Grafana 8 centralizes alert management, simplifies query troubleshooting

Illustration of an alert appearing over someone's laptop
Illustration of an alert appearing over someone's laptop

After having drawn attention to itself for switching project licences earlier this year, Grafana Labs is back with major updates to its tracing backend Tempo and observability platform, Grafana.

Aiming for a better enterprise fit and all-around usability, Grafana 8 is mainly packing improved alerts and additional visualization, while also answering enterprise users’ request for fine-grained access control.

Grafana 8 Alerts are optional, combining alerts managed by Grafana and those from data sources compatible with Prometheus in a centralised view. The associated UI lets users create and edit alerting rules and provides options to generate multiple alert instances from a single rule. Compared to the legacy approach, the new alerting feature isn’t tied to a dashboard and allows managing Loki and Cortex alerting rules as well.

The newly included visualizations are mostly in beta still, and comprise a standalone histogram visualization as well as a bar chart for categorical data in single data frames. Ops folks interested in a closer look at state progression might find the newly added state timeline visualization, which displays discrete state changes over time, or the gridded status history panel useful.

Other areas of improvement include the Explore section, which now sports an inspector for “understanding and troubleshooting” queries. It is made up of a stats, query, JSON, data, and error tab, that give insight into how long queries take and how much they return, which requests were sent, and what raw data was returned. It also allows you to view and copy data and data frame structure JSON.

To help with log navigation, the Grafana team added pagination to Explore’s log component. The whole package should also have become a bit quicker, since the team has reduced the initial download size, and migrated some components to a quicker rendering library. A full list of changes and some details on the integrated data source updates can be found in the project’s documentation.

Grafana Enterprise users, meanwhile, were treated to some access control refinements. These allow extending the Viewer, Editor, and Admin org roles with capabilities to “view and manage Users, Reports, and the Access Control API itself.”

There also seems to have been a need for clarification on restrictions imposed by the product licensing. Additional documentation around counting licensed roles, role updates, and ways to get informed about current permissions and how they affect licensed roles should help in that regard.

Enterprise version 8 also comes with options to export Table panels to attach to report emails, and the capability to cache results of backend data source queries. This buffering of sorts is meant to help get the average load time for dashboards and number of duplicate queries down.

Grafana Tempo, a fairly new tracing project, was announced ready for production use as well, and mainly sports performance and stability improvements. Among other things version 1.0 includes a new endpoint for testing query frontend availability, ingester enhancements, and better WAL Replay that doesn’t depend on rebuilding the WAL.