flooded network

[Mib 15]

Hello Forum!

 

How do I find the reason for a flooded Network?

Can I see somwhere in any logs who is responsible for flooding?

 

Thanks

[akatar 208]

do you have a flooded network?

the usal suspects are battery operated devices with an almost empty battery

[petergebruers 1,268]

I know these... I am not sure if there are any other options

• Power down nodes, stop scripts in a controlled way and see what happens.
• Contact [email protected] and ask to have a look at log files that are not accessible to users.
• Contact a reseller and ask his price to survey your network. Some have tools to sniff your network
• If you have got plenty of time and want to learn more about Z-Wave... Buy one of these and ask assistance from other users (like me) to interpret the results: 

  Please login or register to see this link.

[Mib 15]

@

Please login or register to see this link.

Good question. But I think so.

I have about 100 Fibaro modules (FIBEFGS-222, FIB_FGS-221 , FIB_FGS-211 , FIB_EFGS-212, Rev 02,FGR-222 , FIB_FGWPF-101, FIB_FGK-101, FIB_FGMS-001, FIB_FGD-211, FIB_FGRM-222, Rev 02,FGR-222) in use.

I also have some non Fibaro modules (NODEMSP3101, tapHome Z-Wave, GWRENS310-F, Aeotec Multi-Sensor 6 ZW100-C,  Aeotec  (EU) DSD37-ZWEU, Netatmo Weather Station, Rain Gauge for the Weather Station, Additional indoor Module) in use.

My HC2 is equpped with 4 GB RAM  and an extra antenna.

Normally the system runs quite fine. But at least twice a week I have the problem that I get suddenly dead nodes all over the place. Nothing works anymore.

Then I walk down to the basement where the automatic circuit breakers are and  disconnect nearly all the Fibaro modules from the mains. Then I reboot the HC2 and reconnect all again.

After this process when the HC2 is ready again everything is working.

 

This procedure I got to know when we had a power failure in the building during the night. In the morning I had dead nodes over the whole place.

I was told that this is normal when you have more than 40 devices. All 100 devices can not connect at once to the Home Center and will flood the network. This has to be done groupwise step by step.

 

So I assume that I get a flooded network somehow when dead nodes all over the place pop up.

But who is responsible for it? I have no batteries in use, they were replaced by Switching Power Supplies except Netatmo modules.

 

@

Please login or register to see this link.

I tried to find the device in that way.

Main Problem is that I can not reproduce the issue of the flooded network.

This happens in the morning, during the day, in the evening and at night. No matter if someon is home or not.

Last November when we were in Africa I recognized that the Venetian blinds were closed all the time after one week. My son had drive to the house and shut down everything and reboot the Home Center and power on everything. The next 3 weeks no issues ....

You think Support can see the Problem in the logs? Worth trying it.

Well there is a reseller around here. I give him a call this afternoon.

Plenty of time? Sonds difficult.

 

 

[petergebruers 1,268]

15 minutes ago, Mib said:

Plenty of time? Sonds difficult.

It is difficult... I'd only recommend it if everything else fails...

[Mib 15]

OK reminds me a little  to PRTG and finding some villain ...

[AutoFrank 372]

2 hours ago, Mib said:

I was told that this is normal when you have more than 40 devices. All 100 devices can not connect at once to the Home Center and will flood the network. This has to be done groupwise step by step.

 

Hi @Mib

This is not my experience.

I have about 110 devices and if I have a power outage they all come back online without issue.

My HC2 has the standard RAM and is mostly stable.

 

Your weekly dead nodes...

I suspect you have a device (or maybe a couple) that are intermittently flooding the network that a device reset seems to fix.

I have had a couple of situations like that and the tool that @petergebruersreferenced as helped me in the past identify culprits

It's a slow process and requires some patience. The only other alternative I can think of is to get (pay) a pro installer to place a CIT sniffer on your network for a while and find the device for you. 

 

 

 

[akatar 208]

2 hours ago, Mib said:

 

 

This procedure I got to know when we had a power failure in the building during the night. In the morning I had dead nodes over the whole place.

I was told that this is normal when you have more than 40 devices. All 100 devices can not connect at once to the Home Center and will flood the network.

 

 

 

 

b*llshit, who told you that?

 

 

On ‎14‎-‎10‎-‎2017 at 7:12 PM, petergebruers said:

I know these... I am not sure if there are any other options

•  
• If you have got plenty of time and want to learn more about Z-Wave... Buy one of these and ask assistance from other users (like me) to interpret the results: 

  Please login or register to see this link.

Hey thanks for the sniffer tip, that will be the next thing i buy

sometimes it's easy:

 

(copy past from domoticaforum.eu)

 

Aaaaaand the winner is .... [drum roll] .... Calex LED lamps! I haven't forgotten about posting this:

Please login or register to see this link.

, but I did forget it was about Calex lamps when I bought those nice LED filament lamps. Made in Holland, all certified (

Please login or register to see this link.

) so what could possibly go wrong, especially since I'm no longer using X10 ?!

Well, what could go wrong is erratic power consumption. The electric Load jumps around erratically between 2 and 9 Watt, and everything in between. Not only does this pollute the mains line and the radio spectrum, it also makes the Qubino module flood the Z-Wave network with Load updates, because per default Qubino modules report load change differences larger than 10%. Now perhaps Qubino should have implemented a rate limiter for Load update packets to prevent Z-Wave network congestion, but at least they're not the root cause. And with such a large bandwidth, even changing the 10% to 100% doesn't make much of a difference. The only change that does help, is disabling it by setting it to 0%.

[petergebruers 1,268]

4 minutes ago, akatar said:

(...)

Hey thanks for the sniffer tip, that will be the next thing i buy

I bet you won't regret it, because it will at least tell you which device talks to which device. At the moment, it does not decode traffic in great detail, not like the official tools. But having some clue about how much traffic is going on helped me a lot... If you suspect radio interference (which, in my opinion, is rare...) this tool cannot help you, because it tries to decode a Z-Wave packet and if it does not look like Z-Wave then it does not report anything. What it can do, is report packets that have bad CRC and that gives an indication of interference and collisions.

[Mib 15]

@

Please login or register to see this link.

Well I was told this by this company

Please login or register to see this link.

It worked. If it is not true I'm sorry to write it here. Fact is the power failure was at 3:12 for 3 minutes in the morning. At 7:00 I still had dead nodes all over. Multiple rebooting the Home Center did not change the situation. At 11:00 I called the company to ask for help. They told me to shut down evertihing connected to Fibaro. Then to reboot the HC2 and power on groupwise. This worked - so I never doubted it.

[petergebruers 1,268]

@Mib If your HC was powered and it was doing a polling cycle, while a large part of your modules did not have power, you get lots of dead nodes, because, well, they are actually dead... I posted some explanation in response to @Sankotronic said he had issues with a certain version of HC2 - but the explanation was his UPS + settings of HC2... Does this sound similar to your problem?

 

 

 

4 hours ago, Mib said:

This procedure I got to know when we had a power failure in the building during the night. In the morning I had dead nodes over the whole place.

I was told that this is normal when you have more than 40 devices. All 100 devices can not connect at once to the Home Center and will flood the network. This has to be done groupwise step by step.

 

I cannot confirm this with my 100 physical devices (about half of them mains powered)... I can power down / power up "my house" without getting dead nodes. I'm not saying it could not happen... I have lots of devices that do no power reporting, and I do not start a lot of scripts that try to send to lots of modules at the same time. I am not saying power reporting causes trouble, but suppose you un-power, then power 13 wall plugs at the same time, they'll report power at about the same time. But wait! I've got 13 Fibaro Wall Plugs, and no dead nodes! Nah, that does not crash Z-Wave on its own.

Edited October 16, 2017 by petergebruers

[Mib 15]

I wrote the support a mail. They asked me for changing the firewall settings. So hopefully they will have a look into the home center.

i just have 4 Motion Sensors and one window/door sensor. All working without batteries. 

So the meshed network should be ok on basement, groundfloor, upstairs. 

I had just once the complete building without electricity. And there I had this problem. 

 

[petergebruers 1,268]

@Mib I wonder if your HC2 still had power for some time... If you have 3-phase it is possible that 1 phase has power but not the other 2.

 

This problem is intriguing, so especially for you I did a test:

• I powered down the house (trigger RCB).
• Waited at least 15 minutes
• Restored power
• Wait 2 minutes
• No dead nodes
• Turned on all the lights in the house...
• No dead nodes

...

[Mib 15]

Thank you for testing !

So you got no dead nodes. You are a lucky one.

 

I received a mail from the Support Team.

Please login or register to see this attachment.

Please login or register to see this attachment.

Please login or register to see this attachment.

 

I was once told to uncheck the dead device polling and increase the time for polling.

So let us see if that ist helping.

Wake up dead nodes Skript I had before. It was runnen every 30 minutes. But I stopped it because I read it would cause high traffic. 

 

 

[petergebruers 1,268]

 

Please login or register to see this code.

 

Edit: I accidentally hit "Submit". What I wanted to say... If the script you are referring to is what I think it is, it only resets the 'dead' flag an causes no traffic at all.

 

That code snippet is part of a loop. It tries to turn of dead devices. It really tests them...

Edited October 17, 2017 by petergebruers

[Mib 15]

I once got such a script from you.

I can not check it because fibaro Support is working on my HC2 so far I see on the Firewall (IP adress from Poland)

They are creating a backup right now ...

[Mib 15]

okay this is the result of todays work on my HC2 done by Fibaro support

Please login or register to see this image.

/monthly_2017_10/support.PNG.ff324e8bacf54fff2fcce2e9ad79f4e3.PNG" alt="support.PNG.ff324e8bacf54fff2fcce2e9ad79f4e3.PNG" />

 

I keep you informed

[Mib 15]

after a reboot stucking with Starting Services

 

Please login or register to see this attachment.

 

First attemt to restore with the latest backup failed at the end with well known 503 error.

 

 

[Mib 15]

I give up. I tried 3 different backups. None is working. It always Ends up with 503

 

Please login or register to see this attachment.

Please login or register to see this image.

 

So I have to live with a non intelligent house for tonigth with lights on in garden during night ...

Hopefully Support can solve this and bring the last backup to work again.

 

[Mib 15]

It is working again. Some things are missing but I can live with that.

[petergebruers 1,268]

@Mib I was busy and forgot to post... You seem to have helped other users, by contacting support and let them discover a potential issue with our HC... That is nice. But what happened after that? You had dead nodes and now you suddenly seem to have "things missing"?