[rrd4j] Different storage size on same configuration?

Add-ons Persistence Services
1

I’m using rrd4j for my loggings and have setup the rrd4j.cfg for my needs like:

PreisLogger.def=GAUGE,90,0,3,60
PreisLogger.archives=MAX,.5,1,1440:MAX,.5,5,2016:MAX,.5,15,2668
PreisLogger.items=E10_1,E10_2,E10_3,E5_1,E5_2,E5_3,Diesel_1,Diesel_2,Diesel_3,Temperature,Light_TH_Temperature,Light_TH_Humidity

I would have expected all .rrd files to be of the same size, however the first 9 (E10_1 to Diesel_3) have a size of 49412 while the other are 28280.

When checking with REST API the datapoints for the last 24 houres I get 1440 datapoints (as expected from the archive 1 setting), for the one with a smaller filesize I get only 360 datapoints. That would be a datapoint for every 4 minutes only.

Using DEBUG logging I see that values with bigger filesizes get 2 entries every minute like:

21:08:02.516 [DEBUG] [rsistence.rrd4j.internal.RRD4jService] - Stored 'E10_1' with state '1.2889999999999999236166559057892300188541412353515625' in rrd4j database (again)
21:08:02.722 [DEBUG] [rsistence.rrd4j.internal.RRD4jService] - Stored 'E10_1' with state '1.289' in rrd4j database

while the other only get:

21:09:03.634 [DEBUG] [rsistence.rrd4j.internal.RRD4jService] - Stored 'Light_TH_Temperature' with state '19.7' in rrd4j database

Checking a chart for an item with the smaller filesize I see that a changed value is kept for at least 4 minutes, although the chart (n3-line chart in habpanel) seems to have points for every minute.

I did figure a problem with my .cfg, which was initially build for fuel-prices (E10_1,…). The max and min value are set to 0 and 3. The later logged temperatures and humidities are logging values beyond those. Would that be the reason for the different filesize? The observations of @binderth in here do not go into that direction.

I’m just curious and want to understand!

2

Did some more research. The smaller file size exactly fits the settings for the default-numeric archive setting. A check with the REST API for the last 4 houres shows values for every minute in those cases.
So the exact question is now: Why does rrd4j use the default archive setting although a different setting was desired?
It could be that the desired setting doesn’t fit ( desired to store a value below min or above max) or the item has been set into the rrd4j.persist before it has been added to the rrd4j.cfg.

3

how can you check datapoints via REST?

4

I have installed “REST Documents” via PaperUI.
Having that you can either use the REST API UI or do webcalls like:

http://[OpenHABServer]:8080/rest/persistence/items/[ItemName]?serviceId=rrd4j&starttime=2018-03-31T00%3A00%3A00.000Z

With all data within "[ " “]” from your system, rrd4j as the name of the persistence service and a starttime as you see, I ommited the endtime which will default to now.

However checking datapoints in rrd4j you need to understand that database. When checking my rrd4j as posted above checking for a timeframe of up to 24 houres I would get the points from archive 1, beyond that up to a week archive 2 and longer then that archive 3.

5

As discussed in Here, the solution was: Do not use spaces in the .cfg files and (probably) do not store values outside the max and min. Big thanks go to @binderth