From 1973ebef68c29425b950d7735aa60280fd2f48f5 Mon Sep 17 00:00:00 2001
From: Moe Jette <jette1@llnl.gov>
Date: Tue, 17 Aug 2004 15:52:37 +0000
Subject: [PATCH] Srun cancels a created job if job step creation fails (rather
than leaving it orphaned).
---
NEWS | 8 +++++++-
src/srun/allocate.c | 16 ++++++++++------
src/srun/allocate.h | 4 +++-
src/srun/srun.c | 8 ++++++--
4 files changed, 26 insertions(+), 10 deletions(-)
diff --git a/NEWS b/NEWS
index ddd3a74591e..7df4cc54ab5 100644
--- a/NEWS
+++ b/NEWS
@@ -3,8 +3,14 @@ documents those changes that are of interest to users and admins.
* Changes in SLURM 0.4.0-pre2
=============================
+ -- Fixes for reported problems:
+ - slurm/477: Signal of batch job script (scancel -b) fixed
-- NOTE: "startclean" when transitioning from version 0.4.0-pre1, JOBS ARE LOST
- -- Preserve job's requested processor count info after job is initiated
+ -- Preserve job's requested processor count info after job is initiated
+ (for viewing by squeue and scontrol)
+ -- Added a lots of Blue Gene/L support logic: slurmd executes on a single
+ node to front-end the 512-CPU base-partitions (Blue Gene/L's nodes)
+ -- srun cancels created job if job step creation fails
* Changes in SLURM 0.4.0-pre1
=============================
diff --git a/src/srun/allocate.c b/src/srun/allocate.c
index 4d11e2ac676..d9819cecc5d 100644
--- a/src/srun/allocate.c
+++ b/src/srun/allocate.c
@@ -469,17 +469,20 @@ _step_req_destroy(job_step_create_request_msg_t *r)
}
}
-void
+int
create_job_step(job_t *job)
{
job_step_create_request_msg_t *req = NULL;
job_step_create_response_msg_t *resp = NULL;
- if (!(req = _step_req_create(job)))
- fatal ("Unable to allocate step request message");
-
- if ((slurm_job_step_create(req, &resp) < 0) || (resp == NULL))
- fatal ("Unable to create job step: %m");
+ if (!(req = _step_req_create(job))) {
+ error ("Unable to allocate step request message");
+ return -1;
+ }
+ if ((slurm_job_step_create(req, &resp) < 0) || (resp == NULL)) {
+ error ("Unable to create job step: %m");
+ return -1;
+ }
job->stepid = resp->job_step_id;
job->cred = resp->cred;
@@ -490,5 +493,6 @@ create_job_step(job_t *job)
job_update_io_fnames(job);
_step_req_destroy(req);
+ return 0;
}
diff --git a/src/srun/allocate.h b/src/srun/allocate.h
index 2eef790161a..d92ebc81108 100644
--- a/src/srun/allocate.h
+++ b/src/srun/allocate.h
@@ -74,8 +74,10 @@ uint32_t jobid_from_env(void);
/*
* Create a job step given the job information stored in 'j'
* After returning, 'j' is filled in with information for job step.
+ *
+ * Returns -1 if job step creation failure, 0 otherwise
*/
-void create_job_step(job_t *j);
+int create_job_step(job_t *j);
#endif /* !_HAVE_ALLOCATE_H */
diff --git a/src/srun/srun.c b/src/srun/srun.c
index 804c9ed3ade..530234bc22d 100644
--- a/src/srun/srun.c
+++ b/src/srun/srun.c
@@ -155,7 +155,8 @@ int srun(int ac, char **av)
job = job_create_allocation(resp);
job->old_job = true;
sig_setup_sigmask();
- create_job_step(job);
+ if (create_job_step(job) < 0)
+ exit(1);
slurm_free_resource_allocation_response_msg(resp);
} else if (opt.allocate) {
@@ -191,7 +192,10 @@ int srun(int ac, char **av)
_print_job_information(resp);
job = job_create_allocation(resp);
- create_job_step(job);
+ if (create_job_step(job) < 0) {
+ job_destroy(job, 0);
+ exit(1);
+ }
slurm_free_resource_allocation_response_msg(resp);
}
--
GitLab