From 2d6945c2fd7cd9c7a78a33b71b18bdb2e3af96b0 Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Wed, 13 Aug 2003 20:56:17 +0000
Subject: [PATCH] Don't sleep and retransmit node registration message (causes
slurmd's to hand if race condition starting all daemons).
---
src/common/slurm_protocol_api.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/src/common/slurm_protocol_api.c b/src/common/slurm_protocol_api.c
index 16d4751ed48..3422000515b 100644
--- a/src/common/slurm_protocol_api.c
+++ b/src/common/slurm_protocol_api.c
@@ -800,8 +800,8 @@ _send_and_recv_msg(slurm_fd fd, slurm_msg_t *req, slurm_msg_t *resp,
* opens a connection to the controller, sends the controller a message,
* listens for the response, then closes the connection
* IN request_msg - slurm_msg request
- * OUT response_msg - slurm_msg response
- * RET int - return code
+ * OUT response_msg - slurm_msg response
+ * RET int - return code
*/
int slurm_send_recv_controller_msg(slurm_msg_t *req, slurm_msg_t *resp)
{
@@ -820,8 +820,10 @@ int slurm_send_recv_controller_msg(slurm_msg_t *req, slurm_msg_t *resp)
(resp->msg_type == RESPONSE_SLURM_RC) &&
((((return_code_msg_t *) resp->data)->return_code) ==
ESLURM_IN_STANDBY_MODE) &&
+ (req->msg_type != MESSAGE_NODE_REGISTRATION_STATUS) &&
(slurmctld_conf.backup_controller)) {
- debug("Neither primary nor backup controller responding, sleep and retry");
+ debug("Neither primary nor backup controller responding, "
+ "sleep and retry");
slurm_free_return_code_msg(resp->data);
sleep(slurmctld_conf.slurmctld_timeout +
slurmctld_conf.heartbeat_interval);
--
GitLab