From 8cfe42d94ce9bafee7c9159b7426b1364867ba24 Mon Sep 17 00:00:00 2001
From: Danny Auble <da@llnl.gov>
Date: Fri, 18 Jul 2008 22:53:28 +0000
Subject: [PATCH] svn merge -r14569:14574
https://eris.llnl.gov/svn/slurm/branches/slurm-1.2
---
NEWS | 4 ++--
src/srun/srun.c | 22 ++++++++++++++++------
2 files changed, 18 insertions(+), 8 deletions(-)
diff --git a/NEWS b/NEWS
index 123d2eaabf4..236144c8195 100644
--- a/NEWS
+++ b/NEWS
@@ -407,13 +407,13 @@ documents those changes that are of interest to users and admins.
* Changes in SLURM 1.2.33
=========================
- -- Cancelled or Failed jobs will now report there job and step id on exit
+ -- Cancelled or Failed jobs will now report their job and step id on exit
-- Add SPANK items available to get: SLURM_VERSION, SLURM_VERSION_MAJOR,
SLURM_VERISON_MINOR and SLURM_VERSION_MICRO.
-- Fixed handling of SIGPIPE in srun. Abort job.
-- Fix bug introduced to MVAPICH plugin preventing use of TotalView debugger.
-- Modify slurmctld to get srun/salloc network address based upon the incoming
- message rather than hostname set by the user command (more reliable).
+ message rather than hostname set by the user command (backport of logic in SLURM v1.3).
* Changes in SLURM 1.2.32
=========================
diff --git a/src/srun/srun.c b/src/srun/srun.c
index b47b22ead85..b02333867c9 100644
--- a/src/srun/srun.c
+++ b/src/srun/srun.c
@@ -120,6 +120,7 @@ static int _become_user (void);
static int _call_spank_local_user (srun_job_t *job);
static void _define_symbols(void);
static void _handle_intr();
+static void _handle_pipe(int signo);
static void _handle_signal(int signo);
static void _print_job_information(resource_allocation_response_msg_t *resp);
static void _pty_restore(void);
@@ -241,11 +242,6 @@ int srun(int ac, char **av)
if (!job || create_job_step(job) < 0)
exit(1);
- /* ignore sigpipe for steps. This is already done
- * when allocating nodes elsewhere.
- */
- xsignal(SIGPIPE, ignore_signal);
-
} else {
got_alloc = 1;
/* Combined job allocation and job step launch */
@@ -1080,6 +1076,16 @@ static void _handle_intr()
}
}
+static void _handle_pipe(int signo)
+{
+ static int ending = 0;
+
+ if(ending)
+ return;
+ ending = 1;
+ slurm_step_launch_abort(job->step_ctx);
+}
+
static void _handle_signal(int signo)
{
debug2("got signal %d", signo);
@@ -1116,7 +1122,7 @@ static int _setup_signals()
{
int sigarray[] = {
SIGINT, SIGQUIT, /*SIGTSTP,*/ SIGCONT, SIGTERM,
- SIGALRM, SIGUSR1, SIGUSR2, /* SIGPIPE,*/ 0
+ SIGALRM, SIGUSR1, SIGUSR2, /*SIGPIPE,*/ 0
};
int rc = SLURM_SUCCESS, i=0, signo;
@@ -1125,6 +1131,10 @@ static int _setup_signals()
while ((signo = sigarray[i++]))
xsignal(signo, _handle_signal);
+ /* special case for SIGPIPE since we don't want to print stuff
+ * and get into a locked up state
+ */
+ xsignal(SIGPIPE, _handle_pipe);
return rc;
}
--
GitLab