From ffcbce9de6d3595232c69810fe2f1492dc5319a8 Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Fri, 25 Jun 2004 21:06:49 +0000
Subject: [PATCH] Resync node state to DRAINED or DRAINING on slurmctld restart
(in the event that node state and job count are out of sync).
---
src/slurmctld/read_config.c | 22 ++++++++++++++++++++--
1 file changed, 20 insertions(+), 2 deletions(-)
diff --git a/src/slurmctld/read_config.c b/src/slurmctld/read_config.c
index 8daaca31d2c..24b2da12711 100644
--- a/src/slurmctld/read_config.c
+++ b/src/slurmctld/read_config.c
@@ -154,9 +154,9 @@ static int _build_bitmaps(void)
list_iterator_destroy(job_iterator);
/* scan all nodes and identify which are up, idle and
- * their configuration */
+ * their configuration, resync DRAINED vs. DRAINING state */
for (i = 0; i < node_record_count; i++) {
- uint16_t base_state, no_resp_flag;
+ uint16_t base_state, no_resp_flag, job_cnt;
if (node_record_table_ptr[i].name[0] == '\0')
continue; /* defunct */
@@ -164,6 +164,24 @@ static int _build_bitmaps(void)
(~NODE_STATE_NO_RESPOND);
no_resp_flag = node_record_table_ptr[i].node_state &
NODE_STATE_NO_RESPOND;
+ job_cnt = node_record_table_ptr[i].run_job_cnt +
+ node_record_table_ptr[i].comp_job_cnt;
+
+ if ((base_state == NODE_STATE_DRAINED) &&
+ (job_cnt > 0)) {
+ error("Bad node drain state for %s",
+ node_record_table_ptr[i].name);
+ node_record_table_ptr[i].node_state =
+ NODE_STATE_DRAINING | no_resp_flag;
+ }
+ if ((base_state == NODE_STATE_DRAINING) &&
+ (job_cnt == 0)) {
+ error("Bad node drain state for %s",
+ node_record_table_ptr[i].name);
+ node_record_table_ptr[i].node_state =
+ NODE_STATE_DRAINED | no_resp_flag;
+ }
+
if ((base_state == NODE_STATE_IDLE ) ||
(base_state == NODE_STATE_DOWN ) ||
(base_state == NODE_STATE_DRAINED))
--
GitLab