[Stack Monitoring] Support for error documents #4011
Labels
Integration:elasticsearch
Elasticsearch
Integration:kibana
Kibana
Integration:logstash
Logstash
Team:Infra Monitoring UI - DEPRECATED
Label for the Infrastructure Monitoring UI team. - DEPRECATED - Use Team:obs-ux-infra_services
v8.5.0
Summary
When agent fails to collect metrics from the stack packages an
error.message
field will be populated with the reason, and the document will be indexed in the corresponding data stream likemetrics-elasticsearch.stack_monitoring.cluster_stats
.We should
look into adding support for these documents so we can surface them in the Health api(already added), and also verify that error documents stored in regular indices do not break queries.Two initial options proposed:
error.message
in all datastreams. Note that the error documents may be fetched by a query and cause issues ? We don't really know because metricbeat stores these error documents separately from the monitoring indices so we never fetch themstore the error documents in a separate data stream, with an ingest pipeline ? that would replicate the model we have with metricbeat today: legit data in.monitoring-*
and error data inmetricbeat-*
With the agent/package setup, error docs land in the same data streams so we'll work on adjusting queries to account for that (first option).
error document example
The text was updated successfully, but these errors were encountered: