HOWTO: Persistant DNS Caching with pdnsd.

[karmila]

Thank you for this tutorial. I simply followed your instruction, and up my browsing speed.

I will read all over the post, and maybe back for some question later (actually, i have some questions but i don't want to bother you with asking the same question twice)

[lhaeh]

It says that it fails when trying to restart the daemon:

lhaeh@ubuntu:~$ sudo /etc/init.d/pdnsd restart
 * Stopping pdnsd                                                        [ OK ] 
 * Starting pdnsd                                                        [fail]

I get this from my syslog:

Dec  5 16:09:30 ubuntu pdnsd[13171]: Could not bind tcp socket: Address already in use
Dec  5 16:09:30 ubuntu pdnsd[13171]: Could not bind to udp socket: Address already in use
Dec  5 16:09:30 ubuntu pdnsd[13171]: tcp and udp initialization failed. Exiting.
Dec  5 16:13:03 ubuntu named[3588]: error (network unreachable) resolving 'ns3.mydns.pro/A/IN': 2001:500:1::803f:235#53
Dec  5 16:13:03 ubuntu named[3588]: error (network unreachable) resolving 'ns2.mydns.pro/AAAA/IN': 2001:7fd::1#53
Dec  5 16:13:03 ubuntu named[3588]: error (network unreachable) resolving 'ns4.mydns.pro/AAAA/IN': 2001:503:c27::2:30#53
*SNIP*

However, when I use dig to measure how long it takes to resolve names, I find that caching IS working.


Edit: Seems to be a conflict between it and bind, killing named before trying to run pdnsd lets it run, however then bind has issues when you restart it:

lhaeh@ubuntu:~$ sudo killall named
lhaeh@ubuntu:~$ sudo /etc/init.d/pdnsd restart
 * Stopping pdnsd                                                        [ OK ] 
 * Starting pdnsd                                                        [ OK ]

lhaeh@ubuntu:~$ sudo /etc/init.d/bind9 restart
 * Stopping domain name service... bind9                                        
rndc: connect failed: 127.0.0.1#953: connection refused
                                                                         [ OK ]
 * Starting domain name service... bind9                                        
Failed: Bad server label.
Opening socket /var/cache/pdnsd/pdnsd.status
                                                                         [ OK ]

So, I guess the thing to do is to keep one of them from trying to use port 953, but I don't know how do that.

[zenon222]

I'm having an issue with local windows servers. For example if I navigate to

smb://dave-12183u0zgf/shareddocs/

With pdnsd running, no dice. Perform a

sudo /etc/init.d/pdnsd stop

And I'm back in business.

Ideas??

[ticket]

Trying to get this DNS caching working for natty 11.04.

Almost did it - but I can't get it to 'stick'.

Here's what i did to get it almost working:
(with thanks to all the contributors to this thread):

Install pdnsd.
(Synaptic likes to also install the resolvconf package - this form of the command stops that)

sudo apt-get --no-install-recommends install pdnsd

Edit the pdnsd config file:

gksu gedit /etc/pdnsd.conf

And make the changes suggested in post #1, i.e. comment out the server section for resolv.conf, and modify the other server to point to your favourite DNS server.

Then modify these files:

gksu gedit /etc/dhcp/dhclient.conf
gksu gedit /etc/dhcp3/dhclient.conf

to remove the # in the line:

#prepend domain-name-servers 127.0.0.1;

(i.e. uncomment it).

Note: I could only get 'nameserver 127.0.0.1' to appear in /etc/resolv.conf by editing /etc/dhcp/dhclient.conf.
It is pointless editing /etc/resolv.conf as it is automatically generated.

Finally, check that /etc/default/pdnsd is like this:

# do we start pdnsd ?
START_DAEMON=yes
# auto-mode, overrides /etc/pdsnd.conf if set [see /usr/share/pdnsd/]
#AUTO_MODE=
# optional CLI options to pass to pdnsd(8)
START_OPTIONS=

(If START-DAEMON is commented out, or set to 'no', then it will be impossible to start pdnsd. Also, if AUTO_MODE is set to something, then your custom /etc/pdnsd.conf will be ignored)

Restart the system and test it using 'dig', e.g.
dig www.goog.com (some site you haven't visited before)

Check the dig query worked (no error):

;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 33790

Check the server used was 127.0.0.1

;; SERVER: 127.0.0.1#53(127.0.0.1)

If you dug an unvisited site, the reported query time should be around 50ms:

;; Query time: 24 msec

Do another dig and the query time should become zero.

To check the cache size is increasing with each new web site, use

sudo pdnsd-ctl status

The first few lines of printout will contain:

Cache status:
=============
10240 kB maximum disk cache size.
1207140 of 10496000 bytes (11.5%) memory cache used in 5022 entries.

The contents of my /etc/resolv.conf: (automatically generated)

# Generated by NetworkManager
domain lan
search lan
nameserver 127.0.0.1
nameserver 192.xxx.yyy.zzz

(192.xxx.yyy.zzz is my router address)

However, in my case, pdnsd only worked successfully if, having logged in, I restarted it using:

sudo /etc/init.d/pdnsd restart

Otherwise dig queries would always return a query time of 0 msec and the cache would not get updated.

I tried a couple of ways to automatically restart pdnsd during login,

(a)
Inserted into /etc/rc.local :

/etc/init.d/pdnsd restart

(b)
Put script into /usr/local/bin :

#!/bin/sh
/etc/init.d/pdnsd restart

& put a script launcher into /etc/xdg/autostart

But neither of these helped.

Something is amiss, but I don't know what.

[Motive for using pdnsd: if ever the PTB kill the internet by killing dns servers, a cache will mean you still have a connection]