When a job is requeued during the prologue step, be sure that the
other moms involved in a multinode allocation find out.  Otherwise they
will report errors when the job is rerun on them later.


---

 pbs-2.3.12-pw/src/include/mom_func.h  |    1 +
 pbs-2.3.12-pw/src/resmom/mom_comm.c   |    4 +---
 pbs-2.3.12-pw/src/resmom/start_exec.c |    1 +
 3 files changed, 3 insertions(+), 3 deletions(-)

diff -puN src/include/mom_func.h~prologue-bounce src/include/mom_func.h
--- pbs-2.3.12/src/include/mom_func.h~prologue-bounce	2004-04-17 11:02:33.000000000 -0400
+++ pbs-2.3.12-pw/src/include/mom_func.h	2004-04-17 11:02:33.000000000 -0400
@@ -157,6 +157,7 @@ extern int   run_pelog A_((int which, ch
 extern int   is_joined A_((job *));
 extern void  check_busy A_((double));
 extern void  state_to_server A_((int));
+extern int send_sisters(const job *pjob, int com);
 
 /* Defines for pe_io_type, see run_pelog() */
 
diff -puN src/resmom/mom_comm.c~prologue-bounce src/resmom/mom_comm.c
--- pbs-2.3.12/src/resmom/mom_comm.c~prologue-bounce	2004-04-17 11:02:33.000000000 -0400
+++ pbs-2.3.12-pw/src/resmom/mom_comm.c	2004-04-17 11:02:33.000000000 -0400
@@ -467,9 +467,7 @@ im_compose(stream, jobid, cookie, comman
 **	the job -> pjob.
 */
 int
-send_sisters(pjob, com)
-    job		*pjob;
-    int		com;
+send_sisters(const job *pjob, int com)
 {
 	int		i, num, ret;
 	eventent	*ep;
diff -puN src/resmom/start_exec.c~prologue-bounce src/resmom/start_exec.c
--- pbs-2.3.12/src/resmom/start_exec.c~prologue-bounce	2004-04-17 11:02:33.000000000 -0400
+++ pbs-2.3.12-pw/src/resmom/start_exec.c	2004-04-17 11:03:39.000000000 -0400
@@ -314,6 +314,7 @@ int mom_restart_job(pjob, path)
 void
 exec_bail(job *pjob, int code)
 {
+	(void) send_sisters(pjob, IM_ABORT_JOB);
 	pjob->ji_qs.ji_substate = JOB_SUBSTATE_EXITING;
 	pjob->ji_qs.ji_un.ji_momt.ji_exitstat = code;
 	exiting_tasks = 1;

_