[tim.peters@gmail.com: Re: [Zope-Checkins] CVS: Zope/lib/python/nt_svcutils - service.py:1.1.2.2]

Tim Peters tim.peters at gmail.com
Wed Apr 20 17:52:29 EDT 2005


[Mark Hammond]
>>> However, a service can stop also stop by its own accord.

[Tim Peters]
>> I don't think ZRS ever does that, and am not sure how I could write
>> code to do it.  Does a Zope service ever decide to "stop by its own
>> accord"?

[Mark Hammond]
> By that, I simply mean "terminate with a zero exit code".  Obviously not "of
> its own accord", but it has nothing to do with the service framework.  This
> also includes "terminate with non-zero exit code too many times" - you will
> not be called in this case either.

>>> The question, then, is under what cases should the onStop()
>>> calls be made?  It sounds to me like you want it called only when that
>>> external request comes in.

>> Maybe.  In the end, by whatever means a service gets shut down
>> (external, internal, whatever), I want to see the subclass onStop()
>> method called exactly once.

> My patch does not do that.  If Zope terminates "of its own accord", then
> your hook is never called.  This means that patch is suitable if you need
> onStop to actually trigger the stop process, but not to be notified of
> normal shutdown.
>
> Having that method called exactly once in all circumstances is a little more
> tricky, and would probably involve a maintaining a flag - it must be called
> in SvcStop to *trigger* the shutdown, or if shutdown is "normal" (ie, not
> triggered by us), then it must be called as the main run-loop terminates -
> but not both times.

>> Seems to work fine for the ZRS-related services too, although I didn't
>> get into the "restarting too frequently" branch.  Thank you.  If
>> nobody objects in the next 10 minutes, then, I'll check this in (and
>> port to Zope trunk (2.8) too).

> I did test the "too frequently" branch - but note that as above, that patch
> will *not* have your function be called when Zope just sys.exits.  If that
> is suitable, go for it :)

Bet you were wondering when you'd see new text <wink>.

Thing is, ZRS has nothing in particular to do with Zope.  It runs
independently of Zope, and may run on boxes that don't even have a
Zope installed.  It has no UI at all, it's just a couple of services
that run infinite loops.  The only way it stops on Windows is as a
result of someone clicking "stop" in the Windows Services GUI, or by
running "XYZservice.py stop" from a DOS box, or by getting a stop
request via, e.g., Windows shutdown. In those cases I want onStop()
called exactly once, and that's what's happening now, so I'm happy.

AFAICT, the ZRS services are the only ones we have that override
onStop(); if so, there's no other case to look at.  Good enough for
now, anyway!


More information about the Zope-Checkins mailing list