Possible memory leak in netdata agent v1.37.0-115-gdb0eb4556 on RasPi 3. Help needed to Fix

Since a few days netdata on Raspberry Pi 3 DNS server throws alert for less than 16% RAM free. I checked the metrics and found netdata agent to be suspect

Before that issue netdata was using about 200-300MB RAM. I changed some settings to reduce resource usage, but netdata is still filling up the whole available RAM eventually. It just takes longer now that I set update every = 5.

Are there any known plugins that might cause that high memory usage? Any other way to hardcap netdata RAM usage to a defined amount? It is uploading all metrics to the cloud anyways .

Please help asap. For now, I need to restart netdata service every 6 hours or the system runs out of RAM.

edit: See how it constantly increases RAM usage but not stopping when it reaches the dbengine limit of 128M as defined in my config. This abnormal unusual behavior.

Using these settings:

      update every = 2
      dbengine page cache size MB = 16
      dbengine disk space MB = 128
      dbengine multihost disk space MB = 128

Can you please scroll down to the “Netdata Monitoring” section and look at the “dbengine memory” chart? Add a screenshot here.

Actually, show that chart for as long a retention as you have, it will help us.

Here is the requested chart. Data only avail for last 2 days. I will also monitor further if the today implanted settings of dbengine multihost disk space MB = 128 will change the mem usage.

It was the first time in over a year of netdata usage that it filled the memory. I also ran the update script but I already run the latest version.

Ok, we’re on it. A much improved dbengine version was just released in the nightlies, it’s what’s causing this issue. We think it has to do with some rounding on 32-bit machines, we’re investigating.

Hi. Fixed in netdata#14257. Will be merged today.