I've used htop to list the processes.

I've changed the metric retention settings as follows:

Data: Double, Counter

Raw data 1 Minutes
One minute data 6 Hours
Five minutes data 10 Days
One hour data 90 Days
Six hours data 1 Years
Twelve hours data 5 Years
One day data 10 Years

Others

Binary data 180 Days 
GPS data 180 Days

I've got 4 nodes with 3 sensor values each (humidity, temperature, battery voltage) which send every 5 minutes to a MQTT gateway.

I've got no binary data.

The oldest data are three weeks old, the backup file is about 140kb.

I'm using the follwing java version as descirbed in this tutorial:

VM vendor Oracle Corporation
VM name Java HotSpot(TM) Client VM
VM version 1.8.0_151-b12
Java home /opt/jdk1.8.0_151/jre

When I load my dashboard (3 grouped sensor graphs) my rpi cpu usage is on 100% for some minutes - even if the dashboard is fully loaded.

Additionally I'm wondering if it's normal that mycontroller runs approx 36 java processes at once - most of them have 0.0% cpu usage.

Further more I've found some errors in my log:

2018-01-06 14:34:30,369 ERROR [Acme.Utils.ThreadPool(2)-PooledThread: Acme.Serve.Serve$ServeConnection@823c08] [org.mycontroller.standalone.metrics.engines.McMetricEngine:262] Exception,
java.sql.SQLException: Could not perform raw query for SELECT sensorVariableId, MIN(min) AS min, MAX(max) AS max, SUM(avg*samples)/SUM(samples) AS avg, SUM(samples) AS samples, 1498136632009 AS timestamp FROM metrics_double_type_device WHERE sensorVariableId=101 AND timestamp > 1498136332009 AND timestamp <= 1498136632009 GROUP BY sensorVariableId
	at com.j256.ormlite.misc.SqlExceptionUtil.create(SqlExceptionUtil.java:25)

I've just updated to version 1.1.0 but the behavior was similar with 1.0.0.