Re: systemd fails to poweroff - "A stop job is running for Session 2 of user $USER"
On 20140813_1033+0100, Darac Marjal wrote:
> On Wed, Aug 13, 2014 at 11:15:22AM +1000, Zenaan Harkness wrote:
> > On 8/13/14, Paul E Condon <pecondon@mesanetworks.net> wrote:
> > > I interpret the quoted string in the Subject: header as being flawed
> > > use of English language. 'stop' should be 'stopped'. And, there is a
> >
> > That would definitely be clearer.
> >
> > I was interpreting it as some special systemd shutdown-ey thing which
> > runs around trying to stop things, and that there might be various of
> > these, and one of them has a problem.
>
> Yes, I believe this is the correct interpretation. SystemD will tell all
> services to "stop". It will then wait until all those sevices have
> stopped. Some services will stop immediately, but some need a little
> longer to flush logs, finish servicing a request or whatever. After a
> period of seconds, it appears that SystemD will pop up a message to the
> effect of "I'm still here, still responding. I'm just waiting for
> service X to tell me it has stopped." Given what was said earlier in the
> thread, I suspect this will continue for 90 seconds until it finally
> gives up waiting.
>
> >
> > I.e. "stop job" being a noun.
>
In English, both 'stop job' and 'stopped job' are an adjective
modifying a noun. The noun in both cases is 'job'. 'stop job' is a
noun phrase expressing a type of job, and must be some kind of geeky
usage. OTOH, the noun phrase 'stopped job' is a job that is not
progressing, or not running. But in this context, 'job' must itself
have a geeky, technical jargon meaning. I notice that you render
'systemd' as SystemD, the thought police of systemd object to the
capital D. Be warned. It marks you as not one of the cognoscenti.
There is a lot going on here in the use of language. Systemd people
seem to have developed there own systemd jargon which sounds like UNIX
jargon to the un-initiated. They may believe it is UNIX jargon,
but when closely questioned I think they will reveal a belief system
about UNIX that differs from the mainstream of geeky person's. To them,
it is just UNIX, only better.
> "Stop Command", perhaps? "Stop-Service command" would be even clearer
> (though that literal command doesn't exist). Perhaps a complete
> rewording to "Still waiting for Service $FOO in Session 2 of User $USER
> to stop" would be clearest.
Seems good to me, but we don't have a clear idea of what 'Session 2'
is. It might actually be a concept, which, if properly understood
would be better expressed with a totally different ordering of
the words.
My brother Joe spent his working life at Bell Labs in the same
building as the inventors of UNIX told me decades ago about the go
arounds in the lunch room on topics of word choice in documantation.
All of them had taken old fashioned high school English on there way
to earning their engineering or science degree. They cared about
being understood. He died 2yrs ago. It would have been nice to have
ask him about this situation. I think he actually helped the UNIX
guys by being a audience on whom to test their language.
I think that all commenters would agree that the message is somewhat
confusing. Something is holding up the process of shutting down some
process. There needs to be a standard glossary of terms to use to
express every particular that the message is attempting to communicate
to the user. And, there needs to be some suggestion of what the user
should do about the situation, or where in the operator's manual to
read further information. It really ought to be easier to understand
than the true meaning of the Book of Genesis.
I should stop. I really have very little firm knowledge of systemd,
just opinions that make sense to me. (tm)
Best regards,
--
Paul E Condon
pecondon@mesanetworks.net
Reply to: