On Sun, Feb 16, 2020 at 10:25 PM stan via users
<users(a)lists.fedoraproject.org> wrote:
On Sun, 16 Feb 2020 13:41:06 -0500
Tom Horsley <horsley1953(a)gmail.com> wrote:
>
> An infection seems to be spreading in systemd. First
> I saw dhcpd taking forever to shut down:
>
>
https://bugzilla.redhat.com/show_bug.cgi?id=1768604
>
> Now I just saw the exact same thing with the apache
> httpd service.
>
> I found the systemctl --no-ask-password option, so
> I tried it, now it no longer forks and execs the
> password-agent process, it merely sits like a lump
> timing out for the same amount of time.
>
> Anyone have any idea what leads systemd to decide
> it needs to wait for some mysterious something
> when stopping some services? And why is the number
> of these mysterious services growing? (Does this
> have something to do with me not using NetworkManager?)
I run a custom daemon to gather entropy, and it sleeps most of the
time. If I don't kill -9 it before I shut down, it is sleeping, and I
get the behavior you describe (it doesn't catch the kill signal,
presumably -15, that systemd sends). When I researched the issue, I
got as far as finding that there is a timer that can be set that
defaults to 1 minute 30 seconds. I'm sure it is in the documentation
where this timer is configured, but I never got that far. Setting it
to a lower number for this, and other things, would reduce your
aggravation.
"DefaultTimeoutStartSec" and "DefaultTimeoutStopSec" in
"/etc/systemd/system.conf"