diff --git a/NEWS b/NEWS
index ae679b8be85626be4206bde244817a9103ca6667..b6ae20383e500606b237e308d5970c33616b9301 100644
--- a/NEWS
+++ b/NEWS
@@ -12,6 +12,9 @@ documents those changes that are of interest to users and admins.
-- Remove sched/wiki plugin (use sched/wiki2 for now)
-- Disable pthread_create() for PMI_send when TotalView is running for
better performance.
+ -- fixed certain tests in test suite to not run with bluegene or front-end
+ systems
+ -- removed addresses from slurm_step_layout_t
* Changes in SLURM 1.2.0-pre2
=============================
diff --git a/slurm/slurm.h.in b/slurm/slurm.h.in
index a3a8c12ae5cc9c1b77d0c84ee60a7531eb9dd504..b4d0ced47ebd307b612f4a11e77e8235b1491eff 100644
--- a/slurm/slurm.h.in
+++ b/slurm/slurm.h.in
@@ -506,8 +506,6 @@ typedef struct slurm_step_layout {
uint16_t node_cnt; /* node count */
uint32_t task_cnt; /* total number of tasks in the step */
char *node_list; /* list of nodes in step */
- slurm_addr *node_addr; /* corresponding addresses */
-
/* Array of length "node_cnt". Each element of the array
is the number of tasks assigned to the corresponding node */
uint32_t *tasks;
diff --git a/src/api/job_info.c b/src/api/job_info.c
index 74481eb5ab9a1cd3d3dfd69cd86c2864ac3cac65..9f3ffc7f31b39da5c26cf74a7c8da3a19b618c52 100644
--- a/src/api/job_info.c
+++ b/src/api/job_info.c
@@ -16,7 +16,7 @@
* any later version.
*
* In addition, as a special exception, the copyright holders give permission
- * to link the code of portions of this program with the OpenSSL library under
+ * to link the code of portions of this program with the OpenSSL library under
* certain conditions as described in each individual source file, and
* distribute linked combinations including the two. You must obey the GNU
* General Public License in all respects for all of the code used other than
@@ -345,7 +345,6 @@ slurm_sprint_job_info ( job_info_t * job_ptr, int one_liner )
xstrcat(out, "\n ");
xstrcat(out, select_buf);
}
- xstrcat(out, select_buf);
xstrcat(out, "\n\n");
return out;
diff --git a/src/api/spawn.c b/src/api/spawn.c
index 596a7a59578968ed6ca0ebaca97fe3b990e7010f..2f737b9a76f4e954d7beca0a3384962e08a34faa 100644
--- a/src/api/spawn.c
+++ b/src/api/spawn.c
@@ -59,6 +59,7 @@
#include "src/common/hostlist.h"
#include "src/common/slurm_protocol_api.h"
#include "src/common/slurm_protocol_defs.h"
+#include "src/common/read_config.h"
#include "src/common/xmalloc.h"
#include "src/common/xstring.h"
@@ -109,7 +110,9 @@ extern int slurm_spawn (slurm_step_ctx ctx, int *fd_array)
int task_cnt = 0;
uint32_t *cpus = NULL;
slurm_step_layout_t *step_layout = ctx->step_resp->step_layout;
-
+ hostlist_t hl = NULL;
+ char *name = NULL;
+
if ((ctx == NULL) ||
(ctx->magic != STEP_CTX_MAGIC) ||
(fd_array == NULL)) {
@@ -152,6 +155,7 @@ extern int slurm_spawn (slurm_step_ctx ctx, int *fd_array)
req_array_ptr = xmalloc(sizeof(slurm_msg_t) *
step_layout->node_cnt);
+ hl = hostlist_create(step_layout->node_list);
for (i=0; i<step_layout->node_cnt; i++) {
spawn_task_request_msg_t *r = &msg_array_ptr[i];
slurm_msg_t *m = &req_array_ptr[i];
@@ -179,15 +183,28 @@ extern int slurm_spawn (slurm_step_ctx ctx, int *fd_array)
m->msg_type = REQUEST_SPAWN_TASK;
m->data = r;
-
- memcpy(&m->address, &step_layout->node_addr[i],
- sizeof(slurm_addr));
+ name = hostlist_shift(hl);
+ if(!name) {
+ error("hostlist incomplete for this job request");
+ hostlist_destroy(hl);
+ return SLURM_ERROR;
+ }
+ if(slurm_conf_get_addr(name, &m->address)
+ == SLURM_ERROR) {
+ error("_init_task_layout: can't get addr for "
+ "host %s", name);
+ free(name);
+ hostlist_destroy(hl);
+ return SLURM_ERROR;
+ }
+ free(name);
#if _DEBUG
printf("tid=%d, fd=%d, port=%u, node_id=%u\n",
step_layout->tids[i][0],
fd_array[i], r->io_port, i);
#endif
}
+ hostlist_destroy(hl);
rc = _p_launch(req_array_ptr, ctx);
xfree(msg_array_ptr);
diff --git a/src/common/read_config.c b/src/common/read_config.c
index 16b5368bd152ae96e6410132fef058b58fb9acfe..169592ad676db4ff759310f97f5f72df7a22af09 100644
--- a/src/common/read_config.c
+++ b/src/common/read_config.c
@@ -714,6 +714,7 @@ static int _register_conf_node_aliases(slurm_conf_node_t *node_ptr)
free(hostname);
free(address);
#endif
+
}
/* free allocated storage */
@@ -795,8 +796,8 @@ extern char *slurm_conf_get_nodename(const char *node_hostname)
slurm_conf_lock();
_init_slurmd_nodehash();
-
idx = _get_hash_idx(node_hostname);
+
p = host_to_node_hashtbl[idx];
while (p) {
if (strcmp(p->hostname, node_hostname) == 0) {
diff --git a/src/common/slurm_step_layout.c b/src/common/slurm_step_layout.c
index 12595caeadaa2e8906a65257d493e443f9e3556c..c9687fc5fc1046b50cd0a6bb7e2896dbc2b76018 100644
--- a/src/common/slurm_step_layout.c
+++ b/src/common/slurm_step_layout.c
@@ -158,7 +158,7 @@ slurm_step_layout_t *fake_slurm_step_layout_create(
{
uint32_t cpn = 1;
int cpu_cnt = 0, cpu_inx = 0, i, j;
- char *name = NULL;
+/* char *name = NULL; */
hostlist_t hl = NULL;
slurm_step_layout_t *step_layout =
xmalloc(sizeof(slurm_step_layout_t));
@@ -185,8 +185,8 @@ slurm_step_layout_t *fake_slurm_step_layout_create(
step_layout->node_cnt = node_cnt;
step_layout->tasks = xmalloc(sizeof(uint32_t) * node_cnt);
step_layout->tids = xmalloc(sizeof(uint32_t *) * node_cnt);
- step_layout->node_addr =
- xmalloc(sizeof(slurm_addr) * node_cnt);
+/* step_layout->node_addr = */
+/* xmalloc(sizeof(slurm_addr) * node_cnt); */
step_layout->task_cnt = 0;
for (i=0; i<step_layout->node_cnt; i++) {
@@ -223,26 +223,26 @@ slurm_step_layout_t *fake_slurm_step_layout_create(
}
}
}
- name = hostlist_shift(hl);
- if(!name) {
- error("fake_slurm_step_layout_create: "
- "We don't have the correct nodelist.");
- goto error;
- }
- if(slurm_conf_get_addr(name, &step_layout->node_addr[i]) ==
- SLURM_ERROR) {
- error("fake_slurm_step_layout_create: "
- "we didn't get an addr for host %s.", name);
+/* name = hostlist_shift(hl); */
+/* if(!name) { */
+/* error("fake_slurm_step_layout_create: " */
+/* "We don't have the correct nodelist."); */
+/* goto error; */
+/* } */
+/* if(slurm_conf_get_addr(name, &step_layout->node_addr[i]) == */
+/* SLURM_ERROR) { */
+/* error("fake_slurm_step_layout_create: " */
+/* "we didn't get an addr for host %s.", name); */
- }
- free(name);
+/* } */
+/* free(name); */
}
hostlist_destroy(hl);
return step_layout;
-error:
- hostlist_destroy(hl);
- slurm_step_layout_destroy(step_layout);
- return NULL;
+/* error: */
+/* hostlist_destroy(hl); */
+/* slurm_step_layout_destroy(step_layout); */
+/* return NULL; */
}
@@ -261,9 +261,9 @@ extern slurm_step_layout_t *slurm_step_layout_copy(
layout->node_cnt = step_layout->node_cnt;
layout->task_cnt = step_layout->task_cnt;
- layout->node_addr = xmalloc(sizeof(slurm_addr) * layout->node_cnt);
- memcpy(layout->node_addr, step_layout->node_addr,
- (sizeof(slurm_addr) * layout->node_cnt));
+/* layout->node_addr = xmalloc(sizeof(slurm_addr) * layout->node_cnt); */
+/* memcpy(layout->node_addr, step_layout->node_addr, */
+/* (sizeof(slurm_addr) * layout->node_cnt)); */
layout->tasks = xmalloc(sizeof(uint32_t) * layout->node_cnt);
memcpy(layout->tasks, step_layout->tasks,
@@ -292,8 +292,8 @@ extern void pack_slurm_step_layout(slurm_step_layout_t *step_layout,
packstr(step_layout->node_list, buffer);
pack16(step_layout->node_cnt, buffer);
pack32(step_layout->task_cnt, buffer);
- slurm_pack_slurm_addr_array(step_layout->node_addr,
- step_layout->node_cnt, buffer);
+/* slurm_pack_slurm_addr_array(step_layout->node_addr, */
+/* step_layout->node_cnt, buffer); */
for(i=0; i<step_layout->node_cnt; i++) {
pack32_array(step_layout->tids[i], step_layout->tasks[i],
@@ -323,11 +323,11 @@ extern int unpack_slurm_step_layout(slurm_step_layout_t **layout, Buf buffer)
safe_unpack16(&step_layout->node_cnt, buffer);
safe_unpack32(&step_layout->task_cnt, buffer);
- if (slurm_unpack_slurm_addr_array(&(step_layout->node_addr),
- &uint16_tmp, buffer))
- goto unpack_error;
- if (uint16_tmp != step_layout->node_cnt)
- goto unpack_error;
+/* if (slurm_unpack_slurm_addr_array(&(step_layout->node_addr), */
+/* &uint16_tmp, buffer)) */
+/* goto unpack_error; */
+/* if (uint16_tmp != step_layout->node_cnt) */
+/* goto unpack_error; */
step_layout->tasks = xmalloc(sizeof(uint32_t) * step_layout->node_cnt);
step_layout->tids = xmalloc(sizeof(uint32_t *)
@@ -353,7 +353,7 @@ extern int slurm_step_layout_destroy(slurm_step_layout_t *step_layout)
int i=0;
if(step_layout) {
xfree(step_layout->node_list);
- xfree(step_layout->node_addr);
+/* xfree(step_layout->node_addr); */
xfree(step_layout->tasks);
for (i = 0; i < step_layout->node_cnt; i++) {
xfree(step_layout->tids[i]);
@@ -397,7 +397,7 @@ static int _init_task_layout(slurm_step_layout_t *step_layout,
{
int cpu_cnt = 0, cpu_inx = 0, i;
hostlist_t hl = NULL;
- char *name = NULL;
+/* char *name = NULL; */
uint32_t cpus[step_layout->node_cnt];
if (step_layout->node_cnt == 0)
@@ -405,8 +405,8 @@ static int _init_task_layout(slurm_step_layout_t *step_layout,
if (step_layout->tasks) /* layout already completed */
return SLURM_SUCCESS;
- step_layout->node_addr = xmalloc(sizeof(slurm_addr)
- * step_layout->node_cnt);
+/* step_layout->node_addr = xmalloc(sizeof(slurm_addr) */
+/* * step_layout->node_cnt); */
step_layout->tasks = xmalloc(sizeof(uint32_t)
* step_layout->node_cnt);
step_layout->tids = xmalloc(sizeof(uint32_t *)
@@ -428,22 +428,22 @@ static int _init_task_layout(slurm_step_layout_t *step_layout,
}
for (i=0; i<step_layout->node_cnt; i++) {
- name = hostlist_shift(hl);
- if(!name) {
- error("hostlist incomplete for this job request");
- hostlist_destroy(hl);
- return SLURM_ERROR;
- }
- if(slurm_conf_get_addr(name, &step_layout->node_addr[i])
- == SLURM_ERROR) {
- error("_init_task_layout: can't get addr for "
- "host %s", name);
- free(name);
- continue;
- }
+/* name = hostlist_shift(hl); */
+/* if(!name) { */
+/* error("hostlist incomplete for this job request"); */
+/* hostlist_destroy(hl); */
+/* return SLURM_ERROR; */
+/* } */
+/* if(slurm_conf_get_addr(name, &step_layout->node_addr[i]) */
+/* == SLURM_ERROR) { */
+/* error("_init_task_layout: can't get addr for " */
+/* "host %s", name); */
+/* free(name); */
+/* continue; */
+/* } */
- debug2("host %d = %s", i, name);
- free(name);
+/* debug2("host %d = %s", i, name); */
+/* free(name); */
cpus[i] = cpus_per_node[cpu_inx];
if ((++cpu_cnt) >= cpu_count_reps[cpu_inx]) {
diff --git a/src/plugins/jobacct/common/common_slurmctld.c b/src/plugins/jobacct/common/common_slurmctld.c
index 76be8ee50f72e49da10244b21b4e3f229e365451..8a6dd5fd3f46722efbc879198993aa435db9d08f 100644
--- a/src/plugins/jobacct/common/common_slurmctld.c
+++ b/src/plugins/jobacct/common/common_slurmctld.c
@@ -287,14 +287,14 @@ extern int common_step_start_slurmctld(struct step_record *step)
if(quarter != (uint16_t)NO_VAL
&& nodecard != (uint16_t)NO_VAL)
snprintf(node_list, BUFFER_SIZE,
- "%s.%d.%d", step->step_layout->node_list,
+ "%s.%d.%d", step->job_ptr->nodes,
quarter, nodecard);
else if(quarter != (uint16_t)NO_VAL)
snprintf(node_list, BUFFER_SIZE,
- "%s.%d", step->step_layout->node_list, quarter);
+ "%s.%d", step->job_ptr->nodes, quarter);
else
snprintf(node_list, BUFFER_SIZE, "%s",
- step->step_layout->node_list);
+ step->job_ptr->nodes);
#else
if(!step->step_layout || !step->step_layout->task_cnt) {
@@ -405,14 +405,14 @@ extern int common_step_complete_slurmctld(struct step_record *step)
if(quarter != (uint16_t)NO_VAL
&& nodecard != (uint16_t)NO_VAL)
snprintf(node_list, BUFFER_SIZE,
- "%s.%d.%d", step->step_layout->node_list,
+ "%s.%d.%d", step->job_ptr->nodes,
quarter, nodecard);
else if(quarter != (uint16_t)NO_VAL)
snprintf(node_list, BUFFER_SIZE,
- "%s.%d", step->step_layout->node_list, quarter);
+ "%s.%d", step->job_ptr->nodes, quarter);
else
snprintf(node_list, BUFFER_SIZE, "%s",
- step->step_layout->node_list);
+ step->job_ptr->nodes);
#else
if(!step->step_layout || !step->step_layout->task_cnt) {
@@ -511,7 +511,7 @@ extern int common_suspend_slurmctld(struct job_record *job_ptr)
debug("jobacct init was not called or it failed");
return SLURM_ERROR;
}
-
+
/* tell what time has passed */
if(!now)
now = job_ptr->start_time;
diff --git a/src/sacct/print.c b/src/sacct/print.c
index f2fb7e1f7d231534e473ac76d356243e1e7165d9..5ba577be4abaf7f7640b6e7a04d70232b1faf75d 100644
--- a/src/sacct/print.c
+++ b/src/sacct/print.c
@@ -40,7 +40,7 @@
#include "sacct.h"
#include "src/common/parse_time.h"
#include "slurm.h"
-#define FORMAT_STRING_SIZE 32
+#define FORMAT_STRING_SIZE 50
void _elapsed_time(long secs, long usecs, char *str);
@@ -713,10 +713,10 @@ void print_pages(type_t type, void *object)
switch(type) {
case HEADLINE:
- printf("%-35s", "MaxPages/Node:Task - Ave");
+ printf("%-50s", "MaxPages/Node:Task - Ave");
break;
case UNDERSCORE:
- printf("%-35s", "----------------------------------");
+ printf("%-50s", "----------------------------------");
break;
case JOB:
sacct = job->sacct;
@@ -736,7 +736,7 @@ void print_pages(type_t type, void *object)
sacct.max_pages_id.taskid,
buf2);
}
- printf("%-35s", outbuf);
+ printf("%-50s", outbuf);
break;
case JOBSTEP:
sacct = step->sacct;
@@ -750,7 +750,7 @@ void print_pages(type_t type, void *object)
buf3,
sacct.max_pages_id.taskid,
buf2);
- printf("%-35s", outbuf);
+ printf("%-50s", outbuf);
break;
}
}
@@ -769,10 +769,10 @@ void print_rss(type_t type, void *object)
switch(type) {
case HEADLINE:
- printf("%-32s", "MaxRSS/Node:Task - Ave");
+ printf("%-50s", "MaxRSS/Node:Task - Ave");
break;
case UNDERSCORE:
- printf("%-32s", "--------------------------------");
+ printf("%-50s", "--------------------------------");
break;
case JOB:
sacct = job->sacct;
@@ -792,7 +792,7 @@ void print_rss(type_t type, void *object)
sacct.max_rss_id.taskid,
buf2);
}
- printf("%-32s", outbuf);
+ printf("%-50s", outbuf);
break;
case JOBSTEP:
sacct = step->sacct;
@@ -806,7 +806,7 @@ void print_rss(type_t type, void *object)
buf3,
sacct.max_rss_id.taskid,
buf2);
- printf("%-32s", outbuf);
+ printf("%-50s", outbuf);
break;
}
}
@@ -1044,10 +1044,10 @@ void print_vsize(type_t type, void *object)
switch(type) {
case HEADLINE:
- printf("%-34s", "MaxVSIZE/Node:Task - Ave");
+ printf("%-50s", "MaxVSIZE/Node:Task - Ave");
break;
case UNDERSCORE:
- printf("%-34s", "----------------------------------");
+ printf("%-50s", "----------------------------------");
break;
case JOB:
sacct = job->sacct;
@@ -1066,7 +1066,7 @@ void print_vsize(type_t type, void *object)
sacct.max_vsize_id.taskid,
buf2);
}
- printf("%-34s", outbuf);
+ printf("%-50s", outbuf);
break;
case JOBSTEP:
sacct = step->sacct;
@@ -1080,7 +1080,7 @@ void print_vsize(type_t type, void *object)
buf3,
sacct.max_vsize_id.taskid,
buf2);
- printf("%-34s", outbuf);
+ printf("%-50s", outbuf);
break;
}
}
@@ -1099,10 +1099,10 @@ void print_cputime(type_t type, void *object)
switch(type) {
case HEADLINE:
- printf("%-36s", "MinCPUtime/Node:Task - Ave");
+ printf("%-50s", "MinCPUtime/Node:Task - Ave");
break;
case UNDERSCORE:
- printf("%-36s", "------------------------------------");
+ printf("%-50s", "------------------------------------");
break;
case JOB:
sacct = job->sacct;
@@ -1122,7 +1122,7 @@ void print_cputime(type_t type, void *object)
sacct.min_cpu_id.taskid,
buf2);
}
- printf("%-36s", outbuf);
+ printf("%-50s", outbuf);
break;
case JOBSTEP:
sacct = step->sacct;
@@ -1137,7 +1137,7 @@ void print_cputime(type_t type, void *object)
buf3,
sacct.min_cpu_id.taskid,
buf2);
- printf("%-36s", outbuf);
+ printf("%-50s", outbuf);
break;
}
}
diff --git a/src/slurmctld/job_mgr.c b/src/slurmctld/job_mgr.c
index e428c4c21d0ad97613efbadb21050a435f102105..57d84879cd306b66bb4fd87857f504ee4e7dfef3 100644
--- a/src/slurmctld/job_mgr.c
+++ b/src/slurmctld/job_mgr.c
@@ -4212,7 +4212,8 @@ extern int job_suspend(suspend_msg_t *sus_ptr, uid_t uid,
job_ptr->suspend_time = now;
reply:
- jobacct_g_suspend_slurmctld(job_ptr);
+ if(job_ptr)
+ jobacct_g_suspend_slurmctld(job_ptr);
if (conn_fd >= 0) {
slurm_msg_t_init(&resp_msg);
diff --git a/src/slurmd/slurmd/slurmd.c b/src/slurmd/slurmd/slurmd.c
index c90ec6f7a7b4b2e26d6339e1bee4f92ab752af62..853ee85a94b3dfd35b907120f65b7c84d568bb01 100644
--- a/src/slurmd/slurmd/slurmd.c
+++ b/src/slurmd/slurmd/slurmd.c
@@ -16,7 +16,7 @@
* any later version.
*
* In addition, as a special exception, the copyright holders give permission
- * to link the code of portions of this program with the OpenSSL library under
+ * to link the code of portions of this program with the OpenSSL library under
* certain conditions as described in each individual source file, and
* distribute linked combinations including the two. You must obey the GNU
* General Public License in all respects for all of the code used other than
diff --git a/src/slurmd/slurmstepd/mgr.c b/src/slurmd/slurmstepd/mgr.c
index 61e9b54c99f097527e68ad7018c5cfdba8f66892..7010dc84ed9273521d86572ea2d034ca0578ee89 100644
--- a/src/slurmd/slurmstepd/mgr.c
+++ b/src/slurmd/slurmstepd/mgr.c
@@ -16,7 +16,7 @@
* any later version.
*
* In addition, as a special exception, the copyright holders give permission
- * to link the code of portions of this program with the OpenSSL library under
+ * to link the code of portions of this program with the OpenSSL library under
* certain conditions as described in each individual source file, and
* distribute linked combinations including the two. You must obey the GNU
* General Public License in all respects for all of the code used other than
@@ -1327,8 +1327,15 @@ _send_launch_failure (launch_tasks_request_msg_t *msg, slurm_addr *cli, int rc)
{
slurm_msg_t resp_msg;
launch_tasks_response_msg_t resp;
- int nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
-
+ int nodeid = 0;
+ char *name = NULL;
+#ifndef HAVE_FRONT_END
+ nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
+ name = xstrdup(conf->node_name);
+#else
+ name = xstrdup(msg->complete_nodelist);
+
+#endif
debug ("sending launch failure message: %s", slurm_strerror (rc));
slurm_msg_t_init(&resp_msg);
@@ -1339,12 +1346,12 @@ _send_launch_failure (launch_tasks_request_msg_t *msg, slurm_addr *cli, int rc)
resp_msg.data = &resp;
resp_msg.msg_type = RESPONSE_LAUNCH_TASKS;
- resp.node_name = conf->node_name;
+ resp.node_name = name;
resp.return_code = rc ? rc : -1;
resp.count_of_pids = 0;
slurm_send_only_node_msg(&resp_msg);
-
+ xfree(name);
return;
}
@@ -1366,7 +1373,7 @@ _send_launch_resp(slurmd_job_t *job, int rc)
resp_msg.data = &resp;
resp_msg.msg_type = RESPONSE_LAUNCH_TASKS;
- resp.node_name = conf->node_name;
+ resp.node_name = xstrdup(job->node_name);
resp.return_code = rc;
resp.count_of_pids = job->ntasks;
@@ -1381,6 +1388,7 @@ _send_launch_resp(slurmd_job_t *job, int rc)
xfree(resp.local_pids);
xfree(resp.task_ids);
+ xfree(resp.node_name);
}
@@ -1396,7 +1404,7 @@ _complete_batch_script(slurmd_job_t *job, int err, int status)
req.slurm_rc = err;
slurm_msg_t_init(&req_msg);
- req.node_name = conf->node_name;
+ req.node_name = job->node_name;
req_msg.msg_type= REQUEST_COMPLETE_BATCH_SCRIPT;
req_msg.data = &req;
diff --git a/src/slurmd/slurmstepd/slurmstepd.c b/src/slurmd/slurmstepd/slurmstepd.c
index 08682cecbb15727f77abd5d603926b0468197a40..c022c0242264eadde9dd178ea694377e9533bbd8 100644
--- a/src/slurmd/slurmstepd/slurmstepd.c
+++ b/src/slurmd/slurmstepd/slurmstepd.c
@@ -346,6 +346,9 @@ _step_setup(slurm_addr *cli, slurm_addr *self, slurm_msg_t *msg)
fatal("handle_launch_message: Unrecognized launch/spawn RPC");
break;
}
+ if(!job) {
+ fatal("_step_setup: no job returned");
+ }
job->jmgr_pid = getpid();
job->jobacct = jobacct_g_alloc(NULL);
diff --git a/src/slurmd/slurmstepd/slurmstepd_job.c b/src/slurmd/slurmstepd/slurmstepd_job.c
index c9129641f8dff862825c7a63ccd19a49c7c9cb64..28bddef37247c53c95ef9202b59081c16cc74853 100644
--- a/src/slurmd/slurmstepd/slurmstepd_job.c
+++ b/src/slurmd/slurmstepd/slurmstepd_job.c
@@ -157,15 +157,7 @@ job_create(launch_tasks_request_msg_t *msg, slurm_addr *cli_addr)
xassert(msg != NULL);
xassert(msg->complete_nodelist != NULL);
-
- nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
- debug3("entering job_create on node %d", nodeid);
-
- if(nodeid < 0) {
- error("couldn't find node %s in %s",
- conf->node_name, msg->complete_nodelist);
- return NULL;
- }
+ debug3("entering job_create");
if ((pwd = _pwd_create((uid_t)msg->uid)) == NULL) {
error("uid %ld not found on system", (long) msg->uid);
slurm_seterrno (ESLURMD_UID_NOT_FOUND);
@@ -177,7 +169,20 @@ job_create(launch_tasks_request_msg_t *msg, slurm_addr *cli_addr)
return NULL;
}
job = xmalloc(sizeof(slurmd_job_t));
-
+#ifndef HAVE_FRONT_END
+ nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
+ job->node_name = xstrdup(conf->node_name);
+#else
+ nodeid = 0;
+ job->node_name = xstrdup(msg->complete_nodelist);
+#endif
+ if(nodeid < 0) {
+ error("couldn't find node %s in %s",
+ job->node_name, msg->complete_nodelist);
+ job_destroy(job);
+ return NULL;
+ }
+
job->state = SLURMSTEPD_STEP_STARTING;
job->pwd = pwd;
job->ntasks = msg->tasks_to_launch[nodeid];
@@ -266,18 +271,29 @@ job_spawn_create(spawn_task_request_msg_t *msg, slurm_addr *cli_addr)
int nodeid = NO_VAL;
xassert(msg != NULL);
+ xassert(msg->complete_nodelist != NULL);
debug3("entering job_spawn_create");
- nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
-
if ((pwd = _pwd_create((uid_t)msg->uid)) == NULL) {
error("uid %ld not found on system", (long) msg->uid);
slurm_seterrno (ESLURMD_UID_NOT_FOUND);
return NULL;
}
job = xmalloc(sizeof(slurmd_job_t));
-
+#ifndef HAVE_FRONT_END
+ nodeid = nodelist_find(msg->complete_nodelist, conf->node_name);
+ job->node_name = xstrdup(conf->node_name);
+#else
+ nodeid = 0;
+ job->node_name = xstrdup(msg->complete_nodelist);
+#endif
+ if(nodeid < 0) {
+ error("couldn't find node %s in %s",
+ job->node_name, msg->complete_nodelist);
+ job_destroy(job);
+ return NULL;
+ }
job->state = SLURMSTEPD_STEP_STARTING;
job->pwd = pwd;
job->ntasks = 1; /* tasks to launch always one */
@@ -516,6 +532,7 @@ job_destroy(slurmd_job_t *job)
task_info_destroy(job->task[i]);
list_destroy(job->sruns);
xfree(job->envtp);
+ xfree(job->node_name);
xfree(job->task_prolog);
xfree(job->task_epilog);
xfree(job);
diff --git a/src/slurmd/slurmstepd/slurmstepd_job.h b/src/slurmd/slurmstepd/slurmstepd_job.h
index 16cd1233bbd7ed6f9f85e940224f7dc300dafcb4..1b119a31166c88855d7d0a1baef25c5f40443d01 100644
--- a/src/slurmd/slurmstepd/slurmstepd_job.h
+++ b/src/slurmd/slurmstepd/slurmstepd_job.h
@@ -118,6 +118,8 @@ typedef struct slurmd_job {
char **env; /* job environment */
char **argv; /* job argument vector */
char *cwd; /* path to current working directory */
+ char *node_name; /* node name of node running job
+ * needed for front-end systems */
cpu_bind_type_t cpu_bind_type; /* --cpu_bind= */
char *cpu_bind; /* binding map for map/mask_cpu */
mem_bind_type_t mem_bind_type; /* --mem_bind= */
diff --git a/src/srun/msg.c b/src/srun/msg.c
index 660689ff056251dfd52731e86c3850e96f879510..90020d8ad6030c5a22207e7dcdd3e651b17f7dcf 100644
--- a/src/srun/msg.c
+++ b/src/srun/msg.c
@@ -363,8 +363,9 @@ static void
_process_launch_resp(srun_job_t *job, launch_tasks_response_msg_t *msg)
{
pipe_enum_t pipe_enum = PIPE_HOST_STATE;
- int nodeid = nodelist_find(job->step_layout->node_list,
+ int nodeid = nodelist_find(job->step_layout->node_list,
msg->node_name);
+
if ((nodeid < 0) || (nodeid >= job->nhosts)) {
error ("Bad launch response from %s", msg->node_name);
return;
diff --git a/src/srun/reattach.c b/src/srun/reattach.c
index 9a727315b532aebd3ea3564da077a2d3cd571888..7b32f31f76d9d8d41e655b32c7afe3f68c2c1a88 100644
--- a/src/srun/reattach.c
+++ b/src/srun/reattach.c
@@ -319,12 +319,15 @@ _attach_to_job(srun_job_t *job)
int i;
reattach_tasks_request_msg_t *req = NULL;
slurm_msg_t *msg = NULL;
-
+ hostlist_t hl = NULL;
+ char *name = NULL;
+
req = xmalloc(job->nhosts * sizeof(reattach_tasks_request_msg_t));
msg = xmalloc(job->nhosts * sizeof(slurm_msg_t));
debug("Going to attach to job %u.%u", job->jobid, job->stepid);
+ hl = hostlist_create(job->step_layout->node_list);
for (i = 0; i < job->nhosts; i++) {
reattach_tasks_request_msg_t *r = &req[i];
slurm_msg_t *m = &msg[i];
@@ -342,13 +345,27 @@ _attach_to_job(srun_job_t *job)
slurm_msg_t_init(m);
m->data = r;
m->msg_type = REQUEST_REATTACH_TASKS;
-
- memcpy(&m->address, &job->step_layout->node_addr[i],
- sizeof(slurm_addr));
+ name = hostlist_shift(hl);
+ if(!name) {
+ error("hostlist incomplete for this job request");
+ hostlist_destroy(hl);
+ return SLURM_ERROR;
+ }
+ if(slurm_conf_get_addr(name, &m->address)
+ == SLURM_ERROR) {
+ error("_init_task_layout: can't get addr for "
+ "host %s", name);
+ free(name);
+ hostlist_destroy(hl);
+ return SLURM_ERROR;
+ }
+ free(name);
+ /* memcpy(&m->address, &job->step_layout->node_addr[i], */
+/* sizeof(slurm_addr)); */
}
-
+ hostlist_destroy(hl);
_p_reattach(msg, job);
-
+
return SLURM_SUCCESS;
}
diff --git a/testsuite/expect/test1.80 b/testsuite/expect/test1.80
index 6248842a8a0e275a076e140f320a4d4080d8eb22..8715e68e62e568455cc8db89a81561796fd02fdf 100755
--- a/testsuite/expect/test1.80
+++ b/testsuite/expect/test1.80
@@ -38,8 +38,8 @@ set exit_code 0
print_header $test_id
-if { [test_xcpu] } {
- send_user "\nWARNING: This test is incompatable with XCPU systems\n"
+if { [test_front_end] } {
+ send_user "\nWARNING: This test is incompatable with front end systems\n"
exit 0
}
diff --git a/testsuite/expect/test12.2 b/testsuite/expect/test12.2
index fdc73d59865a8f21551da974302c68b684c476cb..a8691d8071e164949de06eecada43a498908e77a 100755
--- a/testsuite/expect/test12.2
+++ b/testsuite/expect/test12.2
@@ -48,6 +48,10 @@ set ret_code 42
print_header $test_id
+if {[test_bluegene] != 0} {
+ send_user "\nWARNING: This test is incompatable with bluegene systems\n"
+ exit 0
+}
#
# Check if accounting is enabled
#
diff --git a/testsuite/expect/test18.27 b/testsuite/expect/test18.27
index 8bed364f38caff9a8134a6bec8fc8bb269d82007..e20136ce8ea373913442b413331153a0e6b895e3 100755
--- a/testsuite/expect/test18.27
+++ b/testsuite/expect/test18.27
@@ -38,9 +38,9 @@ set exit_code 0
print_header $test_id
-if { [test_xcpu] } {
- send_user "\nWARNING: This test is incompatable with XCPU systems\n"
- exit 0
+if {[test_front_end] != 0} {
+ send_user "\nWARNING: Additional testing is incompatable with front-end systems\n"
+ exit $exit_code
}
#
@@ -91,11 +91,6 @@ if {[string compare $expected_layout $tested_layout] != 0} {
set exit_code 1
}
-if {[test_front_end] != 0} {
- send_user "\nWARNING: Additional testing is incompatable with front-end systems\n"
- exit $exit_code
-}
-
#
# Submit a two node job with cyclic distribution
#
diff --git a/testsuite/expect/test3.7 b/testsuite/expect/test3.7
index abcc862766783ec59c5974d9b17aaf5f059418b2..659cbcd3f2fe28e027c649450f7c6d566147fba0 100755
--- a/testsuite/expect/test3.7
+++ b/testsuite/expect/test3.7
@@ -204,7 +204,6 @@ sleep 5
suspend_job $job_id1 suspend
if {$not_supported == 1} {
exec $scancel $job_id1
- exec $scancel $job_id2
exit 0
}
if {$not_supported == 0} {