Expression Browser

​

Quick access

1. Open your browser and navigate to your Prometheus server, for example: http://<PROMETHEUS_IP>:9090
2. The default view is the Query page. Enter PromQL expressions in the query box and click Execute.

​

Query box and autocomplete

• The query input supports autocomplete for metric names and functions. Toggle autocomplete via the checkbox in the UI.
• Autocomplete speeds up discovery of available metrics and reduces typos.

​

Execute queries and inspect results

• Type a PromQL expression in the Query box and click Execute.
• The UI displays matching series with labels and current values.
• Use the “Evaluation time” dropdown to evaluate queries at a past timestamp.
• Understand vector types:
  • Instant vectors: single sample per series at a specified timestamp (used by the Graph page).
  • Range vectors: sequences of samples over a time range (used by functions like rate()).

# range vector (returns multiple datapoints per series)
node_memory_Active_bytes[5m]

# instant vector (returns the current value(s) for each series)
node_cpu_seconds_total

# Example range-vector output (node_memory_Active_bytes[5m])
node_memory_Active_bytes{instance="192.168.1.168:9100", job="node"}
903811072 @1668416055.132
...
903847936 @1668416340.132

# Example instant-vector output (node_cpu_seconds_total)
node_cpu_seconds_total{cpu="0", instance="192.168.1.168:9100", job="node", mode="idle"} 187180.27
node_cpu_seconds_total{cpu="0", instance="192.168.1.168:9100", job="node", mode="iowait"} 21.29
...
node_cpu_seconds_total{cpu="0", instance="192.168.1.168:9100", job="node", mode="user"} 271.18

​

Panels

• The Query page supports multiple panels so you can run several queries side-by-side.
• Each panel has its own query box and results area — useful for comparing series or testing variations of a query.

​

Graph page — plotting queries

• The Graph tab visualizes query results over time. The Graph expects an instant vector (or an expression that evaluates to an instant vector).
• Supplying a raw range vector directly to the Graph page will produce an error:

Error executing query: invalid expression type "range vector" for range query, must be Scalar or instant Vector

• For time-based rates or derivatives, wrap a range-vector expression with a function that returns an instant vector. Common example for counters:

rate(node_cpu_seconds_total[5m])

• After evaluating to an instant vector, the Graph page will render the time-series. Use the graph time window selector (e.g., 30m, 2h) and the evaluation time control to view historical data.

rate(node_cpu_seconds_total[2m])

{cpu="0", instance="192.168.1.168:9100", job="node", mode="idle"} 0.9926666666666489
{cpu="0", instance="192.168.1.168:9100", job="node", mode="iowait"} 0
...
{cpu="0", instance="192.168.1.168:9100", job="node", mode="user"} 0.0009523809523806274

• The graph legend shows one entry per returned series (for example, different mode label values). Click legend entries to toggle series visibility.

​

Summary and best practices

• Use the Expression Browser for:
  • One-off PromQL evaluations and debugging.
  • Inspecting label sets, sample values, and behavior of functions.
• Remember:
  • The Query page is for ad-hoc evaluation; the Graph page expects instant vectors.
  • Use rate() or similar functions to convert range vectors into plottable instant vectors.
• For production dashboards and alerting, use a dedicated visualization tool and promote validated queries from the Expression Browser into that environment.

• Prometheus Querying Basics
• PromQL Language Overview