Hi - I saw a few posts on this website from people with similar problems to what I seem to be getting, although the resolutions posted for those don't seem to work for me :-( so I'm assuming my problem is slightly different, and I'm making my plea for help separately :-) I'm attempting to get a PPP dial-up connection working on a Mandrake Linux 9.2 system, which is also connected to a home LAN. Suffice it to say, I'm having a spot of trouble which 10+ hours of work on my part has failed to remedy - hence I'm swallowing my pride ;-) and seeking outside help.

Anyway, the problem is this - When initially connected, the link actually works - I can ping internet live servers, and I can use things like FTP. If I type 'route' an IP table with the default route set to tnt1.qsi.net.nz (my ISP's gateway) appears instantly - see below.

Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
tnt1.qsi.net.nz * 255.255.255.255 UH 0 0 0 ppp0
192.168.0.0 * 255.255.255.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default tnt1.qsi.net.nz 0.0.0.0 UG 0 0 0 ppp0

However, if I access anything at all that uses HTTP, with any browser, the connection seems to stop working (although it doesn't disconnect). Sometimes it will work (start loading the website in question) for a few seconds, but then it fails. If I type 'route' after this, the routing table takes quite a long time to appear, and often it won't resolve the name tnt1.qsi.net.nz - it merely shows the IP - like this:

Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
202.89.128.79 * 255.255.255.255 UH 0 0 0 ppp0
192.168.0.0 * 255.255.255.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 202.89.128.79 0.0.0.0 UG 0 0 0 ppp0


I have noticed that when trying to ping an internet live server after the connection has failed, net_monitor often shows a whole load of traffic being broadcast over the LAN, but little to none over the ppp connection - since the IP Routing table is till technically correct (albeit minus name resolution) I don't understand why this is happening.

To complicate matters, sometimes the connection will sporadically start working for a few seconds or more - as an example, it will fail because I try to open a website, but then if I leave that website trying to open (or come back a few minutes later and try to ping an internet server) it will suddenly spur into action for a few seconds (or more) before failing again.

I am using KPPP, and I have specified my ISP's name servers, as well as the domain - I did ask my ISP for help but was told they don't support Linux (thumbs down to them, I say). Anyway. Any help would be greatly appreciated.

Any ideas? :-)


Brett C,
AKL NZ

It sounds most like you are losing the connection to qsi's dns servers...
The delay with route indicates it can't resolv the numbers...
(route -n disables the dns search)
Why is another question... ???

Try the kppp setup and don't ask to use the isp dns servers, rather keep some in /etc/resolv.conf, eg Xtra's or Ihug's dns servers.
do dig www.xtra.co.nz @dns.server.ip.number to check the speed of each dns server and keep a fast one.
Some servers may not talk to you if you aren't their customer, but Xtra & Ihug should be ok. They list their numbers in the Support sections of their home pages.

Hi,
Thx for the quick reply :-)

I tried what you suggested - removing the DNS (along with the domain, and gateway settings,

to avoid complication) from KPPP and specifying the DNS servers of another ISP (Ihug) in the

/etc/resolv.conf file.

The route command now takes a long time to return immediately after connection, and not just

after using HTTP traffic :-P. It is still actually working with DNS, since it was able to

resolve the name of the gateway I was connected to, it just took forever and a day to do it.

Also, when I ping an internet live server, although the actual ping times are very good,

averaging at 376ms for www.google.com, it takes an inordinate length of time between pings -

i.e., to send and receive only 21 packets took 74 seconds.

One thing I should have added - my machine is dual booting with Windows XP, and my

connection to QSI (the ISP) is very stable and reliable in Windows, so I'm assuming that

their DNS servers do work.

I am still getting the problem with the connection destabilising, and totally losing DNS

after accessing anything via HTTP - a route command fails to show the name of the gateway,

and gives only the IP. Also, the issue with it spurting into action for a few seconds and

then dying is a problem. By way of example, I tried to load www.linuxquestions.org to post

this message, and it started to load, but it died before completing the process. I tried to

ping google.com and got a 'host unknown'. I tried to ping it again a few mins later, and it

works (Albeit very slowly, as mentioned above). :-(

I'm wondering if perhaps one of the web server services (I have Webmin & SWAT installed) is

involved in all this somehow (because of the apparent link to HTTP)?

I'm really quite impressed with this new version of Mandrake, but this problem is a real

drag :-(


Brett C,
AKL, NZ