On 07/28/2016 11:53 AM, Tom Horsley wrote:
On Thu, 28 Jul 2016 14:37:41 -0400
David A. De Graaf wrote:
> Have I overlooked something obvious? Is there a way to make systemd
> perform the simple function 'shutdown' smoothly, reliably and quickly?
> If anyone knows how, I would love to hear it.
I can't make systemd itself work, but I've been using an
outside systemd solution for a while now: I setup
an alias for the "reboot" command that arranges to
kill off all the things systemd unreasonably waits
for, then does a real reboot.
Since systemd now has nothing to stop it, it reboots
rather fast (until something new shows up which I have
to track down and add to my list :-).
My current set of things to do before shutdown includes:
umount -l -t nfs -a
apachectl -k stop
kill all the "user deamon" process trees.
The user daemon stuff is handled by a program
described here:
http://tomhorsley.com/game/punch.html
Of course, since (according to the systemd biggest
myths page) systemd is easy to script and not at all
confusing, then it should be trivial for me to script
things so shutdown works this way automatically
without me having to remember to run my alias
in a terminal, but, alas, they don't appear to be
myths at all.
I've said it before, systemd is a huge, cumbersome, useless solution
looking for a problem. The decision to use this crud was one of the most
boneheaded things any distribution (not just RHEL/Fedora) ever chose
to adopt. Unfortunately, we're stuck with it now.
As far as the NFS unmount issue, this was an interesting dialog
that happened in September last year:
https://github.com/systemd/systemd/issues/1257
So, "but I doubt that lazy unmounting is really a comprehensive
solution for anything..." comes from one of the developers. Fine, then
when it times out, do a forced umount. This is pretty obvious. The
server isn't listening and we (the client) are shutting down. Pull the
g*ddamned plug!
I'm with David...but this is only one of a litany of stupid decisions
systemd makes. It seems that developers nowadays are never taught how
(or are simply too lazy) to test for error conditions.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- Political Correctness: The insane doctrine that postulates that it -
- is entirely possible to pick up a turd by the clean end. -
----------------------------------------------------------------------