user_id, char* job_name)
{
List job_queue;
ListIterator job_iterator;
struct job_record *job_ptr = NULL;
job_queue = list_create(NULL);
if (job_queue == NULL)
fatal("list_create memory allocation failure");
job_iterator = list_iterator_create(job_list);
if (job_iterator == NULL)
fatal("list_iterator_create malloc failure");
while ((job_ptr = (struct job_record *) list_next(job_iterator))) {
xassert (job_ptr->magic == JOB_MAGIC);
if (job_ptr->user_id != user_id)
continue;
if (job_name && job_ptr->name &&
strcmp(job_name, job_ptr->name))
continue;
list_append(job_queue, job_ptr);
}
list_iterator_destroy(job_iterator);
return job_queue;
}
static void _job_queue_append(List job_queue, struct job_record *job_ptr,
struct part_record *part_ptr)
{
job_queue_rec_t *job_queue_rec;
job_queue_rec = xmalloc(sizeof(job_queue_rec_t));
job_queue_rec->job_ptr = job_ptr;
job_queue_rec->part_ptr = part_ptr;
list_append(job_queue, job_queue_rec);
}
static void _job_queue_rec_del(void *x)
{
xfree(x);
}
/*
* build_job_queue - build (non-priority ordered) list of pending jobs
* IN clear_start - if set then clear the start_time for pending jobs
* RET the job queue
* NOTE: the caller must call list_destroy() on RET value to free memory
*/
extern List build_job_queue(bool clear_start)
{
List job_queue;
ListIterator job_iterator, part_iterator;
struct job_record *job_ptr = NULL;
struct part_record *part_ptr;
bool job_is_pending;
bool job_indepen = false;
job_queue = list_create(_job_queue_rec_del);
if (job_queue == NULL)
fatal("list_create memory allocation failure");
job_iterator = list_iterator_create(job_list);
if (job_iterator == NULL)
fatal("list_iterator_create memory allocation failure");
while ((job_ptr = (struct job_record *) list_next(job_iterator))) {
xassert (job_ptr->magic == JOB_MAGIC);
job_is_pending = IS_JOB_PENDING(job_ptr);
if (!job_is_pending || IS_JOB_COMPLETING(job_ptr))
continue;
/* ensure dependency shows current values behind a hold */
job_indepen = job_independent(job_ptr, 0);
if (job_is_pending && clear_start)
job_ptr->start_time = (time_t) 0;
if (job_ptr->priority == 0) { /* held */
if ((job_ptr->state_reason != WAIT_HELD) &&
(job_ptr->state_reason != WAIT_HELD_USER)) {
job_ptr->state_reason = WAIT_HELD;
xfree(job_ptr->state_desc);
}
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority);
continue;
} else if ((job_ptr->priority == 1) && !job_indepen &&
((job_ptr->state_reason == WAIT_HELD) ||
(job_ptr->state_reason == WAIT_HELD_USER))) {
/* released behind active dependency? */
job_ptr->state_reason = WAIT_DEPENDENCY;
xfree(job_ptr->state_desc);
}
if (!job_indepen) /* can not run now */
continue;
if (job_ptr->part_ptr_list) {
part_iterator = list_iterator_create(job_ptr->
part_ptr_list);
if (part_iterator == NULL)
fatal("list_iterator_create malloc failure");
while ((part_ptr = (struct part_record *)
list_next(part_iterator))) {
job_ptr->part_ptr = part_ptr;
if (job_limits_check(&job_ptr) !=
WAIT_NO_REASON)
continue;
_job_queue_append(job_queue, job_ptr, part_ptr);
}
list_iterator_destroy(part_iterator);
} else {
if (job_ptr->part_ptr == NULL) {
part_ptr = find_part_record(job_ptr->partition);
if (part_ptr == NULL) {
error("Could not find partition %s "
"for job %u", job_ptr->partition,
job_ptr->job_id);
continue;
}
job_ptr->part_ptr = part_ptr;
error("partition pointer reset for job %u, "
"part %s", job_ptr->job_id,
job_ptr->partition);
}
_job_queue_append(job_queue, job_ptr,
job_ptr->part_ptr);
}
}
list_iterator_destroy(job_iterator);
return job_queue;
}
/*
* job_is_completing - Determine if jobs are in the process of completing.
* RET - True of any job is in the process of completing AND
* CompleteWait is configured non-zero
* NOTE: This function can reduce resource fragmentation, which is a
* critical issue on Elan interconnect based systems.
*/
extern bool job_is_completing(void)
{
bool completing = false;
ListIterator job_iterator;
struct job_record *job_ptr = NULL;
uint16_t complete_wait = slurm_get_complete_wait();
time_t recent;
if ((job_list == NULL) || (complete_wait == 0))
return completing;
recent = time(NULL) - complete_wait;
job_iterator = list_iterator_create(job_list);
while ((job_ptr = (struct job_record *) list_next(job_iterator))) {
if (IS_JOB_COMPLETING(job_ptr) &&
(job_ptr->end_time >= recent)) {
completing = true;
break;
}
}
list_iterator_destroy(job_iterator);
return completing;
}
/*
* set_job_elig_time - set the eligible time for pending jobs once their
* dependencies are lifted (in job->details->begin_time)
*/
extern void set_job_elig_time(void)
{
struct job_record *job_ptr = NULL;
struct part_record *part_ptr = NULL;
ListIterator job_iterator;
slurmctld_lock_t job_write_lock =
{ READ_LOCK, WRITE_LOCK, WRITE_LOCK, READ_LOCK };
lock_slurmctld(job_write_lock);
job_iterator = list_iterator_create(job_list);
while ((job_ptr = (struct job_record *) list_next(job_iterator))) {
part_ptr = job_ptr->part_ptr;
if (!IS_JOB_PENDING(job_ptr))
continue;
if (part_ptr == NULL)
continue;
if ((job_ptr->details == NULL) || job_ptr->details->begin_time)
continue;
if ((part_ptr->state_up & PARTITION_SCHED) == 0)
continue;
if ((job_ptr->time_limit != NO_VAL) &&
(job_ptr->time_limit > part_ptr->max_time))
continue;
if ((job_ptr->details->max_nodes != 0) &&
((job_ptr->details->max_nodes < part_ptr->min_nodes) ||
(job_ptr->details->min_nodes > part_ptr->max_nodes)))
continue;
/* Job's eligible time is set in job_independent() */
if (!job_independent(job_ptr, 0))
continue;
}
list_iterator_destroy(job_iterator);
unlock_slurmctld(job_write_lock);
}
/* Test of part_ptr can still run jobs or if its nodes have
* already been reserved by higher priority jobs (those in
* the failed_parts array) */
static bool _failed_partition(struct part_record *part_ptr,
struct part_record **failed_parts,
int failed_part_cnt)
{
int i;
for (i = 0; i < failed_part_cnt; i++) {
if (failed_parts[i] == part_ptr)
return true;
}
return false;
}
/*
* schedule - attempt to schedule all pending jobs
* pending jobs for each partition will be scheduled in priority
* order until a request fails
* IN job_limit - maximum number of jobs to test now, avoid testing the full
* queue on every job submit (0 means to use the system default,
* SchedulerParameters for default_queue_depth)
* RET count of jobs scheduled
* Note: We re-build the queue every time. Jobs can not only be added
* or removed from the queue, but have their priority or partition
* changed with the update_job RPC. In general nodes will be in priority
* order (by submit time), so the sorting should be pretty fast.
*/
extern int schedule(uint32_t job_limit)
{
List job_queue = NULL;
int error_code, failed_part_cnt = 0, job_cnt = 0, i;
uint32_t job_depth = 0;
job_queue_rec_t *job_queue_rec;
struct job_record *job_ptr;
struct part_record *part_ptr, **failed_parts = NULL;
bitstr_t *save_avail_node_bitmap;
/* Locks: Read config, write job, write node, read partition */
slurmctld_lock_t job_write_lock =
{ READ_LOCK, WRITE_LOCK, WRITE_LOCK, READ_LOCK };
#ifdef HAVE_BG
char *ionodes = NULL;
char tmp_char[256];
static bool backfill_sched = false;
#endif
static time_t sched_update = 0;
static bool wiki_sched = false;
static int sched_timeout = 0;
static int def_job_limit = 100;
time_t now = time(NULL), sched_start;
DEF_TIMERS;
sched_start = now;
if (sched_timeout == 0) {
sched_timeout = slurm_get_msg_timeout() / 2;
sched_timeout = MAX(sched_timeout, 1);
sched_timeout = MIN(sched_timeout, 10);
}
START_TIMER;
if (sched_update != slurmctld_conf.last_update) {
char *sched_params, *tmp_ptr;
char *sched_type = slurm_get_sched_type();
#ifdef HAVE_BG
/* On BlueGene, do FIFO only with sched/backfill */
if (strcmp(sched_type, "sched/backfill") == 0)
backfill_sched = true;
#endif
/* Disable avoiding of fragmentation with sched/wiki */
if ((strcmp(sched_type, "sched/wiki") == 0) ||
(strcmp(sched_type, "sched/wiki2") == 0))
wiki_sched = true;
xfree(sched_type);
sched_params = slurm_get_sched_params();
if (sched_params &&
(tmp_ptr = strstr(sched_params, "default_queue_depth="))) {
/* 01234567890123456789 */
i = atoi(tmp_ptr + 20);
if (i < 0) {
error("ignoring SchedulerParameters: "
"default_queue_depth value of %d", i);
} else {
def_job_limit = i;
}
}
xfree(sched_params);
sched_update = slurmctld_conf.last_update;
}
if (job_limit == 0)
job_limit = def_job_limit;
lock_slurmctld(job_write_lock);
if (!avail_front_end()) {
unlock_slurmctld(job_write_lock);
debug("sched: schedule() returning, no front end nodes are "
"available");
return SLURM_SUCCESS;
}
/* Avoid resource fragmentation if important */
if ((!wiki_sched) && job_is_completing()) {
unlock_slurmctld(job_write_lock);
debug("sched: schedule() returning, some job is still "
"completing");
return SLURM_SUCCESS;
}
#ifdef HAVE_CRAY
/*
* Run a Basil Inventory immediately before scheduling, to avoid
* race conditions caused by ALPS node state change (caused e.g.
* by the node health checker).
* This relies on the above write lock for the node state.
*/
if (select_g_reconfigure()) {
unlock_slurmctld(job_write_lock);
debug4("sched: not scheduling due to ALPS");
return SLURM_SUCCESS;
}
#endif
failed_parts = xmalloc(sizeof(struct part_record *) *
list_count(part_list));
save_avail_node_bitmap = bit_copy(avail_node_bitmap);
debug("sched: Running job scheduler");
job_queue = build_job_queue(false);
while ((job_queue_rec = list_pop_bottom(job_queue, sort_job_queue2))) {
job_ptr = job_queue_rec->job_ptr;
part_ptr = job_queue_rec->part_ptr;
xfree(job_queue_rec);
if ((time(NULL) - sched_start) >= sched_timeout) {
debug("sched: loop taking too long, breaking out");
break;
}
if (job_depth++ > job_limit) {
debug3("sched: already tested %u jobs, breaking out",
job_depth);
break;
}
if (!IS_JOB_PENDING(job_ptr))
continue; /* started in other partition */
if (job_ptr->priority == 0) { /* held */
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority);
continue;
}
/* If a patition update has occurred, then do a limit check. */
if (save_last_part_update != last_part_update) {
int fail_reason = job_limits_check(&job_ptr);
if (fail_reason != WAIT_NO_REASON) {
job_ptr->state_reason = fail_reason;
job_ptr->priority = 1;
continue;
}
} else if ((job_ptr->state_reason == WAIT_PART_TIME_LIMIT) ||
(job_ptr->state_reason == WAIT_PART_NODE_LIMIT)) {
job_ptr->start_time = 0;
job_ptr->priority = 1;
continue;
}
if (job_ptr->part_ptr != part_ptr) {
/* Cycle through partitions usable for this job */
job_ptr->part_ptr = part_ptr;
}
if ((job_ptr->resv_name == NULL) &&
_failed_partition(job_ptr->part_ptr, failed_parts,
failed_part_cnt)) {
if (job_ptr->priority != 1) { /* not system hold */
job_ptr->state_reason = WAIT_PRIORITY;
xfree(job_ptr->state_desc);
}
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u. Partition=%s.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority,
job_ptr->partition);
continue;
}
if (bit_overlap(avail_node_bitmap,
job_ptr->part_ptr->node_bitmap) == 0) {
/* All nodes DRAIN, DOWN, or
* reserved for jobs in higher priority partition */
job_ptr->state_reason = WAIT_RESOURCES;
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u. Partition=%s.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority,
job_ptr->partition);
continue;
}
if (license_job_test(job_ptr, time(NULL)) != SLURM_SUCCESS) {
job_ptr->state_reason = WAIT_LICENSES;
xfree(job_ptr->state_desc);
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority);
continue;
}
if (assoc_mgr_validate_assoc_id(acct_db_conn,
job_ptr->assoc_id,
accounting_enforce)) {
/* NOTE: This only happens if a user's account is
* disabled between when the job was submitted and
* the time we consider running it. It should be
* very rare. */
info("sched: JobId=%u has invalid account",
job_ptr->job_id);
last_job_update = time(NULL);
job_ptr->job_state = JOB_FAILED;
job_ptr->exit_code = 1;
job_ptr->state_reason = FAIL_ACCOUNT;
xfree(job_ptr->state_desc);
job_ptr->start_time = job_ptr->end_time = time(NULL);
job_completion_logger(job_ptr, false);
delete_job_details(job_ptr);
continue;
}
error_code = select_nodes(job_ptr, false, NULL);
if (error_code == ESLURM_NODES_BUSY) {
debug3("sched: JobId=%u. State=%s. Reason=%s. "
"Priority=%u. Partition=%s.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->state_reason),
job_ptr->priority, job_ptr->partition);
bool fail_by_part = true;
#ifdef HAVE_BG
/* When we use static or overlap partitioning on
* BlueGene, each job can possibly be scheduled
* independently, without impacting other jobs of
* different sizes. Therefore we sort and try to
* schedule every pending job unless the backfill
* scheduler is configured. */
if (!backfill_sched)
fail_by_part = false;
#endif
if (fail_by_part) {
/* do not schedule more jobs in this partition
* or on nodes in this partition */
failed_parts[failed_part_cnt++] =
job_ptr->part_ptr;
bit_not(job_ptr->part_ptr->node_bitmap);
bit_and(avail_node_bitmap,
job_ptr->part_ptr->node_bitmap);
bit_not(job_ptr->part_ptr->node_bitmap);
}
} else if (error_code == ESLURM_RESERVATION_NOT_USABLE) {
if (job_ptr->resv_ptr &&
job_ptr->resv_ptr->node_bitmap) {
debug3("sched: JobId=%u. State=%s. "
"Reason=%s. Priority=%u.",
job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_reason_string(job_ptr->
state_reason),
job_ptr->priority);
bit_not(job_ptr->resv_ptr->node_bitmap);
bit_and(avail_node_bitmap,
job_ptr->resv_ptr->node_bitmap);
bit_not(job_ptr->resv_ptr->node_bitmap);
} else {
/* The job has no reservation but requires
* nodes that are currently in some reservation
* so just skip over this job and try running
* the next lower priority job */
debug3("sched: JobId=%u State=%s. "
"Reason=Required nodes are reserved."
"Priority=%u",job_ptr->job_id,
job_state_string(job_ptr->job_state),
job_ptr->priority);
}
} else if (error_code == SLURM_SUCCESS) {
/* job initiated */
debug3("sched: JobId=%u initiated", job_ptr->job_id);
last_job_update = now;
#ifdef HAVE_BG
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
SELECT_JOBDATA_IONODES,
&ionodes);
if(ionodes) {
sprintf(tmp_char,"%s[%s]",
job_ptr->nodes, ionodes);
} else {
sprintf(tmp_char,"%s",job_ptr->nodes);
}
info("sched: Allocate JobId=%u BPList=%s",
job_ptr->job_id, tmp_char);
xfree(ionodes);
#else
info("sched: Allocate JobId=%u NodeList=%s #CPUs=%u",
job_ptr->job_id, job_ptr->nodes,
job_ptr->total_cpus);
#endif
if (job_ptr->batch_flag == 0)
srun_allocate(job_ptr->job_id);
else if (job_ptr->details->prolog_running == 0)
launch_job(job_ptr);
rebuild_job_part_list(job_ptr);
job_cnt++;
} else if ((error_code !=
ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE) &&
(error_code != ESLURM_NODE_NOT_AVAIL) &&
(error_code != ESLURM_ACCOUNTING_POLICY)) {
info("sched: schedule: JobId=%u non-runnable: %s",
job_ptr->job_id, slurm_strerror(error_code));
if (!wiki_sched) {
last_job_update = now;
job_ptr->job_state = JOB_FAILED;
job_ptr->exit_code = 1;
job_ptr->state_reason = FAIL_BAD_CONSTRAINTS;
xfree(job_ptr->state_desc);
job_ptr->start_time = job_ptr->end_time = now;
job_completion_logger(job_ptr, false);
delete_job_details(job_ptr);
}
}
}
save_last_part_update = last_part_update;
FREE_NULL_BITMAP(avail_node_bitmap);
avail_node_bitmap = save_avail_node_bitmap;
xfree(failed_parts);
list_destroy(job_queue);
unlock_slurmctld(job_write_lock);
END_TIMER2("schedule");
return job_cnt;
}
/*
* sort_job_queue - sort job_queue in decending priority order
* IN/OUT job_queue - sorted job queue
*/
extern void sort_job_queue(List job_queue)
{
list_sort(job_queue, sort_job_queue2);
}
/* Note this differs from the ListCmpF typedef since we want jobs sorted
* in order of decreasing priority */
extern int sort_job_queue2(void *x, void *y)
{
job_queue_rec_t *job_rec1 = (job_queue_rec_t *) x;
job_queue_rec_t *job_rec2 = (job_queue_rec_t *) y;
bool has_resv1, has_resv2;
if (slurm_job_preempt_check(job_rec1, job_rec2))
return -1;
if (slurm_job_preempt_check(job_rec2, job_rec1))
return 1;
has_resv1 = (job_rec1->job_ptr->resv_id != 0);
has_resv2 = (job_rec2->job_ptr->resv_id != 0);
if (has_resv1 && !has_resv2)
return -1;
if (!has_resv1 && has_resv2)
return 1;
if (job_rec1->job_ptr->priority < job_rec2->job_ptr->priority)
return 1;
if (job_rec1->job_ptr->priority > job_rec2->job_ptr->priority)
return -1;
return 0;
}
/*
* launch_job - send an RPC to a slurmd to initiate a batch job
* IN job_ptr - pointer to job that will be initiated
*/
extern void launch_job(struct job_record *job_ptr)
{
batch_job_launch_msg_t *launch_msg_ptr;
agent_arg_t *agent_arg_ptr;
/* Initialization of data structures */
launch_msg_ptr = (batch_job_launch_msg_t *)
xmalloc(sizeof(batch_job_launch_msg_t));
launch_msg_ptr->job_id = job_ptr->job_id;
launch_msg_ptr->step_id = NO_VAL;
launch_msg_ptr->uid = job_ptr->user_id;
launch_msg_ptr->gid = job_ptr->group_id;
launch_msg_ptr->ntasks = job_ptr->details->num_tasks;
launch_msg_ptr->nodes = xstrdup(job_ptr->nodes);
launch_msg_ptr->overcommit = job_ptr->details->overcommit;
launch_msg_ptr->open_mode = job_ptr->details->open_mode;
launch_msg_ptr->acctg_freq = job_ptr->details->acctg_freq;
launch_msg_ptr->cpus_per_task = job_ptr->details->cpus_per_task;
launch_msg_ptr->pn_min_memory = job_ptr->details->pn_min_memory;
launch_msg_ptr->restart_cnt = job_ptr->restart_cnt;
if (make_batch_job_cred(launch_msg_ptr, job_ptr)) {
error("aborting batch job %u", job_ptr->job_id);
/* FIXME: This is a kludge, but this event indicates a serious
* problem with OpenSSH and should never happen. We are
* too deep into the job launch to gracefully clean up. */
job_ptr->end_time = time(NULL);
job_ptr->time_limit = 0;
xfree(launch_msg_ptr->nodes);
xfree(launch_msg_ptr);
return;
}
launch_msg_ptr->std_err = xstrdup(job_ptr->details->std_err);
launch_msg_ptr->std_in = xstrdup(job_ptr->details->std_in);
launch_msg_ptr->std_out = xstrdup(job_ptr->details->std_out);
launch_msg_ptr->work_dir = xstrdup(job_ptr->details->work_dir);
launch_msg_ptr->ckpt_dir = xstrdup(job_ptr->details->ckpt_dir);
launch_msg_ptr->restart_dir = xstrdup(job_ptr->details->restart_dir);
launch_msg_ptr->argc = job_ptr->details->argc;
launch_msg_ptr->argv = xduparray(job_ptr->details->argc,
job_ptr->details->argv);
launch_msg_ptr->spank_job_env_size = job_ptr->spank_job_env_size;
launch_msg_ptr->spank_job_env = xduparray(job_ptr->spank_job_env_size,
job_ptr->spank_job_env);
launch_msg_ptr->script = get_job_script(job_ptr);
launch_msg_ptr->environment = get_job_env(job_ptr,
&launch_msg_ptr->envc);
launch_msg_ptr->job_mem = job_ptr->details->pn_min_memory;
launch_msg_ptr->num_cpu_groups = job_ptr->job_resrcs->cpu_array_cnt;
launch_msg_ptr->cpus_per_node = xmalloc(sizeof(uint16_t) *
job_ptr->job_resrcs->cpu_array_cnt);
memcpy(launch_msg_ptr->cpus_per_node,
job_ptr->job_resrcs->cpu_array_value,
(sizeof(uint16_t) * job_ptr->job_resrcs->cpu_array_cnt));
launch_msg_ptr->cpu_count_reps = xmalloc(sizeof(uint32_t) *
job_ptr->job_resrcs->cpu_array_cnt);
memcpy(launch_msg_ptr->cpu_count_reps,
job_ptr->job_resrcs->cpu_array_reps,
(sizeof(uint32_t) * job_ptr->job_resrcs->cpu_array_cnt));
launch_msg_ptr->select_jobinfo = select_g_select_jobinfo_copy(
job_ptr->select_jobinfo);
agent_arg_ptr = (agent_arg_t *) xmalloc(sizeof(agent_arg_t));
agent_arg_ptr->node_count = 1;
agent_arg_ptr->retry = 0;
xassert(job_ptr->batch_host);
agent_arg_ptr->hostlist = hostlist_create(job_ptr->batch_host);
agent_arg_ptr->msg_type = REQUEST_BATCH_JOB_LAUNCH;
agent_arg_ptr->msg_args = (void *) launch_msg_ptr;
/* Launch the RPC via agent */
agent_queue_request(agent_arg_ptr);
}
/*
* make_batch_job_cred - add a job credential to the batch_job_launch_msg
* IN/OUT launch_msg_ptr - batch_job_launch_msg in which job_id, step_id,
* uid and nodes have already been set
* IN job_ptr - pointer to job record
* RET 0 or error code
*/
extern int make_batch_job_cred(batch_job_launch_msg_t *launch_msg_ptr,
struct job_record *job_ptr)
{
slurm_cred_arg_t cred_arg;
job_resources_t *job_resrcs_ptr;
xassert(job_ptr->job_resrcs);
job_resrcs_ptr = job_ptr->job_resrcs;
memset(&cred_arg, 0, sizeof(slurm_cred_arg_t));
cred_arg.jobid = launch_msg_ptr->job_id;
cred_arg.stepid = launch_msg_ptr->step_id;
cred_arg.uid = launch_msg_ptr->uid;
cred_arg.job_hostlist = job_resrcs_ptr->nodes;
cred_arg.job_core_bitmap = job_resrcs_ptr->core_bitmap;
cred_arg.job_mem_limit = job_ptr->details->pn_min_memory;
cred_arg.job_nhosts = job_resrcs_ptr->nhosts;
cred_arg.job_gres_list = job_ptr->gres_list;
/* cred_arg.step_gres_list = NULL; */
#ifdef HAVE_FRONT_END
xassert(job_ptr->batch_host);
cred_arg.step_hostlist = job_ptr->batch_host;
#else
cred_arg.step_hostlist = launch_msg_ptr->nodes;
#endif
cred_arg.step_core_bitmap = job_resrcs_ptr->core_bitmap;
cred_arg.step_mem_limit = job_ptr->details->pn_min_memory;
cred_arg.cores_per_socket = job_resrcs_ptr->cores_per_socket;
cred_arg.sockets_per_node = job_resrcs_ptr->sockets_per_node;
cred_arg.sock_core_rep_count = job_resrcs_ptr->sock_core_rep_count;
launch_msg_ptr->cred = slurm_cred_create(slurmctld_config.cred_ctx,
&cred_arg);
if (launch_msg_ptr->cred)
return SLURM_SUCCESS;
error("slurm_cred_create failure for batch job %u", cred_arg.jobid);
return SLURM_ERROR;
}
static void _depend_list_del(void *dep_ptr)
{
xfree(dep_ptr);
}
/* Print a job's dependency information based upon job_ptr->depend_list */
extern void print_job_dependency(struct job_record *job_ptr)
{
ListIterator depend_iter;
struct depend_spec *dep_ptr;
char *dep_str;
info("Dependency information for job %u", job_ptr->job_id);
if ((job_ptr->details == NULL) ||
(job_ptr->details->depend_list == NULL))
return;
depend_iter = list_iterator_create(job_ptr->details->depend_list);
if (!depend_iter)
fatal("list_iterator_create memory allocation failure");
while ((dep_ptr = list_next(depend_iter))) {
if (dep_ptr->depend_type == SLURM_DEPEND_SINGLETON) {
info(" singleton");
continue;
}
else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER)
dep_str = "after";
else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_ANY)
dep_str = "afterany";
else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_NOT_OK)
dep_str = "afternotok";
else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_OK)
dep_str = "afterok";
else if (dep_ptr->depend_type == SLURM_DEPEND_EXPAND)
dep_str = "expand";
else
dep_str = "unknown";
info(" %s:%u", dep_str, dep_ptr->job_id);
}
list_iterator_destroy(depend_iter);
}
/*
* Determine if a job's dependencies are met
* RET: 0 = no dependencies
* 1 = dependencies remain
* 2 = failure (job completion code not per dependency), delete the job
*/
extern int test_job_dependency(struct job_record *job_ptr)
{
ListIterator depend_iter, job_iterator;
struct depend_spec *dep_ptr;
bool failure = false, depends = false, expands = false;
List job_queue = NULL;
bool run_now;
int count = 0;
struct job_record *qjob_ptr;
if ((job_ptr->details == NULL) ||
(job_ptr->details->depend_list == NULL))
return 0;
count = list_count(job_ptr->details->depend_list);
depend_iter = list_iterator_create(job_ptr->details->depend_list);
if (!depend_iter)
fatal("list_iterator_create memory allocation failure");
while ((dep_ptr = list_next(depend_iter))) {
bool clear_dep = false;
count--;
if ((dep_ptr->depend_type == SLURM_DEPEND_SINGLETON) &&
job_ptr->name) {
/* get user jobs with the same user and name */
job_queue = _build_user_job_list(job_ptr->user_id,
job_ptr->name);
run_now = true;
job_iterator = list_iterator_create(job_queue);
if (job_iterator == NULL)
fatal("list_iterator_create malloc failure");
while ((qjob_ptr = (struct job_record *)
list_next(job_iterator))) {
/* already running/suspended job or previously
* submitted pending job */
if (IS_JOB_RUNNING(qjob_ptr) ||
IS_JOB_SUSPENDED(qjob_ptr) ||
(IS_JOB_PENDING(qjob_ptr) &&
(qjob_ptr->job_id < job_ptr->job_id))) {
run_now = false;
break;
}
}
list_iterator_destroy(job_iterator);
list_destroy(job_queue);
/* job can run now, delete dependency */
if (run_now)
list_delete_item(depend_iter);
else
depends = true;
} else if ((dep_ptr->job_ptr->magic != JOB_MAGIC) ||
(dep_ptr->job_ptr->job_id != dep_ptr->job_id)) {
/* job is gone, dependency lifted */
clear_dep = true;
} else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER) {
if (!IS_JOB_PENDING(dep_ptr->job_ptr)) {
clear_dep = true;
} else
depends = true;
} else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_ANY) {
if (IS_JOB_FINISHED(dep_ptr->job_ptr)) {
clear_dep = true;
} else
depends = true;
} else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_NOT_OK) {
if (!IS_JOB_FINISHED(dep_ptr->job_ptr))
depends = true;
else if (!IS_JOB_COMPLETE(dep_ptr->job_ptr)) {
clear_dep = true;
} else {
failure = true;
break;
}
} else if (dep_ptr->depend_type == SLURM_DEPEND_AFTER_OK) {
if (!IS_JOB_FINISHED(dep_ptr->job_ptr))
depends = true;
else if (IS_JOB_COMPLETE(dep_ptr->job_ptr)) {
clear_dep = true;
} else {
failure = true;
break;
}
} else if (dep_ptr->depend_type == SLURM_DEPEND_EXPAND) {
time_t now = time(NULL);
expands = true;
if (IS_JOB_PENDING(dep_ptr->job_ptr)) {
depends = true;
} else if (IS_JOB_FINISHED(dep_ptr->job_ptr)) {
failure = true;
break;
} else if ((dep_ptr->job_ptr->end_time != 0) &&
(dep_ptr->job_ptr->end_time > now)) {
job_ptr->time_limit = dep_ptr->job_ptr->
end_time - now;
job_ptr->time_limit /= 60; /* sec to min */
}
if (job_ptr->details && dep_ptr->job_ptr->details) {
job_ptr->details->shared =
dep_ptr->job_ptr->details->shared;
}
} else
failure = true;
if (clear_dep) {
char *rmv_dep = xstrdup_printf(
":%u", dep_ptr->job_ptr->job_id);
xstrsubstitute(job_ptr->details->dependency,
rmv_dep, "");
xfree(rmv_dep);
list_delete_item(depend_iter);
}
}
list_iterator_destroy(depend_iter);
if (!depends && !expands && (count == 0))
xfree(job_ptr->details->dependency);
if (failure)
return 2;
if (depends)
return 1;
return 0;
}
/*
* Parse a job dependency string and use it to establish a "depend_spec"
* list of dependencies. We accept both old format (a single job ID) and
* new format (e.g. "afterok:123:124,after:128").
* IN job_ptr - job record to have dependency and depend_list updated
* IN new_depend - new dependency description
* RET returns an error code from slurm_errno.h
*/
extern int update_job_dependency(struct job_record *job_ptr, char *new_depend)
{
int rc = SLURM_SUCCESS;
uint16_t depend_type = 0;
uint32_t job_id = 0;
char *tok = new_depend, *sep_ptr, *sep_ptr2;
List new_depend_list = NULL;
struct depend_spec *dep_ptr;
struct job_record *dep_job_ptr;
char dep_buf[32];
bool expand_cnt = 0;
if (job_ptr->details == NULL)
return EINVAL;
/* Clear dependencies on NULL, "0", or empty dependency input */
job_ptr->details->expanding_jobid = 0;
if ((new_depend == NULL) || (new_depend[0] == '\0') ||
((new_depend[0] == '0') && (new_depend[1] == '\0'))) {
xfree(job_ptr->details->dependency);
if (job_ptr->details->depend_list) {
list_destroy(job_ptr->details->depend_list);
job_ptr->details->depend_list = NULL;
}
return rc;
}
new_depend_list = list_create(_depend_list_del);
if (new_depend_list == NULL)
fatal("list_create: malloc failure");
/* validate new dependency string */
while (rc == SLURM_SUCCESS) {
/* test singleton dependency flag */
if ( strncasecmp(tok, "singleton", 9) == 0 ) {
depend_type = SLURM_DEPEND_SINGLETON;
dep_ptr = xmalloc(sizeof(struct depend_spec));
dep_ptr->depend_type = depend_type;
/* dep_ptr->job_id = 0; set by xmalloc */
/* dep_ptr->job_ptr = NULL; set by xmalloc */
if (!list_append(new_depend_list, dep_ptr)) {
fatal("list_append memory allocation "
"failure for singleton");
}
if ( *(tok + 9 ) == ',' ) {
tok += 10;
continue;
}
else
break;
}
sep_ptr = strchr(tok, ':');
if ((sep_ptr == NULL) && (job_id == 0)) {
job_id = strtol(tok, &sep_ptr, 10);
if ((sep_ptr == NULL) || (sep_ptr[0] != '\0') ||
(job_id == 0) || (job_id == job_ptr->job_id)) {
rc = ESLURM_DEPENDENCY;
break;
}
/* old format, just a single job_id */
dep_job_ptr = find_job_record(job_id);
if (!dep_job_ptr) /* assume already done */
break;
snprintf(dep_buf, sizeof(dep_buf),
"afterany:%u", job_id);
new_depend = dep_buf;
dep_ptr = xmalloc(sizeof(struct depend_spec));
dep_ptr->depend_type = SLURM_DEPEND_AFTER_ANY;
dep_ptr->job_id = job_id;
dep_ptr->job_ptr = dep_job_ptr;
if (!list_append(new_depend_list, dep_ptr))
fatal("list_append memory allocation failure");
break;
} else if (sep_ptr == NULL) {
rc = ESLURM_DEPENDENCY;
break;
}
if (strncasecmp(tok, "afternotok", 10) == 0)
depend_type = SLURM_DEPEND_AFTER_NOT_OK;
else if (strncasecmp(tok, "afterany", 8) == 0)
depend_type = SLURM_DEPEND_AFTER_ANY;
else if (strncasecmp(tok, "afterok", 7) == 0)
depend_type = SLURM_DEPEND_AFTER_OK;
else if (strncasecmp(tok, "after", 5) == 0)
depend_type = SLURM_DEPEND_AFTER;
else if (strncasecmp(tok, "expand", 6) == 0) {
if (!select_g_job_expand_allow()) {
rc = ESLURM_DEPENDENCY;
break;
}
depend_type = SLURM_DEPEND_EXPAND;
} else {
rc = ESLURM_DEPENDENCY;
break;
}
sep_ptr++; /* skip over ":" */
while (rc == SLURM_SUCCESS) {
job_id = strtol(sep_ptr, &sep_ptr2, 10);
if ((sep_ptr2 == NULL) ||
(job_id == 0) || (job_id == job_ptr->job_id) ||
((sep_ptr2[0] != '\0') && (sep_ptr2[0] != ',') &&
(sep_ptr2[0] != ':'))) {
rc = ESLURM_DEPENDENCY;
break;
}
dep_job_ptr = find_job_record(job_id);
if ((depend_type == SLURM_DEPEND_EXPAND) &&
((expand_cnt++ > 0) || (dep_job_ptr == NULL) ||
(!IS_JOB_RUNNING(dep_job_ptr)) ||
(dep_job_ptr->qos_id != job_ptr->qos_id) ||
(dep_job_ptr->part_ptr == NULL) ||
(job_ptr->part_ptr == NULL) ||
(dep_job_ptr->part_ptr != job_ptr->part_ptr))) {
/* Expand only jobs in the same QOS and
* and partition */
rc = ESLURM_DEPENDENCY;
break;
}
if (depend_type == SLURM_DEPEND_EXPAND) {
job_ptr->details->expanding_jobid = job_id;
/* GRES configuration of this job must match
* the job being expanded */
xfree(job_ptr->gres);
job_ptr->gres = xstrdup(dep_job_ptr->gres);
if (job_ptr->gres_list)
list_destroy(job_ptr->gres_list);
gres_plugin_job_state_validate(job_ptr->gres,
&job_ptr->gres_list);
}
if (dep_job_ptr) { /* job still active */
dep_ptr = xmalloc(sizeof(struct depend_spec));
dep_ptr->depend_type = depend_type;
dep_ptr->job_id = job_id;
dep_ptr->job_ptr = dep_job_ptr;
if (!list_append(new_depend_list, dep_ptr)) {
fatal("list_append memory allocation "
"failure");
}
}
if (sep_ptr2[0] != ':')
break;
sep_ptr = sep_ptr2 + 1; /* skip over ":" */
}
if (sep_ptr2[0] == ',')
tok = sep_ptr2 + 1;
else
break;
}
if (rc == SLURM_SUCCESS) {
/* test for circular dependencies (e.g. A -> B -> A) */
if (_scan_depend(new_depend_list, job_ptr->job_id))
rc = ESLURM_CIRCULAR_DEPENDENCY;
}
if (rc == SLURM_SUCCESS) {
xfree(job_ptr->details->dependency);
job_ptr->details->dependency = xstrdup(new_depend);
if (job_ptr->details->depend_list)
list_destroy(job_ptr->details->depend_list);
job_ptr->details->depend_list = new_depend_list;
#if _DEBUG
print_job_dependency(job_ptr);
#endif
} else {
list_destroy(new_depend_list);
}
return rc;
}
/* Return TRUE if job_id is found in dependency_list.
* Execute recursively for each dependent job */
static bool _scan_depend(List dependency_list, uint32_t job_id)
{
bool rc = false;
ListIterator iter;
struct depend_spec *dep_ptr;
xassert(job_id);
xassert(dependency_list);
iter = list_iterator_create(dependency_list);
if (iter == NULL)
fatal("list_iterator_create malloc failure");
while (!rc && (dep_ptr = (struct depend_spec *) list_next(iter))) {
if (dep_ptr->job_id == 0) /* Singleton */
continue;
if (dep_ptr->job_id == job_id)
rc = true;
else if (dep_ptr->job_ptr->details &&
dep_ptr->job_ptr->details->depend_list) {
rc = _scan_depend(dep_ptr->job_ptr->details->
depend_list, job_id);
if (rc) {
info("circular dependency: job %u is dependent "
"upon job %u", dep_ptr->job_id, job_id);
}
}
}
list_iterator_destroy(iter);
return rc;
}
static void _pre_list_del(void *x)
{
xfree(x);
}
/* If there are higher priority queued jobs in this job's partition, then
* delay the job's expected initiation time as needed to run those jobs.
* NOTE: This is only a rough estimate of the job's start time as it ignores
* job dependencies, feature requirements, specific node requirements, etc. */
static void _delayed_job_start_time(struct job_record *job_ptr)
{
uint32_t part_node_cnt, part_cpu_cnt, part_cpus_per_node;
uint32_t job_size_cpus, job_size_nodes, job_time;
uint64_t cume_space_time = 0;
struct job_record *job_q_ptr;
ListIterator job_iterator;
if (job_ptr->part_ptr == NULL)
return;
part_node_cnt = job_ptr->part_ptr->total_nodes;
part_cpu_cnt = job_ptr->part_ptr->total_cpus;
if (part_node_cnt > part_cpu_cnt)
part_cpus_per_node = part_node_cnt / part_cpu_cnt;
else
part_cpus_per_node = 1;
job_iterator = list_iterator_create(job_list);
if (job_iterator == NULL)
fatal("list_iterator_create memory allocation failure");
while ((job_q_ptr = (struct job_record *) list_next(job_iterator))) {
if (!IS_JOB_PENDING(job_q_ptr) || !job_q_ptr->details ||
(job_q_ptr->part_ptr != job_ptr->part_ptr) ||
(job_q_ptr->priority < job_ptr->priority))
continue;
if (job_q_ptr->details->min_nodes == NO_VAL)
job_size_nodes = 1;
else
job_size_nodes = job_q_ptr->details->min_nodes;
if (job_q_ptr->details->min_cpus == NO_VAL)
job_size_cpus = 1;
else
job_size_cpus = job_q_ptr->details->min_nodes;
job_size_cpus = MAX(job_size_cpus,
(job_size_nodes * part_cpus_per_node));
if (job_ptr->time_limit == NO_VAL)
job_time = job_q_ptr->part_ptr->max_time;
else
job_time = job_q_ptr->time_limit;
cume_space_time += job_size_cpus * job_time;
}
list_iterator_destroy(job_iterator);
cume_space_time /= part_cpu_cnt;/* Factor out size */
cume_space_time *= 60; /* Minutes to seconds */
debug2("Increasing estimated start of job %u by %"PRIu64" secs",
job_ptr->job_id, cume_space_time);
job_ptr->start_time += cume_space_time;
}
/* Determine if a pending job will run using only the specified nodes
* (in job_desc_msg->req_nodes), build response message and return
* SLURM_SUCCESS on success. Otherwise return an error code. Caller
* must free response message */
extern int job_start_data(job_desc_msg_t *job_desc_msg,
will_run_response_msg_t **resp)
{
struct job_record *job_ptr;
struct part_record *part_ptr;
bitstr_t *avail_bitmap = NULL, *resv_bitmap = NULL;
uint32_t min_nodes, max_nodes, req_nodes;
int i, rc = SLURM_SUCCESS;
time_t now = time(NULL), start_res, orig_start_time = (time_t) 0;
List preemptee_candidates = NULL, preemptee_job_list = NULL;
job_ptr = find_job_record(job_desc_msg->job_id);
if (job_ptr == NULL)
return ESLURM_INVALID_JOB_ID;
part_ptr = job_ptr->part_ptr;
if (part_ptr == NULL)
return ESLURM_INVALID_PARTITION_NAME;
if ((job_ptr->details == NULL) || (!IS_JOB_PENDING(job_ptr)))
return ESLURM_DISABLED;
if ((job_desc_msg->req_nodes == NULL) ||
(job_desc_msg->req_nodes == '\0')) {
/* assume all nodes available to job for testing */
avail_bitmap = bit_alloc(node_record_count);
bit_nset(avail_bitmap, 0, (node_record_count - 1));
} else if (node_name2bitmap(job_desc_msg->req_nodes, false,
&avail_bitmap) != 0) {
return ESLURM_INVALID_NODE_NAME;
}
/* Consider only nodes in this job's partition */
if (part_ptr->node_bitmap)
bit_and(avail_bitmap, part_ptr->node_bitmap);
else
rc = ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE;
if (job_req_node_filter(job_ptr, avail_bitmap))
rc = ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE;
if (job_ptr->details->exc_node_bitmap) {
bitstr_t *exc_node_mask = NULL;
exc_node_mask = bit_copy(job_ptr->details->exc_node_bitmap);
if (exc_node_mask == NULL)
fatal("bit_copy malloc failure");
bit_not(exc_node_mask);
bit_and(avail_bitmap, exc_node_mask);
FREE_NULL_BITMAP(exc_node_mask);
}
if (job_ptr->details->req_node_bitmap) {
if (!bit_super_set(job_ptr->details->req_node_bitmap,
avail_bitmap)) {
rc = ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE;
}
}
/* Enforce reservation: access control, time and nodes */
if (job_ptr->details->begin_time)
start_res = job_ptr->details->begin_time;
else
start_res = now;
i = job_test_resv(job_ptr, &start_res, false, &resv_bitmap);
if (i != SLURM_SUCCESS)
return i;
bit_and(avail_bitmap, resv_bitmap);
FREE_NULL_BITMAP(resv_bitmap);
/* Only consider nodes that are not DOWN or DRAINED */
bit_and(avail_bitmap, avail_node_bitmap);
if (rc == SLURM_SUCCESS) {
min_nodes = MAX(job_ptr->details->min_nodes,
part_ptr->min_nodes);
if (job_ptr->details->max_nodes == 0)
max_nodes = part_ptr->max_nodes;
else
max_nodes = MIN(job_ptr->details->max_nodes,
part_ptr->max_nodes);
max_nodes = MIN(max_nodes, 500000); /* prevent overflows */
if (!job_ptr->limit_set_max_nodes &&
job_ptr->details->max_nodes)
req_nodes = max_nodes;
else
req_nodes = min_nodes;
preemptee_candidates = slurm_find_preemptable_jobs(job_ptr);
/* The orig_start is based upon the backfill scheduler data
* and considers all higher priority jobs. The logic below
* only considers currently running jobs, so the expected
* start time will almost certainly be earlier and not as
* accurate, but this algorithm is much faster. */
orig_start_time = job_ptr->start_time;
rc = select_g_job_test(job_ptr, avail_bitmap,
min_nodes, max_nodes, req_nodes,
SELECT_MODE_WILL_RUN,
preemptee_candidates,
&preemptee_job_list);
}
if (rc == SLURM_SUCCESS) {
will_run_response_msg_t *resp_data;
resp_data = xmalloc(sizeof(will_run_response_msg_t));
resp_data->job_id = job_ptr->job_id;
#ifdef HAVE_BG
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
SELECT_JOBDATA_NODE_CNT,
&resp_data->proc_cnt);
#else
resp_data->proc_cnt = job_ptr->total_cpus;
#endif
_delayed_job_start_time(job_ptr);
resp_data->start_time = MAX(job_ptr->start_time,
orig_start_time);
resp_data->start_time = MAX(resp_data->start_time, start_res);
job_ptr->start_time = 0; /* restore pending job start time */
resp_data->node_list = bitmap2node_name(avail_bitmap);
if (preemptee_job_list) {
ListIterator preemptee_iterator;
uint32_t *preemptee_jid;
struct job_record *tmp_job_ptr;
resp_data->preemptee_job_id=list_create(_pre_list_del);
if (resp_data->preemptee_job_id == NULL)
fatal("list_create: malloc failure");
preemptee_iterator = list_iterator_create(
preemptee_job_list);
while ((tmp_job_ptr = (struct job_record *)
list_next(preemptee_iterator))) {
preemptee_jid = xmalloc(sizeof(uint32_t));
(*preemptee_jid) = tmp_job_ptr->job_id;
list_append(resp_data->preemptee_job_id,
preemptee_jid);
}
list_iterator_destroy(preemptee_iterator);
}
*resp = resp_data;
} else {
rc = ESLURM_REQUESTED_NODE_CONFIG_UNAVAILABLE;
}
if (preemptee_candidates)
list_destroy(preemptee_candidates);
if (preemptee_job_list)
list_destroy(preemptee_job_list);
FREE_NULL_BITMAP(avail_bitmap);
return rc;
}
/*
* epilog_slurmctld - execute the epilog_slurmctld for a job that has just
* terminated.
* IN job_ptr - pointer to job that has been terminated
* RET SLURM_SUCCESS(0) or error code
*/
extern int epilog_slurmctld(struct job_record *job_ptr)
{
int rc;
pthread_t thread_id_epilog;
pthread_attr_t thread_attr_epilog;
if ((slurmctld_conf.epilog_slurmctld == NULL) ||
(slurmctld_conf.epilog_slurmctld[0] == '\0'))
return SLURM_SUCCESS;
if (access(slurmctld_conf.epilog_slurmctld, X_OK) < 0) {
error("Invalid EpilogSlurmctld: %m");
return errno;
}
slurm_attr_init(&thread_attr_epilog);
pthread_attr_setdetachstate(&thread_attr_epilog,
PTHREAD_CREATE_DETACHED);
while (1) {
rc = pthread_create(&thread_id_epilog,
&thread_attr_epilog,
_run_epilog, (void *) job_ptr);
if (rc == 0) {
slurm_attr_destroy(&thread_attr_epilog);
return SLURM_SUCCESS;
}
if (errno == EAGAIN)
continue;
error("pthread_create: %m");
slurm_attr_destroy(&thread_attr_epilog);
return errno;
}
}
static char **_build_env(struct job_record *job_ptr)
{
char **my_env, *name;
my_env = xmalloc(sizeof(char *));
my_env[0] = NULL;
/* Set SPANK env vars first so that we can overrite as needed
* below. Prevent user hacking from setting SLURM_JOB_ID etc. */
if (job_ptr->spank_job_env_size) {
env_array_merge(&my_env,
(const char **) job_ptr->spank_job_env);
}
#ifdef HAVE_BG
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
SELECT_JOBDATA_BLOCK_ID, &name);
setenvf(&my_env, "MPIRUN_PARTITION", "%s", name);
# ifdef HAVE_BGP
{
uint16_t conn_type = (uint16_t)NO_VAL;
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
SELECT_JOBDATA_CONN_TYPE,
&conn_type);
if (conn_type > SELECT_SMALL) {
/* SUBMIT_POOL over rides
HTC_SUBMIT_POOL */
setenvf(&my_env, "SUBMIT_POOL", "%s", name);
}
}
# endif
xfree(name);
#elif defined HAVE_CRAY
name = select_g_select_jobinfo_xstrdup(job_ptr->select_jobinfo,
SELECT_PRINT_RESV_ID);
setenvf(&my_env, "BASIL_RESERVATION_ID", "%s", name);
xfree(name);
#endif
setenvf(&my_env, "SLURM_JOB_ACCOUNT", "%s", job_ptr->account);
if (job_ptr->details) {
setenvf(&my_env, "SLURM_JOB_CONSTRAINTS",
"%s", job_ptr->details->features);
}
setenvf(&my_env, "SLURM_JOB_DERIVED_EC", "%u",
job_ptr->derived_ec);
setenvf(&my_env, "SLURM_JOB_EXIT_CODE", "%u", job_ptr->exit_code);
setenvf(&my_env, "SLURM_JOB_GID", "%u", job_ptr->group_id);
name = gid_to_string((uid_t) job_ptr->group_id);
setenvf(&my_env, "SLURM_JOB_GROUP", "%s", name);
xfree(name);
setenvf(&my_env, "SLURM_JOBID", "%u", job_ptr->job_id);
setenvf(&my_env, "SLURM_JOB_ID", "%u", job_ptr->job_id);
setenvf(&my_env, "SLURM_JOB_NAME", "%s", job_ptr->name);
setenvf(&my_env, "SLURM_JOB_NODELIST", "%s", job_ptr->nodes);
setenvf(&my_env, "SLURM_JOB_PARTITION", "%s", job_ptr->partition);
setenvf(&my_env, "SLURM_JOB_UID", "%u", job_ptr->user_id);
name = uid_to_string((uid_t) job_ptr->user_id);
setenvf(&my_env, "SLURM_JOB_USER", "%s", name);
xfree(name);
return my_env;
}
static void *_run_epilog(void *arg)
{
struct job_record *job_ptr = (struct job_record *) arg;
uint32_t job_id;
pid_t cpid;
int i, status, wait_rc;
char *argv[2], **my_env;
/* Locks: Read config, job */
slurmctld_lock_t config_read_lock = {
READ_LOCK, READ_LOCK, NO_LOCK, NO_LOCK };
lock_slurmctld(config_read_lock);
argv[0] = xstrdup(slurmctld_conf.epilog_slurmctld);
argv[1] = NULL;
my_env = _build_env(job_ptr);
job_id = job_ptr->job_id;
unlock_slurmctld(config_read_lock);
if ((cpid = fork()) < 0) {
error("epilog_slurmctld fork error: %m");
goto fini;
}
if (cpid == 0) {
#ifdef SETPGRP_TWO_ARGS
setpgrp(0, 0);
#else
setpgrp();
#endif
execve(argv[0], argv, my_env);
exit(127);
}
while (1) {
wait_rc = waitpid(cpid, &status, 0);
if (wait_rc < 0) {
if (errno == EINTR)
continue;
error("epilog_slurmctld waitpid error: %m");
break;
} else if (wait_rc > 0) {
killpg(cpid, SIGKILL); /* kill children too */
break;
}
}
if (status != 0) {
error("epilog_slurmctld job %u epilog exit status %u:%u",
job_id, WEXITSTATUS(status), WTERMSIG(status));
} else
debug2("epilog_slurmctld job %u epilog completed", job_id);
fini: xfree(argv[0]);
for (i=0; my_env[i]; i++)
xfree(my_env[i]);
xfree(my_env);
return NULL;
}
/*
* prolog_slurmctld - execute the prolog_slurmctld for a job that has just
* been allocated resources.
* IN job_ptr - pointer to job that will be initiated
* RET SLURM_SUCCESS(0) or error code
*/
extern int prolog_slurmctld(struct job_record *job_ptr)
{
int rc;
pthread_t thread_id_prolog;
pthread_attr_t thread_attr_prolog;
if ((slurmctld_conf.prolog_slurmctld == NULL) ||
(slurmctld_conf.prolog_slurmctld[0] == '\0'))
return SLURM_SUCCESS;
if (access(slurmctld_conf.prolog_slurmctld, X_OK) < 0) {
error("Invalid PrologSlurmctld: %m");
return errno;
}
if (job_ptr->details)
job_ptr->details->prolog_running = 1;
slurm_attr_init(&thread_attr_prolog);
pthread_attr_setdetachstate(&thread_attr_prolog,
PTHREAD_CREATE_DETACHED);
while (1) {
rc = pthread_create(&thread_id_prolog,
&thread_attr_prolog,
_run_prolog, (void *) job_ptr);
if (rc == 0) {
slurm_attr_destroy(&thread_attr_prolog);
return SLURM_SUCCESS;
}
if (errno == EAGAIN)
continue;
error("pthread_create: %m");
slurm_attr_destroy(&thread_attr_prolog);
return errno;
}
}
static void *_run_prolog(void *arg)
{
struct job_record *job_ptr = (struct job_record *) arg;
uint32_t job_id;
pid_t cpid;
int i, rc, status, wait_rc;
char *argv[2], **my_env;
/* Locks: Read config, job; Write nodes */
slurmctld_lock_t config_read_lock = {
READ_LOCK, READ_LOCK, WRITE_LOCK, NO_LOCK };
bitstr_t *node_bitmap = NULL;
static int last_job_requeue = 0;
lock_slurmctld(config_read_lock);
argv[0] = xstrdup(slurmctld_conf.prolog_slurmctld);
argv[1] = NULL;
my_env = _build_env(job_ptr);
job_id = job_ptr->job_id;
if (job_ptr->node_bitmap) {
node_bitmap = bit_copy(job_ptr->node_bitmap);
for (i=0; i<node_record_count; i++) {
if (bit_test(node_bitmap, i) == 0)
continue;
node_record_table_ptr[i].node_state |=
NODE_STATE_POWER_UP;
}
}
unlock_slurmctld(config_read_lock);
if ((cpid = fork()) < 0) {
error("prolog_slurmctld fork error: %m");
goto fini;
}
if (cpid == 0) {
#ifdef SETPGRP_TWO_ARGS
setpgrp(0, 0);
#else
setpgrp();
#endif
execve(argv[0], argv, my_env);
exit(127);
}
while (1) {
wait_rc = waitpid(cpid, &status, 0);
if (wait_rc < 0) {
if (errno == EINTR)
continue;
error("prolog_slurmctld waitpid error: %m");
break;
} else if (wait_rc > 0) {
killpg(cpid, SIGKILL); /* kill children too */
break;
}
}
if (status != 0) {
bool kill_job = false;
slurmctld_lock_t job_write_lock = {
NO_LOCK, WRITE_LOCK, WRITE_LOCK, NO_LOCK };
error("prolog_slurmctld job %u prolog exit status %u:%u",
job_id, WEXITSTATUS(status), WTERMSIG(status));
lock_slurmctld(job_write_lock);
if (last_job_requeue == job_id) {
info("prolog_slurmctld failed again for job %u",
job_id);
kill_job = true;
} else if ((rc = job_requeue(0, job_id, -1,
(uint16_t)NO_VAL, false))) {
info("unable to requeue job %u: %m", job_id);
kill_job = true;
} else
last_job_requeue = job_id;
if (kill_job) {
srun_user_message(job_ptr,
"PrologSlurmctld failed, job killed");
(void) job_signal(job_id, SIGKILL, 0, 0, false);
}
unlock_slurmctld(job_write_lock);
} else
debug2("prolog_slurmctld job %u prolog completed", job_id);
fini: xfree(argv[0]);
for (i=0; my_env[i]; i++)
xfree(my_env[i]);
xfree(my_env);
lock_slurmctld(config_read_lock);
if (job_ptr->job_id != job_id) {
error("prolog_slurmctld job %u pointer invalid", job_id);
job_ptr = find_job_record(job_id);
if (job_ptr == NULL)
error("prolog_slurmctld job %u now defunct", job_id);
}
if (job_ptr) {
if (job_ptr->details)
job_ptr->details->prolog_running = 0;
if (job_ptr->batch_flag &&
(IS_JOB_RUNNING(job_ptr) || IS_JOB_SUSPENDED(job_ptr)))
launch_job(job_ptr);
}
if (job_ptr && job_ptr->node_bitmap) {
for (i=0; i<node_record_count; i++) {
if (bit_test(job_ptr->node_bitmap, i) == 0)
continue;
node_record_table_ptr[i].node_state &=
(~NODE_STATE_POWER_UP);
}
} else if (node_bitmap) {
for (i=0; i<node_record_count; i++) {
if (bit_test(node_bitmap, i) == 0)
continue;
node_record_table_ptr[i].node_state &=
(~NODE_STATE_POWER_UP);
}
}
unlock_slurmctld(config_read_lock);
FREE_NULL_BITMAP(node_bitmap);
return NULL;
}
/*
* build_feature_list - Translate a job's feature string into a feature_list
* IN details->features
* OUT details->feature_list
* RET error code
*/
extern int build_feature_list(struct job_record *job_ptr)
{
struct job_details *detail_ptr = job_ptr->details;
char *tmp_requested, *str_ptr, *feature = NULL;
int bracket = 0, count = 0, i;
bool have_count = false, have_or = false;
struct feature_record *feat;
if (detail_ptr->features == NULL) /* no constraints */
return SLURM_SUCCESS;
if (detail_ptr->feature_list) /* already processed */
return SLURM_SUCCESS;
tmp_requested = xstrdup(detail_ptr->features);
detail_ptr->feature_list = list_create(_feature_list_delete);
for (i=0; ; i++) {
if (tmp_requested[i] == '*') {
tmp_requested[i] = '\0';
have_count = true;
count = strtol(&tmp_requested[i+1], &str_ptr, 10);
if ((feature == NULL) || (count <= 0)) {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
}
i = str_ptr - tmp_requested - 1;
} else if (tmp_requested[i] == '&') {
tmp_requested[i] = '\0';
if ((feature == NULL) || (bracket != 0)) {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
}
feat = xmalloc(sizeof(struct feature_record));
feat->name = xstrdup(feature);
feat->count = count;
feat->op_code = FEATURE_OP_AND;
list_append(detail_ptr->feature_list, feat);
feature = NULL;
count = 0;
} else if (tmp_requested[i] == '|') {
tmp_requested[i] = '\0';
have_or = true;
if (feature == NULL) {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
}
feat = xmalloc(sizeof(struct feature_record));
feat->name = xstrdup(feature);
feat->count = count;
if (bracket)
feat->op_code = FEATURE_OP_XOR;
else
feat->op_code = FEATURE_OP_OR;
list_append(detail_ptr->feature_list, feat);
feature = NULL;
count = 0;
} else if (tmp_requested[i] == '[') {
tmp_requested[i] = '\0';
if ((feature != NULL) || bracket) {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
}
bracket++;
} else if (tmp_requested[i] == ']') {
tmp_requested[i] = '\0';
if ((feature == NULL) || (bracket == 0)) {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
}
bracket = 0;
} else if (tmp_requested[i] == '\0') {
if (feature) {
feat = xmalloc(sizeof(struct feature_record));
feat->name = xstrdup(feature);
feat->count = count;
feat->op_code = FEATURE_OP_END;
list_append(detail_ptr->feature_list, feat);
}
break;
} else if (tmp_requested[i] == ',') {
info("Job %u invalid constraint %s",
job_ptr->job_id, detail_ptr->features);
xfree(tmp_requested);
return ESLURM_INVALID_FEATURE;
} else if (feature == NULL) {
feature = &tmp_requested[i];
}
}
xfree(tmp_requested);
if (have_count && have_or) {
info("Job %u invalid constraint (OR with feature count): %s",
job_ptr->job_id, detail_ptr->features);
return ESLURM_INVALID_FEATURE;
}
return _valid_feature_list(job_ptr->job_id, detail_ptr->feature_list);
}
static void _feature_list_delete(void *x)
{
struct feature_record *feature = (struct feature_record *)x;
xfree(feature->name);
xfree(feature);
}
static int _valid_feature_list(uint32_t job_id, List feature_list)
{
ListIterator feat_iter;
struct feature_record *feat_ptr;
char *buf = NULL, tmp[16];
int bracket = 0;
int rc = SLURM_SUCCESS;
if (feature_list == NULL) {
debug2("Job %u feature list is empty", job_id);
return rc;
}
feat_iter = list_iterator_create(feature_list);
while ((feat_ptr = (struct feature_record *)list_next(feat_iter))) {
if (feat_ptr->op_code == FEATURE_OP_XOR) {
if (bracket == 0)
xstrcat(buf, "[");
bracket = 1;
}
xstrcat(buf, feat_ptr->name);
if (rc == SLURM_SUCCESS)
rc = _valid_node_feature(feat_ptr->name);
if (feat_ptr->count) {
snprintf(tmp, sizeof(tmp), "*%u", feat_ptr->count);
xstrcat(buf, tmp);
}
if (bracket && (feat_ptr->op_code != FEATURE_OP_XOR)) {
xstrcat(buf, "]");
bracket = 0;
}
if (feat_ptr->op_code == FEATURE_OP_AND)
xstrcat(buf, "&");
else if ((feat_ptr->op_code == FEATURE_OP_OR) ||
(feat_ptr->op_code == FEATURE_OP_XOR))
xstrcat(buf, "|");
}
list_iterator_destroy(feat_iter);
if (rc == SLURM_SUCCESS)
debug("Job %u feature list: %s", job_id, buf);
else
info("Job %u has invalid feature list: %s", job_id, buf);
xfree(buf);
return rc;
}
static int _valid_node_feature(char *feature)
{
int rc = ESLURM_INVALID_FEATURE;
struct features_record *feature_ptr;
ListIterator feature_iter;
/* Clear these nodes from the feature_list record,
* then restore as needed */
feature_iter = list_iterator_create(feature_list);
if (feature_iter == NULL)
fatal("list_inerator_create malloc failure");
while ((feature_ptr = (struct features_record *)
list_next(feature_iter))) {
if (strcmp(feature_ptr->name, feature))
continue;
rc = SLURM_SUCCESS;
break;
}
list_iterator_destroy(feature_iter);
return rc;
}
/* If a job can run in multiple partitions, make sure that the one
* actually used is first in the string. Needed for job state save/restore */
extern void rebuild_job_part_list(struct job_record *job_ptr)
{
ListIterator part_iterator;
struct part_record *part_ptr;
if ((job_ptr->part_ptr_list == NULL) || (job_ptr->part_ptr == NULL))
return;
xfree(job_ptr->partition);
job_ptr->partition = xstrdup(job_ptr->part_ptr->name);
part_iterator = list_iterator_create(job_ptr->part_ptr_list);
if (part_iterator == NULL)
fatal("list_iterator_create malloc failure");
while ((part_ptr = (struct part_record *) list_next(part_iterator))) {
if (part_ptr == job_ptr->part_ptr)
continue;
xstrcat(job_ptr->partition, ",");
xstrcat(job_ptr->partition, part_ptr->name);
}
list_iterator_destroy(part_iterator);
}