Although I'm a total newbie I'll still add my two cents .... that way
someone can correct me if I'm wrong ... I'm still trying to understand it
fully myself ...
1. Either you are probably using dynamic ips in which case once you dial
and the packet that was waiting to get sent (probably did get sent out
????) but had your "unconnected" IP as it's source and so couldn't get a
reply which is why it works the 2nd time ie. it knows the proper IP. ( Hmmm
why doesn't it seem to do that when it reconnects after timing out ???? ).
My packets generally (not always) seem OK because my system triggers a upd
domain packet to look up the host so I guess the packets get lost and it
just requeries it again or rotates it's nameserver ????? someone suggested
putting the same nameserver multiple times in your resolv.conf (ie. if it
times out it tries the same one again). I point to a local nameserver which
forwards to two others so I THINK I get two or three lots of timeouts
before I have to kill and resend the packet.
So first of all why isn't your system sending a domain packet first ? Look
at your DNS setup.
However there appear to be other acceptions. These are:
I've notice if I fire up a browser then it times out for DNS while the
system is coming up and so starts looking for whacko domains. For instance
if I type www.zeta.org.au then by the time the link is up it's looking for
www.zeta.org.au.com or something like that (this is IE 4.0 on Win 95). And
i think the time out value you have talked about is one way of increasing
this value so you get more time for the link to come up.
So is your trigger packet a udp 53 ? If not then it's probably a vital
packet and the FAQ is pretty clear that it's going to get lost if your
using dynamic IPs. If your trigger IS a udp 53 then maybe you should try
either of the two suggestions which are add the nameserver to resolv.conf
multiple times OR recompile the kernel with increased timeouts. Any other
suggestions (maybe the correct ones now ... )
THEN post the results back to this list so I can learn from what you've
done.
2) I don't know about sendmail since I use qmail BUT I have seen references
to this same problem AND the answer was something like "hmmmm, sendmail is
sending packets over the link for some reason ... consult a sendmail guru
to find out why .. ) unfortunately it appears there are no sendmail gurus
on this list.
Well I'm pretty confused now so I'll leave it there.
Wilson Fletcher
----------
From: Jan Csernoch[SMTP:[EMAIL PROTECTED]]
Sent: Friday, 4 February 2000 7:51
To: [EMAIL PROTECTED]
Subject: diald+timeout+sendmail
Hi!
I use Debian 2.1r3. I tried to get diald to work for a long time now. My
problems are:
1. When I try to access some kind of service requiring a dial-up, then
diald
starts dialling, makes the connection, but the service hangs. For example
if
I start 'lynx URL' then I don't get any answer even if the connection gets
established. I have to CTRL-C and reissue the command to succeed. Similarly
if I use a client with netscape then search for the domain gets timed out.
I
have read that if I recompile the kernel with an increased value of a
certain variable in the source, then the time-out problem can get solved.
The problem is that it's not only a time-out problem, it's that the service
can not start for some reason once the connection establishes. It's very
frustrating, because I can not issue commands from cron and I have to
explain the users to reload the pages several times.
2. The second one is that when the system boots and sendmail starts, it
tries to make a connection for some reason. How can I avoid that.
Sorry if my questions are obvious, or already answered.
Thanks for the help in advance
Jan Csernoch
-
To unsubscribe from this list: send the line "unsubscribe linux-diald" in
the body of a message to [EMAIL PROTECTED]
-
To unsubscribe from this list: send the line "unsubscribe linux-diald" in
the body of a message to [EMAIL PROTECTED]
