From: Gary R. Schmidt on
neilsolent wrote:
> Automatic? Careful - that's sounding a lot like a script 8-O
>
More that I am aware of what the systems in my charge are doing, when
they do certain things, what they are able to do, and so on.

Of course, starting in this game before the IBM PC came out /does/ give
certain advantages.

Cheers,
Gary B-)
From: Rainer Weikusat on
neilsolent <n(a)solenttechnology.co.uk> writes:
> On 17 Jan, 16:19, Rainer Weikusat <rweiku...(a)mssgmbh.com> wrote:
>> neilsolent <n...(a)solenttechnology.co.uk> writes:
>> > On 16 Jan, 22:26, David Schwartz <dav...(a)webmaster.com> wrote:

[...]

>> >> Why are you rebooting the machine from a script anyway?
>>
>> > Why write a script to do anything?
>> > We could all just do a lot of typing - it's fun :-)
>>
>> It's a waste of time in this case: If reboots aren't infrequent enough
>> that doing them manually doesn't matter, that hints at another problem
>> which should be solved first.
>
> Depends how many systems are being rebooted. Also, scripts do more
> than save typing - help to prevent typos as well, and save you from
> having to remember a procedure.

'Invoking a particular script' is a procedure, too, and its name can
also be mistyped.

>> Especially, since 'rebooting'
>> necessarily means 'interruption of service'.
>
> No it doesn't. System may be clustered, for example.

Systems may be 'clustered' in order to deal with 'service
interruptions' happening on a single system.
From: neilsolent on

> 'Invoking a particular script' is a procedure, too, and its name can
> also be mistyped.

Of course. But it is generally many orders of magnitude easier to
mistype the contents of a script, than simply the name of script to
invoke it.
Go on admit it, you got that one wrong. It won't hurt.

>
> >> Especially, since 'rebooting'
> >> necessarily means 'interruption of service'.
>
> > No it doesn't. System may be clustered, for example.
>
> Systems may be 'clustered' in order to deal with 'service
> interruptions' happening on a single system.

Sorry, don't get your point there.
From: Scott Lurndal on
neilsolent <n(a)solenttechnology.co.uk> writes:
>On 17 Jan, 19:33, David Schwartz <dav...(a)webmaster.com> wrote:

>> Typically, rebooting a machine requires coordination among multiple
>> humans and multiple services. If you reboot some machines without
>> first disarming their remote monitoring, ...
>>
>> DS
>
>Yes, and broadly I agree. This is your assumption though - that I am
>attempting to reboot systems willy nilly.
>All I was asking was about how to reboot a UNIX system from a C
>program. There are plenty of commerical 3rd party apps out there that
>do this.

I challenge you to name _one_ commercial 3rd party application that
reboots a unix box.

scott
From: David Schwartz on
On Jan 18, 3:54 am, neilsolent <n...(a)solenttechnology.co.uk> wrote:

> > Typically, rebooting a machine requires coordination among multiple
> > humans and multiple services. If you reboot some machines without
> > first disarming their remote monitoring, ...

> Yes, and broadly I agree.

It doesn't seem like you do.

> This is your assumption though - that I am
> attempting to reboot systems willy nilly.

And you continue to validate that assumption. Nothing you've said has
given any indication that that assumption is wrong. And until I know
*why* that assumption is wrong, it's impossible for me to give a
useful answer to your question. That's why I keep pushing it.

> All I was asking was about how to reboot a UNIX system from a C
> program. There are plenty of commerical 3rd party apps out there that
> do this.

No, there aren't. And that is not, in general, a sensible thing to do.
The right way to do it depends on why, in your particular situation,
that's actually sensible.

> Let's say it's up to the user whether he wants to do it or not, I just
> provide the capability.

Then ask the user how to do it.

> In my opinion, there should be standards such that you can just call a
> simple command to do a reboot - and all the complexity of taking care
> of the applications etc is hidden from the user (typically implemented
> by shutdown scripts, or whatever). This is not because the user is
> stupid, but because it's not practical to remember the exact source
> code for every application running when he is aksed to do a reboot at
> 3am in the morning.

I agree with you, but now you're asking a completely different
question. Now your question is about what kind of procedures and
practices a site should follow to ensure that their administration can
do an appropriate reboot should one be needed at 3AM. The answer to
that question will depend on the site and their shutdown/restart
requirements.

The answer will be completely different if it's a guy's personal home
router than if it's a machine that's monitored 24/7 and provides a
business-critical services. It will be different if it's part of a
cluster.

But you are definitely right, sites should implement site-specific
protocols and practices to ensure that safe and appropriate shutdown/
reboot practices are followed, even at 3AM.

But what that process is will not be very much the same between
different sites.

DS