Grafana Associate common mistakes
Grafana Associate · Understand the traps and how to avoid them.
Treating time range and time shift as the same control, so a panel is inspected with an absolute range that silently overrides the dashboard's relative range and the candidate cannot explain why the graph looks empty.
Candidates practise only with the default relative range and never notice the dashboard-level picker versus the panel-level override, or that a shift moves the query window rather than the data.
What to do instead
Before the exam, deliberately build one panel with a relative range (now-6h) and one with a panel-level override and a time shift, then reload the dashboard to see which control wins. On any time-related question, first identify whether the range is set at dashboard or panel level.
Searching for a feature in the wrong place because the candidate assumes the left-hand navigation is identical across Grafana versions, then loses minutes hunting for an item that has moved into a different section.
Study material and screenshots come from different Grafana releases, so muscle memory from one version fails on the exam's version.
What to do instead
Learn the current navigation structure by name rather than position: know that administration, connections (data sources), and alerting are separate destinations reachable from the main menu, and practise locating each by its label rather than by icon position.
Saving a dashboard without checking that the data source is selected explicitly on each panel, so panels fall back to the default data source and return nothing or the wrong series.
The panel editor pre-selects the default data source, so a working query looks correct while the panel is actually pinned to whatever was default at creation time.
What to do instead
After building any panel, open the query editor and confirm the data source dropdown names the intended source, and set a deliberate default data source in connections settings so fallback behaviour is predictable.
Writing a query that returns raw unaggregated rows and then wondering why the visualization shows a single flat value or a broken graph.
Candidates learn query syntax in isolation and do not connect the shape of the returned frame (time series versus table versus instant vector) to what the panel can render.
What to do instead
For each data source, learn which query type produces a time series and which produces a table or instant value. Test the query in the data source's own query inspector or Explore before adding it to a panel, and check the returned frame type.
Assuming variable interpolation works the same everywhere and pasting a variable directly into a query without the correct syntax or quoting for that data source.
Different data sources use different interpolation formats (for example a regex-style multi-value expansion versus a comma-separated list), and candidates reuse one pattern everywhere.
What to do instead
Memorise the interpolation options available in the variable editor and match them to the data source: use the multi-value and regex formatting options when the query expects a list versus a pattern, and preview the interpolated query in the panel inspector.
Choosing a visualization type by appearance rather than by data shape, so a stat panel is fed a multi-series time query or a graph is fed a single instant value.
Candidates pick the panel that looks like the screenshot in their study notes without checking how many fields and rows the query returns.
What to do instead
Before selecting a panel type, read the query result in table view and count fields and rows. Match single latest values to stat or gauge, time series to time series panels, and categorical rows to bar gauge or table.
Leaving thresholds and units at defaults, so a gauge or stat panel shows a value with no unit or colour that contradicts the intended meaning.
The defaults are visually plausible, so nothing looks broken and the candidate never revisits the field options.
What to do instead
For every value-based panel, set the unit explicitly in the standard options and define threshold steps with base, warning, and critical colours. Verify by hovering the rendered value to confirm the formatted display.
Building a template variable whose query depends on another variable but not setting the dependency order, so the dependent dropdown is empty on first load.
Candidates add variables in the order they think of them, and Grafana resolves variables in declaration order, so a child variable queries before its parent has a value.
What to do instead
Order variables so parents are declared before children, and confirm the child's query references the parent variable name. Reload the dashboard from a fresh session to check the initial load behaviour, not just after manually changing the parent.
Sharing a dashboard link that includes the current variable selections and time range when the intent was a clean starting view, so recipients see a filtered or empty dashboard.
The share dialog offers several link options and the default preserves the current state, which is not what most sharing scenarios want.
What to do instead
Use the share dialog deliberately: choose the option that snapshots the current time range and variables only when reproducing a specific view, and otherwise strip them so the recipient starts from the dashboard's saved defaults.
Assuming a dashboard saved in a folder is visible to a colleague and discovering only at review time that folder permissions or the organisation boundary block access.
Candidates test sharing only from their own admin-level account and never verify what a lower-privileged viewer sees.
What to do instead
Learn the permission model: dashboard and folder permissions, organisation membership, and team-based access are distinct layers. Verify sharing by checking the folder's permission list and, where possible, viewing the dashboard as a non-admin user.