Internet Connection 20 minute cut-off
My dial up internet connection never seems to last longer than 20 minutes.
It doesn't disconnect, but data stops being transferred back and forth, usually after about 20 minutes, and I end up having to manually disconnect. My modem is a crappy generic USB winmodem, but the drivers I found for it seem to work quite well, apart from this problem. It's not so bad for general browsing, but it means that I can't download big files or use IRC for any decent length of time. My ISP is OK as well, since it works perfectly with windows. And I don't think it is a permissions thing either since I can dial in as a normal user. I'm using KPPP to dial up, by the way. Help...? :) |
just a thought: it might have something to do with your firewall... take a look at your firewall log to see if something weird is happening at the moment you loose contact...
|
Thanks for that reply, that firewall point makes sense.
Sorry to ask such a silly question, but what firewall does slackware use, and where might I find it's log? |
slackware uses linux's standard firewall - iptables (netfilter)...
on slackware the iptables script is (by defacto standard) placed in /etc/rc.d/rc.firewall... however, one is expected to put the iptables script there on one's own (and make it executable), and there isn't one set by default... when you DO have a firewall script executed, and set to log some packets, the logging will occur by default in this file: /var/log/syslog here's a simple iptables script (for single interface setup) you can use to log dropped input packets in case you don't have any firewall script right now: Code:
#!/bin/sh once the firewall script is running and you are close to the 20 minute mark, open a terminal, become root, and do a: tail /var/log/syslog the "tail" command will give you a "real-time" view of the dropped packets that are getting logged by the firewall... all you gotta do is look and see what's the last ones to get dropped before you loose contact (if any) and post them here... if you wanna look at ALL the dropped packets just do a: cat /var/log/syslog | grep "INPUT DROP" here's the script above, but with explanations (comments) in case you are wondering what the lines mean: Code:
#!/bin/sh http://www.linuxguruz.com/iptables/ also, because it's possible the issue might not be your firewall after all, it would be helpful if you would post your modem's brand and model, as well as the URL to the driver you are using and a brief description of how you installed it... good luck... |
OK, thanks. I'm trying it just now, I'll let you know what results I get.
|
roger that
|
Well, it's been 45 minutes and I still have the connection! Not that I'm complaining of course, it's just a little strange that it should be so variable.
Anyway, /var/log/syslog | grep "INPUT DROP" produced A LOT of information, 300+ lines, so I don't think I'll be posting it. But here is the tail /var/log/syslog information: Code:
bash-2.05b# tail /var/log/syslog |
it's cool, you don't need to post everything... basically the most important parts are the packets dropped right before the connection is lost...
let's keep it under observation for a little while to see how things go... i might have a theory in case the problem has disappeared by using the firewall script... but let's wait and see how things go first... meanwhile, please post the output of this (the whole thing) right after a fresh startup: Code:
ps aux |
Here you go:
Code:
sh-2.05b$ ps aux |
cool, i'll take a look at the processes in a little bit...
oh, and by the way, in my post above when i mentioned "tail" i forgot to post the "-f" option... using "tail" on it's own will output the last 10 lines and exit, but doing a "tail -f" is what will give you the "real-time" output... the "-f" stands for follow... sorry 'bout that... tail -f /var/log/syslog |
Well, I just lost the connection after about 6 minutes, and here's what the tail -f /var/log/syslog had to say:
Code:
Jan 29 18:14:11 darkstar kernel: INPUT DROP: IN=ppp0 OUT= MAC= SRC=66.93.164.72 DST=62.69.52.122 LEN=28 TOS=0x00 PREC=0x00 TTL=242 ID=55851 PROTO=ICMP TYPE=8 CODE=0 ID=512 SEQ=24722 |
Quote:
that's a "ping" which was blocked by your firewall... maybe your ISP does "ping probes", killing connections for hosts that don't send an echo reply... add a rule allowing pings and then re-run the script... here you go (the rule is in bold): Code:
#!/bin/sh now we can wait and see if that did the trick... my fingers are crossed... =) |
as for the processes you are running:
these are the ones which should be deactivated unless you really need them: Quote:
you can stop them from getting auto-started by using pkgtool... just run pkgtool (as root), go into "Setup", select "Services" and hit okay... then remove the Xs from the boxes for rc.sshd, rc.sendmail, and rc.inetd and then okay and okay and exit... now the processes shouldn't get started when you reboot... PS: i also suggest deactivating gpm if you don't use it... i don't think there's an entry for it in the "services" section of pkgtool, but you can disable it by making the rc.gpm file non-exectable with this command: Code:
chmod 644 /etc/rc.d/rc.gpm |
I don't think that new script worked, I was being disconnected after only a few minutes while I was using it. One thing I've noticed since then, is that incomming traffic seems to stop before outgoing. I've been keeping an eye on my gkrellm monitor and it seems that data is still being transferred out of my computer even after I can't recieve any more. I don't know what that means for me though.
Phew...this is getting a bit tedious, I'm sorry to keep bothering you with the same old story, but thanks for your help so far. |
I just noticed something else in the tail -f /var/log/syslog readout. Something called a 'fifo underrun' happens when I am disconnected. I have no idea what that means, does it mean anything to you? Or anyone else?
Code:
Jan 29 20:42:46 darkstar kernel: st7554: fifo underrun! |
All times are GMT -5. The time now is 11:32 AM. |