On Wed, Oct 26, 2016 at 11:26:37AM +0200, Vincent Lefevre wrote:
> On 2016-10-26 10:53:43 +0200, Marc Haber wrote:
> > Can you reproduce this? If so, I'd like you to attach an strace to the
> > atopacct process (strace -f -p <pid of atopacctd> -s 2000 -o
> > strace.atopacctd) and attach the resulting file. I'd like to see
> > whether atopacctd gets the SIGTERM and how it reacts.
>
> I've also added the -t option to get the time:
Great, thanks, I keep forgetting this.
> 910 11:22:14 recvfrom(5, 0x7ffe6905d290, 16000, 0, NULL, NULL) = ?
> ERESTARTSYS (To be restarted if SA_RESTART is set)
> 910 11:22:29 --- SIGTERM {si_signo=SIGTERM, si_code=SI_USER, si_pid=1,
> si_uid=0} ---
> 910 11:22:29 --- SIGCONT {si_signo=SIGCONT, si_code=SI_USER, si_pid=1,
> si_uid=0} ---
> 910 11:22:29 rt_sigreturn({mask=[]}) = 45
> 910 11:22:29 recvfrom(5, <unfinished ...>
> 910 11:22:49 +++ killed by SIGKILL +++
>
> Note that I had reduced the systemd timeout to 20 seconds.
> You can see it above.
So the process receives the SIGTERM and doesn't act on it.
And, incidentally, when trying to obtain an strace of an atopacctd
terminating properly, mine is not terminating as well, so my strace is
identical.
I'll forward this upstream.
Greetings
Marc
--
-----------------------------------------------------------------------------
Marc Haber | "I don't trust Computers. They | Mailadresse im Header
Leimen, Germany | lose things." Winona Ryder | Fon: *49 6224 1600402
Nordisch by Nature | How to make an American Quilt | Fax: *49 6224 1600421
