From 162d164bcbddd4c443f8ac8bfc018695e146e1af Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Wed, 15 Nov 2006 18:47:54 +0000
Subject: [PATCH] Make sure the backfill plugin pthread is terminated at
slurmctld shutdown in order to free all memory (including that associated
with pthread_create).
---
src/plugins/sched/backfill/backfill.c | 21 +++++++++++++------
src/plugins/sched/backfill/backfill.h | 3 +++
src/plugins/sched/backfill/backfill_wrapper.c | 1 +
src/slurmctld/controller.c | 2 +-
4 files changed, 20 insertions(+), 7 deletions(-)
diff --git a/src/plugins/sched/backfill/backfill.c b/src/plugins/sched/backfill/backfill.c
index fdcabfd0a5d..404564b1d28 100644
--- a/src/plugins/sched/backfill/backfill.c
+++ b/src/plugins/sched/backfill/backfill.c
@@ -14,7 +14,7 @@
* "lx[06-08]", we can't start it without possibly delaying the higher
* priority job.
*****************************************************************************
- * Copyright (C) 2003 The Regents of the University of California.
+ * Copyright (C) 2003-2006 The Regents of the University of California.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
* Written by Morris Jette <jette1@llnl.gov>
* UCRL-CODE-217948.
@@ -79,8 +79,9 @@ typedef struct node_space_map {
} node_space_map_t;
/*********************** local variables *********************/
-static bool altered_job = false;
-static bool new_work = false;
+static bool altered_job = false;
+static bool new_work = false;
+static bool stop_backfill = false;
static pthread_mutex_t thread_flag_mutex = PTHREAD_MUTEX_INITIALIZER;
static List pend_job_list = NULL;
@@ -93,7 +94,7 @@ static node_space_map_t node_space[MAX_JOB_CNT + 1];
/* Set __DEBUG to get detailed logging for this thread without
* detailed logging for the entire slurmctld daemon */
#define __DEBUG 0
-#define SLEEP_TIME 2
+#define SLEEP_TIME 1
/*********************** local functions *********************/
static int _add_pending_job(struct job_record *job_ptr,
@@ -162,6 +163,13 @@ static void _diff_tv_str(struct timeval *tv1,struct timeval *tv2,
snprintf(tv_str, len_tv_str, "usec=%ld", delta_t);
}
+/* Terminate backfill_agent */
+extern void stop_backfill_agent(void)
+{
+ stop_backfill = true;
+}
+
+
/* backfill_agent - detached thread periodically attempts to backfill jobs */
extern void *
backfill_agent(void *args)
@@ -175,9 +183,9 @@ backfill_agent(void *args)
if (slurm_get_root_filter())
filter_root = true;
- while (1) {
+ while (!stop_backfill) {
sleep(SLEEP_TIME); /* don't run continuously */
- if (!_more_work())
+ if ((!_more_work()) || stop_backfill)
continue;
gettimeofday(&tv1, NULL);
@@ -210,6 +218,7 @@ backfill_agent(void *args)
schedule(); /* has own locks */
}
}
+ return NULL;
}
/* trigger the attempt of a backfill */
diff --git a/src/plugins/sched/backfill/backfill.h b/src/plugins/sched/backfill/backfill.h
index 2db7e766bbd..e960a3df037 100644
--- a/src/plugins/sched/backfill/backfill.h
+++ b/src/plugins/sched/backfill/backfill.h
@@ -38,6 +38,9 @@
/* backfill_agent - detached thread periodically attempts to backfill jobs */
extern void *backfill_agent(void *args);
+/* Terminate backfill_agent */
+extern void stop_backfill_agent(void);
+
/* trigger the attempt of a backfill */
extern void run_backfill (void);
diff --git a/src/plugins/sched/backfill/backfill_wrapper.c b/src/plugins/sched/backfill/backfill_wrapper.c
index dbf33c67cc7..3663aaf9a1e 100644
--- a/src/plugins/sched/backfill/backfill_wrapper.c
+++ b/src/plugins/sched/backfill/backfill_wrapper.c
@@ -112,6 +112,7 @@ void fini( void )
pthread_mutex_lock( &thread_flag_mutex );
if ( backfill_thread ) {
verbose( "Backfill scheduler plugin shutting down" );
+ stop_backfill_agent();
_cancel_thread( backfill_thread );
backfill_thread = false;
}
diff --git a/src/slurmctld/controller.c b/src/slurmctld/controller.c
index e453404b45e..8c4a2a9c5d4 100644
--- a/src/slurmctld/controller.c
+++ b/src/slurmctld/controller.c
@@ -377,7 +377,7 @@ int main(int argc, char *argv[])
slurm_cred_ctx_destroy(slurmctld_config.cred_ctx);
slurm_conf_destroy();
slurm_api_clear_config();
- sleep(1);
+ sleep(2);
#endif
info("Slurmctld shutdown completing");
--
GitLab