Millions of 'Skipping NAN' warning messages

[rjwills28]

We have recently noticed that our logging service has been flooded by the following warning messages:

2024-10-08 14:12:46 WARN OptimizedWithLastSample:113 - Skipping NAN

The number of messages is on the order of millions:

Both the Optimized and OptimizedWithLastSample post processors log this message if a value in NAN:

epicsarchiverap/src/main/org/epics/archiverappliance/retrieval/postprocessors/Optimized.java

Line 83 in f9b8cd9

Logger.warn("Skipping NAN");

)

Is this log warning really necessary? Other post processors just ignore NANs and only add the event if they are not NANs. e.g.

epicsarchiverap/src/main/org/epics/archiverappliance/retrieval/postprocessors/StandardDeviation.java

Line 40 in f9b8cd9

if(!Double.isNaN(val)) {

[jacomago]

Workaround suggestion for now is to ignore those classes logs, with something like:

For log4j

log4j.logger.org.epics.archiverappliance.retrieval.postprocessors.Optimized=OFF

For log4j2

<Loggers>
    <Logger name="org.epics.archiverappliance.retrieval.postprocessors.Optimized" level="off"/>
</Loggers>

[jacomago]

I wonder why you are getting this though... Do you have a PV producing a lot of null values, or some corrupted data? You could check the access log for the retrieval engine to see which PV/PVs are being requested to see which ones are producing these logs.

[rjwills28]

@jacomago thank you for the suggestion - we'll have a go at configuring that.

I think this was a unique case where a bunch of PVs report NAN when there is no beam and a user must have been running requests in an automated script causing a constant stream of queries for these PVs. We haven't noticed such an issue before.