diff --git a/NEWS b/NEWS
index 19bc21a4f1267c3f91afe73e5e10110f780199a6..2337486492576411c6b6e6feef0386c6486cf805 100644
--- a/NEWS
+++ b/NEWS
@@ -41,6 +41,7 @@ documents those changes that are of interest to users and admins.
by the resp is NULL.
-- Fix switch/federation plugin so backup controller can assume control
repeatedly without leaking or corrupting memory.
+ -- Add new error code (for Maui/Moab scheduler): ESLURM_JOB_HELD
* Changes in SLURM 1.1.9
========================
diff --git a/slurm/slurm_errno.h b/slurm/slurm_errno.h
index 3fcfe7642181280a5cb4f9f090bac7018298f71b..e20de3359598f50a1eb7047c095f7a2a1b7f84a7 100644
--- a/slurm/slurm_errno.h
+++ b/slurm/slurm_errno.h
@@ -1,7 +1,7 @@
/*****************************************************************************\
* slurm_errno.h - error codes and functions for slurm
******************************************************************************
- * Copyright (C) 2002 The Regents of the University of California.
+ * Copyright (C) 2002-2006 The Regents of the University of California.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
* Written by Kevin Tew <tew1@llnl.gov>,
* Jim Garlick <garlick@llnl.gov>, et. al.
@@ -136,6 +136,7 @@ enum {
ESLURM_BATCH_ONLY,
ESLURM_TASKDIST_ARBITRARY_UNSUPPORTED,
ESLURM_TASKDIST_REQUIRES_OVERCOMMIT,
+ ESLURM_JOB_HELD,
/* switch specific error codes, specific values defined in plugin module */
ESLURM_SWITCH_MIN = 3000,
diff --git a/src/common/slurm_errno.c b/src/common/slurm_errno.c
index e84004bc5e03fac573829eb67d76aa153d249fd6..b1dc992d722002b47c125e27658fc5a486783587 100644
--- a/src/common/slurm_errno.c
+++ b/src/common/slurm_errno.c
@@ -1,7 +1,7 @@
/*****************************************************************************\
* slurm_errno.c - error codes and functions for slurm
******************************************************************************
- * Copyright (C) 2002 The Regents of the University of California.
+ * Copyright (C) 2002-2006 The Regents of the University of California.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
* Written by Jim Garlick <garlick@llnl.gov>, et. al.
* UCRL-CODE-217948.
@@ -179,6 +179,8 @@ static slurm_errtab_t slurm_errtab[] = {
"Current SwitchType does not permit arbitrary task distribution"},
{ ESLURM_TASKDIST_REQUIRES_OVERCOMMIT,
"Requested more tasks than available processors" },
+ { ESLURM_JOB_HELD,
+ "Job is in held state, pending scheduler release" },
/* slurmd error codes */
diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index 64b66c54e186d80515f07be0f14dad3a86081238..c479cfd876960e02400d3d1451aeb51308d4e62a 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -1411,8 +1411,9 @@ extern int job_allocate(job_desc_msg_t * job_specs, int immediate, int will_run,
error_code = select_nodes(job_ptr, no_alloc);
- if ((error_code == ESLURM_NODES_BUSY) ||
- (error_code == ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE)) {
+ if ((error_code == ESLURM_NODES_BUSY)
+ || (error_code == ESLURM_JOB_HELD)
+ || (error_code == ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE)) {
/* Not fatal error, but job can't be scheduled right now */
if (immediate) {
job_ptr->job_state = JOB_FAILED;
@@ -2514,7 +2515,7 @@ static int _validate_job_desc(job_desc_msg_t * job_desc_msg, int allocate,
if ((submit_uid != 0) &&
(submit_uid != slurmctld_conf.slurm_user_id)) {
info("attempt by uid %u to set job_id", submit_uid);
- return ESLURM_DUPLICATE_JOB_ID;
+ return ESLURM_INVALID_JOB_ID;
}
if (job_desc_msg->job_id == 0) {
info("attempt by uid %u to set zero job_id", submit_uid);
diff --git a/src/slurmctld/node_scheduler.c b/src/slurmctld/node_scheduler.c
index 0df933363f2f4dfe2d8aca76d76e260f1d8890a8..1e7c68945e9fedf58983c521d3e24646181fd349 100644
--- a/src/slurmctld/node_scheduler.c
+++ b/src/slurmctld/node_scheduler.c
@@ -838,9 +838,10 @@ extern int select_nodes(struct job_record *job_ptr, bool test_only)
if (fail_reason != WAIT_NO_REASON) {
if (detail_ptr)
detail_ptr->wait_reason = fail_reason;
- if (job_ptr->priority != 0) /* not user/admin hold */
- job_ptr->priority = 1; /* sys hold, move to end of queue */
last_job_update = time(NULL);
+ if (job_ptr->priority == 0) /* user/admin hold */
+ return ESLURM_JOB_HELD;
+ job_ptr->priority = 1; /* sys hold, move to end of queue */
return ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE;
}
diff --git a/src/slurmctld/proc_req.c b/src/slurmctld/proc_req.c
index 1b223b1634b90da6fc1457847c991cac68239182..dd59b3a2813b11d6dd740affaf4bd9de1f5edfcb 100644
--- a/src/slurmctld/proc_req.c
+++ b/src/slurmctld/proc_req.c
@@ -458,6 +458,7 @@ static void _slurm_rpc_allocate_resources(slurm_msg_t * msg)
uid_t uid;
int immediate = job_desc_msg->immediate;
bool do_unlock = false;
+ bool job_waiting = false;
struct job_record *job_ptr;
START_TIMER;
@@ -483,9 +484,12 @@ static void _slurm_rpc_allocate_resources(slurm_msg_t * msg)
}
/* return result */
- if ((error_code == SLURM_SUCCESS) ||
- ((immediate == 0) &&
- (error_code == ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE))) {
+ if ((error_code == ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE)
+ || (error_code == ESLURM_JOB_HELD))
+ job_waiting = true;
+
+ if ((error_code == SLURM_SUCCESS)
+ || ((immediate == 0) && job_waiting)) {
xassert(job_ptr);
info("_slurm_rpc_allocate_resources JobId=%u NodeList=%s %s",
job_ptr->job_id, job_ptr->nodes, TIME_STR);
@@ -1767,8 +1771,9 @@ static void _slurm_rpc_submit_batch_job(slurm_msg_t * msg)
}
/* return result */
- if ((error_code != SLURM_SUCCESS) &&
- (error_code != ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE)) {
+ if ((error_code != SLURM_SUCCESS)
+ && (error_code != ESLURM_JOB_HELD)
+ && (error_code != ESLURM_REQUESTED_PART_CONFIG_UNAVAILABLE)) {
info("_slurm_rpc_submit_batch_job: %s",
slurm_strerror(error_code));
slurm_send_rc_msg(msg, error_code);
diff --git a/src/srun/allocate.c b/src/srun/allocate.c
index 95ee43ee11c1e1fd95f2da59b88b6a8ea2d9516d..7062b56676833cc44f5d476617ed867611cd298a 100644
--- a/src/srun/allocate.c
+++ b/src/srun/allocate.c
@@ -121,7 +121,7 @@ allocate_nodes(void)
if ((rc == 0) && (resp->node_list == NULL)) {
if (resp->error_code)
- info("Warning: %s", slurm_strerror(resp->error_code));
+ verbose("Warning: %s", slurm_strerror(resp->error_code));
_wait_for_resources(&resp);
}
/* For diagnosing a node problem, administrators need to sometimes