eibd Daemon Lockup on LM5 Lite with heavy multicast traffic

[MSkjel]

Hi everyone,

We’ve observed an issue on several of our LM5 Lites where the eibd daemon locks up during periods of transient heavy multicast traffic. When this occurs, the process starts consuming CPU cycles but stops reading or writing anything to KNXnet/IP multicast.
Under normal conditions, the LM runs with a load average around 0.5, so it’s not generally under high load. The issue only appears when there’s a short burst of multicast traffic.
Once it happens, the daemon doesn’t recover on its own — the only way to restore functionality is to either reboot the LM or restart eibd manually, e.g.:

Code:

os.execute('/etc/init.d/eibd restart')

This is quite critical for us. So far, I’ve only observed this behavior on devices running the 2023 firmware version. I’ve just updated all units to 2024, but I’m wondering — has anyone else experienced this issue before? 

Thanks in advance for any insights or suggestions! 

[Daniel]

Moat likely you have some loop somewhere. Try to debug it, Install system load app to check what is consuming so much load.

[admin]

Probably it's not locking up but still processing incoming telegrams. If the burst is too big some telegrams might be lost due to the buffer running out of free space. As Daniel said, you need to figure out why and where from this burst of telegrams is coming.