From e5cef28bfa4b646a91c8cd1a465e6d6c7e949f2f Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Wed, 16 Oct 2002 23:02:41 +0000
Subject: [PATCH] Added new error code ESLURM_JOB_PENDING if allocation check
RPC finds job is still not running. srun to wait for job initiation as needed
with polling.
---
src/srun/srun.c | 21 ++++++++++++++++++---
1 file changed, 18 insertions(+), 3 deletions(-)
diff --git a/src/srun/srun.c b/src/srun/srun.c
index ea8f7140fd8..d9c60194b9c 100644
--- a/src/srun/srun.c
+++ b/src/srun/srun.c
@@ -274,6 +274,7 @@ allocate_nodes(void)
int rc, retries;
job_desc_msg_t job;
resource_allocation_response_msg_t *resp;
+ old_job_alloc_msg_t old_job;
slurm_init_job_desc_msg(&job);
@@ -325,9 +326,23 @@ allocate_nodes(void)
}
}
- if (resp->node_list == NULL) {
- info("No nodes allocated. exiting");
- return NULL;
+ if ((rc == 0) && (resp->node_list == NULL)) {
+ if (_verbose || _debug)
+ info ("Job %u queued and waiting for resources", resp->job_id);
+ old_job.job_id = resp->job_id;
+ old_job.uid = (uint32_t) getuid();
+ slurm_free_resource_allocation_response_msg (resp);
+ sleep (2);
+ /* Keep polling until the job is allocated resources */
+ while (slurm_confirm_allocation(&old_job, &resp) == SLURM_FAILURE) {
+ if (slurm_get_errno() == ESLURM_JOB_PENDING)
+ sleep (10);
+ else {
+ error("Unable to confirm resource allocation for job %u: %s",
+ old_job.job_id, slurm_strerror(errno));
+ exit (1);
+ }
+ }
}
return resp;
--
GitLab