[lotuseclat79]

Commonly, I will get a modem hangup using wvdial and will get the following message:

WvDial<*1>: Connect time 27.2 minutes.
WvDial<*1>: Disconnecting at Thu Mar 6 10:09:05 2008
WvDial<*1>: The PPP daemon has died: A modem hung up the phone (exit code = 16)
WvDial<*1>: man pppd explains pppd error codes in more detail.
WvDial<Notice>: Try again and look into /var/log/messages and the wvdial and pppd man pages for more information.

The /var/log/messages file will commonly contain the following output:
Mar 6 14:41:05 ubuntu pppd[10742]: No response to 4 echo-requests
Mar 6 14:41:05 ubuntu pppd[10742]: Serial link appears to be disconnected.
Mar 6 14:41:05 ubuntu pppd[10742]: Connect time 27.2 minutes.
Mar 6 14:41:05 ubuntu pppd[10742]: Sent 513267 bytes, received 3125031 bytes.
Mar 6 14:41:11 ubuntu pppd[10742]: Connection terminated.
Mar 6 14:41:11 ubuntu pppd[10742]: Modem hangup
Mar 6 14:41:11 ubuntu pppd[10742]: Exit.

I have the following questions:
1) Which modem has hung up first?
a) The remote modem at the ISP, or
b) My local modem

2) I have recently (yesterday) set the wvdial.conf file to have Idle Seconds =0 to disable the feature for hangup timeout.

What is the actual cause of the modem hangup? Does the ISP or the Phone Company drop the carrier signal? Setting Idle Seconds = 0 still allows the echo requests even with the hangup timeout feature disabled.

If the actual cause is not due to a carrier signal drop out, then what?

Note: I have a USR 5786E modem. Is there some configuration that I need to specify in its initialization string to keep it alive (I think it is set for 300 seconds - I'll have to check), or given the lack of an Ack response to 4 echo requests (is that due to a busy ISP resource and if so, can the number of echo requests be increased to say 10 over some period of time that makes sense if the ISP is experiencing a high load of traffic that prevents it from responding to the echo requests)?

Tia,

-- Tom

[RobLinux]

AFAIR (it's a while since I used a dialup with a 2 hour max call time) the pppd uses KEEP ALIVE message/response in order to detect dropped connections even if you send no data. I don't think you should need to turn off idling, as that's a feature to save Erlangs when you're not sending data.

Is it always dropping at "27.2 minutes" or is the time variable?

Does someone picking up and disturbing the connection produce similar? I would expect then that the Modem hangs up, and you'ld get a different error message.

It looks to me like the remote end is apperently dying, not sending the PPP responses, and then your pppd is decidng to drop the modem connection. Of course it could be that simply after a while your pppd is not seeing the responses for some reason.

AFAIR the ECHO REQUESTS are at the PPP protocol level, they're not 'pings' , and if there's PPP going on due to high usage, then I doubt that ECHO REQUEST needs to be sent, they're there to keep something going when non-busy, probably so the modems don't get out of sync.

One issue is that the kernel serial code, is much less used now, than in say the mid-90's, I had a problem with 2.2 even in 2000 where excess CPU usage was required, and PPP download throughput would conversely drop on a busy machine, which seemed to slip by, as most serious users had a LAN, and some kind of seperate firewall machine.

So it may be some kind of kernel serial bug, not necessarily the ISP's pppd's fault. Fraid, this response has to be rather speculative, but hopefully it's useful speculation... I'm a bit rusty on the ins and outs of this stuff.

[lotuseclat79]

Hi Rob,

The time is quite variable. The 27.2 minutes was just a snapshot of one of the latest instances.

No one picks up on this line which is a separate land line while I am dialed into the ISP.

Ok, thanks then for your reply!

Cheers,

-- Tom

[lotuseclat79]

Hi Rob,

Checking the USR website, it appears that there is a more up-to-date firmware driver for my modem, 5686E and there is a Linux flasher download which I am going to install. According to a Modem help website in the UK, the possibility exists that the modems on the ISP side that I am am connecting to may not all have been upgraded to v92. The firware update is supposed to negotiate this kind of problem.

When you asked previously about the minutes of connection being the same as 27.2 minutes, sometimes in fact when I first dialup to my ISP, the connection only lasts for 4.0 minutes - perhaps that is the amount of time it takes to negotiate the lack of an Ack on the 4 echo-requests that are not responded. I think this is that kind of situation, and the firmware update is supposed to fix it.

-- Tom

[lotuseclat79]

The UK website that mentioned the information about the ISPs not upgrading their modems as the cause of the V90/V92 dropouts in my last post was from here.

The firmware upgrade seems to be helping. Since installing it this afternoon, I have yet to experience a modem hangup/dropout due to the lack of a response to 4 echo-requests.

I did not have to lower the speed on the connection, and the firmware upgrade worked transparently.

-- Tom

[lotuseclat79]

For the moment there is some improvement.

The frequency of short sessions seems to be less, however, it does not appear that the new firmware is completely impervious to modem hangup on the other end of the connection at the ISP (I assume). This morning I have been disconnected twice. The first session lasted 53.9 minutes, the second session just 11.7 minutes. The third session is still live for over 2.5 hours as I post this message. Both pppd sessions experiencing modem hangup hung up after 4 echo-requests failed to get a response from the remote modem at the ISP.

-- Tom