Hi, I am trying to get automatic reconnection of UMTS connections that are disconnected for example due to low signal to get working. I seems that are some problems, that comes up to time to time again, but I don't see a solution. I am using network manager 0.9 and modem manager 0.5. I have identified the following problems: 1.) Auto reconnection stops after some retries. This makes sense for many types of connections, but not for UMTS connections. I want to reconnect over and over again until signal is good enough again. The patch auto_act.patch changes this behavior. Not sure if this is the best solution, but it works for me. 2.) When pppd ends, for example because of lcp echo request failure, the ppp device gets removed, which in turn causes network manger to move the ttyUSB device in an unmanaged state. So it's now the same as if I remove my USB stick. The only chance to get it working again, is to really unplug the stick and insert it again. Bad luck if you have an internal UMTS modem... The patch no_device_disable.patch changes this behavior. It directly compares the device name to ppp, which might not be the best idea, but I haven't found a more clean way. 3.) If the pppd has already started the first ttyUSB device is in data mode, which is not reseted anymore. In this state, the only chance to get a new connection, is to disable UMTS at all and to reenable it. Then the modem gets initialized again and everything is working fine. So what I think what would be necessary here is that if pppd is stopped that the modem is reseted. I didn't found a way to do it and hope somebody on the list can point me a way. As a fallback I think it would be a good idea if modem-manager, in case it gets an timeout when speaking the ttyUSB port, it just resets the modem and tries it again. Again I didn't find the right place to implement this. Any hints are welcome. 4.) We have an UMTS connection which requires username and password, which are save in a system connection. If pppd fails during handshaking, the users gets ask for a chap password, which he does not know, should not know. The no_ppp_pw.patch changes this behavior. I think it's not a real solution, because it disables asking for a password in case of failure at all, but it shows the problem (and solves the problem for me). I would be very happy if somebody can give me some help how to solve problems in 3.) Regards Gerald
Attachment:
auto_act.patch
Description: Binary data
Attachment:
no_device_disable.patch
Description: Binary data
Attachment:
no_ppp_pw.patch
Description: Binary data