Posts made by Adriaan

Dear all,

Thanks to hints from @Kolja-Vornholt , I found a link that with forking works great for me: Sending the input to my parser with socat. All request will now be handled by their own fork.

Example for Linux:

socat -u TCP4-LISTEN:4445,keepalive,reuseaddr,fork STDOUT | php -f /bin/parseTCPdata4445.php

Thanks to this post:
https://stackoverflow.com/questions/44886399/output-a-linux-command-to-a-url-port-or-scocket-instead-of-writing-it-to-a-file/68299500#68299500

Thanks again, @Kolja-Vornholt

@Adriaan This post continues here: click link

(Why can we not edit posts?)

@Adriaan This thread continues here:

https://forum.iotcreators.com/topic/716/tcp-listener-seems-to-be-killed-or-stalled/9

@Kolja-Vornholt Yes. I will make a nice clean one from 1 hour or so. Ok?

@Kolja-Vornholt Yes, a quick analysis hints to that, but I need more time.

Is there any way in which I can use wireshark to check pcap file for FIN messages, AND in relation to an IP number? (

Again, I am sorry, not a big expert here
Again: Thanks for your help.

Hello Kolja,

I have a tcap file, but that contains many request. I would not know what to show. I do have the idea though that the button does not always sends the [F.] flag.

I will run the following for a while:

tcpdump -i <interface> "tcp[tcpflags] & (tcp-fin) != 0" >> file.txt

@Adriaan

Well, at least I can reproduce the issue. On the sending side on Linux, the -N flag does the thing.

echo "Hello there" | nc machine.domain <port> -N -v 2>&1

When connecting to a remote host, and using strace -r -f netcat -k -l <port> , setting the -N flag results in

 0.000596 read(4, "", 16308)        = 0
 0.000354 shutdown(4, SHUT_RD)      = 0

Without the -N flag, listener remains listening, and hence not closing the connection. This is the last line then:

0.000404 poll([{fd=0, events=POLLIN}, {fd=4, events=0}, {fd=4, events=POLLIN}, {fd=1, events=0}], 4, -1

Next step is that I am going to look if I can change something on the listener side as well.

Greetz,

Adriaan

I have this strange experience.

In Holland, I am testing NB-IOT and CAT M1 SOS buttons, that work 100% on a Chinese platform, but as soon as I point the button to my clients Windows Azure server or my own Linux server, it keeps the TCP listener occuped. I guess it’s because the TCP connection is not properly closed, but I am not an expert here.

I have tcpdumped my ass off, here is an strace of the processes:

On my Linux, my listener is started like this:

strace netcat -k -l

Then in a normal sequence, I get this (strace output):

     0.000640 accept4(3, …
    20.274326 poll([{fd=0, events=POLLIN}, …
     0.003220 read(4, "MT;6;**TEST WEBSRV …
     0.000537 read(4, "", 16309)        = 0
     0.000304 shutdown(4, SHUT_RD)      = 0
     0.000282 write(1, "MT;6;**TEST WEBSRV …
     0.000354 close(4)                  = 0

This is the problem causing one:

     0.000519 accept4(3, …
    12.447784 poll([{fd=0, events=POLLIN}, …
     0.092840 read(4, "MT;6;862785045411272;…
     0.000411 poll([{fd=0, events=POLLIN}, …
     0.000518 write(1, "MT;6;862785045411272;…
     0.000341 poll([{fd=0, events=POLLIN}, {fd=4, events=0}, {fd=4, events=POLLIN}, {fd=1, events=0}], 4, -1

–> And the it waits forever, keeping my TCP listener occupied.

I see the latter is not doing read() and shutdown(). I think that has to do with the button. Or can it be the T-Mobile network?

I don’t get it. Maybe someone here?

Thanks, Adriaan

@Kolja-Vornholt Thanks. I have tcpdumped my ass off already. Anything particular that I can look for while using tcpdump?

Hello all, I have this strange experience.

We have an NB-IOT and CAT M1 SOS button, that works 100% on a Chinese platform, but as soon as I point the button to my clients Windows Azure server or my own Linux server, it seems to kill or stall the TCP listener.

I think is has something to do with the messages being malformed, or TTL or EOL or EOF or so. But I am not an expert.

Do you have any idea how I can analyze this problem?

On my Linux, my listener is started like this:

netcat -k -l <port>

and then parsed by a script.

I have a couple of cronjobs running on 3 machines that emulate the NB-IoT message. That works 100% fine, over long time.

Yet, as soon as I point the button to my Linux system, it stalls.

I don’t get it. Maybe someone here?

Thanks, Adriaan

@Kolja-Vornholt Thank you, Kolja!

I am going to look at python. Still, I would expect that when a device has done it’s work, it should nicely close the TCP session and the listerer should be ‘ready for next connection’.

(It’s logical that a listener van not handle 2 devices on the same port at the same time, of course)

Do you also have experience with the listener getting ‘corrupted’? My listener seems perfectly stable as long as my IoT device does not connect to it

Hello all,

Recently I picked up a new NB-IoT project, and started playing around again, with Linux and some NB-IoT and CAT-M1 devices.

There are somethings I don’t understand. Maybe someone here does.

I created some listeners and parsers on a Linux machine. I configure 2 devices to connect to them. It is all quick and dirty coding, but I get the devices to connect and send data.

On linux, I listen to TCP and I redirect incoming data to a file with
with the command

netcat -k -l [port] > netcat.pcap

I start the listener with tmux so that I can detach from the session and keep the listener running headless

• It seems that one listener can only serve one device. Is that correct? I am used to Apache Webserver, that can handle as many devices (browsers) that you want. But that is HTTP. If I point 2 devices to the same port, I get only the data from 1 device. Or am I doing something wrong / thinking in the wrong direction?

• Another question. each device connects to it’s own port. For current testing, that is good enough. (remember, I run the listener headless.) Yet it also seems the listener can get ‘corrupted’ or so. I then need to restart the netcat listener. Does anyone have any experience with this?

Thank you.

Adriaan

Hi guys, is there any way in which we can use our existing SIM cards on the new platform? Thanks!

@andre-rodenburg Ah, ik zie deze nu pas. Ga ik de volgende keer proberen.

@afzal_m @techniek

Laat maar, ik denk dat het wat anders was: mijn compaan zit wellicht in een “NB-IOT dal” of farraday kooi. Thuis en op mijn kantoor werkt mijn automated script nu prima, ook stand-alone.

Maar…wat ik nou zo raar vind is dat alle waarden uit de logging wel goed zijn, ook bij hem. Dus we krijgen IP adres, etc: dit werkt allemaal goed:

AT+CSQ // altijd iets als 21,99 of 17,99,

AT+CGATT // geeft 1

AT+CGPADDR

AT+NPING=“172.16.14.22”

@felixdonkers said in - NB-IoTest -:

ATI9

YES, works, dank je wel!

@techniek Hoe krijg je die application_version eruit? Dank, adriaan

@afzal_m said in Probleem met AT+NMGS=3,"AABBAA":

Yeap, that’s right. The Neul Messaging Protocol (NMGS) is not supported anymore on these modules. However, there are two solutions:

1. You can try to add the NMGS CoAP library on your module
2. Our just wait for us, before the summer we will be ready to support UDP messages on our network and then you will be able to use the AT+NSOST command to send messages

@afzal_m geldt dit ook voor de Ublox Sara-N200? Een paar weken geleden kon ik nog prima messages verzenden en ontvangen, nu krijgen wij ook +CME ERROR: 50.

Thanks, Adriaan

Same problem here. Any solution?