On Tue, Nov 11, 2014 at 11:48:18AM -0800, Daniel Jung wrote:
I thought the following line meant it nss got SIGTERM no?
(Wed Oct 29 15:44:43:453191 2014) [sssd] [get_ping_config] (0x0100): Time
between SIGTERM and SIGKILL for [nss]: [60]
(Wed Oct 29 15:44:43:453206 2014) [sssd] [start_service] (0x0100): Queueing
service nss for startup
No, this is merely a configuration notice that if a process is sent
SIGTERM and doesn't quit in 60 seconds after that, it's forcibly shut
down with SIGKILL.
The reason we try SIGTERM first is that in some cases (like, database
write in progress), SIGKILL might corrupt the database. SIGTERM allows
for a more graceful shutdown.
So if you're seeing issues, I'm afraid these are not the symptoms...