I have a dashboard panel showing GC count (jvm.gc.count.) I have it broken down by labels.name, which is how I get separate counts for young and old GCs. However the legend also shows labels.name for other documents, such as the ones with heap pool usage. So instead of young and old GC counts, I have also have eden, survivor, and old gen pools. (There are no values for those because those docs don't have jvm.gc.count.) I have tried to filter them out by including the filter (on the panel) of jvm.gc.count:* but it doesn't seem to have any effect.
One clue might be that when I use the "Explore data in Discover" link, it shows the "jvm.gc.count:*" filter, as well as the Dashboard-wide filters, but the font has a strikethrough and the data table includes the "wrong" documents. When I hover over it with my mouse, it says it's disabled. When I enable it (in Discover), the data (in Discover) is correct.
I see nothing in the panel settings or visualization config that indicates the filter can be enabled or disabled.
I think this is the best route to investigate on.
The "Explore data in discover" collects all existing filters within a visualization as they are (enabled or disabled) and propagates them to Discover.
There's probably a disabled filter somewhere in the visualization to enable. On the top left bar is there anything? Can you share your configuration?
The "Filter your data using KQL syntax" box is empty.
The dashboard has 3 controls: service name, environment, and a custom label we use to identify the app. (Each app has multiple services.)
If I go to "edit visualization" for the specific panel, horizontal axis is timestamp, vertical axis if counter rate of jvm.gc.count per minute, breakdown is top 5 value by labels.name.
If I choose advanced for jvm.gc.count, I have "filter by jvm.gc.count:*"
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.
