Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.networking > #678 > unrolled thread

Only some IPs accessable?

Back to article view | Back to comp.os.linux.networking

Contents

  Only some IPs accessable? "NoHtmlMailsPlease" <UsePlainText@dog.edu> - 2011-10-07 22:42 +0200
    Re: Only some IPs accessable? Frank Boehm <fraboehm@gmx.de> - 2011-10-08 07:39 +0000
    Re: Only some IPs accessable? Chris Davies <chris-usenet@roaima.co.uk> - 2011-10-08 14:24 +0100
      Re: Only some IPs accessable? Ivan Shmakov <ivan@gray.siamics.net> - 2011-10-08 23:18 +0700
        Re: Only some IPs accessable? Chris Davies <chris-usenet@roaima.co.uk> - 2011-10-08 21:29 +0100
      Re: Only some IPs accessable? "NoHtmlMailsPlease" <UsePlainText@dog.edu> - 2011-10-12 15:09 +0200
        Re: Only some IPs accessable? Ralph Spitzner <rasp@spitzner.org> - 2011-10-12 17:24 +0200
          Re: Only some IPs accessable? Chris Davies <chris-usenet@roaima.co.uk> - 2011-10-13 10:55 +0100
        Re: Only some IPs accessable? Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> - 2011-10-12 22:22 +0200
        Re: Only some IPs accessable? Chris Davies <chris-usenet@roaima.co.uk> - 2011-10-13 11:01 +0100
          Re: Only some IPs accessable? "NoHtmlMailsPlease" <UsePlainText@dog.edu> - 2011-10-14 23:00 +0200
            Re: Only some IPs accessable? Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> - 2011-10-15 11:46 +0200
              Re: Only some IPs accessable? floyd@apaflo.com (Floyd L. Davidson) - 2011-10-15 04:59 -0800
                Re: Only some IPs accessable? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2011-10-15 16:39 +0000
                  Re: Only some IPs accessable? floyd@apaflo.com (Floyd L. Davidson) - 2011-10-15 08:50 -0800
                    Re: Only some IPs accessable? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2011-10-15 16:58 +0000
                      Re: Only some IPs accessable? Richard Kettlewell <rjk@greenend.org.uk> - 2011-10-15 19:07 +0100
                Re: Only some IPs accessable? Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> - 2011-10-15 19:50 +0200
                  Re: Only some IPs accessable? floyd@apaflo.com (Floyd L. Davidson) - 2011-10-15 12:30 -0800
                    Re: Only some IPs accessable? Ralph Spitzner <rasp@spitzner.org> - 2011-10-19 18:14 +0200

#678 — Only some IPs accessable?

Networking is quiet complex?

After I got my 10 years old pppd-script modified sufficiently
to connect the fixed-wireless-modem, as confirmed by 
 /var/log/messages     reporting: 
CHAP authenticated,
<a local IP> allocated,
2 DNSs allocated;
I still wasn't able to: lynx <URL>,
except by: 
chroot <an Old Installation> lynx <URL>.

Ie. I used Slackware13 to ppp,
and then switched to the old/proven installation
  to fetch http via lynx.
--------
These are my test results under Slackware13:-

First I suspected the DNSs allocated by the ISP:
>Oct  5 18:20:44 labeas pppd[4599]: secondary DNS address 41.164.0.11
   but both pinged OK:
> nameserver 41.164.0.11
> 7 packets transmitted, 7 received, 0% packet loss, time 6002ms
---------
Then I used the DNS on a URL:
->  nslookup     krugman.blogs.nytimes.com  41.164.0.11   ==
> Server:41.164.0.11
> Address:41.164.0.11#53
> 
> Non-authoritative answer:
> krugman.blogs.nytimes.comcanonical name = blog.nytimes.com.
> Name:blog.nytimes.com
> Address: 199.239.137.32
------------
So, then I used *that* IP
-> lynx 199.239.137.32  ==

> Looking up 199.239.137.32 first
> Looking up 199.239.137.32
> Making HTTP connection to 199.239.137.32
> Sending HTTP request.
> HTTP request sent; waiting for response.
> Alert!: Unexpected network read error; connection aborted.
> Can't Access `http://199.239.137.32/'
> Alert!: Unable to access document.
> 
> lynx: Can't access startfile 
------

Although the <old proven system> which fetches http,
if I `chroot` to it, has different DNSs in /etc/*;
which I think I set manually, years ago,
and that is setup for a different ISP,
than this new wireless one,
it seems that DNS is *NOT* the problem.

Let's try, via 
`chroot <oldSystem> lynx 199.239.137.32` 
== Alert!: Unexpected network read error; connection aborted.

See if, in same session, can do:
`chroot <oldSystem> lynx http://krugman.blogs.nytimes.com`
== OK.
------
-> chroot /mnt/p6 lynx http://199.239.137.32

Looking up 199.239.137.32
199.239.137.32
Making HTTP connection to 199.239.137.32
Sending HTTP request.
HTTP request sent; waiting for response.
Alert!: Unexpected network read error; connection aborted.
Can't Access `http://199.239.137.32/'
Alert!: Unable to access document.

lynx: Can't access startfile 
--------
This one is very confusing:-
-> lynx 184.106.20.99 ==
      Posterous - 404 - The page you requested could not be found (p1 of 2)
   [posterous-small.png]
     *
     * Login
     * Get your own Space >

Sorry, we couldnt find what you were asking for.
---------------
So 184.106.20.99 arrived at the correct location,
 but not other `lynx <IPs>`

Can someone please advise.

BTW, I've never confirmed this Slackware13 installation
as being setup for inet, since I was not able to previously
connect linux to the wireless-modem, and used FC1 at another
location on a dial-up.

== Chris Glur.
 

[toc] | [next] | [standalone]

#679

NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:
> Networking is quiet complex?
> 
> > Alert!: Unexpected network read error; connection aborted.
> == Alert!: Unexpected network read error; connection aborted.
> Alert!: Unexpected network read error; connection aborted.

> Can someone please advise.
> 
> BTW, I've never confirmed this Slackware13 installation
> as being setup for inet, since I was not able to previously
> connect linux to the wireless-modem, and used FC1 at another
> location on a dial-up.

you don't have software related problems, this is hardware related, 
check your cables first

check if you have got an IP with: ifconfig

check your routing: route
there is a default route?
Destination 0.0.0.0

reachable nameserver in /etc/resolv.conf? 
(seems to be ok for both of your slackware installations)

root@sif: /etc # ping -c10 www.google.com
PING www.l.google.com (74.125.39.105) 56(84) bytes of data.
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=1 ttl=55 time=47.0 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=2 ttl=55 time=46.2 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=3 ttl=55 time=47.1 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=4 ttl=55 time=46.6 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=5 ttl=55 time=47.0 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=6 ttl=55 time=46.7 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=7 ttl=55 time=46.8 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=8 ttl=55 time=47.0 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=9 ttl=55 time=48.1 ms
64 bytes from fx-in-f105.1e100.net (74.125.39.105): icmp_seq=10 ttl=55 time=46.8 ms

--- www.l.google.com ping statistics ---
10 packets transmitted, 10 received, 0% packet loss, time 9012ms
rtt min/avg/max/mdev = 46.213/46.965/48.147/0.520 ms

e.g. 100 transmitted pings, how much are lost?


root@sif: /etc # traceroute www.google.com
traceroute: Warning: www.google.com has multiple addresses; using
74.125.39.103
traceroute to www.l.google.com (74.125.39.103), 30 hops max, 38 byte
packets
 1  speedport.baldar.de (192.168.0.234)  0.765 ms  0.371 ms  0.494 ms
 2  217.0.116.148 (217.0.116.148)  105.937 ms  40.831 ms  40.442 ms
 3  217.0.74.58 (217.0.74.58)  43.468 ms  42.234 ms  43.199 ms
 4  217.239.39.22 (217.239.39.22)  44.231 ms  43.581 ms  88.910 ms
 5  80.150.170.42 (80.150.170.42)  44.945 ms 80.150.170.38 (80.150.170.38)  44.387 ms 80.150.170.34 (80.150.170.34)  45.216 ms
 6  72.14.238.46 (72.14.238.46)  44.178 ms 72.14.238.44 (72.14.238.44)
 43.602 ms 72.14.238.46 (72.14.238.46)  50.175 ms
 7  72.14.236.68 (72.14.236.68)  45.675 ms  45.871 ms  44.654 ms
 8  209.85.254.118 (209.85.254.118)  44.705 ms  44.669 ms  45.440 ms
 9  * 209.85.249.162 (209.85.249.162)  51.393 ms *
10  fx-in-f103.1e100.net (74.125.39.103)  45.965 ms  44.872 ms  45.979 ms

I expect you to find lost packages or unusual delays, 
sometimes hardware is realy broken

network in your lan is ok? you can ping an access point, 
or another computer without problems?


cu Frank

-- 
"now for the happy time between the moment the lie is told 
and the moment it is found out" -- Homer Simpson

[toc] | [prev] | [next] | [standalone]

#680

NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:
> After I got my 10 years old pppd-script modified sufficiently
> to connect the fixed-wireless-modem, as confirmed by 
> /var/log/messages     reporting: 
> CHAP authenticated,
> <a local IP> allocated,
> 2 DNSs allocated;
> I still wasn't able to: lynx <URL>,

> First I suspected the DNSs allocated by the ISP:
>Oct  5 18:20:44 labeas pppd[4599]: secondary DNS address 41.164.0.11
>   but both pinged OK:

Ping shows connectivity, so that's good (but bear in mind that lack of
ping does not prove lack of connectivity).

> Then I used the DNS on a URL:
> ->  nslookup     krugman.blogs.nytimes.com  41.164.0.11   ==

Being slightly picky: krugman.blogs.nytimes.com isn't a URL, it's a
domain. (The URL form would be http://krugman.blogs.nytimes.com/)


> -> lynx 199.239.137.32  ==

You can't guarantee that replacing a domain with an IP address like that
will work these days, due to virtual hosting configurations (multiple
websites sharing the same IP address). In fact, when I try that exact
command here it also fails.

If you really want to bypass DNS lookups (although you've proven they're
working) you're going to need to pop the IP address in your hosts file.


> -> lynx 184.106.20.99 ==

Where did you get this 184... address from?


> So 184.106.20.99 arrived at the correct location,
> but not other `lynx <IPs>`

This is due to the use of virtual hosting. Basically, you can't do that
that and expect it to work. You *must* be prepared to use domain names
in the domain name component of your URL, e.g. http://www.roaima.co.uk/
instead of http://188.222.203.136/


Personally, I think I'd go back to basics and ignore lynx initially -
it's confusing you and therefore muddying the water.

1. Can you ping your assigned DNS servers? (Yes)
2. Can you resolve hosts from at least one of the assigned DNS servers? (Yes)

3. Has your /etc/resolv.conf been updated with these DNS servers?

I suspect the answer to #3 is "no", in which case you just need to update
your ppp configuration to tell it to do so. However, in case it's yes
I've provided the following additional suggestions -

4. Can you ping 188.222.203.136?

5. Can you ping test.www.roaima.co.uk?

6. If you "lynx http://test.www.roaima.co.uk/", do you get my test page?

If yes then I'd say that the networking layer is fine. Otherwise,
exclude DNS and try again -

7. Add this line to your /etc/hosts file:

188.222.203.136 test.www.roaima.co.uk

Then, when you "lynx http://test.www.roaima.co.uk/" do you get my test
page this time?

If the answer to this is no, then it's time to start digging really
seriously into files such as /etc/nsswitch.conf (what does the "files"
line read as?) and possibly checking your networking libraries.

Chris

[toc] | [prev] | [next] | [standalone]

#681

>>>>> Chris Davies <chris-usenet@roaima.co.uk> writes:
>>>>> NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:

[…]

 >> So 184.106.20.99 arrived at the correct location, but not other
 >> `lynx <IPs>`

 > This is due to the use of virtual hosting. Basically, you can't do
 > that that and expect it to work. You *must* be prepared to use domain
 > names in the domain name component of your URL,
 > e.g. http://www.roaima.co.uk/ instead of http://188.222.203.136/

	When using GNU Wget, there's the --header= option:

--cut: (wget) HTTP Options --
`--header=HEADER-LINE'
     Send HEADER-LINE along with the rest of the headers in each HTTP
     request.  The supplied header is sent as-is, which means it must
     contain name and value separated by colon, and must not contain
     newlines.

[…]

     As of Wget 1.10, this option can be used to override headers
     otherwise generated automatically.  This example instructs Wget to
     connect to localhost, but to specify `foo.bar' in the `Host'
     header:

          wget --header="Host: foo.bar" http://localhost/
--cut: (wget) HTTP Options --

	It's my opinion that it'd be an improvement for Lynx to support
	a similar feature.

[…]

 > If yes then I'd say that the networking layer is fine. Otherwise,
 > exclude DNS and try again -

 > 7. Add this line to your /etc/hosts file:

 > 188.222.203.136 test.www.roaima.co.uk

 > Then, when you "lynx http://test.www.roaima.co.uk/" do you get my
 > test page this time?

 > If the answer to this is no, then it's time to start digging really
 > seriously into files such as /etc/nsswitch.conf (what does the
 > "files" line read as?)

	I believe that it should read “the ‘hosts’ line” instead.  Like:

$ grep -E -- ^hosts /etc/nsswitch.conf 
hosts:          files dns
$ 

 > and possibly checking your networking libraries.

-- 
FSF associate member #7257

[toc] | [prev] | [next] | [standalone]

#682

Ivan Shmakov <ivan@gray.siamics.net> wrote:
>        When using GNU Wget, there's the --header= option:

Yes. I'd tried that myself while composing my suggestions and couldn't
get it to work on the nytimes.com site. I'm pretty sure this /should/
have worked, but didn't:

wget -S -O- --no-proxy --header='Host: krugman.blogs.nytimes.com' http://199.239.137.32


> > If the answer to this is no, then it's time to start digging really
> > seriously into files such as /etc/nsswitch.conf (what does the
> > "files" line read as?)

>        I believe that it should read “the ‘hosts’ line” instead.  Like:

> $ grep -E -- ^hosts /etc/nsswitch.conf 
> hosts:          files dns

Yes, you're correct. Thank you for picking up that silly mistake of
mine. Hopefully the OP will see this in time.

Chris

[toc] | [prev] | [next] | [standalone]

#693

"Chris Davies" <chris-usenet@roaima.co.uk> wrote in message 
news:ogv5m8xn5a.ln2@news.roaima.co.uk...
> NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:
>> After I got my 10 years old pppd-script modified sufficiently
>> to connect the fixed-wireless-modem, as confirmed by
>> /var/log/messages     reporting:
>> CHAP authenticated,
>> <a local IP> allocated,
>> 2 DNSs allocated;
>> I still wasn't able to: lynx <URL>,
>
>> First I suspected the DNSs allocated by the ISP:
>>Oct  5 18:20:44 labeas pppd[4599]: secondary DNS address 41.164.0.11
>>   but both pinged OK:
>
[snip]

> Personally, I think I'd go back to basics and ignore lynx initially -
> it's confusing you and therefore muddying the water.
>
> 1. Can you ping your assigned DNS servers? (Yes)
> 2. Can you resolve hosts from at least one of the assigned DNS servers? 
> (Yes)
>
> 3. Has your /etc/resolv.conf been updated with these DNS servers?
>
> I suspect the answer to #3 is "no", in which case you just need to update
> your ppp configuration to tell it to do so. However, in case it's yes
> I've provided the following additional suggestions -
>
> 4. Can you ping 188.222.203.136?
>
> 5. Can you ping test.www.roaima.co.uk?
>
> 6. If you "lynx http://test.www.roaima.co.uk/", do you get my test page?
>
> If yes then I'd say that the networking layer is fine. Otherwise,
> exclude DNS and try again -
>
> 7. Add this line to your /etc/hosts file:
>
> 188.222.203.136 test.www.roaima.co.uk
>
> Then, when you "lynx http://test.www.roaima.co.uk/" do you get my test
> page this time?
>
> If the answer to this is no, then it's time to start digging really
> seriously into files such as /etc/nsswitch.conf (what does the "files"
> line read as?) and possibly checking your networking libraries.
==================== 2nd test log ===============
-> wget -V  == GNU Wget 1.11.4

3,1-> ping -c9  188.222.203.136  ==
--- 188.222.203.136 ping statistics ---
9 packets transmitted, 9 received, 0% packet loss, time 8012ms
rtt min/avg/max/mdev = 502.310/515.514/531.036/9.126 ms

3,2-> ping -c9 test.www.roaima.co.uk ==
ping: unknown host test.www.roaima.co.uk

4,1-> traceroute www.google.com  ==
traceroute: unknown host www.google.com

4,2->  lynx http://test.www.roaima.co.uk/ == ...
Unable to locate remote host test.www.roaima.co.uk.

----------> add: "188.222.203.136 test.www.roaima.co.uk" to /etc/hosts
3,1-> lynx http://test.www.roaima.co.uk/ ==
                             Test page - test.www.roaima.co.uk

   This is a test page to prove vHost connectivity to test.www.roaima.co.uk. 
If you had
   used an IP address instead of the name you would not have arrived here.

===> DN Service is failing ??
-> cat /etc/ppp/resolv.conf  ==
nameserver 41.160.0.36
nameserver 41.164.0.11
====> these are dynamically provided by this new/ISP2

3,1-> ping -c9 41.160.0.36 == 0% packet loss,
3,2-> ping -c9 41.164.0.11 == 0% packet loss,

==> see DNS:IP for old/proven/working installation
-> cat /mnt/p6/etc/ppp/resolv.conf ==
nameserver 168.210.2.2
nameserver 196.14.239.2

4,1-> ping -c9 196.14.239.2 == 0% packet loss,

==> Does new/unproven/IPS2's DNS KNOW: krugman.blogs.nytimes.com/
4,2 -> nslookup krugman.blogs.nytimes.com/ 41.164.0.11 ==
Server:		41.164.0.11
Address:	41.164.0.11#53
** server can't find krugman.blogs.nytimes.com/: NXDOMAIN

==> Does old/proven/IPS1's DNS KNOW: krugman.blogs.nytimes.com/
4,3 -> nslookup krugman.blogs.nytimes.com/  196.14.239.2 ==
Server:		196.14.239.2
Address:	196.14.239.2#53
** server can't find krugman.blogs.nytimes.com/: NXDOMAIN

==> BTW the way I *DO* get <krugman> on this new/suspected-unconfigured 
Slak13 is:
-> chroot /mnt/p6 lynx krugman.blogs.nytimes.com
==> where /mnt/p6 is the old/proven <Mandrake Installation>.

== TIA

 

[toc] | [prev] | [next] | [standalone]

#694

NoHtmlMailsPlease wrote:
>
[loads of DNS related BS goes here]

What's that to do with Slackware ?

	-rasp




-- 
I'm in the news chair and I don't know anything.
     -Michael Caine @ Chris Moyles 29.9.2010

[toc] | [prev] | [next] | [standalone]

#697

Ralph Spitzner <rasp@spitzner.org> wrote:
> What's that to do with Slackware ?

Since the OP's running slackware, maybe they thought that the problem
they're having *might* be a distro-related issue? Just a thought.

Chris

[toc] | [prev] | [next] | [standalone]

#695

"NoHtmlMailsPlease" <UsePlainText@dog.edu> wrote:
> ==> see DNS:IP for old/proven/working installation
> -> cat /mnt/p6/etc/ppp/resolv.conf ==
> nameserver 168.210.2.2
> nameserver 196.14.239.2

You have two DNS servers there which used to work fine.
 
> ==> Does old/proven/IPS1's DNS KNOW: krugman.blogs.nytimes.com/
> 4,3 -> nslookup krugman.blogs.nytimes.com/  196.14.239.2 ==
> Server:		196.14.239.2
> Address:	196.14.239.2#53
> ** server can't find krugman.blogs.nytimes.com/: NXDOMAIN

Ok, so 196.14.239.2 seems to be broken, but what about 168.210.2.2?
 
> ==> BTW the way I *DO* get <krugman> on this new/suspected-unconfigured 
> Slak13 is:
> -> chroot /mnt/p6 lynx krugman.blogs.nytimes.com

So at least one of the DNS servers in your old resolv.conf seems to be
working.

regards Henrik
-- 
The address in the header is only to prevent spam. My real address is:
hc123(at)poolhem.se Examples of addresses which go to spammers:
root@localhost postmaster@localhost

[toc] | [prev] | [next] | [standalone]

#698

>> 3. Has your /etc/resolv.conf been updated with these DNS servers?

NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:
> -> cat /etc/ppp/resolv.conf  ==
> nameserver 41.160.0.36
> nameserver 41.164.0.11
> ====> these are dynamically provided by this new/ISP2

But what about /etc/resolv.conf (not /etc/ppp/resolv.conf)? It's *this*
file that gets used by the system for name resolution.

The /etc/ppp/resolv.conf file contains the values derived via ppp, which
may or may not be relevant. So it's optional whether they get installed
into the true resolv.conf. In my distro (Debian, not Slackware TBH),
the ppp "up" script installs /etc/ppp/resolv.conf into /etc/resolv.conf.

It may be that this scripting is broken or doesn't exist, or perhaps
you could get away with a simple "ln -fs /etc/ppp/resolv.conf
/etc/resolv.conf".

Chris

[toc] | [prev] | [next] | [standalone]

#699

"Chris Davies" <chris-usenet@roaima.co.uk> wrote in message
news:2hpim8x384.ln2@news.roaima.co.uk...
>>> 3. Has your /etc/resolv.conf been updated with these DNS servers?
>
> NoHtmlMailsPlease <UsePlainText@dog.edu> wrote:
>> -> cat /etc/ppp/resolv.conf  ==
>> nameserver 41.160.0.36
>> nameserver 41.164.0.11
>> ====> these are dynamically provided by this new/ISP2
>
> But what about /etc/resolv.conf (not /etc/ppp/resolv.conf)? It's *this*
> file that gets used by the system for name resolution.
>
> The /etc/ppp/resolv.conf file contains the values derived via ppp, which
> may or may not be relevant. So it's optional whether they get installed
> into the true resolv.conf. In my distro (Debian, not Slackware TBH),
> the ppp "up" script installs /etc/ppp/resolv.conf into /etc/resolv.conf.

Here:Slak13 `ls /etc/res* -l`
shows 3 OLD files. /etc/resolv.conf == 2 lines copied manually now
search neotel.co.za
nameserver 207.132.16.5

and /etc/ppp/resolv.conf  is new/NOW with the DNSs of this problematic ISP.

>
> It may be that this scripting is broken or doesn't exist, or perhaps
> you could get away with a simple "ln -fs /etc/ppp/resolv.conf
> /etc/resolv.conf".
>
WOW!! Linking /etc/resolv.conf to /etc/ppp/resolv.conf
 does it !!

Thanks man.
I've just fetched my email, which needed Slak's X.
I had been tearing my guts out trying to get it via
`chroot <the old proven installations>` via telnet.

I suspect that I skipped the <config inet/ppp step> during installing
Slak, because it seemed unlikely that ANY linux would connect to the
fixed-wireless-telephone's wireless, which is only specified to work with
Win.   Since it uses the zeroCD method, whereby the PC's USB initially see
a simulated CD, which it downloads and runs to drive the modem;
the modem and driver could be a completely secret standard.
Ie. since the modem tells the Win-PC how to dance, it could be an
unknown protocol.

The wireless modem only connects to linux after it's been 'primed' by Win7.
The normal zeroCD protocol, is that the PC sends some byte/s to the
device, to make it switch over from a CD to a modem, once the PC has
d/l-ed the driver-software from the simulated CD.

Having climbed this big mountain with your help, I now need a rest,
and then I'd like to understand how/why/where linux puts the DNS-IPs
given by the ISP, and how they are then used.

Thanks,

== Chris Glur.

> Chris

PS. we note that I got no help from Slak-boys.
 

[toc] | [prev] | [next] | [standalone]

#700

"NoHtmlMailsPlease" <UsePlainText@dog.edu> wrote:
> and then I'd like to understand how/why/where linux puts the DNS-IPs
> given by the ISP, and how they are then used.

The answer is really simple. The DNS servers used are those listed in
/etc/resolv.conf

How they are used might be a little bit more complicated. Having more than
one DNS server listed in /etc/resolv.conf might be good for redundancy,
but it might also slow things down. For some reason it seems as if the
system waits for all servers to answer even if the first server gives a
correct answer.

regards Henrik
-- 
The address in the header is only to prevent spam. My real address is:
hc123(at)poolhem.se Examples of addresses which go to spammers:
root@localhost postmaster@localhost

[toc] | [prev] | [next] | [standalone]

#701

Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
>"NoHtmlMailsPlease" <UsePlainText@dog.edu> wrote:
>> and then I'd like to understand how/why/where linux puts the DNS-IPs
>> given by the ISP, and how they are then used.
>
>The answer is really simple. The DNS servers used are those listed in
>/etc/resolv.conf
>
>How they are used might be a little bit more complicated. Having more than
>one DNS server listed in /etc/resolv.conf might be good for redundancy,
>but it might also slow things down. For some reason it seems as if the
>system waits for all servers to answer even if the first server gives a
>correct answer.

That should not happen.  This quote is from the
RESOLV.CONF(5) man page:

  "On a normally configured system this file should not
  be necessary.  The only name server to be queried will
  be on the local machine; the domain name is determined
  from the hostname and the domain search path is
  constructed from the domain name.

  The different configuration options are:

  nameserver Name server IP address
     Internet address (in dot notation) of a name server
     that the resolver should query.  Up to MAXNS
     (currently 3, see <resolv.h>) name servers may be
     listed, one per keyword.  If there are multiple
     servers, the resolver library queries them in the
     order listed.  If no nameserver entries are
     present, the default is to use the name server on
     the local machine.  (The algorithm used is to try a
     name server, and if the query times out, try the
     next, until out of name servers, then repeat trying
     all the name servers until a maximum number of
     retries are made.)"

The gist is that there is no point in listing more than
3 name servers.  But in any event the first one is the
only one that will ever be queried if the first query is
successful, which also assumes that either there is no
local DNS server or that it has failed.

The reason for more than one server to be listed is so
that a single server need not have 100% availability for
DNS to be 100% available.

-- 
Floyd L. Davidson                <http://www.apaflo.com/>
Ukpeagvik (Barrow, Alaska)              floyd@apaflo.com

[toc] | [prev] | [next] | [standalone]

#702

["Followup-To:" header set to comp.os.linux.networking.]
On Sat, 2011-10-15, Floyd L. Davidson wrote:
> Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
>>"NoHtmlMailsPlease" <UsePlainText@dog.edu> wrote:
>>> and then I'd like to understand how/why/where linux puts the DNS-IPs
>>> given by the ISP, and how they are then used.
>>
>>The answer is really simple. The DNS servers used are those listed in
>>/etc/resolv.conf
>>
>>How they are used might be a little bit more complicated. Having more than
>>one DNS server listed in /etc/resolv.conf might be good for redundancy,
>>but it might also slow things down. For some reason it seems as if the
>>system waits for all servers to answer even if the first server gives a
>>correct answer.
>
> That should not happen.  This quote is from the
> RESOLV.CONF(5) man page:
...
> The gist is that there is no point in listing more than
> 3 name servers.  But in any event the first one is the
> only one that will ever be queried if the first query is
> successful,

Where "successful" includes "you got a response, and it said there is
no such name in DNS".

> which also assumes that either there is no
> local DNS server or that it has failed.

What do you mean here?  As far as I can tell, resolv.conf is used even
if there is a local name server running.

/Jorgen

-- 
  // Jorgen Grahn <grahn@  Oo  o.   .     .
\X/     snipabacken.se>   O  o   .

[toc] | [prev] | [next] | [standalone]

#703

Jorgen Grahn <grahn+nntp@snipabacken.se> wrote:
>On Sat, 2011-10-15, Floyd L. Davidson wrote:
>>
>> That should not happen.  This quote is from the
>> RESOLV.CONF(5) man page:
>...
>> The gist is that there is no point in listing more than
>> 3 name servers.  But in any event the first one is the
>> only one that will ever be queried if the first query is
>> successful,
>
>Where "successful" includes "you got a response, and it said there is
>no such name in DNS".
>
>> which also assumes that either there is no
>> local DNS server or that it has failed.
>
>What do you mean here?  As far as I can tell, resolv.conf is used even
>if there is a local name server running.

From the text of the  man page as quoted in my previous article:

  "On a normally configured system this file should not
  be necessary.  The only name server to be queried will
  be on the local machine; [...]"

-- 
Floyd L. Davidson                <http://www.apaflo.com/>
Ukpeagvik (Barrow, Alaska)              floyd@apaflo.com

[toc] | [prev] | [next] | [standalone]

#704

["Followup-To:" header set to comp.os.linux.networking.]
On Sat, 2011-10-15, Floyd L. Davidson wrote:
> Jorgen Grahn <grahn+nntp@snipabacken.se> wrote:
>>On Sat, 2011-10-15, Floyd L. Davidson wrote:
>>>
>>> That should not happen.  This quote is from the
>>> RESOLV.CONF(5) man page:
>>...
>>> The gist is that there is no point in listing more than
>>> 3 name servers.  But in any event the first one is the
>>> only one that will ever be queried if the first query is
>>> successful,
>>
>>Where "successful" includes "you got a response, and it said there is
>>no such name in DNS".
>>
>>> which also assumes that either there is no
>>> local DNS server or that it has failed.
>>
>>What do you mean here?  As far as I can tell, resolv.conf is used even
>>if there is a local name server running.
>
> From the text of the  man page as quoted in my previous article:
>
>   "On a normally configured system this file should not
>   be necessary.  The only name server to be queried will
>   be on the local machine; [...]"

Yes, but that only means localhost is queried if the file doesn't exist,
not that the file is ignored if there's a nameserver on localhost.
Do you agree?

(I didn't know about that sane default, so I have a local DNS and a
resolv.conf which points to localhost.)

/Jorgen

-- 
  // Jorgen Grahn <grahn@  Oo  o.   .     .
\X/     snipabacken.se>   O  o   .

[toc] | [prev] | [next] | [standalone]

#706

Jorgen Grahn <grahn+nntp@snipabacken.se> writes:
> On Sat, 2011-10-15, Floyd L. Davidson wrote:

>> From the text of the  man page as quoted in my previous article:
>>
>>   "On a normally configured system this file should not
>>   be necessary.  The only name server to be queried will
>>   be on the local machine; [...]"
>
> Yes, but that only means localhost is queried if the file doesn't exist,
> not that the file is ignored if there's a nameserver on localhost.
> Do you agree?

That's correct.  After reading /etc/resolv.conf if no name servers are
found to be configured, 127.0.0.1 is added to the list.

I don't know if that's "normally" configured - I suspect that
numerically at least, nonempty resolv.conf is the more frequent
configuration.

-- 
http://www.greenend.org.uk/rjk/

[toc] | [prev] | [next] | [standalone]

#705

floyd@apaflo.com (Floyd L. Davidson) wrote:

> Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
>> For some reason it seems as if the system waits for all servers to
>> answer even if the first server gives a correct answer.

> But in any event the first one is the only one that will ever be queried
> if the first query is successful,

Thats what one might expect, but I have seen examples of when queries go
to all servers listed even though the first one gives a correct answer and
I have seen this slow down DNS queries when some DNS server is down even
if the first server is up.

It did come as a surpries to me when running strace to see that all
servers were queried even though the first server gave the right answer.

Now, when trying to recreate that I only got partial success. I have no
DNS server running on localhost:

-8<---------------------
bash-3.1# cat /etc/resolv.conf
# Generated by dhcpcd for interface wlan0
search Belkin
nameserver 192.168.30.254

bash-3.1# dig www.google.com @192.168.30.254

; <<>> DiG 9.4-ESV-R4-P1 <<>> www.google.com @192.168.30.254
;; global options:  printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 16474
;; flags: qr aa rd; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 0
;; WARNING: recursion requested but not available

;; QUESTION SECTION:
;www.google.com.                        IN      A

;; ANSWER SECTION:
www.google.com.         86400   IN      A       209.85.169.105

;; Query time: 98 msec
;; SERVER: 192.168.30.254#53(192.168.30.254)
;; WHEN: Sat Oct 15 19:39:55 2011
;; MSG SIZE  rcvd: 48

bash-3.1# dig www.google.com @127.0.0.1     

; <<>> DiG 9.4-ESV-R4-P1 <<>> www.google.com @127.0.0.1
;; global options:  printcmd
;; connection timed out; no servers could be reached

bash-3.1# time nslookup www.google.com
Server:         192.168.30.254
Address:        192.168.30.254#53

Name:   www.google.com
Address: 209.85.169.105


real    0m0.104s
user    0m0.004s
sys     0m0.004s
bash-3.1# emacs -nw /etc/resolv.conf
bash-3.1# cat /etc/resolv.conf
# Generated by dhcpcd for interface wlan0
search Belkin
nameserver 192.168.30.254
nameserver 127.0.0.1


bash-3.1# time nslookup www.google.com
;; Got recursion not available from 192.168.30.254, trying next server
;; connection timed out; no servers could be reached


real    0m15.113s
user    0m0.004s
sys     0m0.000s
-8<---------------------

In the example above "nslookup www.google.com" takes 0.1s with a single
working nameserver in resolv.conf. Adding a broken nameserver increases
that time to 15s and also gives no answer. This test was done on Slackware
12.2.

The above example does not behave exactly as I remembered the problem, but
at least it shows there is something fishy with nslookup.

regards Henrik
-- 
The address in the header is only to prevent spam. My real address is:
hc123(at)poolhem.se Examples of addresses which go to spammers:
root@localhost postmaster@localhost

[toc] | [prev] | [next] | [standalone]

#708

Henrik Carlqvist <Henrik.Carlqvist@deadspam.com> wrote:
>Now, when trying to recreate that I only got partial success. I have no
>DNS server running on localhost:
>
>-8<---------------------
>bash-3.1# cat /etc/resolv.conf
># Generated by dhcpcd for interface wlan0
>search Belkin
>nameserver 192.168.30.254

...

>bash-3.1# time nslookup www.google.com
>Server:         192.168.30.254
>Address:        192.168.30.254#53
>
>Name:   www.google.com
>Address: 209.85.169.105
>
>real    0m0.104s
>user    0m0.004s
>sys     0m0.004s

That is as expected.

>bash-3.1# emacs -nw /etc/resolv.conf
>bash-3.1# cat /etc/resolv.conf
># Generated by dhcpcd for interface wlan0
>search Belkin
>nameserver 192.168.30.254
>nameserver 127.0.0.1
>
>bash-3.1# time nslookup www.google.com
>;; Got recursion not available from 192.168.30.254, trying next server
>;; connection timed out; no servers could be reached
>
>real    0m15.113s
>user    0m0.004s
>sys     0m0.000s
>-8<---------------------

On my system the same configuration results in exactly
the same time as with only a single valid nameserver
entry.  If I put the invalid entry first it takes about
1 second longer.  I only get the 15 second time out if
there are no valid entries.

>In the example above "nslookup www.google.com" takes 0.1s with a single
>working nameserver in resolv.conf. Adding a broken nameserver increases
>that time to 15s and also gives no answer. This test was done on Slackware
>12.2.

I'm using Slackware 13.37, but have never seen it any different.

>The above example does not behave exactly as I remembered the problem, but
>at least it shows there is something fishy with nslookup.

Yep, that's downright weird!

-- 
Floyd L. Davidson                <http://www.apaflo.com/>
Ukpeagvik (Barrow, Alaska)              floyd@apaflo.com

[toc] | [prev] | [next] | [standalone]

#717

Floyd L. Davidson wrote:
> Henrik Carlqvist<Henrik.Carlqvist@deadspam.com>  wrote:
[...]
>> The above example does not behave exactly as I remembered the problem, but
>> at least it shows there is something fishy with nslookup.
>
> Yep, that's downright weird!
>

Just 'dig' it :-)

	-rasp

[toc] | [prev] | [standalone]

Back to top | Article view | comp.os.linux.networking

csiph-web