Hello,
First of all I wish you a happy new year 
I’m trying to build a dedicated server for Grafana (with the last 11.4 version) under Oracle Linux 8. I successfully installed Grafana on this server and downloaded the latest version of the openHistorian-grafana datasource plugin (2.1.0), which I placed in the appropriate directory. I configured the plugin as specified in the documentation, and it returns an OK result when I click the ‘Save & Test’ button.
However, when I try a simple query to actually test its functionality, it returns the following error:
Surprisingly, when using the query inspector, I notice that the traceId field of the query is set to ‘Undefined’, even though the query does contain data :
Could this be an incompatibility between Grafana 11.4 and the Datasource plugin 2.1.0? Or perhaps a configuration setting I might have missed or misconfigured?
Any help would be appreciate
Regards
Edit 1 : the issue is the same with a simple “Time Series” visualisation : the visualisation show an error due to this ‘undefined’ traceID that can not be parsed. The request answer contains the data, but the visualisation doesn’t work :
Edit 2 : I’ve tried with a nightly build openhistorian version (2.8.441.0) for the historian part, and the same OL8 standalone Grafana server : it’s now working well. I still have the “query error” message due to the ‘undefined traceID’ ; but the graph is now shown :
I now think that my previous historical version of openhistorian (2.8.284) is not compatible with Grafana 11.4, even if I use the last openHistorian datasource plugin.
I noticed that for functions Evaluate() and Slicesum(), the interval value ‘0’ is no longer allowed; the smallest accepted value is ‘0.001’. I have therefore adjusted my queries accordingly for testing. I don’t know if this change is the cause, but I have observed that now some queries implementing these functions take much longer than with the version I am currently using (which utilizes the Grafana service integrated into openHistorian, version 9.5.1). Some even result in a ‘gateway timeout’ error message. More standard visualizations (displaying frequency or signal magnitude/amplitude), which do not involve these types of functions requiring the notion of intervals, display within a similar timeframe to version 9.5.1.
here is an example of a request concerned by this behaviour :
label(“$PMU1”,Evaluate(0.001, {Sqrt(((P1+P2+P3)^2)+((Q1+Q2+Q3)^2))}, P1=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IA Active Power%’ AND Description LIKE ‘%$PMU1%’; P2=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IB Active Power%’ AND Description LIKE ‘%$PMU1%’; P3=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IC Active Power%’ AND Description LIKE ‘%$PMU1%’; Q1=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IA Reactive Power%’ AND Description LIKE ‘%$PMU1%’; Q2=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IB Reactive Power%’ AND Description LIKE ‘%$PMU1%’; Q3=FILTER ActiveMeasurements WHERE Description LIKE ‘%-IC Reactive Power%’ AND Description LIKE ‘%$PMU1%’))
This is another one :
label(“$PMU1”,slicesum(0.001,FILTER ActiveMeasurements WHERE Description LIKE ‘% Active Power%’ AND Description LIKE ‘%$PMU1%’ ORDER BY Description ASC));
For the last one, requesting 30 minutes data takes 30 seconds to update the graph. With 1 hour data request, the graph show a “Gateway timeout” error after running for 20 seconds. Performance issue ?