--- UNIX admin <[EMAIL PROTECTED]> wrote:

> > RIGHT...for a modified sendmail that does not
> change
> > at all save for security exploits...hmm...you
> really
> > have a closed mind. The only 'engineering' needed
> > here
> > would be how to quickly replace the OS with a more
> > efficient one and making sure that the OS gets its
> > security holes plugged quickly without breaking
> > anything.
> 
> "Quickly replace"? You mean, walk around with a CD
> or DVD, log onto the console, run install?

ROTFL. No. But that is what I would have to do with
OpenSolaris at the moment. Where is this flar?

> 
> Now it's my turn to write "yawn"...
> 
> And what would you do if your environment grew?
> Would you still be running around, installing
> systems? And if you had your customers requiring all
> kinds of different things from you, all at the same
> time?

I had kickstart, pxe, tftp, a repository hosted on ftp
and dhcp for this at that time with plans to move to
images. If I were to get twenty new boxes, I just
needed to add twenty new entries for dhcp and
kickstart. All twenty would have been done in 30 mins
and ready to run. Too bad I never got the chance to
get an image system in place.

> 
> What you describe is easy with one person, if you
> have up to 10 systems, 20 tops. Honestly, it's
> really not fun to go around and install systems
> manually, let alone configure them.

Yeah, especially when the data centre is in San Jose,
California and I am in Hong Kong. I would really have
to be Flash to get that done.

> 
> Boring.
> 
> > Those methodics are so exclusive to Solaris.
> 
> Incorrect. You can do the same thing with HP-UX and
> IRIX.
> 
> Actually, it would work for FreeBSD and Linux, too,
> up to a point. Some things could not be carried over
> simply because Linux lacks engineering and
> architecture to make them feasible.

Sure. Prove that you cannot use an image system with
Linux.

> 
> > Huh? What is going to beat automatic updates?
> 
> A flash of the system that lasts about all of 45
> seconds. Meanwhile, the other half of the cluster is
> still serving, without interruption to the service.
> It is possible to do; I've demoed it recently.

What makes you think that with an apt/dpkg or yum/rpm
system will result in any interruption of the service
when the systems in the cluster do their updates? I
have never suffered interruption of service due to
systems being updated.

> 
> And BTW the cluster is also configured
> automatically, no manual configuration is needed or
> involved.

Ooh, wow, that is so out of this world.

> 
> > yeah, nothing like flar exists outside of solaris.
> > Which planet do you live on? You would never
> imagine
> > images + automatic updates because you cannot do
> that
> > in Solaris. All you can do is maintain staging box
> > and
> > flar out when you are happy with staging box.
> 
> Ignite-UX? Roboinst? Come on, cut me some slack.
> I've been in your shoes, doing Linux boxes.

Right...what are HP Unix and Irix tools doing here?

> 
> And I don't see why I couldn't do automatic updates
> on Solaris. A cleverly written Makefile will take
> care of that in a jiffy. One doesn't even need a
> fancy tool like yum or apt-get.
> 
> Actually, I've got an even better idea: just flash
> the system with an incremental flar. Much faster.
> You'd most likely argue this would cause a
> disruption of service. But when you update running
> services on a system, you have to restart them
> anyway. So I don't see the difference. Of course,
> the disruption of service is a moot point if I have
> a cluster.

NO. I do not have to restart any service for my mail
boxes after a system update.


Send instant messages to your online friends http://uk.messenger.yahoo.com 
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to