The roll-over policy was based on this index name, but this configuration is no longer working in the newer version (8.7.1). Can anyone tell me how to create an APM index in Elasticsearch based on the APM client event name in APM server version 8.7.1?
@mralians in 8.x, APM Server writes documents to data streams with a specific format: <type>-<dataset>-<namespace>. At the moment, the only thing that can be customised is the namespace. Everything else is based on the data.
Custom application metrics are sent to service-specific data streams.
Can you tell us a little bit about your reasons for splitting indices by service.name? Is it so you can keep metrics for some services longer than others?
In the previous version, I had APM indices based on the processor.name, for example, apm-transaction-2022.03. Based on this index name pattern, I could define rollover policies, which allowed me to obtain different APM indices based on the processor.name with different rollover policies.
This is not currently possible. It is in our backlog, and there have been some recent changes to make it possible. I can't say when it will happen, but we do intend to add support for namespacing data streams.
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.