Hi all,
I’m running openHAB 3.1.1 on a Raspberry Pi (openHABian, Raspbian Buster) and recently experienced a very strange issue.
Symptoms
- Network became extremely slow after openHAB startup.
- Ping from LAN clients to my UniFi gateway increased from <1 ms to 800–1300 ms.
- Internet access became noticeably slow.
- Stopping openHAB immediately restored normal network performance.
Troubleshooting
I initially suspected:
- Raspberry Pi hardware
- SD card corruption
- Ethernet cable / switch port
- Homematic binding
- Hue binding
- Amazon Echo Control binding
- Rules
However:
- Same behaviour occurred on two different Raspberry Pis.
- Different cable and switch port made no difference.
- Disabling Rules did not help.
- Stopping Homematic did not help.
- Stopping Hue did not help.
- Stopping Amazon Echo Control did not help.
- Stopping UPnP Discovery did not help.
Interesting findings
top -H showed the most active threads as:
- SocketListener(…)
- JmDNS(…)
The openHAB Java process consumed significant CPU time, and the activity seemed related to mDNS/JmDNS.
I then stopped the following bundles:
bundle:stop 166
bundle:stop 182
bundle:stop 188
where:
- 166 = Configuration mDNS Discovery
- 182 = REST mDNS Announcer
- 188 = mDNS Service
After testing the bundles individually, the issue appears to be related specifically to:
188 = openHAB Core :: Bundles :: mDNS Service
When bundle 188 is stopped, network latency immediately returns to normal and remains stable.
Environment
- openHAB 3.1.1
- openHABian
- Raspberry Pi
- UniFi Cloud Gateway
- Homematic
- Hue
- Amazon Echo Control
- InfluxDB
- Grafana
Questions
- Has anyone seen similar behaviour from the mDNS Service bundle?
- Are there known issues with JmDNS and UniFi networks?
- Could IPv6 or a recently updated network device trigger such behaviour even though the openHAB installation itself has not changed?
- Is this a known issue that has been fixed in newer openHAB releases?
Thanks!