Open weltenwort opened 1 month ago
Changing the values
or adding/removing buckets can cause the problem to disappear or reappear. So it seems to be very dependent on the specific metric that the change point runs on.
Pinging @elastic/ml-core (Team:ML)
Elasticsearch Version
8.16.0-SNAPSHOT
Installed Plugins
No response
Java Version
bundled
OS Version
Linux 6.5.0-1024-gcp #26~22.04.1-Ubuntu SMP Fri Jun 14 18:48:45 UTC 2024 x86_64 GNU/Linux
Problem Description
When run on some sets of documents the
change_point
aggregation throws anIllegalArgumentException
. The Observability Logs UX team is trying to use the aggregation to detect change points in log documents. I was unable to detect a pattern to the failures, though, as slight modifications to the buckets cause it to disappear or re-appear:Steps to Reproduce
I originally encountered this when running on millions of log entries, but I could reduce it to this synthetic scenario:
Logs (if relevant)
No response