New Problem with KNX-Binding: close connection - maximum send attemps

Add-ons Bindings
, ,
21

this will mostly be a networking problem on the virtual switch of the VM host … also don’t understand some configuration here why people still use devices topology address is like working with channels instead of items … also my config is

Bridge	knx:ip:tunnel "Knx tunnel" [
	ipAddress="[my tunnel ip]",
	type="TUNNEL"
]

Bridge	knx:ip:router "Knx router" [
	type="ROUTER"
]
22

I already suspect the network because the problems only occurred after the update and restart of the host server.

also don’t understand some configuration here why people still use devices topology address is like working with channels instead of items … also my config is

I do not understand that.
My configuration is

UID: knx:ip:KNXGateway
label: KNX/IP Gateway
thingTypeUID: knx:ip
configuration:
  useNAT: false
  readRetriesLimit: 3
  ipAddress: 192.168.10.112
  autoReconnectPeriod: 60
  localIp: 192.168.10.113
  type: TUNNEL
  readingPause: 50
  localSourceAddr: 0.0.0
  portNumber: 3671
  responseTimeout: 15
location: RD-KG-Verteiler
23

What is that for a syntax - also in the manuals. Is this the configuration syntax for openHAB before version 3?

24

well i was pointing to this below
please read https://www.openhab.org/addons/bindings/knx/#device-things

Thing device dimmaktor_1_2_25 [
		address="1.2.25",
		fetch=false,
		pingInterval=300,
		readInterval=3600

then your config should look like this for UI config

UID: knx:ip:6e68bf5e4d
label: KNX/IP Gateway
thingTypeUID: knx:ip
configuration:
  useNAT: false
  readRetriesLimit: 3
  ipAddress: 192.168.1.blabla
  autoReconnectPeriod: 60
  type: TUNNEL
  localSourceAddr: 0.0.1
  readingPause: 50
  portNumber: 3671
  responseTimeout: 10

Do not out set Network address of the local host ip address.
I just made a test on a debian running openhabian script with ooenhab 3.3 with the config trough UI listed above and the VM is running on xcp-ng and no disconnect.

I believe there was an update to hyper-v that broke some stuff try reverting to a previous update of the windows server to check

25

that is still file based configuration Wich i still prefer

26

Check if the Windows NIC is allowed to go to sleep. Also check the Firewall-Log of Windows.
I would suspect something along those lines…
What NIC (Vendor & Model) do you use in the Windows machine?

27

Ok, which file? My thing.json looks different.

28

Check if the Windows NIC is allowed to go to sleep.

Ok, I changed the settings. But a server may not switch off its NICs.

Also check the Firewall-Log of Windows.

The local firewall is switched of.

What NIC (Vendor & Model) do you use in the Windows machine?

two Intel I350 Gigabit Network Connection paired as NIC team

30

what are you running as VM ?

31

Have you tried use nat true?

32

Linux CentOS 8 and openHAB 3.1.0 as a service.

Do you mean to use only one device for all KNX devices?

There were the following updates:

  • 2022-07 Servicing Stack Update für Windows Server 2016 für x64-basierte Systeme (KB5016058)
  • 2022-07 Cumulative Update for Windows Server 2016 for x64-based Systems (KB5015808)

I’ve thought about rolling back the updates but they don’t seem relevant to me and I haven’t tried yet.

34

Yes but without any difference.

35

what about this https://docs.microsoft.com/answers/questions/941917/networking-issue-from-kb5015874-on-server-2012-r2.html

it’s the second update for your windows server 2016
KB5015808

36

Good idea. Are you sure? But it says KB5015874 for Windows server 2012.
I will look for deinstallation at Windows server 2016…
… done but nothing changes.

I have run the virtual machine on my local computer (Windows 10) aber the problem ist still the same.

37

ok then its something with the network switch in your house or CentOS. can you spin a debian with openhabian VM and just add the tunnel with one knx channel

38

Well, i ment that the NICs should NOT be allowed to go to sleep :wink: Unfortunately, sometimes the drivers of te NICs do set this to allow, despite the fact that it is a server.

Good.

The HW is fine. Your problem might be related to the teaming though…I came across weird connection problems on windows server a couple of times, which had been related to NIC teaming. Never really figured out what the root cause was, but getting rid of the teaming solved the connection problems in all cases. Maybe you wanna give this a try, too.

39

I got exactlly the same problem on this constellation Hyper-V / openhab 3.3 running in linux vm (ubuntu)
Anybody fixxed this issue on server 2016 ?

40

Good idea, but i don’t want to deactivate the team because it’ my server and the test running on my Windows 10 client Hyper-V without NIC-team creates exactly the same error.

So I gave up and reactivated the windows installation from the test phase, transferred the openHAB configuration and it has been running stably for days - easy.

41

You’re not alone!

Some update to Windows / Hyper-V definitely broke something.
I ran Home Assistant (yes, HA, not OpenHAB) with a KNX integration flawlessly for 2 years, on a Hyper-V VM running in Windows Server, with I350 NIC.

Then, about 2 months ago, without me changing anything in the systems, I started getting those “Connection Closed” notices and all KNX switches became unresponsive.

If I do a clean install of HA+KNX to a new Hyper-V virtual machine, it still doesn’t work (Connection Closed).
If I restore my existing HA+KNX configuration to a VMWare VM, or RPI… it works.

So, here we have 2 cases where HyperV + KNX stopped working all of the sudden. One running in OpenHAB, the other in Home Assistant.

I tried changing all settings in the Network Adapter, but didn’t get it to work.
Think I’ll try a different NIC.

Did you manage to get your KNX integration to work again?

42

there was ( should be fixed in the meantime as far as I understand the end of the thread ) a patch under suspect causing this: