From e05a1cf6570c2c777e0245cd688d23b209860f2d Mon Sep 17 00:00:00 2001
From: Mehdi Dogguy <mehdi@debian.org>
Date: Mon, 8 Sep 2014 21:34:16 +0200
Subject: [PATCH] Imported Upstream version 2.1.9
---
META | 4 +-
NEWS | 39 +-
doc/man/man1/sacct.1 | 38 +-
doc/man/man1/salloc.1 | 3 +-
doc/man/man1/sbatch.1 | 3 +-
doc/man/man1/scancel.1 | 6 +
doc/man/man1/sreport.1 | 4 +
doc/man/man1/srun.1 | 5 +-
doc/man/man5/bluegene.conf.5 | 6 +-
etc/slurmdbd.conf.example | 2 +-
slurm.spec | 6 +-
src/common/job_resources.c | 5 +-
src/common/node_select.h | 7 +-
.../filetxt/filetxt_jobacct_process.c | 4 +-
.../mysql/accounting_storage_mysql.c | 30 +-
.../mysql/mysql_jobacct_process.c | 13 +-
.../pgsql/pgsql_jobacct_process.c | 8 +-
.../jobacct_gather/aix/jobacct_gather_aix.c | 2 +-
.../linux/jobacct_gather_linux.c | 30 +-
.../multifactor/priority_multifactor.c | 2 +-
src/plugins/proctrack/linuxproc/kill_tree.c | 8 +-
src/plugins/sched/backfill/backfill.c | 2 +-
.../select/bluegene/plugin/bg_job_place.c | 72 ++--
src/plugins/select/bluegene/plugin/bluegene.c | 3 +-
src/plugins/select/linear/select_linear.c | 13 +-
src/sacct/options.c | 4 +
src/sacct/print.c | 16 +-
src/sacct/sacct.h | 4 +-
src/sacctmgr/sacctmgr.c | 12 +-
src/salloc/opt.c | 5 +
src/sbatch/opt.c | 5 +
src/slurmctld/controller.c | 9 +-
src/slurmctld/job_mgr.c | 43 --
src/slurmctld/job_scheduler.c | 8 +-
src/slurmctld/node_mgr.c | 22 +-
src/slurmctld/node_scheduler.c | 48 +--
src/slurmctld/slurmctld.h | 7 -
src/slurmd/slurmd/req.c | 11 +-
src/slurmd/slurmstepd/mgr.c | 3 +-
src/sreport/sreport.c | 25 ++
src/srun/allocate.c | 1 +
src/sstat/print.c | 6 +-
src/sstat/sstat.c | 13 +-
src/sview/common.c | 83 +++-
src/sview/sview.c | 12 +-
src/sview/sview.h | 2 +
testsuite/expect/Makefile.am | 1 +
testsuite/expect/README | 2 +
testsuite/expect/globals | 31 ++
testsuite/expect/test21.21 | 243 +++++++++++-
testsuite/expect/test21.27 | 2 +-
testsuite/expect/test21.28 | 366 ++++++++++++++++++
52 files changed, 1020 insertions(+), 279 deletions(-)
create mode 100755 testsuite/expect/test21.28
diff --git a/META b/META
index 3ef94ea78..aef01630c 100644
--- a/META
+++ b/META
@@ -3,9 +3,9 @@
Api_revision: 0
Major: 2
Meta: 1
- Micro: 8
+ Micro: 9
Minor: 1
Name: slurm
Release: 1
Release_tags: dist
- Version: 2.1.8
+ Version: 2.1.9
diff --git a/NEWS b/NEWS
index 85c97fe69..6e070ec89 100644
--- a/NEWS
+++ b/NEWS
@@ -1,6 +1,43 @@
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 2.1.9
+========================
+ -- In select/linear - Fix logic to prevent over-subscribing memory with shared
+ nodes (Shared=YES or Shared=FORCE).
+ -- Fix for handling -N and --ntasks-per-node without specifying -n with
+ salloc and sbatch.
+ -- Fix jobacct_gather/linux if not polling on tasks to give tasks time to
+ start before doing initial gather.
+ -- When changing priority with the multifactor plugin we make sure we update
+ the last_job_update variable.
+ -- Fixed sview for gtk < 2.10 to display correct debug level at first.
+ -- Fixed sview to not select too fast when using a mouse right click.
+ -- Fixed sacct to display correct timelimits for jobs from accounting.
+ -- Fixed sacct when running as root by default query all users as documented.
+ -- In proctrack/linuxproc, skip over files in /proc that are not really user
+ processes (e.g. "/proc/bus").
+ -- Fix documentation bug for slurmdbd.conf
+ -- Fix slurmctld to update qos preempt list without restart.
+ -- Fix bug in select/cons_res that in some cases would prevent a preempting job
+ from using of resources already allocated to a preemptable running job.
+ -- Fix for sreport in interactive mode to honor parsable/2 options.
+ -- Fixed minor bugs in sacct and sstat commands
+ -- BLUEGENE - Fixed issue if the slurmd becomes unresponsive and you have
+ blocks in an error state accounting is correct when the slurmd comes
+ back up.
+ -- Corrected documentation for -n option in srun/salloc/sbatch
+ -- BLUEGENE - when running a willrun test along with preemption the bluegene
+ plugin now does the correct thing.
+ -- Fix possible memory corruption issue which can cause slurmctld to abort.
+ -- BLUEGENE - fixed small memory leak when setting up env.
+ -- Fixed deadlock if using accounting and cluster changes size in the
+ database. This can happen if you mistakenly have multiple primary
+ slurmctld's running for a single cluster, which should rarely if ever
+ happen.
+ -- Fixed sacct -c option.
+ -- Critical bug fix in sched/backfill plugin that caused in memory corruption.
+
* Changes in SLURM 2.1.8
========================
-- Update BUILD_NOTES for AIX and bgp systems on how to get sview to
@@ -5037,4 +5074,4 @@ documents those changes that are of interest to users and admins.
-- Change directory to /tmp in slurmd if daemonizing.
-- Logfiles are reopened on reconfigure.
-$Id: NEWS 20218 2010-05-07 21:29:56Z jette $
+$Id: NEWS 20444 2010-06-08 22:20:32Z jette $
diff --git a/doc/man/man1/sacct.1 b/doc/man/man1/sacct.1
index fd452fc79..8e0d41d64 100644
--- a/doc/man/man1/sacct.1
+++ b/doc/man/man1/sacct.1
@@ -625,25 +625,26 @@ was submitted. The format of the output is identical to that of the end field.
How long the job was suspended for.
.TP
-\f3systemcpu\fP
-The amount of system CPU time. (If job was running on multiple cpus this
-is a combination of all the times so this number could be much larger
-than the elapsed time.)
-The format of the output is identical to that of the
-\f3elapsed\fP
-field.
+\f3SystemCPU\fP
+The amount of system CPU time used by the job or job step. The format
+of the output is identical to that of the \f3elapsed\fP field.
+
+NOTE: SystemCPU provides a measure of the task's parent process and
+does not include CPU time of child processes.
.TP
\f3timelimit\fP
What the timelimit was/is for the job.
.TP
-\f3totalcpu\fP
-The total amount CPU time actually used by the job, not just
-accounted for (which most likely is a higher number). (If job was
-running on multiple cpus this is a combination of all the times so
-this number could be much larger than the elapsed time.) The format of
-the output is identical to that of the elapsed field.
+\f3TotalCPU\fP
+The sum of the SystemCPU and UserCPU time used by the job or job step.
+The total CPU time of the job may exceed the job's elapsed time for
+jobs that include multiple job steps. The format of the output is
+identical to that of the \f3elapsed\fP field.
+
+NOTE: TotalCPU provides a measure of the task's parent process and
+does not include CPU time of child processes.
.TP
\f3uid\fP
@@ -654,11 +655,12 @@ The user identifier of the user who ran the job.
The user name of the user who ran the job.
.TP
-\f3usercpu\fP
-The amount of user CPU time. (If job was running on multiple cpus
-this is a combination of all the times so this number could be
-much larger than the elapsed time.) The format of the output is
-identical to that of the elapsed field.
+\f3UserCPU\fP
+The amount of user CPU time used by the job or job step. The format
+of the output is identical to that of the \f3elapsed\fP field.
+
+NOTE: UserCPU provides a measure of the task's parent process and does
+not include CPU time of child processes.
.TP
\f3wckey\fP
diff --git a/doc/man/man1/salloc.1 b/doc/man/man1/salloc.1
index beb5d4b2d..414d43818 100644
--- a/doc/man/man1/salloc.1
+++ b/doc/man/man1/salloc.1
@@ -626,8 +626,7 @@ salloc does not launch tasks, it requests an allocation of resources and
executed some command. This option advises the SLURM controller that job
steps run within this allocation will launch a maximum of \fInumber\fR
tasks and sufficient resources are allocated to accomplish this.
-The default is one task per socket or core (depending upon the value
-of the \fISelectTypeParameters\fR parameter in slurm.conf), but note
+The default is one task per node, but note
that the \fB\-\-cpus\-per\-task\fR option will change this default.
.TP
diff --git a/doc/man/man1/sbatch.1 b/doc/man/man1/sbatch.1
index 53e286a59..c62b53894 100644
--- a/doc/man/man1/sbatch.1
+++ b/doc/man/man1/sbatch.1
@@ -655,8 +655,7 @@ sbatch does not launch tasks, it requests an allocation of resources and
submits a batch script. This option advises the SLURM controller that job
steps run within this allocation will launch a maximum of \fInumber\fR
tasks and sufficient resources are allocated to accomplish this.
-The default is one task per socket or core (depending upon the value
-of the \fISelectTypeParameters\fR parameter in slurm.conf), but note
+The default is one task per node, but note
that the \fB\-\-cpus\-per\-task\fR option will change this default.
.TP
diff --git a/doc/man/man1/scancel.1 b/doc/man/man1/scancel.1
index ce3e2b6cf..4c6169fde 100644
--- a/doc/man/man1/scancel.1
+++ b/doc/man/man1/scancel.1
@@ -173,6 +173,12 @@ If a signal value of "KILL" (the default value) is to be sent to an entire
job, this will result in the job's termination and its resource allocation
being released.
.LP
+Specifying no \-\-signal option will send a SIGTERM and
+wait the KillWait duration as defined in the slurm.conf file before sending the
+SIGKILL giving time for the running job/step(s) to clean up. To
+immediately kill a job, you can specify \-\-signal=KILL which will
+bypass the SIGTERM.
+.LP
Cancelling a job step will not result in a job being terminated.
The job must be cancelled to release a resource allocation.
diff --git a/doc/man/man1/sreport.1 b/doc/man/man1/sreport.1
index 3ec9547ed..866fde6b0 100644
--- a/doc/man/man1/sreport.1
+++ b/doc/man/man1/sreport.1
@@ -61,6 +61,10 @@ Identical to the \fBquit\fR command.
\fBhelp\fP
Display a description of sreport options and commands.
+.TP
+\fBnonparsable\fP
+Return output to normal after parsable or parsable2 has been set.
+
.TP
\fBparsable\fP
Output will be | delimited with an ending '|'.
diff --git a/doc/man/man1/srun.1 b/doc/man/man1/srun.1
index 8fd76ecc5..9524159ef 100644
--- a/doc/man/man1/srun.1
+++ b/doc/man/man1/srun.1
@@ -728,8 +728,7 @@ and without delaying the initiation of the job.
\fB\-n\fR, \fB\-\-ntasks\fR=<\fInumber\fR>
Specify the number of tasks to run. Request that \fBsrun\fR
allocate resources for \fIntasks\fR tasks.
-The default is one task per socket or core (depending upon the value
-of the \fISelectTypeParameters\fR parameter in slurm.conf), but note
+The default is one task per node, but note
that the \fB\-\-cpus\-per\-task\fR option will change this default.
.TP
@@ -921,7 +920,7 @@ This option may be used to spread several job steps out among the
nodes of the current job. If \fB\-r\fR is used, the current job
step will begin at node \fIn\fR of the allocated nodelist, where
the first node is considered node 0. The \fB\-r\fR option is not
-permitted along with \fB\-w\fR or \fB\-x\fR, and will be silently
+permitted along with \fB\-w\fR or \fB\-x\fR, and will be
ignored when not running within a prior allocation (i.e. when
SLURM_JOB_ID is not set). The default for \fIn\fR is 0. If the
value of \fB\-\-nodes\fR exceeds the number of nodes identified
diff --git a/doc/man/man5/bluegene.conf.5 b/doc/man/man5/bluegene.conf.5
index 1a016c205..9853319ee 100644
--- a/doc/man/man5/bluegene.conf.5
+++ b/doc/man/man5/bluegene.conf.5
@@ -13,9 +13,9 @@ Parameter names are case insensitive.
Any text following a "#" in the configuration file is treated
as a comment through the end of that line.
The size of each line in the file is limited to 1024 characters.
-Changes to the configuration file take effect upon restart of
-SLURM daemons, daemon receipt of the SIGHUP signal, or execution
-of the command "scontrol reconfigure" unless otherwise noted.
+Changes to the configuration file take only effect upon restart of
+the slurmctld daemon. "scontrol reconfigure" does nothing with this file.
+Changes will only take place after a restart of the controller.
.LP
There are some differences between Bluegene/L and Bluegene/P in respects to the contents of the bluegene.conf file.
diff --git a/etc/slurmdbd.conf.example b/etc/slurmdbd.conf.example
index 3d609b084..6518d6423 100644
--- a/etc/slurmdbd.conf.example
+++ b/etc/slurmdbd.conf.example
@@ -33,7 +33,7 @@ PidFile=/var/run/slurmdbd.pid
StorageType=accounting_storage/mysql
#StorageHost=localhost
#StoragePort=1234
-StoragePassword=password
+StoragePass=password
StorageUser=slurm
#StorageLoc=slurm_acct_db
diff --git a/slurm.spec b/slurm.spec
index d03fd7ce5..40eb41123 100644
--- a/slurm.spec
+++ b/slurm.spec
@@ -83,14 +83,14 @@
%endif
Name: slurm
-Version: 2.1.8
+Version: 2.1.9
Release: 1%{?dist}
Summary: Simple Linux Utility for Resource Management
License: GPL
Group: System Environment/Base
-Source: slurm-2.1.8.tar.bz2
+Source: slurm-2.1.9.tar.bz2
BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}
URL: https://computing.llnl.gov/linux/slurm/
@@ -352,7 +352,7 @@ Gives the ability for SLURM to use Berkeley Lab Checkpoint/Restart
#############################################################################
%prep
-%setup -n slurm-2.1.8
+%setup -n slurm-2.1.9
%build
%configure --program-prefix=%{?_program_prefix:%{_program_prefix}} \
diff --git a/src/common/job_resources.c b/src/common/job_resources.c
index ba73b6e09..4a8b81698 100644
--- a/src/common/job_resources.c
+++ b/src/common/job_resources.c
@@ -224,10 +224,9 @@ extern void reset_node_bitmap(job_resources_t *job_resrcs_ptr,
{
if (job_resrcs_ptr) {
if (job_resrcs_ptr->node_bitmap)
- bit_free(job_resrcs_ptr->node_bitmap);
+ FREE_NULL_BITMAP(job_resrcs_ptr->node_bitmap);
if (new_node_bitmap) {
- job_resrcs_ptr->node_bitmap =
- bit_copy(new_node_bitmap);
+ job_resrcs_ptr->node_bitmap = bit_copy(new_node_bitmap);
job_resrcs_ptr->nhosts = bit_set_count(new_node_bitmap);
} else
job_resrcs_ptr->nhosts = 0;
diff --git a/src/common/node_select.h b/src/common/node_select.h
index 378ad7ed0..0ea1fcbc1 100644
--- a/src/common/node_select.h
+++ b/src/common/node_select.h
@@ -184,6 +184,7 @@ extern int select_g_alter_node_cnt (enum select_node_cnt type, void *data);
#define SELECT_MODE_WILL_RUN 0x0002
#define SELECT_MODE_PREEMPT_FLAG 0x0100
+#define SELECT_MODE_CHECK_FULL 0x0200
#define SELECT_IS_MODE_RUN_NOW(_X) \
((_X & SELECT_MODE_BASE) == SELECT_MODE_RUN_NOW)
@@ -197,9 +198,11 @@ extern int select_g_alter_node_cnt (enum select_node_cnt type, void *data);
#define SELECT_IS_PREEMPT_SET(_X) \
(_X & SELECT_MODE_PREEMPT_FLAG)
+#define SELECT_IS_CHECK_FULL_SET(_X) \
+ (_X & SELECT_MODE_CHECK_FULL)
+
#define SELECT_IS_TEST(_X) \
- ((SELECT_IS_MODE_TEST_ONLY(_X) || SELECT_IS_MODE_WILL_RUN(_X)) \
- && !SELECT_IS_PREEMPT_SET(_X))
+ (SELECT_IS_MODE_TEST_ONLY(_X) || SELECT_IS_MODE_WILL_RUN(_X))
#define SELECT_IS_PREEMPTABLE_TEST(_X) \
((SELECT_IS_MODE_TEST_ONLY(_X) || SELECT_IS_MODE_WILL_RUN(_X)) \
diff --git a/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c b/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c
index ebd380e45..73bf93005 100644
--- a/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c
+++ b/src/plugins/accounting_storage/filetxt/filetxt_jobacct_process.c
@@ -705,10 +705,8 @@ static int _parse_line(char *f[], void **data, int len)
(*job)->nodes = xstrdup(f[F_NODES]);
for (i=0; (*job)->nodes[i]; i++) { /* discard trailing <CR> */
- if (isspace((*job)->nodes[i])) {
+ if (isspace((*job)->nodes[i]))
(*job)->nodes[i] = '\0';
- info("got here");
- }
}
if (!strcmp((*job)->nodes, "(null)")) {
xfree((*job)->nodes);
diff --git a/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c b/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c
index 52eac1bcb..c47073e7b 100644
--- a/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c
+++ b/src/plugins/accounting_storage/mysql/accounting_storage_mysql.c
@@ -6098,9 +6098,10 @@ extern List acct_storage_p_modify_qos(mysql_conn_t *mysql_conn, uint32_t uid,
ret_list = list_create(slurm_destroy_char);
while((row = mysql_fetch_row(result))) {
acct_qos_rec_t *qos_rec = NULL;
+ int id = atoi(row[2]);
+
if(preempt_bitstr) {
- if(_preemption_loop(mysql_conn,
- atoi(row[2]), preempt_bitstr))
+ if(_preemption_loop(mysql_conn, id, preempt_bitstr))
break;
}
object = xstrdup(row[0]);
@@ -6113,6 +6114,7 @@ extern List acct_storage_p_modify_qos(mysql_conn_t *mysql_conn, uint32_t uid,
}
qos_rec = xmalloc(sizeof(acct_qos_rec_t));
+ qos_rec->id = id;
qos_rec->name = xstrdup(object);
qos_rec->grp_cpus = qos->grp_cpus;
@@ -6136,27 +6138,29 @@ extern List acct_storage_p_modify_qos(mysql_conn_t *mysql_conn, uint32_t uid,
list_iterator_create(qos->preempt_list);
char *new_preempt = NULL;
- qos->preempt_bitstr = bit_alloc(g_qos_count);
+ qos_rec->preempt_bitstr = bit_alloc(g_qos_count);
if(row[1] && row[1][0])
- bit_unfmt(qos->preempt_bitstr, row[1]+1);
+ bit_unfmt(qos_rec->preempt_bitstr, row[1]+1);
while((new_preempt = list_next(new_preempt_itr))) {
bool cleared = 0;
if(new_preempt[0] == '-') {
- bit_clear(qos->preempt_bitstr,
+ bit_clear(qos_rec->preempt_bitstr,
atoi(new_preempt+1));
} else if(new_preempt[0] == '+') {
- bit_set(qos->preempt_bitstr,
+ bit_set(qos_rec->preempt_bitstr,
atoi(new_preempt+1));
} else {
if(!cleared) {
cleared = 1;
- bit_nclear(qos->preempt_bitstr,
- 0,
- bit_size(qos->preempt_bitstr)-1);
+ bit_nclear(
+ qos_rec->preempt_bitstr,
+ 0,
+ bit_size(qos_rec->
+ preempt_bitstr)-1);
}
- bit_set(qos->preempt_bitstr,
+ bit_set(qos_rec->preempt_bitstr,
atoi(new_preempt));
}
}
@@ -10618,7 +10622,11 @@ extern int jobacct_storage_p_job_start(mysql_conn_t *mysql_conn,
global_last_rollup = check_time;
slurm_mutex_unlock(&rollup_lock);
-
+ /* If the times here are later than the daily_rollup
+ or monthly rollup it isn't a big deal since they
+ are always shrunk down to the beginning of each
+ time period.
+ */
query = xstrdup_printf("update %s set hourly_rollup=%d, "
"daily_rollup=%d, monthly_rollup=%d",
last_ran_table, check_time,
diff --git a/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c b/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c
index 0280748e0..65f591106 100644
--- a/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c
+++ b/src/plugins/accounting_storage/mysql/mysql_jobacct_process.c
@@ -898,6 +898,7 @@ extern List mysql_jobacct_process_get_jobs(mysql_conn_t *mysql_conn, uid_t uid,
"t1.nodelist",
"t1.node_inx",
"t1.kill_requid",
+ "t1.timelimit",
"t1.qos",
"t2.user",
"t2.cluster",
@@ -934,6 +935,7 @@ extern List mysql_jobacct_process_get_jobs(mysql_conn_t *mysql_conn, uid_t uid,
JOB_REQ_NODELIST,
JOB_REQ_NODE_INX,
JOB_REQ_KILL_REQUID,
+ JOB_REQ_TIMELIMIT,
JOB_REQ_QOS,
JOB_REQ_USER_NAME,
JOB_REQ_CLUSTER,
@@ -1326,6 +1328,7 @@ extern List mysql_jobacct_process_get_jobs(mysql_conn_t *mysql_conn, uid_t uid,
job->priority = atoi(row[JOB_REQ_PRIORITY]);
job->req_cpus = atoi(row[JOB_REQ_REQ_CPUS]);
job->requid = atoi(row[JOB_REQ_KILL_REQUID]);
+ job->timelimit = atoi(row[JOB_REQ_TIMELIMIT]);
job->qos = atoi(row[JOB_REQ_QOS]);
job->show_full = 1;
@@ -1398,6 +1401,8 @@ extern List mysql_jobacct_process_get_jobs(mysql_conn_t *mysql_conn, uid_t uid,
continue;
step = create_jobacct_step_rec();
+ step->tot_cpu_sec = 0;
+ step->tot_cpu_usec = 0;
step->job_ptr = job;
if(!job->first_step_ptr)
job->first_step_ptr = step;
@@ -1455,11 +1460,9 @@ extern List mysql_jobacct_process_get_jobs(mysql_conn_t *mysql_conn, uid_t uid,
atoi(step_row[STEP_REQ_USER_USEC]);
step->sys_cpu_sec = atoi(step_row[STEP_REQ_SYS_SEC]);
step->sys_cpu_usec = atoi(step_row[STEP_REQ_SYS_USEC]);
- job->tot_cpu_sec +=
- step->tot_cpu_sec +=
+ step->tot_cpu_sec +=
step->user_cpu_sec + step->sys_cpu_sec;
- job->tot_cpu_usec +=
- step->tot_cpu_usec +=
+ step->tot_cpu_usec +=
step->user_cpu_usec + step->sys_cpu_usec;
step->sacct.max_vsize =
atoi(step_row[STEP_REQ_MAX_VSIZE]);
@@ -1582,6 +1585,7 @@ extern int mysql_jobacct_process_archive(mysql_conn_t *mysql_conn,
"nodelist",
"node_inx",
"kill_requid",
+ "timelimit",
"qos"
};
@@ -1676,6 +1680,7 @@ extern int mysql_jobacct_process_archive(mysql_conn_t *mysql_conn,
JOB_REQ_NODELIST,
JOB_REQ_NODE_INX,
JOB_REQ_KILL_REQUID,
+ JOB_REQ_TIMELIMIT,
JOB_REQ_QOS,
JOB_REQ_COUNT
};
diff --git a/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c b/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c
index 033d517ef..3069680f8 100644
--- a/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c
+++ b/src/plugins/accounting_storage/pgsql/pgsql_jobacct_process.c
@@ -593,6 +593,8 @@ no_cond:
xfree(query);
for(j = 0; j < PQntuples(step_result); j++) {
step = create_jobacct_step_rec();
+ step->tot_cpu_sec = 0;
+ step->tot_cpu_usec = 0;
step->job_ptr = job;
if(!job->first_step_ptr)
job->first_step_ptr = step;
@@ -654,11 +656,9 @@ no_cond:
PQgetvalue(step_result, j, STEP_REQ_SYS_SEC));
step->sys_cpu_usec = atoi(
PQgetvalue(step_result, j, STEP_REQ_SYS_USEC));
- job->tot_cpu_sec +=
- step->tot_cpu_sec +=
+ step->tot_cpu_sec +=
step->user_cpu_sec + step->sys_cpu_sec;
- job->tot_cpu_usec +=
- step->tot_cpu_usec +=
+ step->tot_cpu_usec +=
step->user_cpu_usec + step->sys_cpu_usec;
step->sacct.max_vsize = atoi(
PQgetvalue(step_result, j,
diff --git a/src/plugins/jobacct_gather/aix/jobacct_gather_aix.c b/src/plugins/jobacct_gather/aix/jobacct_gather_aix.c
index 51908c692..32a1e0304 100644
--- a/src/plugins/jobacct_gather/aix/jobacct_gather_aix.c
+++ b/src/plugins/jobacct_gather/aix/jobacct_gather_aix.c
@@ -346,7 +346,7 @@ static void _acct_kill_job(void)
static void *_watch_tasks(void *arg)
{
- while(!jobacct_shutdown) { /* Do this until shutdown is requested */
+ while(!jobacct_shutdown) { /* Do this until shutdown is requested */
if(!jobacct_suspended) {
_get_process_data(); /* Update the data */
}
diff --git a/src/plugins/jobacct_gather/linux/jobacct_gather_linux.c b/src/plugins/jobacct_gather/linux/jobacct_gather_linux.c
index 823bfe371..c4b42643b 100644
--- a/src/plugins/jobacct_gather/linux/jobacct_gather_linux.c
+++ b/src/plugins/jobacct_gather/linux/jobacct_gather_linux.c
@@ -336,9 +336,10 @@ static void _get_process_data() {
jobacct->min_cpu = jobacct->tot_cpu =
MAX(jobacct->min_cpu,
(prec->usec + prec->ssec));
- debug2("%d mem size %u %u time %u",
- jobacct->pid, jobacct->max_rss,
- jobacct->max_vsize, jobacct->tot_cpu);
+ debug2("%d mem size %u %u time %u(%u+%u)",
+ jobacct->pid, jobacct->max_rss,
+ jobacct->max_vsize, jobacct->tot_cpu,
+ prec->usec, prec->ssec);
break;
}
}
@@ -436,8 +437,8 @@ static int _get_process_data_line(int in, prec_t *prec) {
prec->pages = majflt;
prec->usec = utime;
prec->ssec = stime;
- prec->vsize = vsize / 1024; /* convert from bytes to KB */
- prec->rss = rss * getpagesize() / 1024; /* convert from pages to KB */
+ prec->vsize = vsize / 1024; /* convert from bytes to KB */
+ prec->rss = rss * getpagesize() / 1024;/* convert from pages to KB */
return 1;
}
@@ -622,7 +623,8 @@ extern void jobacct_gather_p_change_poll(uint16_t frequency)
pthread_t _watch_tasks_thread_id;
/* create polling thread */
slurm_attr_init(&attr);
- if (pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED))
+ if (pthread_attr_setdetachstate(&attr,
+ PTHREAD_CREATE_DETACHED))
error("pthread_attr_setdetachstate error %m");
if (pthread_create(&_watch_tasks_thread_id, &attr,
@@ -682,11 +684,21 @@ extern int jobacct_gather_p_add_task(pid_t pid, jobacct_id_t *jobacct_id)
extern struct jobacctinfo *jobacct_gather_p_stat_task(pid_t pid)
{
- _get_process_data();
- if(pid)
+ if(pid) {
+ _get_process_data();
return jobacct_common_stat_task(pid, task_list);
- else
+ } else {
+ /* In this situation, we are just trying to get a
+ * basis of information since we are not pollng. So
+ * we will give a chance for processes to spawn before we
+ * gather information. This should largely eliminate the
+ * the chance of having /proc open when the tasks are
+ * spawned, which would prevent a valid checkpoint/restart
+ * with some systems */
+ _task_sleep(1);
+ _get_process_data();
return NULL;
+ }
}
extern struct jobacctinfo *jobacct_gather_p_remove_task(pid_t pid)
diff --git a/src/plugins/priority/multifactor/priority_multifactor.c b/src/plugins/priority/multifactor/priority_multifactor.c
index aa3bda1ae..2d1954259 100644
--- a/src/plugins/priority/multifactor/priority_multifactor.c
+++ b/src/plugins/priority/multifactor/priority_multifactor.c
@@ -816,7 +816,7 @@ static void *_decay_thread(void *no_data)
job_ptr->priority =
_get_priority_internal(start_time, job_ptr);
-
+ last_job_update = time(NULL);
debug2("priority for job %u is now %u",
job_ptr->job_id, job_ptr->priority);
}
diff --git a/src/plugins/proctrack/linuxproc/kill_tree.c b/src/plugins/proctrack/linuxproc/kill_tree.c
index 41836c113..4f32c3827 100644
--- a/src/plugins/proctrack/linuxproc/kill_tree.c
+++ b/src/plugins/proctrack/linuxproc/kill_tree.c
@@ -144,7 +144,7 @@ static int get_myname(char *s)
return 0;
}
-static xppid_t **_build_hashtbl()
+static xppid_t **_build_hashtbl(void)
{
DIR *dir;
struct dirent *de;
@@ -165,11 +165,13 @@ static xppid_t **_build_hashtbl()
while ((de = readdir(dir)) != NULL) {
num = de->d_name;
+ if ((num[0] < '0') || (num[0] > '9'))
+ continue;
ret_l = strtol(num, &endptr, 10);
- if(errno == ERANGE)
+ if(errno == ERANGE) {
error("couldn't do a strtol on str %s(%d): %m",
num, ret_l);
-
+ }
if (endptr == NULL || *endptr != 0)
continue;
sprintf(path, "/proc/%s/stat", num);
diff --git a/src/plugins/sched/backfill/backfill.c b/src/plugins/sched/backfill/backfill.c
index af28f7821..d936e648a 100644
--- a/src/plugins/sched/backfill/backfill.c
+++ b/src/plugins/sched/backfill/backfill.c
@@ -563,7 +563,7 @@ static void _attempt_backfill(void)
continue;
}
- if (node_space_recs == max_backfill_job_cnt) {
+ if (node_space_recs >= max_backfill_job_cnt) {
/* Already have too many jobs to deal with */
break;
}
diff --git a/src/plugins/select/bluegene/plugin/bg_job_place.c b/src/plugins/select/bluegene/plugin/bg_job_place.c
index e8139e4a0..8ff04f891 100644
--- a/src/plugins/select/bluegene/plugin/bg_job_place.c
+++ b/src/plugins/select/bluegene/plugin/bg_job_place.c
@@ -315,7 +315,7 @@ static bg_record_t *_find_matching_block(List block_list,
} else if((bg_record->job_running != NO_JOB_RUNNING)
&& (bg_record->job_running != job_ptr->job_id)
&& (bg_conf->layout_mode == LAYOUT_DYNAMIC
- || (!SELECT_IS_TEST(query_mode)
+ || (SELECT_IS_MODE_RUN_NOW(query_mode)
&& bg_conf->layout_mode != LAYOUT_DYNAMIC))) {
debug("block %s in use by %s job %d",
bg_record->bg_block_id,
@@ -520,7 +520,7 @@ static int _check_for_booted_overlapping_blocks(
if(is_test && overlapped_list
&& found_record->job_ptr
&& bg_record->job_running == NO_JOB_RUNNING) {
- debug2("found over lapping block %s "
+ debug2("found overlapping block %s "
"overlapped %s with job %u",
found_record->bg_block_id,
bg_record->bg_block_id,
@@ -666,7 +666,13 @@ static int _dynamically_request(List block_list, int *blocks_added,
debug2("going to create %d", request->size);
list_of_lists = list_create(NULL);
- if(user_req_nodes)
+ /* If preempt is set and we are checking full system it means
+ we altered the block list so only look at it.
+ */
+ if(SELECT_IS_PREEMPT_SET(query_mode)
+ && SELECT_IS_CHECK_FULL_SET(query_mode)) {
+ list_append(list_of_lists, block_list);
+ } else if(user_req_nodes)
list_append(list_of_lists, bg_lists->job_running);
else {
list_append(list_of_lists, block_list);
@@ -699,8 +705,7 @@ static int _dynamically_request(List block_list, int *blocks_added,
while((bg_record = list_pop(new_blocks))) {
if(block_exist_in_list(block_list, bg_record))
destroy_bg_record(bg_record);
- else if(SELECT_IS_PREEMPTABLE_TEST(
- query_mode)) {
+ else if(SELECT_IS_TEST(query_mode)) {
/* Here we don't really want
to create the block if we
are testing.
@@ -893,7 +898,7 @@ static int _find_best_block_match(List block_list,
* works we will have can look and see the earliest
* the job can start. This doesn't apply to Dynamic mode.
*/
- if(is_test
+ if(is_test && SELECT_IS_CHECK_FULL_SET(query_mode)
&& bg_conf->layout_mode != LAYOUT_DYNAMIC)
overlapped_list = list_create(NULL);
@@ -906,8 +911,7 @@ static int _find_best_block_match(List block_list,
overlap_check,
overlapped_list,
query_mode);
- if(!bg_record && is_test
- && bg_conf->layout_mode != LAYOUT_DYNAMIC
+ if(!bg_record && overlapped_list
&& list_count(overlapped_list)) {
ListIterator itr =
list_iterator_create(overlapped_list);
@@ -921,7 +925,7 @@ static int _find_best_block_match(List block_list,
list_iterator_destroy(itr);
}
- if(is_test && bg_conf->layout_mode != LAYOUT_DYNAMIC)
+ if(overlapped_list)
list_destroy(overlapped_list);
/* set the bitmap and do other allocation activities */
@@ -995,8 +999,10 @@ static int _find_best_block_match(List block_list,
continue;
}
-
- if(is_test) {
+ /* Only look at the full system if we aren't going to
+ preempt jobs later and look.
+ */
+ if(is_test && SELECT_IS_CHECK_FULL_SET(query_mode)) {
List new_blocks = NULL;
List job_list = list_create(NULL);
ListIterator itr = NULL;
@@ -1294,7 +1300,8 @@ static void _build_select_struct(struct job_record *job_ptr,
}
}
-static List _get_preemptables(bg_record_t *bg_record, List preempt_jobs)
+static List _get_preemptables(uint16_t query_mode, bg_record_t *bg_record,
+ List preempt_jobs)
{
List preempt = NULL;
ListIterator itr;
@@ -1321,10 +1328,10 @@ static List _get_preemptables(bg_record_t *bg_record, List preempt_jobs)
break;
}
if(job_ptr) {
- list_append(preempt, job_ptr);
+ list_push(preempt, job_ptr);
/* info("going to preempt %u running on %s", */
/* job_ptr->job_id, found_record->bg_block_id); */
- } else {
+ } else if(SELECT_IS_MODE_RUN_NOW(query_mode)) {
error("Job %u running on block %s "
"wasn't in the preempt list, but needs to be "
"preempted for queried job to run on block %s",
@@ -1422,6 +1429,8 @@ extern int submit_job(struct job_record *job_ptr, bitstr_t *slurm_block_bitmap,
if (preemptee_candidates && preemptee_job_list
&& list_count(preemptee_candidates))
local_mode |= SELECT_MODE_PREEMPT_FLAG;
+ else
+ local_mode |= SELECT_MODE_CHECK_FULL;
if(bg_conf->layout_mode == LAYOUT_DYNAMIC)
slurm_mutex_lock(&create_dynamic_mutex);
@@ -1487,8 +1496,8 @@ extern int submit_job(struct job_record *job_ptr, bitstr_t *slurm_block_bitmap,
block_list = copy_bg_list(bg_lists->main);
slurm_mutex_unlock(&block_state_mutex);
- /* just remove the preemptable jobs now since we are treating
- this as a run now deal */
+ /* First look at the empty space, and then remove the
+ preemptable jobs and try again. */
preempt:
list_sort(block_list, (ListCmpF)bg_record_sort_aval_inc);
@@ -1538,18 +1547,6 @@ preempt:
job_ptr->select_jobinfo,
SELECT_JOBDATA_NODE_CNT,
&bg_record->node_cnt);
-
- /* This is a fake record so we need to
- * destroy it after we get the info from
- * it. if it was just testing then
- * we added this record to the
- * block_list. If this is the case
- * it will be set below, but set
- * blocks_added to 0 since we don't
- * want to sync this with the list. */
- if(!blocks_added)
- destroy_bg_record(bg_record);
- blocks_added = 0;
} else {
if((bg_record->ionodes)
&& (job_ptr->part_ptr->max_share <= 1))
@@ -1596,12 +1593,29 @@ preempt:
if(*preemptee_job_list)
list_destroy(*preemptee_job_list);
*preemptee_job_list = _get_preemptables(
- bg_record, preemptee_candidates);
+ local_mode, bg_record,
+ preemptee_candidates);
+ }
+ if(!bg_record->bg_block_id) {
+ /* This is a fake record so we need to
+ * destroy it after we get the info from
+ * it. If it was just testing then
+ * we added this record to the
+ * block_list. If this is the case
+ * it will be handled if se sync the
+ * lists. But we don't want to do
+ * that so we will set blocks_added to
+ * 0 so it doesn't happen. */
+ if(!blocks_added)
+ destroy_bg_record(bg_record);
+ blocks_added = 0;
}
} else {
error("we got a success, but no block back");
}
} else if(!preempt_done && SELECT_IS_PREEMPT_SET(local_mode)) {
+ debug2("doing preemption");
+ local_mode |= SELECT_MODE_CHECK_FULL;
avail_cpus += _remove_preemptables(
block_list, preemptee_candidates);
preempt_done = true;
diff --git a/src/plugins/select/bluegene/plugin/bluegene.c b/src/plugins/select/bluegene/plugin/bluegene.c
index 93f5ddfc6..922909d51 100644
--- a/src/plugins/select/bluegene/plugin/bluegene.c
+++ b/src/plugins/select/bluegene/plugin/bluegene.c
@@ -1,7 +1,7 @@
/*****************************************************************************\
* bluegene.c - blue gene node configuration processing module.
*
- * $Id: bluegene.c 19755 2010-03-16 19:15:43Z da $
+ * $Id: bluegene.c 20411 2010-06-03 23:44:06Z da $
*****************************************************************************
* Copyright (C) 2004 The Regents of the University of California.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
@@ -164,7 +164,6 @@ extern bool blocks_overlap(bg_record_t *rec_a, bg_record_t *rec_b)
return true;
}
-
if (!bit_overlap(rec_a->bitmap, rec_b->bitmap))
return false;
diff --git a/src/plugins/select/linear/select_linear.c b/src/plugins/select/linear/select_linear.c
index 69e76d730..39cd508f1 100644
--- a/src/plugins/select/linear/select_linear.c
+++ b/src/plugins/select/linear/select_linear.c
@@ -89,15 +89,6 @@ int node_record_count;
time_t last_node_update;
struct switch_record *switch_record_table;
int switch_record_cnt;
-#ifdef HAVE_AIX
-/* On AIX dlopen's RTLD_LAZY flag does NOT work, so we must define
- * actual functions here in order to load this plugin from anywhere
- * other than slurmctld */
-void job_preempt_remove(uint32_t job_id)
-{
- ;
-}
-#endif
struct select_nodeinfo {
uint16_t magic; /* magic number */
@@ -598,6 +589,10 @@ static int _job_count_bitmap(struct node_cr_record *node_cr_ptr,
} else
job_mem = job_memory_node;
}
+ if ((alloc_mem + job_mem) > avail_mem) {
+ bit_clear(jobmap, i);
+ continue;
+ }
}
if (node_cr_ptr[i].exclusive_cnt != 0) {
diff --git a/src/sacct/options.c b/src/sacct/options.c
index 77d4d41be..bc0eac979 100644
--- a/src/sacct/options.c
+++ b/src/sacct/options.c
@@ -988,6 +988,10 @@ void parse_command_line(int argc, char **argv)
|| job_cond->used_nodes)
all_users=1;
+ /* set all_users for user root if not requesting any */
+ if(!job_cond->userid_list && !params.opt_uid)
+ all_users = 1;
+
if(all_users) {
if(job_cond->userid_list && list_count(job_cond->userid_list)) {
list_destroy(job_cond->userid_list);
diff --git a/src/sacct/print.c b/src/sacct/print.c
index baad19f00..d0be850d1 100644
--- a/src/sacct/print.c
+++ b/src/sacct/print.c
@@ -1126,7 +1126,6 @@ void print_fields(type_t type, void *object)
case JOBSTEP:
tmp_int = step->sys_cpu_sec;
tmp_int2 = step->sys_cpu_usec;
-
break;
case JOBCOMP:
@@ -1145,10 +1144,18 @@ void print_fields(type_t type, void *object)
case PRINT_TIMELIMIT:
switch(type) {
case JOB:
-
+ if (job->timelimit == INFINITE)
+ tmp_char = "UNLIMITED";
+ else if (job->timelimit == NO_VAL)
+ tmp_char = "Partition_Limit";
+ else if(job->timelimit) {
+ char tmp1[128];
+ mins2time_str(job->timelimit,
+ tmp1, sizeof(tmp1));
+ tmp_char = tmp1;
+ }
break;
case JOBSTEP:
-
break;
case JOBCOMP:
tmp_char = job_comp->timelimit;
@@ -1237,12 +1244,11 @@ void print_fields(type_t type, void *object)
switch(type) {
case JOB:
tmp_int = job->user_cpu_sec;
- tmp_int2 = job->tot_cpu_usec;
+ tmp_int2 = job->user_cpu_usec;
break;
case JOBSTEP:
tmp_int = step->user_cpu_sec;
tmp_int2 = step->user_cpu_usec;
-
break;
case JOBCOMP:
diff --git a/src/sacct/sacct.h b/src/sacct/sacct.h
index 0bec9148d..b4cb40da0 100644
--- a/src/sacct/sacct.h
+++ b/src/sacct/sacct.h
@@ -68,10 +68,10 @@
#define BRIEF_FIELDS "jobid,state,exitcode"
#define BRIEF_COMP_FIELDS "jobid,uid,state"
#define DEFAULT_FIELDS "jobid,jobname,partition,account,alloccpus,state,exitcode"
-#define DEFAULT_COMP_FIELDS "jobid,uid,jobname,partition,nnodes,nodes,state,end"
+#define DEFAULT_COMP_FIELDS "jobid,uid,jobname,partition,nnodes,nodelist,state,end"
#define LONG_FIELDS "jobid,jobname,partition,maxvmsize,maxvmsizenode,maxvmsizetask,avevmsize,maxrss,maxrssnode,maxrsstask,averss,maxpages,maxpagesnode,maxpagestask,avepages,mincpu,mincpunode,mincputask,avecpu,ntasks,alloccpus,elapsed,state,exitcode"
-#define LONG_COMP_FIELDS "jobid,uid,jobname,partition,nnodes,nodes,state,start,end,timelimit"
+#define LONG_COMP_FIELDS "jobid,uid,jobname,partition,nnodes,nodelist,state,start,end,timelimit"
#define BUFFER_SIZE 4096
#define STATE_COUNT 10
diff --git a/src/sacctmgr/sacctmgr.c b/src/sacctmgr/sacctmgr.c
index 0489d0fd0..81eaab60c 100644
--- a/src/sacctmgr/sacctmgr.c
+++ b/src/sacctmgr/sacctmgr.c
@@ -814,14 +814,14 @@ sacctmgr [<OPTION>] [<COMMAND>] \n\
WithCoordinators, WithRawQOS, and WOPLimits \n\
add account - Clusters=, Description=, Fairshare=, \n\
GrpCPUMins=, GrpCPUs=, GrpJobs=, GrpNodes=, \n\
- GrpSubmitJob=, GrpWall=, MaxCPUMins=, MaxJobs=,\n\
- MaxNodes=, MaxSubmitJobs=, MaxWall=, Names=, \n\
- Organization=, Parent=, and QosLevel \n\
+ GrpSubmitJob=, GrpWall=, MaxCPUMins=, MaxCPUs=,\n\
+ MaxJobs=, MaxNodes=, MaxSubmitJobs=, MaxWall=, \n\
+ Names=, Organization=, Parent=, and QosLevel= \n\
modify account - (set options) Description=, Fairshare=, \n\
GrpCPUMins=, GrpCPUs=, GrpJobs=, GrpNodes=, \n\
- GrpSubmitJob=, GrpWall=, MaxCPUMins=, MaxJobs=,\n\
- MaxNodes=, MaxSubmitJobs=, MaxWall=, Names=, \n\
- Organization=, Parent=, and QosLevel= \n\
+ GrpSubmitJob=, GrpWall=, MaxCPUMins=, MaxCPUs=,\n\
+ MaxJobs=, MaxNodes=, MaxSubmitJobs=, MaxWall=, \n\
+ Names=, Organization=, Parent=, and QosLevel= \n\
(where options) Clusters=, Descriptions=, \n\
Names=, Organizations=, Parent=, and QosLevel= \n\
delete account - Clusters=, Descriptions=, Names=, \n\
diff --git a/src/salloc/opt.c b/src/salloc/opt.c
index 3253ec573..a45c5c0a5 100644
--- a/src/salloc/opt.c
+++ b/src/salloc/opt.c
@@ -1196,6 +1196,11 @@ static bool _opt_verify(void)
verified = false;
}
+ if ((opt.ntasks_per_node > -1) && (!opt.nprocs_set)) {
+ opt.nprocs = opt.min_nodes * opt.ntasks_per_node;
+ opt.nprocs_set = 1;
+ }
+
if (opt.mincpus < opt.cpus_per_task)
opt.mincpus = opt.cpus_per_task;
diff --git a/src/sbatch/opt.c b/src/sbatch/opt.c
index 8c963d051..1f9155a03 100644
--- a/src/sbatch/opt.c
+++ b/src/sbatch/opt.c
@@ -1981,6 +1981,11 @@ static bool _opt_verify(void)
_fullpath(&opt.ifname, opt.cwd);
_fullpath(&opt.ofname, opt.cwd);
+ if ((opt.ntasks_per_node > 0) && (!opt.nprocs_set)) {
+ opt.nprocs = opt.min_nodes * opt.ntasks_per_node;
+ opt.nprocs_set = 1;
+ }
+
if (opt.mincpus < opt.cpus_per_task)
opt.mincpus = opt.cpus_per_task;
diff --git a/src/slurmctld/controller.c b/src/slurmctld/controller.c
index 408df3f46..b05e78e7a 100644
--- a/src/slurmctld/controller.c
+++ b/src/slurmctld/controller.c
@@ -1019,7 +1019,10 @@ static int _accounting_cluster_ready()
int procs = 0;
bitstr_t *total_node_bitmap = NULL;
char *cluster_nodes = NULL;
+ slurmctld_lock_t node_read_lock = {
+ NO_LOCK, NO_LOCK, READ_LOCK, NO_LOCK };
+ lock_slurmctld(node_read_lock);
node_ptr = node_record_table_ptr;
for (i = 0; i < node_record_count; i++, node_ptr++) {
if (node_ptr->name == '\0')
@@ -1047,6 +1050,7 @@ static int _accounting_cluster_ready()
bit_nset(total_node_bitmap, 0, node_record_count-1);
cluster_nodes = bitmap2node_name(total_node_bitmap);
FREE_NULL_BITMAP(total_node_bitmap);
+ unlock_slurmctld(node_read_lock);
rc = clusteracct_storage_g_cluster_procs(acct_db_conn,
slurmctld_cluster_name,
@@ -1160,9 +1164,6 @@ static void *_slurmctld_background(void *no_data)
* (Might kill jobs on nodes set DOWN) */
slurmctld_lock_t node_write_lock = {
READ_LOCK, WRITE_LOCK, WRITE_LOCK, NO_LOCK };
- /* Locks: Read node */
- slurmctld_lock_t node_read_lock = {
- NO_LOCK, NO_LOCK, READ_LOCK, NO_LOCK };
/* Locks: Write node */
slurmctld_lock_t node_write_lock2 = {
NO_LOCK, NO_LOCK, WRITE_LOCK, NO_LOCK };
@@ -1335,9 +1336,7 @@ static void *_slurmctld_background(void *no_data)
/* Report current node state to account for added
* or reconfigured nodes */
last_node_acct = now;
- lock_slurmctld(node_read_lock);
_accounting_cluster_ready();
- unlock_slurmctld(node_read_lock);
}
/* Reassert this machine as the primary controller.
diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index f7c373206..634119c59 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -8079,46 +8079,3 @@ _read_job_ckpt_file(char *ckpt_file, int *size_ptr)
*size_ptr = data_size;
return data;
}
-
-/*
- * Preempt a job using the proper job removal mechanism (checkpoint, requeue).
- * Do not use this function for job suspend/resume. This is handled by the
- * gang module.
- */
-extern void job_preempt_remove(uint32_t job_id)
-{
- int rc = SLURM_SUCCESS;
- uint16_t preempt_mode = slurm_get_preempt_mode();
- checkpoint_msg_t ckpt_msg;
-
- preempt_mode &= (~PREEMPT_MODE_GANG);
- if (preempt_mode == PREEMPT_MODE_REQUEUE) {
- rc = job_requeue(0, job_id, -1);
- if (rc == SLURM_SUCCESS) {
- info("preempted job %u has been requeued", job_id);
- }
- } else if (preempt_mode == PREEMPT_MODE_CANCEL) {
- (void) job_signal(job_id, SIGKILL, 0, 0);
- } else if (preempt_mode == PREEMPT_MODE_CHECKPOINT) {
- memset(&ckpt_msg, 0, sizeof(checkpoint_msg_t));
- ckpt_msg.op = CHECK_VACATE;
- ckpt_msg.job_id = job_id;
- rc = job_checkpoint(&ckpt_msg, 0, -1);
- if (rc == SLURM_SUCCESS) {
- info("preempted job %u has been checkpointed", job_id);
- }
- } else {
- error("Invalid preempt_mode: %u", preempt_mode);
- }
-
- if (rc != SLURM_SUCCESS) {
- rc = job_signal(job_id, SIGKILL, 0, 0);
- if (rc == SLURM_SUCCESS)
- info("preempted job %u had to be killed", job_id);
- else {
- info("preempted job %u kill failure %s",
- job_id, slurm_strerror(rc));
- }
- }
-}
-
diff --git a/src/slurmctld/job_scheduler.c b/src/slurmctld/job_scheduler.c
index a2ffe2c0f..11dfee940 100644
--- a/src/slurmctld/job_scheduler.c
+++ b/src/slurmctld/job_scheduler.c
@@ -509,7 +509,8 @@ extern void sort_job_queue(struct job_queue *job_queue, int job_queue_size)
job_queue[i].job_ptr = job_queue[top_prio_inx].job_ptr;
job_queue[i].job_priority = job_queue[top_prio_inx].
job_priority;
-
+ job_queue[i].part_priority = job_queue[top_prio_inx].
+ part_priority;
job_queue[top_prio_inx].job_ptr = tmp_job_ptr;
job_queue[top_prio_inx].job_priority = tmp_job_prio;
@@ -1130,14 +1131,15 @@ static char **_build_env(struct job_record *job_ptr)
#ifdef HAVE_CRAY_XT
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
- SELECT_JOBDATA_RESV_ID, &name);
+ SELECT_JOBDATA_RESV_ID, &name);
setenvf(&env, "BASIL_RESERVATION_ID", "%s", name);
xfree(name);
#endif
#ifdef HAVE_BG
select_g_select_jobinfo_get(job_ptr->select_jobinfo,
- SELECT_JOBDATA_BLOCK_ID, &name);
+ SELECT_JOBDATA_BLOCK_ID, &name);
setenvf(&my_env, "MPIRUN_PARTITION", "%s", name);
+ xfree(name);
#endif
setenvf(&my_env, "SLURM_JOB_ACCOUNT", "%s", job_ptr->account);
if (job_ptr->details) {
diff --git a/src/slurmctld/node_mgr.c b/src/slurmctld/node_mgr.c
index 34130d412..ec70dd080 100644
--- a/src/slurmctld/node_mgr.c
+++ b/src/slurmctld/node_mgr.c
@@ -1347,11 +1347,6 @@ extern int validate_node_specs(slurm_node_registration_status_msg_t *reg_msg)
#endif
}
} else {
- uint16_t err_cpus = 0;
- select_g_select_nodeinfo_get(node_ptr->select_nodeinfo,
- SELECT_NODEDATA_SUBCNT,
- NODE_STATE_ERROR,
- &err_cpus);
if (IS_NODE_UNKNOWN(node_ptr)) {
last_node_update = now;
reset_job_priority();
@@ -1365,8 +1360,7 @@ extern int validate_node_specs(slurm_node_registration_status_msg_t *reg_msg)
node_flags;
node_ptr->last_idle = now;
}
- if (!err_cpus
- && !IS_NODE_DRAIN(node_ptr)
+ if (!IS_NODE_DRAIN(node_ptr)
&& !IS_NODE_FAIL(node_ptr)) {
xfree(node_ptr->reason);
clusteracct_storage_g_node_up(
@@ -1392,8 +1386,7 @@ extern int validate_node_specs(slurm_node_registration_status_msg_t *reg_msg)
reg_msg->node_name);
reset_job_priority();
trigger_node_up(node_ptr);
- if (!err_cpus
- && !IS_NODE_DRAIN(node_ptr)
+ if (!IS_NODE_DRAIN(node_ptr)
&& !IS_NODE_FAIL(node_ptr)) {
xfree(node_ptr->reason);
clusteracct_storage_g_node_up(
@@ -1579,11 +1572,6 @@ extern int validate_nodes_via_front_end(
#endif
}
} else {
- uint16_t err_cpus = 0;
- select_g_select_nodeinfo_get(node_ptr->select_nodeinfo,
- SELECT_NODEDATA_SUBCNT,
- NODE_STATE_ERROR,
- &err_cpus);
if (reg_hostlist)
(void) hostlist_push_host(
reg_hostlist, node_ptr->name);
@@ -1603,8 +1591,7 @@ extern int validate_nodes_via_front_end(
node_flags;
node_ptr->last_idle = now;
}
- if (!err_cpus
- && !IS_NODE_DRAIN(node_ptr)
+ if (!IS_NODE_DRAIN(node_ptr)
&& !IS_NODE_FAIL(node_ptr)) {
xfree(node_ptr->reason);
clusteracct_storage_g_node_up(
@@ -1626,8 +1613,7 @@ extern int validate_nodes_via_front_end(
node_ptr->last_idle = now;
}
trigger_node_up(node_ptr);
- if (!err_cpus
- && !IS_NODE_DRAIN(node_ptr)
+ if (!IS_NODE_DRAIN(node_ptr)
&& !IS_NODE_FAIL(node_ptr)) {
xfree(node_ptr->reason);
clusteracct_storage_g_node_up(
diff --git a/src/slurmctld/node_scheduler.c b/src/slurmctld/node_scheduler.c
index 77052471f..0d8654b08 100644
--- a/src/slurmctld/node_scheduler.c
+++ b/src/slurmctld/node_scheduler.c
@@ -609,14 +609,12 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
bitstr_t *avail_bitmap = NULL, *total_bitmap = NULL;
bitstr_t *backup_bitmap = NULL;
bitstr_t *possible_bitmap = NULL;
- bitstr_t *partially_idle_node_bitmap = NULL;
int max_feature, min_feature;
bool runable_ever = false; /* Job can ever run */
bool runable_avail = false; /* Job can run with available nodes */
bool tried_sched = false; /* Tried to schedule with avail nodes */
static uint32_t cr_enabled = NO_VAL;
bool preempt_flag = false;
- select_type_plugin_info_t cr_type = SELECT_TYPE_INFO_NONE;
int shared = 0, select_mode;
if (test_only)
@@ -643,34 +641,14 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
shared = _resolve_shared_status(job_ptr->details->shared,
part_ptr->max_share, cr_enabled);
job_ptr->details->shared = shared;
+ if (cr_enabled)
+ job_ptr->cr_enabled = cr_enabled; /* CR enabled for this job */
/* If job preemption is enabled, then do NOT limit the set of available
* nodes by their current 'sharable' or 'idle' setting */
if (slurm_get_preempt_mode() != PREEMPT_MODE_OFF)
preempt_flag = true;
- if (cr_enabled) {
- /* Determine which nodes might be used by this job based upon
- * its ability to share resources */
- job_ptr->cr_enabled = cr_enabled; /* CR enabled for this job */
-
- cr_type = (select_type_plugin_info_t) slurmctld_conf.
- select_type_param;
-
- /* Set the partially_idle_node_bitmap to reflect the
- * idle and partially idle nodes */
- error_code = select_g_get_info_from_plugin (SELECT_BITMAP,
- job_ptr, &partially_idle_node_bitmap);
- if (error_code != SLURM_SUCCESS) {
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
- return error_code;
- }
- debug3("Job %u shared %d CR type %d num_procs %d nbits %d",
- job_ptr->job_id, shared, cr_enabled, cr_type,
- job_ptr->num_procs,
- bit_set_count(partially_idle_node_bitmap));
- }
-
if (job_ptr->details->req_node_bitmap) { /* specific nodes required */
/* We have already confirmed that all of these nodes have a
* usable configuration and are in the proper partition.
@@ -680,7 +658,6 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
job_ptr->details->req_node_bitmap);
}
if (total_nodes > max_nodes) { /* exceeds node limit */
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
return ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE;
}
@@ -688,33 +665,20 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
/* Should we check memory availability on these nodes? */
if (!bit_super_set(job_ptr->details->req_node_bitmap,
avail_node_bitmap)) {
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
return ESLURM_NODE_NOT_AVAIL;
}
- if (partially_idle_node_bitmap) {
- if (!bit_super_set(job_ptr->details->req_node_bitmap,
- partially_idle_node_bitmap)) {
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
- return ESLURM_NODES_BUSY;
- }
- }
-
if (!preempt_flag) {
if (shared) {
if (!bit_super_set(job_ptr->details->
req_node_bitmap,
share_node_bitmap)) {
- FREE_NULL_BITMAP(
- partially_idle_node_bitmap);
return ESLURM_NODES_BUSY;
}
} else {
if (!bit_super_set(job_ptr->details->
req_node_bitmap,
idle_node_bitmap)) {
- FREE_NULL_BITMAP(
- partially_idle_node_bitmap);
return ESLURM_NODES_BUSY;
}
}
@@ -759,11 +723,6 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
}
bit_and(node_set_ptr[i].my_bitmap, avail_node_bitmap);
- if (partially_idle_node_bitmap) {
- bit_and(node_set_ptr[i].my_bitmap,
- partially_idle_node_bitmap);
- }
-
if (!preempt_flag) {
if (shared) {
bit_and(node_set_ptr[i].my_bitmap,
@@ -840,7 +799,6 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
avail_nodes = 0;
break;
}
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
FREE_NULL_BITMAP(total_bitmap);
FREE_NULL_BITMAP(possible_bitmap);
*select_bitmap = avail_bitmap;
@@ -870,7 +828,6 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
preemptee_job_list);
if ((pick_code == SLURM_SUCCESS) &&
(bit_set_count(avail_bitmap) <= max_nodes)) {
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
FREE_NULL_BITMAP(total_bitmap);
FREE_NULL_BITMAP(possible_bitmap);
*select_bitmap = avail_bitmap;
@@ -949,7 +906,6 @@ _pick_best_nodes(struct node_set *node_set_ptr, int node_set_size,
} else {
FREE_NULL_BITMAP(possible_bitmap);
}
- FREE_NULL_BITMAP(partially_idle_node_bitmap);
return error_code;
}
diff --git a/src/slurmctld/slurmctld.h b/src/slurmctld/slurmctld.h
index e25764399..191598346 100644
--- a/src/slurmctld/slurmctld.h
+++ b/src/slurmctld/slurmctld.h
@@ -860,13 +860,6 @@ extern int job_fail(uint32_t job_id);
*/
extern int job_node_ready(uint32_t job_id, int *ready);
-/*
- * Preempt a job using the proper job removal mechanism (checkpoint, requeue).
- * Do not use this function for job suspend/resume. This is handled by the
- * gang module.
- */
-extern void job_preempt_remove(uint32_t job_id);
-
/*
* job_restart - Restart a batch job from checkpointed state
*
diff --git a/src/slurmd/slurmd/req.c b/src/slurmd/slurmd/req.c
index 1cc41c4a4..29efe3a52 100644
--- a/src/slurmd/slurmd/req.c
+++ b/src/slurmd/slurmd/req.c
@@ -1111,11 +1111,11 @@ _rpc_batch_job(slurm_msg_t *msg)
*/
#ifdef HAVE_BG
select_g_select_jobinfo_get(req->select_jobinfo,
- SELECT_JOBDATA_BLOCK_ID, &resv_id);
+ SELECT_JOBDATA_BLOCK_ID, &resv_id);
#endif
#ifdef HAVE_CRAY_XT
select_g_select_jobinfo_get(req->select_jobinfo,
- SELECT_JOBDATA_RESV_ID, &resv_id);
+ SELECT_JOBDATA_RESV_ID, &resv_id);
#endif
rc = _run_prolog(req->job_id, req->uid, resv_id,
req->spank_job_env, req->spank_job_env_size);
@@ -3336,9 +3336,11 @@ _run_prolog(uint32_t jobid, uid_t uid, char *resv_id,
char *my_prolog;
char **my_env = _build_env(jobid, uid, resv_id, spank_job_env,
spank_job_env_size);
+ /* a long prolog is expected on bluegene systems */
+#ifndef HAVE_BG
time_t start_time = time(NULL), diff_time;
static uint16_t msg_timeout = 0;
-
+#endif
slurm_mutex_lock(&conf->config_mutex);
my_prolog = xstrdup(conf->prolog);
slurm_mutex_unlock(&conf->config_mutex);
@@ -3349,6 +3351,7 @@ _run_prolog(uint32_t jobid, uid_t uid, char *resv_id,
xfree(my_prolog);
_destroy_env(my_env);
+#ifndef HAVE_BG
diff_time = difftime(time(NULL), start_time);
if (msg_timeout == 0)
msg_timeout = slurm_get_msg_timeout();
@@ -3356,7 +3359,7 @@ _run_prolog(uint32_t jobid, uid_t uid, char *resv_id,
error("prolog for job %u ran for %d seconds",
jobid, diff_time);
}
-
+#endif
return error_code;
}
diff --git a/src/slurmd/slurmstepd/mgr.c b/src/slurmd/slurmstepd/mgr.c
index a32336089..4b38bbc8a 100644
--- a/src/slurmd/slurmstepd/mgr.c
+++ b/src/slurmd/slurmstepd/mgr.c
@@ -1,6 +1,6 @@
/*****************************************************************************\
* src/slurmd/slurmstepd/mgr.c - job manager functions for slurmstepd
- * $Id: mgr.c 20036 2010-04-15 18:52:13Z da $
+ * $Id: mgr.c 20276 2010-05-18 16:51:08Z da $
*****************************************************************************
* Copyright (C) 2002-2007 The Regents of the University of California.
* Copyright (C) 2008-2009 Lawrence Livermore National Security.
@@ -910,6 +910,7 @@ job_manager(slurmd_job_t *job)
*/
if(!conf->job_acct_gather_freq)
jobacct_gather_g_stat_task(0);
+
/* Send job launch response with list of pids */
_send_launch_resp(job, 0);
diff --git a/src/sreport/sreport.c b/src/sreport/sreport.c
index a7fc6ae3a..7f44d7d2b 100644
--- a/src/sreport/sreport.c
+++ b/src/sreport/sreport.c
@@ -516,6 +516,30 @@ _process_command (int argc, char *argv[])
argv[0]);
}
exit_flag = 1;
+ } else if (strncasecmp (argv[0], "nonparsable",
+ MAX(command_len, 4)) == 0) {
+ if (argc > 1) {
+ exit_code = 1;
+ fprintf (stderr, "too many arguments for keyword:%s\n",
+ argv[0]);
+ }
+ print_fields_parsable_print = 0;
+ } else if (strncasecmp (argv[0], "parsable",
+ MAX(command_len, 8)) == 0) {
+ if (argc > 1) {
+ exit_code = 1;
+ fprintf (stderr, "too many arguments for keyword:%s\n",
+ argv[0]);
+ }
+ print_fields_parsable_print = PRINT_FIELDS_PARSABLE_ENDING;
+ } else if (strncasecmp (argv[0], "parsable2",
+ MAX(command_len, 9)) == 0) {
+ if (argc > 1) {
+ exit_code = 1;
+ fprintf (stderr, "too many arguments for keyword:%s\n",
+ argv[0]);
+ }
+ print_fields_parsable_print = PRINT_FIELDS_PARSABLE_NO_ENDING;
} else if ((strncasecmp (argv[0], "reservation",
MAX(command_len, 2)) == 0)
|| (strncasecmp (argv[0], "resv",
@@ -649,6 +673,7 @@ sreport [<OPTION>] [<COMMAND>] \n\
Valid <COMMAND> values are: \n\
exit Terminate sreport \n\
help Print this description of use. \n\
+ nonparsable Return output to normal \n\
parsable Output will be | delimited with an ending '|' \n\
parsable2 Output will be | delimited without an ending '|' \n\
quiet Print no messages other than error messages. \n\
diff --git a/src/srun/allocate.c b/src/srun/allocate.c
index 31f76c3bb..712cece8b 100644
--- a/src/srun/allocate.c
+++ b/src/srun/allocate.c
@@ -806,6 +806,7 @@ create_job_step(srun_job_t *job, bool use_all_cpus)
(difftime(time(NULL), begin_time) > opt.immediate))) ||
((rc != ESLURM_NODES_BUSY) && (rc != ESLURM_PORTS_BUSY) &&
(rc != ESLURM_PROLOG_RUNNING) &&
+ (rc != SLURM_PROTOCOL_SOCKET_IMPL_TIMEOUT) &&
(rc != ESLURM_DISABLED))) {
error ("Unable to create job step: %m");
return -1;
diff --git a/src/sstat/print.c b/src/sstat/print.c
index 86ca691c6..f07a8ecd1 100644
--- a/src/sstat/print.c
+++ b/src/sstat/print.c
@@ -1,5 +1,5 @@
/*****************************************************************************\
- * print.c - print functions for sacct
+ * print.c - print functions for sstat
*
* $Id: print.c 7541 2006-03-18 01:44:58Z da $
*****************************************************************************
@@ -93,7 +93,7 @@ void print_fields(jobacct_step_rec_t *step)
switch(field->type) {
case PRINT_AVECPU:
- tmp_char = _elapsed_time((int)step->sacct.ave_cpu, 0);
+ tmp_char = _elapsed_time((long)step->sacct.ave_cpu, 0);
field->print_routine(field,
tmp_char,
@@ -206,7 +206,7 @@ void print_fields(jobacct_step_rec_t *step)
(curr_inx == field_count));
break;
case PRINT_MINCPU:
- tmp_char = _elapsed_time((int)step->sacct.min_cpu, 0);
+ tmp_char = _elapsed_time((long)step->sacct.min_cpu, 0);
field->print_routine(field,
tmp_char,
(curr_inx == field_count));
diff --git a/src/sstat/sstat.c b/src/sstat/sstat.c
index b3cc2d49f..3f0fe8037 100644
--- a/src/sstat/sstat.c
+++ b/src/sstat/sstat.c
@@ -132,7 +132,8 @@ int _sstat_query(slurm_step_layout_t *step_layout, uint32_t job_id,
jobacct_msg = (stat_jobacct_msg_t *)
ret_data_info->data;
if(jobacct_msg) {
- debug2("got it back for job %d",
+ debug2("got it back from %s for job %d",
+ ret_data_info->node_name,
jobacct_msg->job_id);
jobacct_gather_g_2_sacct(
&temp_sacct,
@@ -144,14 +145,16 @@ int _sstat_query(slurm_step_layout_t *step_layout, uint32_t job_id,
case RESPONSE_SLURM_RC:
rc = slurm_get_return_code(ret_data_info->type,
ret_data_info->data);
- error("there was an error with the request rc = %s",
- slurm_strerror(rc));
+ error("there was an error with the request to "
+ "%s rc = %s",
+ ret_data_info->node_name, slurm_strerror(rc));
break;
default:
rc = slurm_get_return_code(ret_data_info->type,
ret_data_info->data);
- error("unknown return given %d rc = %s",
- ret_data_info->type, slurm_strerror(rc));
+ error("unknown return given from %s: %d rc = %s",
+ ret_data_info->node_name, ret_data_info->type,
+ slurm_strerror(rc));
break;
}
}
diff --git a/src/sview/common.c b/src/sview/common.c
index 287f60ee1..3fa146a34 100644
--- a/src/sview/common.c
+++ b/src/sview/common.c
@@ -30,11 +30,35 @@
#include "src/sview/sview.h"
#include "src/common/parse_time.h"
+static bool menu_right_pressed = false;
+
typedef struct {
GtkTreeModel *model;
GtkTreeIter iter;
} treedata_t;
+/* These next 2 functions are here to make it so we don't magically
+ * click on something before we really want to in a menu.
+ */
+static gboolean _menu_button_pressed(GtkWidget *widget, GdkEventButton *event,
+ gpointer extra)
+{
+ if(event->button == 3) {
+ menu_right_pressed = true;
+ return true;
+ }
+ return false;
+}
+
+static gboolean _menu_button_released(GtkWidget *widget, GdkEventButton *event,
+ gpointer extra)
+{
+ if(event->button == 3 && !menu_right_pressed)
+ return true;
+ menu_right_pressed = false;
+ return false;
+}
+
static gboolean _entry_changed(GtkWidget *widget, void *msg)
{
global_entry_changed = 1;
@@ -372,8 +396,7 @@ static void _popup_state_changed(GtkCheckMenuItem *menuitem,
(display_data->refresh)(NULL, display_data->user_data);
}
-static void _selected_page(GtkMenuItem *menuitem,
- display_data_t *display_data)
+static void _selected_page(GtkMenuItem *menuitem, display_data_t *display_data)
{
treedata_t *treedata = (treedata_t *)display_data->user_data;
@@ -510,6 +533,13 @@ extern void make_fields_menu(popup_info_t *popup_win, GtkMenu *menu,
if(popup_win && popup_win->spec_info->type == INFO_PAGE)
return;
+ g_signal_connect(G_OBJECT(menu), "button-press-event",
+ G_CALLBACK(_menu_button_pressed),
+ NULL);
+ g_signal_connect(G_OBJECT(menu), "button-release-event",
+ G_CALLBACK(_menu_button_released),
+ NULL);
+
for(i=0; i<count; i++) {
while(display_data++) {
if(display_data->id == -1)
@@ -550,6 +580,14 @@ extern void make_options_menu(GtkTreeView *tree_view, GtkTreePath *path,
GtkWidget *menuitem = NULL;
treedata_t *treedata = xmalloc(sizeof(treedata_t));
treedata->model = gtk_tree_view_get_model(tree_view);
+
+ g_signal_connect(G_OBJECT(menu), "button-press-event",
+ G_CALLBACK(_menu_button_pressed),
+ NULL);
+ g_signal_connect(G_OBJECT(menu), "button-release-event",
+ G_CALLBACK(_menu_button_released),
+ NULL);
+
if (!gtk_tree_model_get_iter(treedata->model, &treedata->iter, path)) {
g_error("make menus error getting iter from model\n");
return;
@@ -1556,3 +1594,44 @@ extern void sview_widget_modify_bg(GtkWidget *widget, GtkStateType state,
/* g_print("%d 3 took %s\n", grid_button->inx, TIME_STR); */
}
+
+extern void sview_radio_action_set_current_value(GtkRadioAction *action,
+ gint current_value)
+{
+#ifdef GTK2_USE_RADIO_SET
+ gtk_radio_action_set_current_value(action, current_value);
+#else
+ GSList *slist, *group;
+ int i=0;
+ /* gtk_radio_action_set_current_value wasn't added to
+ GTK until 2.10, it turns out this is what is required to
+ set the correct value.
+ */
+ g_return_if_fail(GTK_IS_RADIO_ACTION(action));
+ if((group = gtk_radio_action_get_group(action))) {
+ /* for some reason groups are set backwards like a
+ stack, g_slist_reverse will fix this but takes twice
+ as long so just figure out the length, they add 1
+ to it sense 0 isn't a number and then subtract the
+ value to get the augmented in the stack.
+ */
+ current_value = g_slist_length(group) - 1 - current_value;
+ if(current_value < 0) {
+ g_warning("Radio group does not contain an action "
+ "with value '%d'\n", current_value);
+ return;
+ }
+
+ for (slist = group; slist; slist = slist->next) {
+ if(i == current_value) {
+ gtk_toggle_action_set_active(
+ GTK_TOGGLE_ACTION(slist->data), TRUE);
+ g_object_set(action, "value",
+ current_value, NULL);
+ return;
+ }
+ i++;
+ }
+ }
+#endif
+}
diff --git a/src/sview/sview.c b/src/sview/sview.c
index 997667d04..1c0158989 100644
--- a/src/sview/sview.c
+++ b/src/sview/sview.c
@@ -344,7 +344,6 @@ static void _reconfigure(GtkToggleAction *action)
static void _get_current_debug(GtkRadioAction *action)
{
-#ifdef GTK2_USE_RADIO_SET
static int debug_level = 0;
static slurm_ctl_conf_info_msg_t *slurm_ctl_conf_ptr = NULL;
static GtkAction *debug_action = NULL;
@@ -360,15 +359,8 @@ static void _get_current_debug(GtkRadioAction *action)
changed so we need to make it happen here */
if(debug_level == 0)
debug_inited = 1;
- gtk_radio_action_set_current_value(GTK_RADIO_ACTION(debug_action),
- debug_level);
-#else
- /* Since gtk_radio_action_set_current_value wasn't added to
- GTK until 2.10 we have to just not set the first one here
- correctly unless we are running with it.
- */
- debug_inited = 1;
-#endif
+ sview_radio_action_set_current_value(GTK_RADIO_ACTION(debug_action),
+ debug_level);
}
static void _set_debug(GtkRadioAction *action,
diff --git a/src/sview/sview.h b/src/sview/sview.h
index eb0b34208..b868dbd35 100644
--- a/src/sview/sview.h
+++ b/src/sview/sview.h
@@ -505,4 +505,6 @@ extern void add_display_treestore_line_with_font(
char *font);
extern void sview_widget_modify_bg(GtkWidget *widget, GtkStateType state,
const GdkColor color);
+extern void sview_radio_action_set_current_value(GtkRadioAction *action,
+ gint current_value);
#endif
diff --git a/testsuite/expect/Makefile.am b/testsuite/expect/Makefile.am
index 044904a01..d42c29a18 100644
--- a/testsuite/expect/Makefile.am
+++ b/testsuite/expect/Makefile.am
@@ -325,6 +325,7 @@ EXTRA_DIST = \
test21.25 \
test21.26 \
test21.27 \
+ test21.28 \
test22.1 \
test22.2 \
test23.1 \
diff --git a/testsuite/expect/README b/testsuite/expect/README
index dce472d94..6cdc2c56d 100644
--- a/testsuite/expect/README
+++ b/testsuite/expect/README
@@ -530,6 +530,8 @@ test21.23 sacctmgr wckey
test21.24 sacctmgr dump file
test21.25 sacctmgr show config
test21.26 sacctmgr +=, -= modify QoS test
+test21.27 sacctmgr problems test
+test21.25 sacctmgr abort delete
test22.# Testing of sreport commands and options.
These also test the sacctmgr archive dump/load functions.
diff --git a/testsuite/expect/globals b/testsuite/expect/globals
index a0c3e2e7c..8fe0b067c 100755
--- a/testsuite/expect/globals
+++ b/testsuite/expect/globals
@@ -641,6 +641,37 @@ proc test_assoc_enforced { } {
return $assoc_enforced
}
+################################################################
+#
+# Proc: test_assoc_enforced
+#
+# Purpose: Determine if we need an association to run a job.
+# This is based upon
+# the value of AccountingStorageEnforce in the slurm.conf.
+#
+# Returns level of association enforcement, 0 if none
+#
+################################################################
+proc test_limits_enforced { } {
+ global scontrol number
+
+ log_user 0
+ set enforced 0
+ spawn $scontrol show config
+ expect {
+ -re "AccountingStorageEnforce *= (\[a-z]+),limits" {
+ set enforced 1
+ exp_continue
+ }
+ eof {
+ wait
+ }
+ }
+
+ log_user 1
+ return $enforced
+}
+
################################################################
#
# Proc: test_power_save
diff --git a/testsuite/expect/test21.21 b/testsuite/expect/test21.21
index 2d1d41f52..a309b4ea7 100755
--- a/testsuite/expect/test21.21
+++ b/testsuite/expect/test21.21
@@ -2,7 +2,7 @@
############################################################################
# Purpose: Test of SLURM functionality
# sacctmgr add an account to this cluster and try using it with
-# salloc, sbatch and srun.
+# salloc, sbatch and srun. We also test limits here as well.
#
# Output: "TEST: #.#" followed by "SUCCESS" if test was successful, OR
# "FAILURE: ..." otherwise with an explanation of the failure, OR
@@ -32,6 +32,7 @@
# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
############################################################################
source ./globals
+source ./globals_accounting
set test_id "21.21"
set exit_code 0
@@ -40,6 +41,237 @@ set test_acct "test_acct"
set timeout 60
print_header $test_id
+proc _test_limits { } {
+ global test_acct file_in srun sbatch squeue scancel bin_id number bin_sleep bin_rm
+ # test maxcpumin maxcpu maxjob maxnode maxsubmit maxwall
+ set maxcpumin 10
+ set maxcpu 4
+ set maxjob 2
+ set maxnode 1
+ set maxsubmit 4
+ set maxwall 10
+ set exit_code 0
+
+ incr exit_code [mod_acct "" "" "$test_acct" "" "" "" "" "" "" "" "" "" "" "" "" $maxjob $maxnode $maxsubmit $maxwall "" "" ""]
+ if { $exit_code } {
+ return $exit_code
+ }
+ #
+ # Spawn a job testing maxnode
+ #
+ set matches 0
+ spawn $srun -v -N$maxnode --account=$test_acct $bin_id
+ expect {
+ -re "launching ($number)" {
+ incr matches
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: srun not responding\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $matches != 1 } {
+ send_user "\nFAILURE: job dion't launch with correct limit\n"
+ set exit_code 1
+ return $exit_code
+ }
+ #
+ # Spawn a job testing maxnode+1
+ #
+ spawn $srun -N[expr $maxnode + 1] --account=$test_acct $bin_id
+ expect {
+ -re "Job violates accounting policy" {
+ send_user "\nThis error is expected, not a problem\n"
+ exp_continue
+ }
+ -re "launching ($number)" {
+ send_user "\nFAILURE: job should not have run.\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: srun not responding\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $exit_code } {
+ return $exit_code
+ }
+
+ #
+ # Spawn a job testing maxwall
+ #
+ set matches 0
+ spawn $srun -v -t$maxwall --account=$test_acct $bin_id
+ expect {
+ -re "launching ($number)" {
+ incr matches
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: srun not responding\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $matches != 1 } {
+ send_user "\nFAILURE: job dion't launch with correct limit\n"
+ set exit_code 1
+ return $exit_code
+ }
+ #
+ # Spawn a job testing maxwall+1
+ #
+ spawn $srun -t[expr $maxwall + 1] --account=$test_acct $bin_id
+ expect {
+ -re "Job violates accounting policy" {
+ send_user "\nThis error is expected, not a problem\n"
+ exp_continue
+ }
+ -re "launching ($number)" {
+ send_user "\nFAILURE: job should not have run.\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: srun not responding\n"
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $exit_code } {
+ return $exit_code
+ }
+
+ # This should overwrite the old file_in which has already been used,
+ # so no big deal.
+ make_bash_script $file_in "
+ $bin_sleep 5
+ "
+
+ # test job max cnt and submit
+ for {set inx 0} {$inx < $maxsubmit} {incr inx} {
+ set job_id($inx) 0
+ set mypid [spawn $sbatch -N1 -n1 --account=$test_acct --output=/dev/null --error=/dev/null -t5 $file_in]
+ expect {
+ -re "Submitted batch job ($number)" {
+ set job_id($inx) $expect_out(1,string)
+ exp_continue
+ }
+ -re "Unable to contact" {
+ send_user "\nFAILURE: slurm appears to be down\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: sbatch not responding\n"
+ slow_kill $mypid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { !$job_id($inx) } {
+ send_user "\nFAILURE: sbatch didn't return jobid\n"
+ set exit_code 1
+ exit
+ }
+
+ if { $exit_code } {
+ break
+ }
+ }
+
+ if { $exit_code } {
+ return $exit_code
+ }
+
+ # then submit one more over the limit and it should fail
+ set mypid [spawn $sbatch -N1 -n1 --account=$test_acct --output=/dev/null --error=/dev/null -t5 $file_in]
+ expect {
+ -re "Job violates accounting policy" {
+ send_user "\nThis error is expected, not a problem\n"
+ exp_continue
+ }
+ -re "Submitted batch job ($number)" {
+ send_user "\nFAILURE: job should not have run.\n"
+ set exit_code 1
+ exp_continue
+ }
+ -re "Unable to contact" {
+ send_user "\nFAILURE: slurm appears to be down\n"
+ set exit_code 1
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: sbatch not responding\n"
+ slow_kill $mypid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ if { $exit_code } {
+ return $exit_code
+ }
+
+ set matches 0
+ set mypid [spawn $squeue -o "\%i \%t \%r"]
+ expect {
+ -re "($job_id(2)|$job_id(3)).PD.AssociationJobLimit" {
+ incr matches
+ exp_continue
+ }
+ -re "($job_id(0)|$job_id(1)).R.None" {
+ incr matches
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: squeue not responding\n"
+ slow_kill $mypid
+ set exit_code 1
+ }
+ eof {
+ wait
+ }
+ }
+
+ spawn $scancel --quiet --account=$test_acct
+ expect {
+ eof {
+ wait
+ }
+ }
+
+
+ if { $matches != 4 } {
+ send_user "\nFAILURE: job's are in the state expected got $matches\n"
+ set exit_code 1
+ return $exit_code
+ }
+
+ return $exit_code
+}
+
#
# Check accounting config and bail if not found.
#
@@ -249,6 +481,15 @@ if {$job_id == 0} {
}
}
+
+#
+# Check to see if limits are enforced.
+#
+if { [test_limits_enforced] == 1 } {
+ set exit_code [_test_limits]
+}
+
+
#
# Use sacctmgr to delete the test account
#
diff --git a/testsuite/expect/test21.27 b/testsuite/expect/test21.27
index 81fa81a04..38d87ef73 100755
--- a/testsuite/expect/test21.27
+++ b/testsuite/expect/test21.27
@@ -66,7 +66,7 @@ if { [test_account_storage] == 0 } {
}
if { [string compare [check_accounting_admin_level] "Administrator"] } {
- send_user "\nWARNING: This test can't be run without being an Accounting administrator.\nUse sacctmgr mod user \$USER_NAME admin=admin.\n"
+ send_user "\nWARNING: This test can't be run without being an Accounting administrator.\nUse sacctmgr mod user \$USER_NAME set admin=admin.\n"
exit 0
}
diff --git a/testsuite/expect/test21.28 b/testsuite/expect/test21.28
new file mode 100755
index 000000000..4f777f204
--- /dev/null
+++ b/testsuite/expect/test21.28
@@ -0,0 +1,366 @@
+#!/usr/bin/expect
+############################################################################
+# Purpose: Test of SLURM functionality
+# sacctmgr delete N (abort) test
+#
+#
+# Output: "TEST: #.#" followed by "SUCCESS" if test was successful, OR
+# "FAILURE: ..." otherwise with an explanation of the failure, OR
+# anything else indicates a failure mode that must be investigated.
+############################################################################
+# Copyright (C) 2008 Lawrence Livermore National Security.
+# Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
+# Written by Joseph Donaghy <donaghy1@llnl.gov>
+# CODE-OCEC-09-009. All rights reserved.
+#
+# This file is part of SLURM, a resource management program.
+# For details, see <https://computing.llnl.gov/linux/slurm/>.
+# 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.
+#
+# 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.
+############################################################################
+source ./globals_accounting
+
+set test_id "21.28"
+set exit_code 0
+set tc1 tcluster1
+set tc2 tcluster2
+set tc3 tcluster3
+set fs1 2500
+set fs2 1700
+set fs3 1
+set mc1 1000000
+set mc2 700000
+set mc3 1
+set mj1 50
+set mj2 70
+set mj3 1
+set mn1 300
+set mn2 700
+set mn3 1
+set mw1 01:00:00
+set mw2 00:07:00
+set mw3 00:01:00
+set clu cluster
+set cl1 1tmach
+set cl2 2tmach
+set cl3 3tmach
+set acc account
+set acc account
+set nams names
+set nm1 testaccta1
+set nm2 testaccta2
+set nm3 testaccta3
+set des Description
+set ds1 testaccounta1
+set ds2 testacct
+set org Organization
+set or1 accountorga1
+set or2 acttrg
+set qs QosLevel
+set qs1 normal
+set par parent
+set usr user
+set us1 tuser1
+set us2 tuser2
+set us3 tuser3
+set al AdminLevel
+set aln None
+set ala Administrator
+set alo Operator
+set dac DefaultAccount
+set pts Partitions
+set fs fairshare
+set mc maxcpu
+set mj maxjob
+set mn maxnode
+set mw maxwall
+set dbu debug
+set access_err 0
+set lista "$nm1 $nm2 $nm3"
+set listc "$tc1 $tc2 $tc3"
+set listu "$us1 $us2 $us3"
+#set user_name "id -u -n"
+
+print_header $test_id
+
+set timeout 60
+
+#
+# Check accounting config and bail if not found.
+#
+if { [test_account_storage] == 0 } {
+ send_user "\nWARNING: This test can't be run without a usable AccountStorageType\n"
+ exit 0
+}
+
+if { [string compare [check_accounting_admin_level] "Administrator"] } {
+ send_user "\nWARNING: This test can't be run without being an Accounting administrator.\nUse sacctmgr mod user \$USER_NAME admin=admin.\n"
+ exit 0
+}
+
+#make sure we have a clean system and permission to do this work
+remove_user "" "" "$us1,$us2,$us3"
+remove_acct "" "$nm1,$nm2,$nm3"
+remove_cluster "$tc1,$tc2,$tc3"
+if {$access_err != 0} {
+ send_user "\nWARNING: not authorized to perform this test\n"
+ exit $exit_code
+}
+
+#add cluster
+incr exit_code [add_cluster "$tc1,$tc2,$tc3" "" "" "" "" "" "" "" "" "" "" "" ""]
+if { $exit_code } {
+ remove_user "" "" "$us1,$us2,$us3"
+ remove_acct "" "$nm1,$nm2,$nm3"
+ remove_cluster "$tc1,$tc2,$tc3"
+ exit $exit_code
+}
+
+#add accounts
+incr exit_code [add_acct "$tc1,$tc2,$tc3" "" "$nm1,$nm2,$nm3" "" "" "" "" "" "" "" "" "" "" "" "" "" "" "" ""]
+if { $exit_code } {
+ remove_user "" "" "$us1,$us2,$us3"
+ remove_acct "" "$nm1,$nm2,$nm3"
+ remove_cluster "$tc1,$tc2,$tc3"
+ exit $exit_code
+}
+
+#add users
+# account adminlevel cluster defaultaccount fs maxcpu maxjob maxnodes maxwall name
+incr exit_code [add_user "$tc1,$tc2,$tc3" "$nm1,$nm2,$nm3" $us1,$us2,$us3 "" "$alo" "$nm2" "" "" "" "$fs1" "" "" "" "" "" "" "$mc1" "" "$mj1" "$mn1" "" "$mw1" ]
+if { $exit_code } {
+ remove_user "" "" "$us1,$us2,$us3"
+ remove_acct "" "$nm1,$nm2,$nm3"
+ remove_cluster "$tc1,$tc2,$tc3"
+ exit $exit_code
+}
+
+################################################################
+#
+# Proc: list_test_users
+#
+# Purpose: Use sacctmgr to list the test user additions
+#
+# Returns: Number representing the number of entries found.
+#
+# Input: None
+#
+################################################################
+proc list_test_users { } {
+ global sacctmgr
+ global us1 us2 us3 nm1 nm2 nm3 tc1 tc2 tc3 fs1 mc1 mj1 mn1 mw1 alo
+ set matches 0
+ set my_pid [spawn $sacctmgr -n -p list user format="User,DefaultA,Admin,clus,acc,fair,maxcpum,maxj,maxn,maxw" names=$us1,$us2,$us3 withassoc]
+ expect {
+ -re "There was a problem" {
+ send_user "FAILURE: there was a problem with the sacctmgr command\n"
+ }
+ -re "(($us1|$us2|$us3).$nm2.$alo.($tc1|$tc2|$tc3).($nm1|$nm2|$nm3).$fs1.$mc1.$mj1.$mn1.$mw1.)" {
+ incr matches
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: sacctmgr list associations not responding\n"
+ slow_kill $my_pid
+ }
+ eof {
+ wait
+ }
+ }
+
+ return $matches
+}
+################################################################
+################################################################
+#
+# Proc: list_test_object
+#
+# Purpose: Use sacctmgr to list the test object deletions
+#
+# Returns: Number representing the number of entries found.
+#
+# Input: None
+#
+################################################################
+proc list_test_object { to1 to2 to3 tname} {
+ global sacctmgr
+ set tmatches 0
+ set my_pid [spawn $sacctmgr -n -p list $tname $to1,$to2,$to3]
+ expect {
+ -re "There was a problem" {
+ send_user "FAILURE: there was a problem with the sacctmgr command\n"
+ }
+ -re "($to1|$to2|$to3)" {
+ incr tmatches
+ exp_continue
+ }
+ timeout {
+ send_user "\nFAILURE: sacctmgr list associations not responding\n"
+ slow_kill $my_pid
+ }
+ eof {
+ wait
+ }
+ }
+
+ return $tmatches
+}
+################################################################
+#
+# Begin here with abort delete test
+#
+foreach atest $lista {
+set my_pid [spawn $sacctmgr del $acc $atest]
+ expect {
+ "(N/y):" {
+ send "N\r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list $acc timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+
+if { [list_test_users] != 27 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not abort delete of accounts.\n"
+ exit 1
+ }
+
+#
+foreach ctest $listc {
+set my_pid [spawn $sacctmgr del $clu $ctest]
+ expect {
+ "(N/y):" {
+ send "N\r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list cluster timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+
+if { [list_test_users] != 27 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not abort delete of $clu\n"
+ exit 1
+ }
+
+#
+foreach utest $listu {
+set my_pid [spawn $sacctmgr del $usr $atest]
+ expect {
+ "(N/y):" {
+ send "N\r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list $usr timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+
+if { [list_test_users] != 27 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not abort delete of users.\n"
+ exit 1
+ }
+
+################################################################
+#
+# Begin here with delete test
+#
+#
+foreach udtest $listu {
+set my_pid [spawn $sacctmgr del $usr $udtest]
+ expect {
+ "(N/y):" {
+ send "y\r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list $usr timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+if { [list_test_object $us1 $us2 $us3 user] != 0 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not delete users.\n"
+ exit 1
+ }
+
+#
+
+foreach adtest $lista {
+set my_pid [spawn $sacctmgr del $acc $adtest]
+ expect {
+ "(N/y):" {
+ send "y\r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list $acc timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+
+if { [list_test_object $nm1 $nm2 $nm3 account] != 0 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not delete accounts.\n"
+ exit 1
+ }
+
+#
+foreach cdtest $listc {
+set my_pid [spawn $sacctmgr del $clu $cdtest]
+ expect {
+ "(N/y):" {
+ send "y/r"
+ exp_continue
+ timeout {
+ send_user "\nFAILURE: sacctmgr list cluster timing out.\n"
+ slow_kill $my_pid
+ }
+ }
+ }
+}
+
+
+if { [list_test_object $tc1 $tc2 $tc3 cluster] != 0 } {
+ send_user "\nFAILURE: This test$test_id concludes that sacctmgr did not delete $clu\n"
+ exit 1
+ }
+
+# This is the end below here
+
+incr exit_code [remove_user "" "" "$us1,$us2,$us3"]
+incr exit_code [remove_acct "" "$nm1,$nm2,$nm3"]
+incr exit_code [remove_cluster "$tc1,$tc2,$tc3"]
+
+if {$exit_code == 0} {
+ send_user "\nSUCCESS\n"
+} else {
+ send_user "\nFAILURE\n"
+}
+exit $exit_code
+
--
GitLab