OT?: Emailing from within filepro

[Bill Vermillion]

On Sat, May 29, 2004 at 02:00:07PM -0400, Fairlight thus spoke:
> When asked his whereabouts on Sat, May 29, 2004 at 12:00:59PM -0500,
> Tom Aldridge took the fifth, drank it, and then slurred:
> > Jay R. Ashworth wrote:
> > > On Thu, May 27, 2004 at 11:57:43AM -0400, Fairlight wrote:
> > >> In the relative spacial/temporal region of
> > >> Thu, May 27, 2004 at 11:04:42AM -0400, Kenneth Brody achieved the
> > >> spontaneous generation of the following:

> > >>>> Recursive queries supported by this server
> > >>>>  Query for aldridgeinc.com type=255 class=1
> > >>>>   aldridgeinc.com MX (Mail Exchanger) Priority: 1
> > >>>>   gate.aldridgeinc.com aldridgeinc.com SOA (Zone of Authority)
> > >>>>         Primary NS: gate.aldridgeinc.com
> > >>>>         Responsible person: sid at aldridgeinc@com
> > >>>>         serial:200381703
> > >>>>         refresh:20s
> > >>>>         retry:20s
> > >>>>         expire:20s
> > >>>>         minimum-ttl:20s
> > >> 
> > >> Dear God in heaven.  WHO on earth sets those times to 20s?!  I had to
> > >> reread that five times to make sure I actually saw that right, and I
> > >> -still- couldn't believe my eyes.  Unfortunately, I just looked up
> > >> the SOA myself and it's actually -true-.  Good gravy.
> > > 
> > > Well-trained adminstrators who are about to move a service from one
> > > server to another.

> > > :-P

> > Moving from one server to another was the reason. Forgot to
> > change it back.

> But 20s?! I used to update my zone pretty frequently, so I have
> a 1hr min ttl, and even that's considered short, but at least
> it's not as extreme as 20s. With that, you effectively defeat
> caching.

And I guess that's what started this whole thread, when I couldn't
find the server or IP.   One link I went through appeared to obey
the 20s times while another seemed to cache things longer than they
were supposed to. 

> I -still- would never put the primary on the private side of
> the demarc on DSL. Actually, I would never run a business
> purely on DSL, purely because of the regulations. It's simply a
> bad idea for anything mission critical. I know a lot of people
> that dropped T1's and fractionals for DSL due to the cost.
> You'll make up for that when your telco costs you a few grand
> more than you save during an outage they don't get around to
> fixing soon enough. It's a foolhardy move at best.

I have one site I maintain that moved from T1 to DSL.  But they do
save a lot of money.  Not all businesses will lose 'a few grand'
during and outage.  DSL comes directly from a Telco, not a 3rd
party, and the one time in 2 years there was an outage it was fixed
quickly.  So you can't really make a blanket statement like the
above as not all would be classified 'mission critical'.

The only strange thing about that site is the public IP space they
have on their internal network.  The started out as a .US domain
and they have their onn IP block allocated to them from ARIN as
they are serving public/education with their services.  

But you won't find any DSL provider who will route your private
blocks, and few major providers who will do so for only a /24
address.

Since all the machines were numbered, the only change was made was
that the static public IP they were given gets remapped at the
router to another public IP.  But since it is NATed to that block
no one in the outside world sees it.   But the several hundred
dollars/month saving does add up.  I've seen one outage in the 2.5
years I've been subcontracted to handle it.  So they have saved
well over $21,000.  

So ever decision needs to be evaluated on an individual basis.

> There are business "plans", but that's still only a contract
> with the carrier. The FCC won't intervene because there's no
> mandate for restoration of service built into the DSL tarriffs.
> With a T1, you at least have some leverage.

The problems with T1s is when you start getting service from CLECs
and they don't have the infrastructure needed to support the T1s
they sell.  On CLEC that the ISP I was tech on used our IP space
for their customers and I had 12 T1s routing their customers to the
world.  If they had a failure after the normal 9-5 and it involved
the CLEC and they called us, I'd have to call the one person they
had that handled that and wake him up and then he'd have to drive
in to fix it.  

OTOH working with telco based providers I've never had a problem
getting through [sometimes a hold but usually not long] and have
had people dispatched late a night. We had one outage that came
up late evening and I was on the phone with different repair
people as they were dispatched to different crossover points
until they found the bad switch. We wrapped that up about 6AM and
both techs [two transport providers were involved] had been on
the clock since about 9AM the previous day.   

But using the argument of having the FCC to provide leverage is
not going to get anything fixed any faster. You need a reliable
service with enough people to fix problems when they occur.

> I'd second Jay's suggestion for Sid. Heck, it doesn't even take
> a whole book to learn enough to keep yourself out of this much
> trouble:

>      man named
>      man named.conf

I'll posit that it takes more than man pages if someone doesn't
understand the structure and working of DNS.  The cricket book
is about the best for that.

Bill
-- 
Bill Vermillion - bv @ wjv . com