From 1ed38f267e7044b249fb29f5d90ab9100d2a4edc Mon Sep 17 00:00:00 2001
From: Morris Jette <jette@schedmd.com>
Date: Fri, 18 Mar 2016 08:13:10 -0700
Subject: [PATCH] Fix for srun abort on SIGSTOP+SIGCONT
Avoid possibly aborting srun that gets simultaneous SIGSTOP+SIGCONT while
creating the job step. The result is that the signal hanlder gets a
argument (the signal received) of zero.
Here's a log, window 1:
$ srun hostname
srun: Job step creation temporarily disabled, retrying
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 18
srun: I Got signal 0
srun: Cancelled pending job step
Window 2:
$ kill -STOP 18696 ; kill -CONT 18696
$ kill -STOP 18696 ; kill -CONT 18696
$ kill -STOP 18696 ; kill -CONT 18696
....
bug 2494
---
NEWS | 2 ++
src/api/step_ctx.c | 3 ++-
src/srun/libsrun/allocate.c | 3 ++-
3 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/NEWS b/NEWS
index 07caefd5807..c99c0c1f40d 100644
--- a/NEWS
+++ b/NEWS
@@ -60,6 +60,8 @@ documents those changes that are of interest to users and administrators.
-- Update gang scheduling data structures when job changes in size.
-- Associations - prevent hash table corruption if uid initially unset for
a user, which can cause slurmctld to crash if that user is deleted.
+ -- Avoid possibly aborting srun that gets simultaneous SIGSTOP+SIGCONT while
+ creating the job step.
* Changes in Slurm 15.08.8
==========================
diff --git a/src/api/step_ctx.c b/src/api/step_ctx.c
index 0302c6652f6..fe3be8b3e96 100644
--- a/src/api/step_ctx.c
+++ b/src/api/step_ctx.c
@@ -77,7 +77,8 @@ static int destroy_step = 0;
static void _signal_while_allocating(int signo)
{
debug("Got signal %d", signo);
- if (signo == SIGCONT)
+ /* NOTE: Near simultaneous SIGSTOP+SIGCONT can result in signo == 0 */
+ if ((signo == SIGCONT) || (signo == 0))
return;
destroy_step = signo;
diff --git a/src/srun/libsrun/allocate.c b/src/srun/libsrun/allocate.c
index d74df0b5834..c861c9532d4 100644
--- a/src/srun/libsrun/allocate.c
+++ b/src/srun/libsrun/allocate.c
@@ -118,7 +118,8 @@ static void *_safe_signal_while_allocating(void *in_data)
int signo = *(int *)in_data;
debug("Got signal %d", signo);
- if (signo == SIGCONT)
+ /* NOTE: Near simultaneous SIGSTOP+SIGCONT can result in signo == 0 */
+ if ((signo == SIGCONT) || (signo == 0))
return NULL;
destroy_job = 1;
--
GitLab