OpenNetHomeServer

[Mastiff]

Yeah, that may be it. The same thing happened again, so now I did a full restart of the VM, and then we'll see what happens.

[Mastiff]

Nope. I'm going berzerk here soon... After a couple of hours it stops showing events every time! I have tried to download the latest version of the plug-in in case the one I had was messed up. I have commented out the code you commented out to avoid loss of contact, and I have had one loss of contact during the night, and that was exactly 6 minutes before it stopped receiving stuff. But I have not had the timeout for socket set to that, I had it to 60. I am starting to think that there may be an error in the plug-in instance in the EG XML file since the problem happens in both versions of the VM, so I have now deleted the plug-in instance directly in the XML and then copied it from an older version of the XML (I have tons of backups). I'm also deleting the Tellstick instance and I am adding that again, so we'll see if that helps.

Oh, btw: How do I make a timer in Python? I was thinking LUA, there it's so easy:

Code: Select all

if GirderUIRTTimer
then
GirderUIRTTimer:Destroy()
GirderUIRTTimer=nil
end

GirderUIRTTimer = gir.CreateTimer(DoSomethingOn Start, DoSomethingOnTimer, DoSomethingOnCancel, 0)
GirderUIRTTimer:Arm(60000)

So here I have a timer that every time it's fired, it removes it's own previous instance first, so I don't have many of the sam etimer (so when an event comes in) and then starts again on 6000 milliseconds. Can I do the same in Python?

[Mastiff]

Here are the two plug-in instances, the new one first:

Code: Select all

<Plugin id="8" Identifier="NetHomeServer" Guid="{7B2DF95B-DC3E-4D56-9785-0E13365C568F}" File="NetHomeServer">
            gAIoWA0AAAAxOTIuMTY4LjAuMTAwcQBNRR+ITVgCdHEBLg==
        </Plugin>

Code: Select all

<Plugin Identifier="NetHomeServer" Guid="{7B2DF95B-DC3E-4D56-9785-0E13365C568F}" File="NetHomeServer">
            gAIoWA0AAAAxOTIuMTY4LjAuMTAwcQBNRR+ITVgCdHEBLg==
        </Plugin>

As you can see it says plugin i="8" before the identifier on the version that I know creates problems, but not on the one I know was working when I made the backup. Any idea what that's about?

[Mastiff]

Nope, not that either! I really don't understand what's happening, the logging just stops coming in after a while! It's driving me crazy... And I really don't have time for this either! It was about the worst time it could start acting up!

Edit: The only thing I can thing about is that it's something with the XML in either EG or NHS, since I have tried using backups on everything else. But finding out what it is would probably be about impossible.... They are both very large and complex. Can you make some print code that can debug this?

[Mastiff]

It seems like the only way to get the logging back without restarting EG completely is to change the IP address in the plug-in. Is it possible to do this with Python, so I can have a script that alternates localhost with 127.0.0.1? Then I can go from locallhost to 127.0.0.1 the first time it stops, and then back from 127.0.0.1 to localhost the next time and so on. And then the timer to check for events, of course. I'm getting desperate, I have four days left here, and I really don't have time to mess with this.

[krambriw]

Hold out a minute...I have some news, just typing it

[Mastiff]

Ooooh...exciting!

[krambriw]

A lot of problems it seems...

My configuration for the plugin looks like this bbut I do run a rather old EG version (1691, 1,5 years old):

<Plugin Identifier="NetHomeServer" Guid="{7B2DF95B-DC3E-4D56-9785-0E13365C568F}" File="NetHomeServer">
gAIoWA4AAAAxOTIuMTY4LjEwLjI1M3EATUUfiEsUdHEBLg==
</Plugin>

Anyway, I have worked a lot on the error handling so I have a new version you can try (keep a copy of the one you have now). This one should work as discussed with a timer. So if there is a socket error, a 60 second timer is started, If a valid message is read before the timer fires, it is cancelled. If fired, a new connection trial is performed.

I have also tested how it is handling a longer break a few minutes and it seems repairing itself fine when the network (cable) is restored

Also,
- you can go back to configure the ip address as you like I think
- you can try to set the time out value in the plugin to 60 seconds, the same as I have
- I cannot understand why it stops receiving events except if there was a glitch in the code that would prevent 'things' from being restarted correctly
- this version is printing some additional info in the log, this can easily be commented out once we now it is working ok in your system as well


Edit: found a problem with the attached version, removed...will be back

[Mastiff]

Yeah I guess the computer really don't want to change masters... Thanks! I have that code running now, and I've turned off my own self healing system, so I'll let you know when something comes up.

[Mastiff]

OK, I'll wait. Btw please make new posts for these, or I won't see the update. I need the mail warnings when I'm all over the house like a tornado, packing stuff up.

[krambriw]

This version is now running in my system


Also a possible reason to why events stopped coming in: I saw the same and it was ONH that could not send any longer, the ONH log was filled with

2015.10.28 05:17:35.685 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889282, TID16)
2015.10.28 05:17:35.689 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889283, TID16)
2015.10.28 05:17:35.693 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889284, TID16)
2015.10.28 05:17:35.696 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889285, TID16)
2015.10.28 05:17:35.700 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889286, TID16)
2015.10.28 05:17:35.703 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889287, TID16)
2015.10.28 05:17:35.707 WARNING:TCPCommandPort could not send event, TCP stream blocked (TCPCommandPort 12889288, TID16)

A restart of ONH and events started to pop in again

So this can happen on the ONH side, I believe it is because running out of ports/resources or whatever

[Mastiff]

Thanks, downloaded. And I see that stuff too, now that I look in the ONH log. So I'm guessing that's what happens when EG stops getting events. So is it possible to find out why this happens? Should I contact Stefan?

[krambriw]

I think I now have the same problem as you experience and I am somehow making ONH blocking the sending...

[Mastiff]

Here it breaks:

Code: Select all

17:47:55   socket.error, reconnect to host
17:47:55   Connection with NetHomeServer TCP interface failed
17:47:55   OpenNetHome.Connection with NetHomeServer TCP interface failed
17:47:55   Keep Alive thread ended
17:47:56   connecting to host
17:47:56   OpenNetHome.Connection with NetHomeServer TCP interface established
17:47:56   Keep Alive thread started

Code: Select all

17:49:55   socket.error, reconnect to host
17:49:55   Connection with NetHomeServer TCP interface failed
17:49:55   OpenNetHome.Connection with NetHomeServer TCP interface failed
17:49:55         Keep Alive thread ended
17:49:56   RFXtrx.Type: Viking 02811 id: 41216 ' temperature: +20.8 deg C signal: 5 battery: 9'
17:49:56            **********************************
17:49:56            **Ny temperatursensor i RFXtrx-en**
17:49:56            **********************************
17:49:56            ************* 41216 *************
17:49:56            **********************************
17:49:56   connecting to host
17:49:56   OpenNetHome.Connection with NetHomeServer TCP interface established
17:49:56   Keep Alive thread started

But nothing happens after that, I just get a mail warning about those every two minutes or so (since I have deactivated my self-healing). So now I'm doing a manual heal. Oh, and the ONH log is full of those TCP Stream blocked things.

[Mastiff]

Had to exit EG completely this time, along with ONH.