[Info-vax] New OpenSSL update from HP

[lists at openmailbox.org]

On Mon, 15 Jun 2015 13:33:26 +0000 (UTC)
Simon Clubley via Info-vax <info-vax at rbnsn.com> wrote:

> On 2015-06-15, <lists at openmailbox.org> <lists at openmailbox.org> wrote:
> > On Mon, 15 Jun 2015 11:06:47 +0000 (UTC)
> > Simon Clubley via Info-vax <info-vax at rbnsn.com> wrote:
> >
> >> On 2015-06-15, David Froble <davef at tsoft-inc.com> wrote:
> >> >
> >> > Ok, I've never written a web server, yet.  If I was to do so, I'd 
> >> > seriously look at detached processes rather than sub-processes.
> >> > Why? Because once you assign the worker process a task, perhaps
> >> > you'd want it to complete, regardless of whatever the web server
> >> > does.
> >> >
> >> 
> >> What happens if you need to shutdown/restart the web server and need
> >> to _guarantee_ that all related processes have terminated as part of
> >> of the shutdown process ?
> >
> > From this question I understand there are no appropriate signaling and
> > synchronization mechanisms in POSIX?
> >
> 
> You may have missed the implications when David talked about making
> them detached processes; detached processes continue executing even
> when the process which created them disappears.

I saw the discussion but I didn't understand whether this was specifically
about a VMS-based webserver that David might write or whether it was a
generic webserver running on UNIX. I'm not familiar with either environment
which is why when I saw your question I wondered whether the signalling
and/or synchronization mechanisms exist in POSIX- since you often speak
about Linux and relate your experiences with various programming issues in
that environment but I would have the same question about VMS. I didn't
have the proper context and probably should have asked initially. But I was
suprised to see the question and it caught me off guard.

> In such a case you need a mechanism to identify all these processes
> and kill them and you can't rely on the primary web server process
> sending a shutdown message to the detached process; the detached
> process may be stuck and not responding to application initiated
> messages.

I understood that part but not being familiar with the environment I didn't
understand why doing this correctly would be difficult or unusual.

Why couldn't the primary server send a message to the detached processes?
Is there not a signaling mechanism and known best practices appropriate for
this sort of thing?

I'm not sure what a "detached process" is but in MVS we would use a special
control block and the POST facility to signal a subtask (or other,
non-subtask unit of work) to cancel and if it was necessary for integrity
we could a few different mechanisms to be sure the task ended. We would
create a list of those control blocks in some commonly-accessible piece of
storage so that we wouldn't need a "discovery" process to determine which
are the processes we want to kill.

This all seems pretty standard, so I didn't know whehter your question was
simply a design question as in "have you thought about this?" or whether
your question was a little deeper than that and meant "Hmmm, I don't see a
way to do that..."

-- 
Please DO NOT COPY ME on mailing list replies. I read the mailing list.
RSA 4096 fingerprint 7940 3F02 16D3 AFEE F2F8  ACAA 557C 4B36 98E4 4D49