<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> </head> <body text="#000000" bgcolor="#FFFFFF"> Hey, <blockquote type="cite" cite="mid:CAAP7uc+m76f=St5BF59JMbycL73ZqZ85RFPGAY=mwYm8vMBkiQ@mail.gmail.com"> <pre wrap="">On Fri, Dec 8, 2017 at 6:29 PM, Marcel <a class="moz-txt-link-rfc2396E" href="mailto:mahik18@gmx.de"><mahik18@gmx.de></a> wrote: </pre> <blockquote type="cite"> <pre wrap="">Do you think it could help if I provide you another log of an execution of my script with specific parameters? Or should I go through my logs searching for specific parts? I'm not really sure what I should look for. Is there anything I could do to stabilize the network switching? Maybe do some stuff between the connects? Or quickly restart the mm? Would it help to quickly stop the mm and then start again? Or is it even the modem which is busy, so the mm doesn't have to do anything with the error? </pre> </blockquote> <pre wrap="">At this point I would maybe suggest to run the steps manually e.g. in minicom; and try to see why the registration in that last operator takes so long. But one thing I would definitely do is leave more than 15s for the whole connection process, especially when registration in a different network is involved. </pre> </blockquote> Aah, sorry I answered to that in my last mail already(summarized: already tried timeouts of 50 seconds and tried to connect up to 8 times with that 50 seconds timeout, I also don't really know which AT commands are used for simple-connect [or even any of the mmcli commands] which is why I actually used mmcli): --> Hey, <blockquote type="cite" style="color: #000000;">In your logs I see a connection attempt with an explicit operator id (26203) at 1511990328.384995. The logic goes through the manual registration process, trying to get registered in that specific network. At 1511990345.704504 I see another connection attempt to the same operator, that is approx 17s after the previous attempt. Are you waiting some specific time between connection attempts? </blockquote> Yes, I'm using the simple-connect function with a 15 sec timeout. The first operator 26201 connects without any problems at all, then when disconnecting and connecting to 26202, the first, maybe also the second try to simple-connect fails. But then when I disconnect again from 26202 and try to connect to 26203 again it just won't connect. It seems like the mm is still busy doing some other stuff in the background. At the end of my script, I disconnect again and disable the modem. Disconnecting works fine, but the modem doesn't disable, but seems to hang on state 'disabling'. Around 1 minute after my script finished the mm seems to be done with its background stuff and disables the modem. <blockquote type="cite" style="color: #000000;">Those two attempts seem too close to each other. You should leave much more time for the connection attempt to go on, because re-registration in a different operator, especially when doing it manually, may take some time. The logs show that the registration in the operator didn't happen in those 17s, and it looks like the actual COPS command used to request manual registration didn't reply in that time either. Are you able to manually reproduce these issues running the AT commands in minicom? </blockquote> Well I just tried a script setup with 8 tries to connect each with a 50 seconds timeout, same result, operators 26201 and 26202 worked perfectly, 26203 didn't succeed at any of the 8 connect attempts. I also tried a setup with a 15 seconds timeout and a 60 second sleep between each operator with the same result. And right now I tried a setup where I also used the 3gpp-register-in-operator function, but that didn't work either. Weell about the reproduction with AT commands: I'm don't really know which commands I would have to use, I already thought about just doing the whole script with AT commands, but then I recognized that it's a lot more complicated than I thought and so I just used mmcli which already is very simple and efficient. So about that, I'm sorry, but no, I unfortunately can't reproduce the issues, sorry. (PS: I'm sorry for always answering to your mails directly without ccing the mm list!) </body> </html>