Mailing List Archive

On Feb 14, 10:49, Randy Bush wrote:
} Subject: Re: root server update
} > So now everyone will start sending test DNS requests to the servers
} > and overload them even further.
}
} What do you expect for folk who call people at home because they can't
} figure out how to call a NOC?

the appropriate NOC number isn't published anywhere i could find (and i
looked- it's what i do). what did you do to try to fix or notify someone
of this (rather major) problem?

welcome to the new "it's not my problem, and i'm not logged in anyway"
internet operators group.

-Qarin Van Brink
queue@scruz.net
- - - - - - - - - - - - - - - - -

hm. i count 10% of initial queries failing with a domain doesn't exist
error, and that information then being cached so subsequent queries
always fail, as being, as i said, a rather major problem. earthshaking,
no; major, yes. many servers, in this instance, helps, but because the
problem manifested itself as a root server not being *down*, but
returning actually false information, the redundancy of the root
servers meant that the odds were lower of failure, not that failure was
avoided. a lot of mail was bouncing ("host unknown", on valid hosts),
and a lot of queries were failing.

am i "soooo important"? no. and i can flush my nameserver's cache to
take care of a 1 in 10 problem looking up a site i'm trying to look up.
but my users, and yours, are so important.

and, happily, mr. fields, on his way to bed in maryland last night,
wasn't so condescending.

-Qarin Van Brink
queue@scruz.net


On Feb 14, 11:20, Randy Bush wrote:
} Subject: Re: root server update
} > what did you do to try to fix or notify someone of this (rather major)
} > problem?
}
} major? funny, but i thought that was why we had many servers.
}
} and, not realizing it was a major earthshaking event, i sent email. silly
} me.
}
} randy
}-- End of excerpt from Randy Bush


- - - - - - - - - - - - - - - - -

> Date: Thu, 13 Feb 1997 20:57:59 -0800 (PST)
> From: matthew@scruz.net (Matthew Kaufman)
> Subject: root server update
> To: nanog@merit.edu


> now, to add an infrequent check of the root nameservers to our homebrew NMS,
> so our NOC operators can see these things happen as they happen, not after
> the fact.

So now everyone will start sending test DNS requests to the servers
and overload them even further.
>
> -matthew kaufman
> matthew@scruz.net
>
>

Dave Nordlund d-nordlund@ukans.edu
University of Kansas 913/864-0450
Computing Services FAX 913/864-0485
Lawrence, KS 66045 KANREN
- - - - - - - - - - - - - - - - -

> So now everyone will start sending test DNS requests to the servers
> and overload them even further.

What do you expect for folk who call people at home because they can't
figure out how to call a NOC?

Welcome to the new "I am sooooo important that thinking is not required"
internet.

randy
- - - - - - - - - - - - - - - - -

> what did you do to try to fix or notify someone of this (rather major)
> problem?

major? funny, but i thought that was why we had many servers.

and, not realizing it was a major earthshaking event, i sent email. silly
me.

randy
- - - - - - - - - - - - - - - - -

The root name servers each take between 500 and 1000 queries per second.
So testing them from time to time isn't going to hurt much.

However, we don't need to test them from very many places, since we would
just overload the trouble reporting system, which tends to be biologically
limited. How about if each root name server just implements RFC 2010 and
does their own monitoring, and InterNIC does the fallback monitoring, and
we stop arguing about global issues on a north american mailing list?
- - - - - - - - - - - - - - - - -

In article <199702140457.UAA08984@scruz.net> you write:
>After verifying that the only problem server remaining was H, we
>managed to track down the keeper of H around 20:15 PST, he had nameservice
>down at 20:39 PST and back running with correct results now at 20:47 PST.
>[...]
>now, to add an infrequent check of the root nameservers to our homebrew NMS,
>so our NOC operators can see these things happen as they happen, not after
>the fact.

Should I even ask this? Where is the Internic's NMS for the
root-servers to sound an alarm and get someone to act?

And don't let them say that the resources aren't there: "Seventy
percent of the [domain] fees collected will be retained by Network
Solutions to cover operating costs..."

--
Bob Kupiec GES / WNA
http://www.ges.com/~kupiec Princeton, NJ
- - - - - - - - - - - - - - - - -

Randy,

I suspect it would help some of us small incompetent ISPs if you could
provide a pointer to where *you* would have successfully found the NOC
phone number. That way we could all deal with problems right away without
adding more traffic to this list, and we could add this to one of these
"how to be an ISP" FAQs, too.

-matthew kaufman
matthew@scruz.net

Original message <m0vvShi-0007zZC@rip.psg.com>
From: randy@psg.com (Randy Bush)
Date: Feb 14, 10:49
Subject: Re: root server update
>
> > So now everyone will start sending test DNS requests to the servers
> > and overload them even further.
>
> What do you expect for folk who call people at home because they can't
> figure out how to call a NOC?
>
> Welcome to the new "I am sooooo important that thinking is not required"
> internet.
>
> randy
>-- End of excerpt from Randy Bush


- - - - - - - - - - - - - - - - -

On Fri, 14 Feb 1997, Matthew Kaufman wrote:

> I suspect it would help some of us small incompetent ISPs if you could
> provide a pointer to where *you* would have successfully found the NOC
> phone number.

OK, everbody. Do this right NOW! Check the whois data on your NIC handle
and check the whois data on your domain name, especially the technical
contact. Als check the NIC handle for the tech contact. I'll bet most
of you discover contact info that is missing or inaccurate. In my case
whois MD130 shows the wrong area code because it has changed recently.

If you want to do a better job of keeping this info up to date, set up a
cron job to do whois queries on the relevant records and email them to you
once a month on some random day (not the 1st of the month please). If this
information were accurate and complete there would be a lot less flailing
around when people need NOC contact info.

Michael Dillon - Internet & ISP Consulting
Memra Software Inc. - Fax: +1-250-546-3049
http://www.memra.com - E-mail: michael@memra.com

- - - - - - - - - - - - - - - - -

The root name servers each take between 500 and 1000 queries per second.
So testing them from time to time isn't going to hurt much.
However, we don't need to test them from very many places, since we would
just overload the trouble reporting system, which tends to be biologically
limited. How about if each root name server just implements RFC 2010 and
does their own monitoring, and InterNIC does the fallback monitoring, and
we stop arguing about global issues on a north american mailing list?


- - - - - - - - - - - - - - - - -

The root name servers each take between 500 and 1000 queries per second.
So testing them from time to time isn't going to hurt much.
However, we don't need to test them from very many places, since we would
just overload the trouble reporting system, which tends to be biologically
limited. How about if each root name server just implements RFC 2010 and
does their own monitoring, and InterNIC does the fallback monitoring, and
we stop arguing about global issues on a north american mailing list?


- - - - - - - - - - - - - - - - -

The root name servers each take between 500 and 1000 queries per second.
So testing them from time to time isn't going to hurt much.
However, we don't need to test them from very many places, since we would
just overload the trouble reporting system, which tends to be biologically
limited. How about if each root name server just implements RFC 2010 and
does their own monitoring, and InterNIC does the fallback monitoring, and
we stop arguing about global issues on a north american mailing list?


- - - - - - - - - - - - - - - - -

The root name servers each take between 500 and 1000 queries per second.
So testing them from time to time isn't going to hurt much.
However, we don't need to test them from very many places, since we would
just overload the trouble reporting system, which tends to be biologically
limited. How about if each root name server just implements RFC 2010 and
does their own monitoring, and InterNIC does the fallback monitoring, and
we stop arguing about global issues on a north american mailing list?


- - - - - - - - - - - - - - - - -