Share Z-wave dongle over IP (USB over IP using ser2net / socat ) guide

[JohnChiasson]

Hi Vossi,

I think this thread was more about debugging zwave dongles, but the techniques apply to any serial device sharing over network. I don’t know if the windows versions of socat / ncat / nmap will work but the 115200 baudrate is correct for the alarmdecoder serial device. You are right though, I have some experience in windows serial devices and slower baud rates seem to help. I might try that with alarmdecoder.

I have also tried using it with Home Assistant, and have experienced no hangs there, so my suspicions are in the OpenHab AlarmDecoder binding hanging.
Thanks for the feedback.

[a.tibi (Ahmad Yazan Tibi)]

Is it working as same as Homeseer Z-NET?

[ei_Gelb_Geek (Kevin)]

Thanks for pointing out how to get socat / ser2net into a docker container

Can you perhaps briefly describe how you build the socat connection on the NAS? I have now just made a script in the task scheduler when booting.

#!/bin/sh
socat pty,link=/dev/ttyUSB0,raw,user=docker,mode=777 tcp:10.10.80.220:6666 &

I do not think that’s the perfect solution, even if it works.

[ArdKuijpers]

In my configuration, I also added a boot-up task in Task Scheduler. The Run command is:
nohup /volume1/docker/openhab/diskstation-configuration/scripts/receive_zwave.sh &
and my script (receive_zwave.sh) is similar to yours, with the addition of a while loop to restart socat when it might fail:

#!/bin/bash
DEVICE=ZWave
REMOTEIP=192.168.249.13
REMOTEPORT=3333
LOGFILE=/volume1/docker/openhab/diskstation-configuration/scripts/socat-$DEVICE.log
while sleep 1; do
        TIMESTAMP=`date +%Y%m%d_%H%M`
    cp $LOGFILE ${LOGFILE%.log}-${TIMESTAMP}.log
        echo "% Start logging on ${TIMESTAMP}" > $LOGFILE
    /usr/local/bin/socat -d -d -d -lf $LOGFILE pty,link=/dev/tty$DEVICE,raw,user=openhab,group=openhab,mode=777 tcp:$REMOTEIP:$REMOTEPORT
done

[ei_Gelb_Geek (Kevin)]

Thanks

[Laufeyjarson (Lou Erickson)]

systemctl would be the right answer here.

I wrote a tutorial on how to use AlarmDecoder with SSL; it had similar requirements. Look at how it used systemctl to set up.

[ei_Gelb_Geek (Kevin)]

No … Synology Diskstation … Docker … No systemd…

[Laufeyjarson (Lou Erickson)]

Sorry, missed that. Saw the other discussion of systemctl and thought that’s the part that was missing.

[dastrix80 (Kris K)]

Hi All,

Just wondering if anyone has managed to fix the issue assocaited with the Zwave binding being stuck if the ser2net/socat link breaks?

I find the I/O errors pop up on OpenHab. A restart of socat/ser2net fixes the link, but then one must manually restart the binding for ZWave to return.

Be keen to know how your systems have self healed using ser2net/socat

thanks

[vossivossi]

@dastrix80

I find the I/O errors pop up on OpenHab. A restart of socat/ser2net fixes the link, but then one must manually restart the binding for ZWave to return.

Be keen to know how your systems have self healed using ser2net/socat

I meanwhile managed to have this stable but with a Windows based OH server in my setup. I am now using Raspis with ser2net to feed the serial connection into the LAN and on the Windows machine I am using com0com and hub4com as clients (it took me days of testing and tweaking before I successfully managed to have the Linux serial server talking to the Windows serial client).
The virtual COM ports created by com0com on the Windows machine remain persistent even if the connection is interrupted. So there is no exception thrown on the OH side anymore. And when the connection recovers everything just continues to work as before the interruption without any need of restart of the Z-Wave binidng anymore. For me that is the best solution I ever had (an I tested all kind of solutions for nearly 2.5 years to have that connection stable).

[dastrix80 (Kris K)]

Cool, Windows isn’t a solution for me though. I wonder why it changes using Windows? It appears these tools do not auto recover in linux

[dastrix80 (Kris K)]

@vossivossi

Yes, of course. However I have to say that I am running OH on Windows so my solution might not be suitable 1:1 for other configurations.

So basically I created a “zwaverestart”- thing for the excec binding (which you need to install in OH):

/*exec*/
Thing exec:command:zwaverestart "Command: zwaverestart" [command="restartzwave.bat", interval=0, timeout=20, autorun=false]

This “thing” calls an executable batch file “restartzwave.bat” on the command line level (this is of course Windows specific). I did this after many tries as I was not able to put the complete command line successfully in the thing. So the file restartzwave.bat has this content:

plink  openhab@127.0.0.1 -P 8101 -pw habopen -m c:\zwave\bundle.txt

This is the call of Karaf Console via SSH (in Windows done with plink (a Open SSH implemenation for Widnows), which needs to be installed first), here with the default password “habopen” and the command itself is again in another text file (C:\zwave\bundle.txt). Again I did this hack with the second file because I was not able to get it to run with the command text directly in the same line (might also be a Windows specific problem).

So finally, in the file bundle.txt I have the command for Karaf console:

bundle:restart org.openhab.binding.zwave

Hopefully this helps you to get further.

However, there is one more important point: How do you know that the rule needs to be fired? How do you know that your Z-Wave controller actually seems to be offline? I implemented this in the following way:
All my Z-Wave based items are members of the same group ZWAVE. Whenever one of the members receives an update the controller must be still be working. So I defined an item OH_ZWAVEUP_LCH to store the last timestamp when a Z-Wave item was updated:

rule "MemberUpdatezwave"
when
    Member of zwave received update
then
       postUpdate(OH_ZWAVEUP_LCH, new DateTimeType())	
end

I have a then a time based cron rule every 15 minutes which checks if the last timestamp of Zwave updates is older than 15 minutes. If this is the case, I send the command for the excec thing to trigger:

sendCommand(RESTARTZW, ON)

I know that this is a bit complicated and there might be of course other ways to implement it. But for me it works very reliable for more than a year now. Since I have this setup I do not worry about a lost USB-IP connection as it only affects my system for 15 minutes, then the Z-Wave binding restarts and everything works again.

Hi vossi, how do you stop the thousands of zwave updats to the group? or do you only make your zwave switches part of the group?

[vossivossi]

@dastrix80

Hi vossi, how do you stop the thousands of zwave updats to the group? or do you only make your zwave switches part of the group?

Actually I have 3 groups as I do have 3 Z-Wave sticks (and therefore 3 networks). ALL items of one network are members of the specific group. So the time stamp is updated very frequently (which is on purpose as I want to monitor this “heartbeat”).

[dastrix80 (Kris K)]

Wow, thats ALOT of traffic! like every second

[vossivossi]

What’s the problem? Are you running on a Pi with SD Card?

[dastrix80 (Kris K)]

No, ive got PLENTY of grunt, but i just dont want my karaf log full of the updates every second.

So you’re using 3 controllers on one OH2 system with 2 remote machines connecting the controller to OH2 via socat/ser2net?

[vossivossi]

@dastrix80

So you’re using 3 controllers on one OH2 system with 2 remote machines connecting the controller to OH2 via socat/ser2net?

3 remote machines. I separated the sticks completely physically from the OH-Server. They are all connected over IP.

[dastrix80 (Kris K)]

Interesting you must have a very distributed building. And very reliable from your previous posts. Are they all MAIN controllers? I assume then different nodes are paired with the respective controllers

[vossivossi]

@dastrix80

Are they all MAIN controllers? I assume then different nodes are paired with the respective controllers

Yes, all are MAIN controllers. The pairing needs to be done with the correct controller for that part of the building. Unfortunately the OH Z-Wave binding does NOT support a selective inclusion mode only for one controller. Instead all my 3 controllers go into pairing mode when launched via Habmin. If the device is located only within range of one controller this will work. If there are 2 controllers close to the device, I “disconnect” (stop ser2net) the other controller and then click on network wide inclusion.

[snoekieboe (Roel)]

For those interested…

i’ve updated my initial post with a setup for running multiple socat instances on the OpenHAB machine connecting to multiple ser2net nodes in your network simultaniously.
This might be helpful if you require e.g. multiple Z-wave nodes to cover a large or complex installation. I’ve tested it and it seems rock solid. Autostart of multiple socat instances at (re)boot and logging into separate log files for each instance (/var/log/socatx.log)

br

Roel