From 584fd444b095e6b4bb84e90118b79197bc458d3c Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Thu, 29 Mar 2007 19:05:54 +0000
Subject: [PATCH] svn merge -r11263:11266
https://eris.llnl.gov/svn/slurm/branches/slurm-1.1
---
NEWS | 2 ++
etc/init.d.slurm | 31 ++++++++++++++++++++++++++++---
2 files changed, 30 insertions(+), 3 deletions(-)
diff --git a/NEWS b/NEWS
index 63d8e430a9a..2b6ad4ab8d8 100644
--- a/NEWS
+++ b/NEWS
@@ -264,6 +264,8 @@ documents those changes that are of interest to users and admins.
=========================
- Insure that slurm_signal_job_step() is defined in srun for mvapich
and mpichgm error conditions.
+ - Modify /etc/init.d/slurm restart command to wait for daemon to terminate
+ before starting a new one
- Permit job steps to be started on draining nodes that have already
been allocated to that job.
diff --git a/etc/init.d.slurm b/etc/init.d.slurm
index 045a38e9094..7e807576e34 100644
--- a/etc/init.d.slurm
+++ b/etc/init.d.slurm
@@ -93,6 +93,7 @@ startall() {
done
}
+#
# status() with slight modifications to take into account
# instantiations of job manager slurmd's, which should not be
# counted as "running"
@@ -124,7 +125,11 @@ slurmstatus() {
fi
done
elif [ "$rpid" != "" -a "$pid" = "" ]; then
- echo $"${base} dead but pid file exists"
+# Due to change in user id, pid file may persist
+# after slurmctld terminates
+ if [ "$base" != "slurmctld" ] ; then
+ echo $"${base} dead but pid file exists"
+ fi
return 1
fi
@@ -140,6 +145,27 @@ slurmstatus() {
return 3
}
+#
+# stop slurm daemons,
+# wait for termination to complete (up to 10 seconds), then restart
+#
+slurmrestart() {
+ for prog in `$BINDIR/scontrol show daemons`; do
+ stop $prog
+
+ for i in 1 2 3 4
+ do
+ sleep $i
+ slurmstatus $prog
+ if [ $? != 0 ]; then
+ break
+ fi
+ done
+
+ optvar=`echo ${prog}_OPTIONS | tr "a-z" "A-Z"`
+ start $prog ${!optvar}
+ done
+}
#
# The pathname substitution in daemon command assumes prefix and
@@ -168,8 +194,7 @@ case "$1" in
done
;;
restart)
- $0 stop
- $0 start
+ slurmrestart
;;
condrestart)
if [ -f /var/lock/subsys/slurm ]; then
--
GitLab