Environment:
Elasticsearch: 7.12.0
Metricbeat: 7.12.0
Grafana: 7.5.1
I just upgrade my Metricbeat to 7.12.0 following Elasticsearch 7.12.0, using the same dashboard from ELK stack version 7.10.1, the specific dashboard for event.dataset: cpu with specific hostname broken that can’t display the graph. Then I downgraded the metricbeat to 7.10.0, the graph appear again.
I saw in the Kibana Discovery, the data still ingested but in grafana not displayed. BTW, I use Grafana 7.5.1.
I checked with query inspector, the result is empty…
Is somebody have the same issue?
Regards,
Fadjar Tandabawana
The same here.
We was using grafana with ELK 5.6 for years. Alerts, graphs, etc.
Now we installed a new cluster with ELK 7.12 while keeping the same grafana instance (7.4.3).
We are now using ILM policy in elasticsearch so index are rotated based on a policy that determines sizes, age, etc. This makes index names to be in the format: my_index-00001, my_index-00002, and so on…
We configured grafana datasource index pattern as my_index-* and also my_index, since there is no way to set a pattern based on numeric increment but just date based. When we try to build a new basic graph, Grafana can see the fields but time series return with 0 count. Even using the same existent graphs that uses the old ELK, after change its datasource to ELK 7.12 cluster (with the same data structure) it fails the same way.
Does this have someting to do with ILM namespace.?
