Filestream metrics might not work correctly in Kubernetes #37925
Labels
bug
Team:Elastic-Agent
Label for the Agent team
Team:Elastic-Agent-Data-Plane
Label for the Agent Data Plane team
There are some situations the Filestream input will have its metrics collection disabled in Kubernetes, that happens when:
This is somehow common when running Filebeat with autodiscover on Kubernetes, the autodiscover code might stop and start the same input, for the same files. This has no effect in the data collection, however due to a bug (#31767) on how we keep track of IDs and start/stop filestream inputs we never clean the ID registry leading to two issues:
metricsID = ""
).While the erroneous log message is annoying, disabling metrics collection is a bigger issue that requires attention.
How to reproduce
filebeat-deplyment.yml
belowfilebeat-deployment.yml
I haven't looked enough into that to fully assess how the metrics are working (or not), but when I SSH into the container and run
curl localhost:5066/inputs/ | jq
I get a valid output with metrics that are changing over time.However I haven't managed to access how correct they are. I'll update this issue once I have more information.
The text was updated successfully, but these errors were encountered: