> ## Documentation Index
> Fetch the complete documentation index at: https://private-7c7dfe99-parallel-read-in-order-multi-part.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

> Evaluates a prometheus query using data from a TimeSeries table.

# prometheusQuery

Evaluates a prometheus query using data from a TimeSeries table.

<h2 id="syntax">
  Syntax
</h2>

```sql theme={null}
prometheusQuery('db_name', 'time_series_table', 'promql_query', evaluation_time)
prometheusQuery(db_name.time_series_table, 'promql_query', evaluation_time)
prometheusQuery('time_series_table', 'promql_query', evaluation_time)
```

<h2 id="arguments">
  Arguments
</h2>

* `db_name` - The name of the database where a TimeSeries table is located.
* `time_series_table` - The name of a TimeSeries table.
* `promql_query` - A query written in [PromQL syntax](https://prometheus.io/docs/prometheus/latest/querying/basics/).
* `evaluation_time` - The evaluation timestamp, with millisecond or finer precision. To evaluate a query at the current time, use `now()` as `evaluation_time`.

<h2 id="returned-value">
  Returned value
</h2>

The function returns different columns depending on the result type of the query passed to parameter `promql_query`:

| Result Type | Result Columns | Example |
| - | - | - |
| vector | tags Array(Tuple(String, String)), timestamp DateTime64(S, TZ), value Float64 | prometheusQuery(mytable, 'up') |
| matrix | tags Array(Tuple(String, String)), samples Array(Tuple(DateTime64(S, TZ), Float64)) | prometheusQuery(mytable, 'up\[1m]') |
| scalar | timestamp DateTime64(S, TZ), value Float64 | prometheusQuery(mytable, '1h30m') |
| string | timestamp DateTime64(S, TZ), value String | prometheusQuery(mytable, '"abc"') |

The values are always `Float64` regardless of the type of the values in the TimeSeries table. The scale `S` of the timestamps is the scale
of the timestamps in the table, but not less than 3 (milliseconds). The time zone `TZ` is the time zone of `evaluation_time` if it has
type `DateTime` or `DateTime64` with a time zone, otherwise it is the time zone of the timestamps in the table.

The `samples` column is named `time_series` if the `TimeSeries` table has [version](/reference/engines/table-engines/integrations/time-series#schema-versioning) 2 or earlier.

<h2 id="supported-promql-features">
  Supported PromQL Features
</h2>

<h3 id="selectors">
  Selectors
</h3>

Instant selectors, range selectors, label matchers (`=`, `!=`, `=~`, `!~`), offset modifiers, `@` timestamp modifiers, and subqueries.

<h3 id="functions">
  Functions
</h3>

| Category | Functions |
| - | - |
| Range | `rate`, `irate`, `delta`, `idelta`, `increase`, `last_over_time`, `first_over_time`, `sum_over_time`, `avg_over_time`, `count_over_time`, `max_over_time`, `min_over_time`, `ts_of_max_over_time`, `ts_of_min_over_time`, `ts_of_last_over_time`, `ts_of_first_over_time`, `deriv`, `changes`, `resets`, `present_over_time`, `absent_over_time`, `quantile_over_time`, `stddev_over_time`, `stdvar_over_time`, `mad_over_time`, `predict_linear` |
| Math | `abs`, `sgn`, `floor`, `ceil`, `sqrt`, `exp`, `ln`, `log2`, `log10`, `rad`, `deg`, `round`, `clamp`, `clamp_min`, `clamp_max`, `min_of`, `max_of` |
| Trig | `sin`, `cos`, `tan`, `asin`, `acos`, `atan`, `sinh`, `cosh`, `tanh`, `asinh`, `acosh`, `atanh` |
| DateTime | `day_of_week`, `day_of_month`, `days_in_month`, `day_of_year`, `minute`, `hour`, `month`, `year` |
| Label | `label_replace`, `label_join` |
| Type | `scalar`, `vector` |
| Histogram | `histogram_quantile` |
| Other | `time`, `pi`, `absent`, `timestamp` |

**Note**: `histogram_quantile` uses linear interpolation on classic histogram buckets (identified by the `le` label). Native histograms are not supported. The `phi` (quantile level) argument must be a constant scalar. Expressions that vary per step, such as `histogram_quantile(time() / 1000, ...)`, are rejected with a `NOT_IMPLEMENTED` exception.

**Note**: `min_of(a, b)` and `max_of(a, b)` return the smaller or larger of two scalar values. Both arguments must be scalars.

**Note**: `min_of`, `max_of`, `ts_of_min_over_time`, `ts_of_max_over_time`, `ts_of_last_over_time`, `first_over_time`, `ts_of_first_over_time` and `mad_over_time` are experimental functions in Prometheus (enabled there with `--enable-feature=promql-experimental-functions`); ClickHouse evaluates them without requiring that flag.

<h3 id="operators">
  Operators
</h3>

Arithmetic (`+`, `-`, `*`, `/`, `%`, `^`, `atan2`) and comparison (`==`, `!=`, `<`, `>`, `<=`, `>=` with optional `bool`) binary operators, with `on()`/`ignoring()` and `group_left()`/`group_right()` modifiers.

Logical set operators `and`, `or`, and `unless`, with `on()`/`ignoring()` modifiers.

Unary operators `+` and `-`.

<h3 id="aggregation-operators">
  Aggregation Operators
</h3>

`sum`, `avg`, `min`, `max`, `count`, `count_values`, `stddev`, `stdvar`, `group`, `quantile`, `topk`, `bottomk`, `limitk` — with optional `by()` or `without()` modifiers.

<h2 id="example">
  Example
</h2>

```sql theme={null}
SELECT * FROM prometheusQuery(mytable, 'rate(http_requests{job="prometheus"}[10m])[1h:10m]', now())
```