Support defered initiation of job (e.g. srun --begin=11:30 ...).

NEWS

 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 0.7.0-pre1
+=============================
+ -- Support defered initiation of job (e.g. srun --begin=11:30 ...).
+
 * Changes in SLURM 0.6.0-pre6
 =============================
  -- Added logic to return scheduled nodes to Maui scheduler (David

doc/man/man1/scontrol.1

-.TH SCONTROL "1" "August 2005" "scontrol 0.6" "Slurm components"
+.TH SCONTROL "1" "September 2005" "scontrol 0.7" "Slurm components"
 
 .SH "NAME"
 scontrol \- Used view and modify Slurm configuration and state.
@@ -222,6 +222,24 @@ Set the job's count of required processors to the specified value.
 Set the job's ability to share nodes with other jobs. Possible values are
 "YES" and "NO".
 .TP
+\fIStartTime\fP=<time_spec>
+Set the job's earliest initiation time.
+It accepts times of the form \fIHH:MM:SS\fR to run a job at
+a specific time of day (seconds are optional).
+(If that time is already past, the next day is assumed.)
+You may also specify \fImidnight\fR, \fInoon\fR, or
+\fIteatime\fR (4pm) and you can have a time-of-day suffixed
+with \fIAM\fR or \fIPM\fR for running in the morning or the evening.
+You can also say what day the job will be run, by giving
+a date in the form \fImonth-name\fR day with an optional year,
+or giving a date of the form \fIMMDDYY\fR or \fIMM/DD/YY\fR
+or \fIDD.MM.YY\fR. You can also
+give times like \fInow + count time-units\fR, where the time-units
+can be \fIminutes\fR, \fIhours\fR, \fIdays\fR, or \fIweeks\fR
+and you can tell SLURM to run the job today with the keyword
+\fItoday\fR and to run the job tomorrow with the keyword
+\fItomorrow\fR.
+.TP
 \fITimeLimit\fP=<minutes>
 Set the job's time limit to the specified value.
 .TP

doc/man/man1/srun.1

 \." $Id$
 .\"
-.TH SRUN "1" "August 2005" "srun 0.6" "slurm components"
+.TH SRUN "1" "September 2005" "srun 0.7" "slurm components"
 .SH "NAME"
 srun \- run parallel jobs
 .SH SYNOPSIS
@@ -75,6 +75,26 @@ dependency and these jobs may belong to different users.
 The value may be changed after job submission using the 
 \fBscontrol\fR command.
 .TP
+\fB\-\-begin\fR=\fItime\fR
+Defer initiation of this job until the specified time.
+It accepts times of the form \fIHH:MM:SS\fR to run a job at 
+a specific time of day (seconds are optional).
+(If that time is already past, the next day is assumed.) 
+You may also specify \fImidnight\fR, \fInoon\fR, or 
+\fIteatime\fR (4pm) and you can have a time-of-day suffixed 
+with \fIAM\fR or \fIPM\fR for running in the morning or the evening.  
+You can also say what day the job will be run, by giving 
+a date in the form \fImonth-name\fR day with an optional year,
+or giving a date of the form \fIMMDDYY\fR or \fIMM/DD/YY\fR 
+or \fIDD.MM.YY\fR. You can also 
+give times like \fInow + count time-units\fR, where the time-units
+can be \fIminutes\fR, \fIhours\fR, \fIdays\fR, or \fIweeks\fR 
+and you can tell SLURM to run the job today with the keyword
+\fItoday\fR and to run the job tomorrow with the keyword
+\fItomorrow\fR.
+The value may be changed after job submission using the
+\fBscontrol\fR command.
+.TP
 \fB\-U\fR, \fB\-\-account\fR=\fIaccount\fR
 Change resource use by this job to specified account.
 The \fIaccount\fR is an arbitrary string. The may 

slurm/slurm.h.in

@@ -154,7 +154,8 @@ enum job_wait_reason {
 	WAIT_PART_NODE_LIMIT,	/* request exceeds partition node limit */
 	WAIT_PART_TIME_LIMIT,	/* request exceeds partition time limit */
 	WAIT_PART_STATE,	/* requested partition is down */
-	WAIT_HELD		/* job is held, priority==0 */
+	WAIT_HELD,		/* job is held, priority==0 */
+	WAIT_TIME		/* job waiting for specific begin time */
 };
 
 #ifdef HAVE_BGL
@@ -306,6 +307,7 @@ typedef struct job_descriptor {	/* For submit, allocate, and update requests */
 				 * 0 otherwise, default=0. Only useful
 				 * when Consumable Resources are
 				 * enabled */
+	time_t begin_time;	/* delay initiation until this time */
 /*
  * The following parameters are only meaningful on a Blue Gene
  * system at present. Some will be of value on other system.

src/api/init_msg.c

@@ -80,6 +80,7 @@ void slurm_init_job_desc_msg(job_desc_msg_t * job_desc_msg)
 	job_desc_msg->work_dir    = NULL;
 	job_desc_msg->host        = NULL;
 	job_desc_msg->port        = 0;
+	job_desc_msg->begin_time  = 0;
 #if SYSTEM_DIMENSIONS
 {
 	int i;

src/common/slurm_protocol_defs.c

@@ -462,6 +462,8 @@ extern char *job_reason_string(enum job_wait_reason inx)
 			return "PartitionDown";
 		case WAIT_HELD:
 			return "JobHeld";
+		case WAIT_TIME:
+			return "BeginTime";
 		default:
 			return "?";
 	}

src/common/slurm_protocol_pack.c

@@ -2043,6 +2043,7 @@ _pack_job_desc_msg(job_desc_msg_t * job_desc_ptr, Buf buffer)
 	pack16(job_desc_ptr->port, buffer);
 	packstr(job_desc_ptr->host, buffer);
 	packstr(job_desc_ptr->network, buffer);
+	pack_time(job_desc_ptr->begin_time, buffer);
 
 	if (select_g_alloc_jobinfo (&jobinfo) == SLURM_SUCCESS) {
 #if SYSTEM_DIMENSIONS
@@ -2120,6 +2121,7 @@ _unpack_job_desc_msg(job_desc_msg_t ** job_desc_buffer_ptr, Buf buffer)
 	safe_unpack16(&job_desc_ptr->port, buffer);
 	safe_unpackstr_xmalloc(&job_desc_ptr->host, &uint16_tmp, buffer);
 	safe_unpackstr_xmalloc(&job_desc_ptr->network, &uint16_tmp, buffer);
+	safe_unpack_time(&job_desc_ptr->begin_time, buffer);
 
 	if (select_g_alloc_jobinfo (&job_desc_ptr->select_jobinfo)
 	||  select_g_unpack_jobinfo(job_desc_ptr->select_jobinfo, buffer))

src/scontrol/scontrol.c

@@ -67,6 +67,7 @@
 #include "src/common/hostlist.h"
 #include "src/common/log.h"
 #include "src/common/parse_spec.h"
+#include "src/common/parse_time.h"
 #include "src/common/read_config.h"
 #include "src/common/slurm_protocol_api.h"
 #include "src/common/xmalloc.h"
@@ -1495,6 +1496,9 @@ _update_job (int argc, char *argv[])
 					(uint16_t) strtol(&argv[i][11], 
 							(char **) NULL, 10);
 		}
+		else if (strncasecmp(argv[i], "StartTime=", 10) == 0) {
+			job_msg.begin_time = parse_time(&argv[i][10]);
+		}
 		else {
 			exit_code = 1;
 			fprintf (stderr, "Invalid input: %s\n", argv[i]);

src/slurmctld/job_mgr.c

@@ -114,6 +114,7 @@ static int  _list_find_job_old(void *job_entry, void *key);
 static int  _load_job_details(struct job_record *job_ptr, Buf buffer);
 static int  _load_job_state(Buf buffer);
 static int  _load_step_state(struct job_record *job_ptr, Buf buffer);
+static void _make_time_str (time_t *time, char *string);
 static void _pack_job_details(struct job_details *detail_ptr, Buf buffer);
 static int  _purge_job_record(uint32_t job_id);
 static void _purge_lost_batch_jobs(int node_inx, time_t now);
@@ -693,6 +694,7 @@ void _dump_job_details(struct job_details *detail_ptr, Buf buffer)
 	pack32((uint32_t) detail_ptr->min_procs, buffer);
 	pack32((uint32_t) detail_ptr->min_memory, buffer);
 	pack32((uint32_t) detail_ptr->min_tmp_disk, buffer);
+	pack_time(detail_ptr->begin_time, buffer);
 	pack_time(detail_ptr->submit_time, buffer);
 
 	packstr(detail_ptr->req_nodes, buffer);
@@ -716,7 +718,7 @@ static int _load_job_details(struct job_record *job_ptr, Buf buffer)
 	uint32_t min_nodes, max_nodes, min_procs;
 	uint16_t argc = 0, req_tasks, shared, contiguous, name_len;
 	uint32_t min_memory, min_tmp_disk, total_procs;
-	time_t submit_time;
+	time_t begin_time, submit_time;
 	int i;
 
 	/* unpack the job's details from the buffer */
@@ -731,6 +733,7 @@ static int _load_job_details(struct job_record *job_ptr, Buf buffer)
 	safe_unpack32(&min_procs, buffer);
 	safe_unpack32(&min_memory, buffer);
 	safe_unpack32(&min_tmp_disk, buffer);
+	safe_unpack_time(&begin_time, buffer);
 	safe_unpack_time(&submit_time, buffer);
 
 	safe_unpackstr_xmalloc(&req_nodes, &name_len, buffer);
@@ -774,6 +777,7 @@ static int _load_job_details(struct job_record *job_ptr, Buf buffer)
 	job_ptr->details->min_procs = min_procs;
 	job_ptr->details->min_memory = min_memory;
 	job_ptr->details->min_tmp_disk = min_tmp_disk;
+	job_ptr->details->begin_time = begin_time;
 	job_ptr->details->submit_time = submit_time;
 	job_ptr->details->req_nodes = req_nodes;
 	job_ptr->details->exc_nodes = exc_nodes;
@@ -1198,10 +1202,11 @@ void dump_job_desc(job_desc_msg_t * job_specs)
 	dependency = (job_specs->dependency != NO_VAL) ?
                         (long) job_specs->dependency : -1L;
 	debug3("   host=%s port=%u dependency=%ld account=%s",
-	       job_specs->host, job_specs->port,
-	       dependency, job_specs->account);
+		job_specs->host, job_specs->port,
+		dependency, job_specs->account);
 
-	debug3("   network=%s", job_specs->network);
+	_make_time_str(&job_specs->begin_time, buf);
+	debug3("   network=%s begin=%s", job_specs->network, buf);
 
 	select_g_sprint_jobinfo(job_specs->select_jobinfo, 
 		buf, sizeof(buf), SELECT_PRINT_MIXED);
@@ -1209,6 +1214,21 @@ void dump_job_desc(job_desc_msg_t * job_specs)
 		debug3("   %s", buf);
 }
 
+static void _make_time_str (time_t *time, char *string)
+{
+	struct tm time_tm;
+
+	localtime_r (time, &time_tm);
+	if ( *time == (time_t) 0 ) {
+		sprintf( string, "N/A" );
+	} else {
+		sprintf ( string, "%2.2u/%2.2u-%2.2u:%2.2u:%2.2u",
+			(time_tm.tm_mon+1), time_tm.tm_mday,
+			time_tm.tm_hour, time_tm.tm_min, time_tm.tm_sec);
+	}
+}
+
+
 
 /* 
  * init_job_conf - initialize the job configuration tables and values. 
@@ -2221,6 +2241,7 @@ _copy_job_desc_to_job_record(job_desc_msg_t * job_desc,
 		detail_ptr->out = xstrdup(job_desc->out);
 	if (job_desc->work_dir)
 		detail_ptr->work_dir = xstrdup(job_desc->work_dir);
+	detail_ptr->begin_time = job_desc->begin_time;
 
 	if (select_g_alloc_jobinfo(&job_ptr->select_jobinfo))
 		return SLURM_ERROR;
@@ -2613,7 +2634,14 @@ void pack_job(struct job_record *dump_job_ptr, Buf buffer)
 	pack32(dump_job_ptr->alloc_sid, buffer);
 	pack32(dump_job_ptr->time_limit, buffer);
 
-	pack_time(dump_job_ptr->start_time, buffer);
+	if (IS_JOB_PENDING(dump_job_ptr)) {
+		if (dump_job_ptr->details)
+			pack_time(dump_job_ptr->details->begin_time,
+				buffer);
+		else
+			pack_time((time_t) 0, buffer);
+	} else
+		pack_time(dump_job_ptr->start_time, buffer);
 	pack_time(dump_job_ptr->end_time, buffer);
 	pack32(dump_job_ptr->priority, buffer);
 
@@ -3215,6 +3243,13 @@ int update_job(job_desc_msg_t * job_specs, uid_t uid)
 		}
 	}
 
+	if (job_specs->begin_time) {
+		if (IS_JOB_PENDING(job_ptr) && detail_ptr)
+			detail_ptr->begin_time = job_specs->begin_time;
+		else
+			error_code = ESLURM_DISABLED;
+	}
+
 #if SYSTEM_DIMENSIONS
 	if (job_specs->geometry[0] != (uint16_t) NO_VAL) {
 		int i, tot = 1;
@@ -3682,7 +3717,8 @@ extern void job_completion_logger(struct job_record  *job_ptr)
 }
 
 /*
- * job_independent - determine if this job has a depenentent job pending
+ * job_independent - determine if this job has a depenendent job pending
+ *	or if the job's scheduled begin time is in the future
  * IN job_ptr - pointer to job being tested
  * RET - true if job no longer must be defered for another job
  */
@@ -3691,6 +3727,11 @@ extern bool job_independent(struct job_record *job_ptr)
 	struct job_record *dep_ptr;
 	struct job_details *detail_ptr = job_ptr->details;
 
+	if (detail_ptr && (detail_ptr->begin_time > time(NULL))) {
+		detail_ptr->wait_reason = WAIT_TIME;
+		return false;	/* not yet time */
+	}
+		
 	if (job_ptr->dependency == 0)
 		return true;
 

src/slurmctld/slurmctld.h

@@ -261,6 +261,7 @@ struct job_details {
 	uint32_t total_procs;		/* number of allocated processors, 
 					   for accounting */
 	time_t submit_time;		/* time of submission */
+	time_t begin_time;		/* start after this time */
 	char *work_dir;			/* pathname of working directory */
 	char **argv;			/* arguments for a batch job script */
 	uint16_t argc;			/* count of argv elements */
@@ -739,6 +740,7 @@ extern int job_complete (uint32_t job_id, uid_t uid, bool requeue,
 
 /*
  * job_independent - determine if this job has a depenentent job pending
+ *	or if the job's scheduled begin time is in the future
  * IN job_ptr - pointer to job being tested
  * RET - true if job no longer must be defered for another job
  */

src/srun/allocate.c

@@ -373,6 +373,8 @@ job_desc_msg_create_from_opts (char *script)
 	j->dependency     = opt.dependency;
 	j->exclusive      = opt.exclusive;
 	j->group_id       = opt.gid;
+	if (opt.begin)
+		j->begin_time = opt.begin;
 	if (opt.network)
 		j->network = xstrdup(opt.network);
 	if (opt.account)

src/srun/opt.c

@@ -57,6 +57,7 @@
 
 #include "src/common/list.h"
 #include "src/common/log.h"
+#include "src/common/parse_time.h"
 #include "src/common/slurm_protocol_api.h"
 #include "src/common/uid.h"
 #include "src/common/xmalloc.h"
@@ -105,6 +106,7 @@
 #define LONG_OPT_PROPAGATE 0x116
 #define LONG_OPT_PROLOG   0x117
 #define LONG_OPT_EPILOG   0x118
+#define LONG_OPT_BEGIN    0x119
 
 /*---- forward declarations of static functions  ----*/
 
@@ -771,6 +773,7 @@ void set_options(const int argc, char **argv, int first)
 		{"propagate",        optional_argument, 0, LONG_OPT_PROPAGATE},
 		{"prolog",           required_argument, 0, LONG_OPT_PROLOG},
 		{"epilog",           required_argument, 0, LONG_OPT_EPILOG},
+		{"begin",            required_argument, 0, LONG_OPT_BEGIN},
 		{NULL,               0,                 0, 0}
 	};
 	char *opt_string = "+a:Abc:C:d:D:e:g:Hi:IjJ:kKlm:n:N:"
@@ -1158,6 +1161,9 @@ void set_options(const int argc, char **argv, int first)
 			xfree(opt.epilog);
 			opt.epilog = xstrdup(optarg);
 			break;
+		case LONG_OPT_BEGIN:
+			opt.begin = parse_time(optarg);
+			break;
 		}
 	}
 }
@@ -1572,6 +1578,10 @@ static void _opt_list()
 	info("network        : %s", opt.network);
 	info("propagate      : %s",
 	     opt.propagate == NULL ? "NONE" : opt.propagate);
+	if (opt.begin) {
+		info("begin          : %s", 
+			asctime(localtime(&opt.begin)));
+	}
 	str = print_commandline();
 	info("remote command : `%s'", str);
 	xfree(str);
@@ -1650,6 +1660,7 @@ static void _help(void)
 "      --mpi=type              specifies version of MPI to use\n"
 "      --prolog=program        run \"program\" before launching job step\n"
 "      --epilog=program        run \"program\" after launching job step\n"
+"      --begin=time            defer job until HH:MM DD/MM/YY\n"
 "\n"
 "Allocate only:\n"
 "  -A, --allocate              allocate resources and spawn a shell\n"

src/srun/opt.h

 /*****************************************************************************\
  *  opt.h - definitions for srun option processing
+ *  $Id$
  *****************************************************************************
  *  Copyright (C) 2002 The Regents of the University of California.
  *  Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
@@ -31,6 +32,7 @@
 #  include "config.h"
 #endif
 
+#include <time.h>
 #include <sys/types.h>
 #include <unistd.h>
 
@@ -159,6 +161,7 @@ typedef struct srun_options {
 	int conn_type;		/* --conn-type 			*/
 	char *prolog;           /* --prolog                     */
 	char *epilog;           /* --epilog                     */
+	time_t begin;		/* --begin			*/
 } opt_t;
 
 opt_t opt;