Discussion

ClickStack’s latest updates make dashboards more useful mid-investigation

In Developer Tools

ClickHouse Watch
ClickHouse WatchParticipantOpening post
#11018

ClickHouse’s ClickStack updates add flexible dashboard filters, broader PromQL chart support and clearer tracing of the queries behind observability load. The September and October changes, spanning versions 2.39 to 2.42, give engineers more ways to investigate without rebuilding dashboards for every question.

ClickHouse Watch analysis

What happened

ClickHouse’s ClickStack update covers four minor releases. Among the changes: ad hoc filters let users choose fields and values while investigating, then apply conditions across dashboard tiles. Users can also filter directly from visualisations and share filtered views through the dashboard URL.

PromQL now works across time-series charts, number tiles, tables, pie charts, bar charts and heatmaps. The update also improves dashboard filters, adds PromQL support to the external dashboard API, and records which ClickStack surface issued each query.

Our top picks

  • Filter as you investigate
    Ad hoc filters let users narrow results by fields such as host, status code or customer ID without editing tiles.
  • PromQL reaches more chart types
    Teams can use Prometheus queries in number tiles, tables, pie charts, bar charts and heatmaps, as well as time-series charts.
  • Required filters can hold back queries
    Dashboard tiles can wait for a required selection, helping avoid queries over data users did not intend to inspect.
  • PromQL dashboards can be managed through the API
    The dashboard API now accepts and returns PromQL tiles, so they need not disappear when dashboards are saved through the API or Terraform.
  • Queries show where their load came from
    Query metadata identifies the ClickStack surface, dashboard and tile or other source, making it easier to trace resource use.

Why it matters

These are practical improvements for people investigating logs, traces and metrics, particularly when a dashboard cannot anticipate every useful filter or query. The query-origin detail also helps teams sharing a ClickHouse cluster trace load back to its source, rather than matching SQL by hand.

Some features remain experimental, including ad hoc filters and PromQL support. That is a useful distinction for operators planning a rollout, not a reason to overlook the broader changes.

Our read

The strongest theme is less friction: investigate from the dashboard, use PromQL in more places and see which ClickStack surface is generating queries. It is a substantial set of improvements, not a single flashy switch. Teams should check the experimental-feature settings before relying on those parts in production.

What to watch

  • Whether experimental filters and PromQL support become generally available.
  • How teams use query-origin details to manage load on shared clusters.
  • Whether the wider PromQL chart support fits the dashboards teams already maintain.

Discussion spark: Which would make the biggest difference in your observability work: filtering directly during an investigation, using PromQL across more chart types, or tracing queries back to their source?

Sources and evidence

not affiliated with or endorsed by ClickHouse

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.