09.07.2020, 12:25 (This post was last modified: 09.07.2020, 12:34 by KNXUser.)
Hello Forum Users,
i have a strange problem with a Spacelynk on a customers site.
The device is running there since 2 years and worked fine all the time (HW is 2.0, FW was 2.3 all the time).
But since last week, there is the problem, that the device is restarting itself every 5 to 90 minutes.
When i look into the system massages, i can see the restarts and the time there.
There are a lot of trends recorded in the device and with any restart the data for the last periode is lost.
The device itself stands in no direct relationship with other ethernet participants and all the data is comming from the KNX connection itself.
The ethernet is only connectet, that the customer can view the visualisation and the trends on the device.
Looking at trends or the visualisation from an other device stands in no direct relationship with the restarts.
What i did was upgrading the firmware to 2.5 with no effect.
I changes the device to a Homelynk (HW 3.1 FW 2.5) with the same effect, still restarting, so i installed the old Spacelynk again.
Then i disconnected the ethernet cable over the night an in the morning there were no restarts in the systemlog, the device was running all the night long.
After connecting the ethernet again, there was the first restart after 20 minutes.
So the problem is, that something comming from the ethernet causes the device to quit its job and resart.
What could that be?
The customers site is a factory with hundreds of ethernet participants.
The only indication i have, is the "Error counter" on the system page for the network connection (see picture attached).
This counter runs up to values over 27000 and i guess this could stand in relationship with the restarts.
I believe that there was some kind of modification in the customers network (a new device connected for example ...),
causing special traffic on the ethernet which can not be handled by Spacelynk.
I am standing in close contact to the customers network administration but they do also not know what could have happened since last week.
Has anyone had a mailfunction like this bevore or does anyone have an idea, what to do to solve the problem?
I thought about using special Software like "Wireshark" to look inside the ethernet traffic but i am not sure if i could interpret well, what i would see there.
How big is the TX/RX bytes value in the interface list? Try installing Processes app and see if there's a service that's consuming a lot CPU. You can use WireShark to check if there's a flood of multicast and/or broadcast packets. Another thing to try is to change the IP address of the device if there's unicast flood.
Hello again and thanks for the quick answer to my question.
Attached is a picture with several information concerning IP.
The Prozesses App will be installed tomorrow, when i am at the custumers site again
(because till now, there is no web access for the Spacelynk).
Maybe there is allredy something to see in the new attachement.
Thanks and kind Regards
Michael
Yes this value is realy high but i can not imagine this could cause the problem.
Could this high reparts be caused because the functionality for acknowledgement of all telegrams is not activated (see attachement)?
The problem is a reboot of Spacelynk only when LAN is connected, it works fine when LAN is disconnected.
And when i look at the values, i can see that there is not a single telegram coming from IP or is goint to IP (IP RX/TX value is 0).
13.07.2020, 11:43 (This post was last modified: 13.07.2020, 11:44 by Erwin van der Zwart.)
You probably have another KNX IP interface in the installation (same line or backbone), when connecting to LAN you create the KNX loop, you can either remove that IP interface from the KNX network or disable KNX IP Features in the KNX settings, if you do the last option you need to connect to the other IP interface with ETS..
(13.07.2020, 11:45)Daniel. Wrote: Usually this is happening when there are 2 routers on the same TP network. One of which is spaceLYnk. Try disabling KNX IP features.
(13.07.2020, 11:45)Daniel. Wrote: Usually this is happening when there are 2 routers on the same TP network. One of which is spaceLYnk. Try disabling KNX IP features.
Since the load average is low the cause of the reboot is not some process. One possibility is that is a hardware error or a software error connected to the Ethernet errors. Can you try replacing Ethernet cable?
Yes, there is another IP-Router in the switch cabinet of the HVAC-Supply ... but it is there since 2 jears and my problem appeared just 2 weeks ago ... nothing has changed since 2 jears.
If there is a loop from KNX into IP and from IP back into KNX, there should be a value counted for the IP RX/TX statistic, shouldn´t it?
But the value is 0 (as it can be seen in my older attachement), so there is no telegram counted and so there is no loop, right?
Yes i can replace the flexible ethernet cable, i will be at the customers site tomorrow and will do this, as well as the installation of the Prozesses App.
Could anyone respond again to the "loop" problem, of which i think it could not be becouse of the IP RX/TX counter showing a 0?
13.07.2020, 12:03 (This post was last modified: 13.07.2020, 12:06 by Erwin van der Zwart.)
I don't think you can say that, when the system is overloaded you migh not even see the counting..
Is the other IP router connected to the same line/backbone? If yes then you always had this issue as the telegrams go from TP to IP and the other interface receives it on IP and puts it on the TP again and this process repeats infinity ...
Can you try to disable the KNX IP features to see the results..
13.07.2020, 12:11 (This post was last modified: 13.07.2020, 12:18 by KNXUser.)
Yes, the other IP-Router is in the same subline.
And yes, i can also disable the IP features in Spacelynk.
... or i activate the filter functionality from IP > local and from local > IP with no KNX adress in the filtertabels,
then the loop-telegrams should in this device not pass through.
14.07.2020, 07:43 (This post was last modified: 14.07.2020, 07:44 by KNXUser.)
I am back from the customers site ...
I changed the ethernet cable to a new one.
I disabled "IP Features" in Spacelynk.
The problem is still the same ... restarts every 10 to 120 minutes when ethernet is connected.
The KNX traffic also is still the same (with the same high repeats), see attachement.
The counter for IP RX/TX still is at 0.
What i couldnt do, is to install the Processes App, because i couldnt get an internet connection.
Can someone give me the Processes App as an *.ipk file, so i can install it lokal on the device?
Hello again and thanks to Daniel for providing the App, which is called "System load" and is a little bit different to the "Processes" App.
The values from both pages can be seen attached ... the device is nearly running in idle mode.
So i think this is the wrong path for finding the reason for its restarts.
So if there are no furter ideas, where to look and check, i have to go inside the networt with Wireshark.
When i do so, is there something on which i should have a special look?
And what can be the trigger coming out of the ethernet and cause Spacelynk to restart?
Is there some kind of fail safe mechanism inside the ethernet controller, when a special event happens, which triggers the restart of Spacelynk?
What about the failure counter on the ehernet config page, which i mentioned in my first comment and which goes up to values of 27000?
Is there a value, above which a restart is triggered?
A bad thing for the customer is, that the data for his trends are always lost, when the device restarts bevore the data is saved to the SD-Card.
Is there any possibility to make the time for storing shorter (i heared about 30 minutes is the default), to get more data to the SD-Card until this special problem is solved?
14.07.2020, 15:44 (This post was last modified: 14.07.2020, 15:45 by KNXUser.)
I will check this with the customer and will come back to you soon ... maybe it is possible via teamviewer.
Can you please answer my question, if it is possible to change the interval time for storing the trend data to the SD card?
So there would not be such a high loss due to the restarts.
