Dashboard YAML Templates

Uptrace dashboards can be defined as YAML templates. Templates are used for built-in dashboards that ship with Uptrace and for importing/exporting user-created dashboards.

When Uptrace receives new metrics, it checks available templates and automatically creates matching dashboards.

Template structure

Every template starts with a schema: v2 header followed by metadata, then the layout of the dashboard and optional metric_monitors.

yaml

schema: v2
name: 'Hostmetrics: Overview'
description: Tracks host CPU, memory, and system load.
tags: [otel, infra]
version: v26.11.05
setup_link: https://example.com/setup
doc_link: https://example.com/docs

# Optional: minimum query interval
min_interval: 1m

# Optional: shift the query time window
time_offset: 5m

# Optional: only apply when these metrics exist
metrics_match:
  present:
    - metric: system_cpu_utilization
    - library: io.opentelemetry.instrumentation.system
      metric: system_cpu_time

layout:
  items: [] # the widgets or the groups of the dashboard

metric_monitors: [] # bundled alert monitors

Top-level fields

Field	Type	Required	Description
`schema`	string	yes	Must be `"v2"`
`name`	string	yes	Human-readable dashboard title
`description`	string	no	Summary of the dashboard's purpose, shown in the template list
`tags`	list	yes	Categorization labels (see Tags)
`version`	string	yes	Revision in `vYY.MM.DD` format (e.g., `v26.11.05`)
`setup_link`	string	no	URL to instrumentation/setup instructions
`doc_link`	string	no	URL to documentation
`min_interval`	duration	no	Minimum query interval (e.g., `1m`, `5m`)
`time_offset`	duration	no	Shifts query time window (e.g., `5m`)
`default_time_period`	duration	no	Time period the dashboard shows when it opens without a stated range (e.g., `1h`)
`tooltips_connected`	bool	no	Whether a hover on one chart marks the same time on every chart. Defaults to `true`
`max_width`	string	no	Page width: `none` (default), `small`, `medium`, `large`, or `extra_large`
`variables`	list	no	Attribute names the filter bar of the dashboard offers as variables
`metrics_match`	object	no	Metrics that must, or must not, exist before the template is applied (see Metric requirements)
`layout`	object	no	The root of the layout (see Layout)
`metric_monitors`	list	no	Metric monitors bundled with the template (see Metric monitors)

Metrics and queries

Metrics and queries appear throughout the template (tables, grid items, monitors). They follow the same pattern everywhere.

Metrics

The metrics field declares metric aliases using MQL syntax:

yaml

metrics:
  - system_cpu_utilization as $cpu_util
  - system_memory_usage as $mem_usage

Each entry takes the form metric_name as $alias. The alias (prefixed with $) is used in queries.

Query

The query field is a list of MQL clauses. Splitting across lines improves readability -- clauses are joined into a single query during processing.

yaml

query:
  - group by host_name
  - avg($cpu_util) as cpu_avg
  - sum($mem_usage{state="used"}) as mem_used
  - where device !~ "loop"

Common clause patterns:

Aggregations: avg($x), sum($x), max($x), min($x), p50($x), p75($x), p90($x), p99($x)
Rate functions: perMin(sum($x)), perSec(sum($x))
Counting: histogram_count($x), uniq($x, attr_name)
Grouping: group by attr_name, group by attr1, attr2
Filtering: where attr = "value", where attr !~ "pattern"
Aliasing: sum($x) as my_alias
Inline grouping: avg($x) as y group by state
Attribute filtering: $metric{state="used"}, $metric{direction=read}
Arithmetic: sum($a) / sum($b) as ratio

The layout key holds the root of the dashboard. The root and each group are a body: a list of items with the fields that arrange them. An item is a widget or a group, and its type selects which: omit it or state chart for a chart, a widget type for a widget of that type, or group for a group. A body holds widgets or groups, never both.

yaml

layout:
  mode: tabs
  items:
    - type: group
      title: CPU
      items:
        - title: CPU utilization
          metrics:
            - system_cpu_utilization as $cpu_util
          query:
            - avg($cpu_util)

    - type: group
      title: Memory
      items:
        - title: Memory utilization
          metrics:
            - system_memory_utilization as $mem_util
          query:
            - avg($mem_util)

Modes

The mode of a body states how it arranges its items:

Mode	Holds	Description
`auto`	widgets	A grid of columns. Each widget takes one cell, or the full width with `full_width: true`. All grid rows have one height
`custom`	widgets	Each widget in the cell its `grid_pos` states, on a grid of 24 columns
`rows`	groups	Groups one under another, each with a header that collapses it
`tabs`	groups	Groups as tabs. The dashboard shows one group at a time

Without mode, a body that holds groups is rows, a body whose widgets state grid_pos is custom, and any other body is auto. When a body that states no mode lists widgets beside groups, Uptrace puts each run of widgets in a group with no header. A rows or tabs body holds groups only, and Uptrace refuses a widget in it.

Body fields

Field	Type	Description
`mode`	string	`auto`, `custom`, `rows`, or `tabs` (see Modes)
`items`	list	The widgets or the groups of the body, in the order the body shows them
`columns`	int	Columns of an `auto` body, 1 to 12
`column_width_mode`	string	Narrowest column of an `auto` body: `narrow`, `standard`, `wide`, `auto`, or `custom`
`column_width`	int	Narrowest column in pixels when `column_width_mode` is `custom`
`row_height_mode`	string	Height of the grid rows of an `auto` body: `short`, `standard`, `tall`, or `custom`
`row_height`	int	Grid row height in pixels when `row_height_mode` is `custom`
`fit_content`	bool	Whether a widget that can fit its content grows or shrinks to it
`fit_min_height_mode`	string	Least height of a widget that fits its content: `none`, `short`, `standard`, `tall`, or `custom`
`fit_min_height`	int	Least height in pixels when `fit_min_height_mode` is `custom`
`fit_max_height_mode`	string	Most height of a widget that fits its content: `short`, `standard`, `tall`, or `custom`
`fit_max_height`	int	Most height in pixels when `fit_max_height_mode` is `custom`
`match_tallest`	bool	Whether all widgets of a grid row take the height of the tallest one when `fit_content` is on

Groups, rows and tabs

A group is an item with type: group. It is a body too, so it takes all the body fields, and it holds widgets or other groups. A group in a rows body is a row with a header, and a group in a tabs body is a tab.

yaml

- type: group
  title: Gauges
  columns: 6
  column_width_mode: narrow
  items:
    - title: Availability
      type: stat
      metrics:
        - httpcheck_status as $status
      query:
        - sum($status{http_status_class="2xx"}) / sum($status) as availability
      columns:
        availability: { unit: utilization }

Field	Type	Required	Description
`type`	string	yes	Always `group`
`title`	string	no	Heading above the widgets. Omit it for a group with no heading
`description`	string	no	Explanation shown below the heading
`collapsed`	bool	no	Show the group closed when the dashboard loads. Only a group with a header can close
`hide_header`	bool	no	Show no header. The group then only arranges its widgets
`items`	list	yes	The widgets or the groups of the group

Table widgets and linked grids

A table widget in the root can link a grid: a layout of its own that a row of the table opens, filtered to the group by values of that row. The grid key of the table widget holds it. It is a body, so it takes the same fields as layout.

yaml

layout:
  items:
    - type: table
      title: Hosts
      metrics:
        - system_cpu_utilization as $cpu_util
        - system_memory_usage as $mem_usage
      query:
        - group by host_name
        - avg($cpu_util) as cpu_util
        - sum($mem_usage{state="used"}) as mem_used
      columns:
        cpu_util: { unit: utilization }
        mem_used: { unit: bytes }
      grid_variables: [deployment_environment, service_name]
      grid:
        items:
          - type: group
            title: General
            items:
              - title: CPU utilization
                metrics:
                  - system_cpu_utilization as $cpu_util
                query:
                  - avg($cpu_util)

Only a table widget in the root links a grid. The grid keys of a table widget:

Field	Type	Description
`grid`	object	The linked grid. An empty mapping (`grid: {}`) links an empty grid
`grid_query`	string	A `where` clause that each widget of the grid starts from, joined with the row a reader opened
`grid_variables`	list	Attribute names the filter bar of the grid offers, beside the grouping attributes of the table
`grid_include`	list	Other templates whose root items are appended to the grid (see Composition)

See Table for the other fields of a table widget.

Widgets beside the table

The root can hold other widgets beside the table, for example a summary above it. They are ordinary widgets of the root:

yaml

layout:
  items:
    - title: Number of hosts
      type: text
      metrics:
        - system_memory_usage as $mem_usage
      query:
        - uniq($mem_usage, host_name) as num_host
      text: ${num_host} hosts

    - type: table
      title: Hosts
      metrics:
        - system_memory_usage as $mem_usage
      query:
        - group by host_name
        - sum($mem_usage{state="used"}) as mem_used

The type field of an item selects the widget. When type is omitted, the widget is a chart.

Common fields

All widget types share these fields:

Field	Type	Required	Description
`title`	string	yes	Display heading. An empty title (`title: ""`) shows the widget with none
`description`	string	no	Explanation shown below the title
`type`	string	no	One of: `chart` (default), `table`, `stat`, `gauge`, `pie`, `heatmap`, `text`
`full_width`	bool	no	In an `auto` body, take the full width (`true`) or one cell (`false`). A table defaults to `true`, other types to `false`
`grid_pos`	object	no	In a `custom` body, the cell on one line: `{w: 12, h: 20, x: 0, y: 0}`. A widget states `grid_pos` or `full_width`, not both
`height`	string	no	In a `custom` body, `content` (fit the content) or `fixed` (the rows `grid_pos` states)

Chart

Charts are the default grid item type. They render time-series data as line, bar, or scatter plots.

yaml

- title: CPU time
  # type: chart (default, can be omitted)
  metrics:
    - system_cpu_time as $cpu_time
  query:
    - perMin(sum($cpu_time)) as cpu_time group by state
  fill_opacity: 0.1

Chart-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses
`chart`	string	Chart type: `line` (default), `bar`, `scatter`
`tooltip`	object	`{mode, sort_order}`: what the tooltip shows, and in which order
`sort_by`	list	Columns the legend table sorts the series by
`max_data_points`	int	Most datapoints the chart reads
`min_interval`	duration	Lower bound for the grouping step
`fill_opacity`	float	Area fill opacity, 0 to 1
`stack`	string	Stacking mode: `""` (none) or `"all"`
`columns`	map	Per-metric display settings with `unit` and `color`
`legend`	object	Legend configuration (see below)
`properties`	list	Visual properties (see Visual properties)
`overrides`	list	Per-timeseries overrides (see Timeseries overrides)

Legend configuration

yaml

legend:
  type: table     # "none", "list", or "table"
  placement: bottom # "bottom" or "right"
  values: [avg, min, max, last]
  items_per_page: 10

Table widgets show query results in rows with sortable columns.

yaml

- title: Slowest groups
  type: table
  metrics:
    - uptrace_tracing_spans as $spans
  query:
    - group by _group_id
    - group by _system
    - p50($spans)

Table-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses
`columns`	map	Per-column display settings (see Column settings)
`overrides`	list	Per-column visual property overrides (see Column overrides)
`items_per_page`	int	Rows on one page of the table. Defaults to 5
`compact_table`	bool	Show the rows with less padding. Defaults to `true`
`lines_per_row`	int	Lines of text one row shows. Defaults to 1
`nulls_mode`	string	How a cell with no value shows: `keep`, `replace`, or `connect`
`sort_by`	list	Columns the table sorts its rows by
`grid`, `grid_query`, `grid_variables`, `grid_include`		The linked grid (see Table widgets and linked grids)

Column settings (shorthand)

The columns map provides a compact syntax for column formatting:

yaml

columns:
  cpu_util: { unit: utilization }
  mem_used: { unit: bytes, color: red }
  availability: { unit: utilization, agg_func: avg }
  max_latency: { unit: milliseconds, agg_func: max, display: bar }

Available column properties:

Property	Description
`unit`	Display unit (e.g., `bytes`, `utilization`, `seconds`, `nanoseconds`, `milliseconds`, `1`, or custom like `span/min`)
`color`	Display color
`agg_func`	Aggregation function (e.g., `sum`, `avg`, `max`, `last`)
`display`	Column display mode: `value` (default), `sparkline`, or `bar`. Can also be a map: `{mode: bar, min: 0, max: 1}`

Column overrides (structured)

The overrides format provides more control:

yaml

overrides:
  - column: cpu_util
    properties:
      - name: unit
        value: utilization
      - name: color
        value: red

Stat

Stat widgets show a single aggregated value as text, optionally with value mappings for status labels.

yaml

- title: Status
  type: stat
  metrics:
    - httpcheck_status as $status
  query:
    - sum($status{http_status_class="2xx"})
  value_mappings:
    - op: any
      text: UNKNOWN
      color: gray
    - op: eq
      value: 0
      text: DOWN
      color: red
    - op: gte
      value: 1
      text: UP
      color: green

Stat-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses
`columns`	map	Per-column settings with `unit` and `agg_func`
`overrides`	list	Per-column visual property overrides
`value_mappings`	list	Maps values to labels and colors

Value mappings

Value mappings are evaluated in order, and the last matching rule wins, so a lower rule overrides the rules above it. Put the catch-all (any) first, and the more specific rules below it.

Field	Type	Description
`op`	string	Comparison operator: `any`, `eq`, `lt`, `lte`, `gt`, `gte`
`value`	number	Threshold value (not required for `any`)
`text`	string	Display label. An empty label keeps the number
`color`	string	Display color

Gauge

Gauges draw a single value on a dial, against the scale they state.

yaml

- title: CPU utilization
  type: gauge
  metrics:
    - system_cpu_utilization as $cpu
  query:
    - avg($cpu) as cpu
  columns:
    cpu: { unit: utilization }
  scale_min: 0
  scale_max: 1

Gauge-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses
`columns`	map	Per-column settings with `unit` and `agg_func`
`overrides`	list	Per-column visual property overrides
`value_mappings`	list	Maps values to labels and colors (see Value mappings)
`scale_min`	number	Value the dial starts at. Defaults to 0
`scale_max`	number	Value the dial ends at. Must be greater than `scale_min`
`shape`	string	`arc` (default) or `circle`
`bar_cap`	string	`flat` (default) or `rounded`
`bar_width_factor`	number	Bar thickness as a fraction of the radius. Defaults to 0.3
`sparkline`	bool	Show a small trend chart beside the value

Pie

Pie widgets show the share of each group, or of each column when the query has no grouping.

yaml

- title: Memory by state
  type: pie
  metrics:
    - system_memory_usage as $mem
  query:
    - avg($mem) as mem
    - group by state
  columns:
    mem: { unit: bytes }

Pie-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses. With a grouping, state one value column
`columns`	map	Per-column settings with `unit` and `agg_func`
`overrides`	list	Per-column visual property overrides
`shape`	string	`donut` (default) or `pie`
`slice_style`	string	`flat` (default) or `rounded`
`slice_gap`	bool	Part the slices with a small gap. Defaults to `true`
`color_scheme`	string	`default`, `carbon`, `sequential-reds`, `sequential-greens`, or `sequential-blues`
`max_slices`	int	Most slices the pie shows. Defaults to 8
`other_slice`	bool	Sum the smallest slices above `max_slices` into one slice named Other. Defaults to `true`
`legend`	object	Legend configuration
`labels`	object	Slice label configuration

Heatmap

Heatmaps visualize the distribution of a single histogram metric over time.

yaml

- title: Span duration heatmap
  type: heatmap
  metric: uptrace_tracing_spans
  unit: milliseconds

Heatmap-specific fields

Field	Type	Description
`metric`	string	Single metric name (not an alias)
`unit`	string	Display unit
`query`	list	Optional MQL query clauses for filtering

Text

Text widgets render a Go template string with metric data, useful for summary counts and labels.

yaml

- title: Host count
  type: text
  metrics:
    - process_runtime_go_goroutines as $goroutines
  query:
    - uniq($goroutines, host_name) as num_host
  text: ${num_host} hosts

Text-specific fields

Field	Type	Description
`metrics`	list	MQL metric expressions
`query`	list	MQL query clauses
`text`	string	Template string using `${column_name}` placeholders
`columns`	map	Per-column settings with `unit` and `agg_func`
`overrides`	list	Per-column visual property overrides

Visual properties

Visual properties control chart appearance. They can be set via shorthand fields (chart, fill_opacity, stack) or the properties list.

yaml

properties:
  - name: chartType
    value: bar
  - name: fillOpacity
    value: 0.3
  - name: stack
    value: all

Property	Type	Values
`chartType`	string	`line`, `bar`, `scatter`
`stack`	string	`""` (off), `all`
`nullsMode`	string	How a time bucket with no value is drawn: `keep` (a gap), `replace` (a zero), `connect` (a line across the gap)
`showPoints`	string	When a line chart marks its points: `auto`, `always`, `never`
`lineWidth`	number	Line thickness in pixels, 1 to 10
`fillOpacity`	number	0 to 1
`symbolSize`	number	Point marker size in pixels, 1 to 20
`symbol`	string	Marker name, such as `circle` or `triangle`
`color`	string	A CSS color
`colorScheme`	string	Color scheme name
`colorMode`	string	Which part of a stat is colored: `none`, `value`, `backgroundGradient`, `backgroundSolid`
`textSize`	number	Stat value size in pixels, 1 to 200
`unit`	string	Display unit
`aggFunc`	string	Aggregation function
`columnDisplay`	object	`{mode: value \| sparkline \| bar, min?, max?}`

nullsMode replaces the older boolean connectNulls.

Timeseries overrides

Chart widgets support per-timeseries overrides that apply visual properties to specific series:

yaml

overrides:
  - matchers:
      - target: metric   # "metric" or "timeseries"
        value: cpu_idle
    properties:
      - name: unit
        value: utilization
      - name: color
        value: blue

The target field specifies what to match:

metric -- matches by metric/column name
timeseries -- matches by timeseries label

Metric monitors

The metric_monitors section bundles alert monitors with the dashboard. Each entry uses the same format as a metric monitor and adds a key:

yaml

metric_monitors:
  - key: cpu_usage
    name: CPU usage
    type: threshold
    metrics:
      - system_cpu_load_average_15m as $load_avg_15m
      - system_cpu_time as $cpu_time
    query:
      - avg($load_avg_15m) / uniq($cpu_time, cpu) as cpu_util
      - group by host_name
    column:
      unit: utilization
    max_value: 3
    num_eval_points: 10
    trend_agg_func: last

Monitor fields

Field	Type	Required	Description
`key`	string	yes	Unique monitor identifier within the template
`name`	string	no	Human-readable title. Defaults to a name made from `key`
`type`	string	yes	Detector: `threshold`, `anomaly`, `outlier`, or `forecast`
`metrics`	list	yes	MQL metric expressions
`query`	list	yes	MQL query clauses
`status`	string	no	`active` (default) or `paused`
`column`	string/object	no	Column to evaluate: a name, or `{name: "col", unit: "ms"}`
`resolution`	duration	no	Data point interval
`num_eval_points`	int	no	Number of recent data points to evaluate, 1 to 15
`window_rule`	string	no	`at_all_times`, `at_least`, `on_average`, or `in_total` (the last two for `threshold` only)
`min_breach_points`	int	no	Number of points that must breach, for `at_least`
`absent_points`	string	no	Missing data handling: `zero`, `ignore`, or `alert`
`time_offset`	duration	no	Shifts the evaluation window, for late data
`trend_agg_func`	string	no	Aggregation function for trend detection: `sum`, `avg`, `median`, `last`
`trend_sensitivity`	string	no	Sensitivity of trend detection
`repeat_interval`	object	no	Notification repeat interval configuration

The detector settings sit at the top level beside type. A setting of another type is accepted and ignored.

Type	Settings
`threshold`	`min_value`, `max_value` (at least one), `recovery: {min_value, max_value}`
`anomaly`	`sensitivity`, `direction`, `seasonality`, `min_change`, `min_volume`, `preset`
`outlier`	`sensitivity`, `direction`, `min_peers`, `min_change`, `keep_firing_for`
`forecast`	`min_value`, `max_value` (at least one), `horizon`, `seasonality`

See Alerting for the meaning of each setting.

yaml

# Threshold with a recovery band inside the range
type: threshold
min_value: 100
max_value: 1000
recovery:
  min_value: 110
  max_value: 900

yaml

# Anomaly detection
type: anomaly
sensitivity: medium     # very_low, low, medium, high, very_high
direction: both         # both, above, below
min_change: 10          # optional: smallest move, in metric units

Older templates use a nested detector block (type: manual or type: auto). Uptrace still reads it and rewrites it to the current shape, but new templates should use the top-level type.

Template composition

A table widget can append the root items of other templates to its grid with grid_include. This enables modular dashboard composition:

yaml

# Parent: uptrace.dotnet.10.all.yml
schema: v2
name: '.NET: All'
tags: [otel, app]
version: v26.11.05

layout:
  items:
    - type: table
      title: Services
      metrics:
        - process_runtime_dotnet_gc_heap_size as $heap_size
      query:
        - group by service_name
        - sum($heap_size) as heap_size
      grid_include:
        - uptrace.dotnet.20.gc
        - uptrace.dotnet.30.runtime
        - uptrace.dotnet.40.thread_pool

Each entry references another template by filename (without .yml).

Metric requirements

The metrics_match field controls when a template is activated. It appears at the top level of the template:

yaml

metrics_match:
  present:
    - metric: system_cpu_utilization
    - library: io.opentelemetry.instrumentation.system
      metric: system_cpu_time
      attrs: [cpu]
  missing:
    - metric: system_cpu_logical_count

present lists metrics the project must emit, and every entry must match. missing lists metrics the project must not emit. Each entry has these fields:

Field	Type	Description
`metric`	string or list	Metric name. A list matches any one of the names, for example after a rename
`library`	string	Instrumentation library name (optional)
`attrs`	list	Attribute keys the metric must carry (optional)

A file can hold several YAML documents separated by ---. They are variants of the same dashboard for different metric sets, and Uptrace applies the first one whose metrics_match passes. Only the first document states the metadata (schema, name, tags, version, and so on). A variant without metrics_match matches every project, so it must be the last one. Older templates use require_metrics, which Uptrace still reads and rewrites to metrics_match.

Template ID and filename

The template ID is derived from the filename:

uptrace.hostmetrics.10.overview.yml becomes ID uptrace.hostmetrics.overview
The numeric suffix (e.g., 10) is stripped from the ID and used as a display priority

Validation

Templates are validated in two ways:

YAML parsing with DisallowUnknownField catches typos and unknown keys.
JSON Schema validation against schema.json ensures structural correctness.

To validate templates:

bash

go run -tags=cloud ./cmd/cloud dashboard validate_templates

Earlier template formats

Templates written for earlier releases still import. Uptrace rewrites each earlier spelling to the current one, and an export always writes the current format:

Earlier spelling	Current spelling
`table` at the top level	A table widget at the end of `layout.items`
`grid_sections` (or `grid_rows`)	Groups in `rows` mode: in the `grid` of that table widget, or in the root when there is no table
`table_grid_items`	Widgets in `layout.items`, before the table widget
`include_grid_sections`, `grid_query`, `grid_variables` at the top level	`grid_include`, `grid_query`, `grid_variables` on the table widget
`type: gauge` with no gauge field	`type: stat`
`require_metrics`	`metrics_match`

Older monitors with a nested detector block are rewritten too (see Metric monitors).

Full example

yaml

schema: v2
name: 'HTTP Check: Endpoints'
description: Monitors HTTP endpoint availability and response times.
tags: [otel, infra]
version: v26.11.05
setup_link: https://example.com/setup

layout:
  fit_content: true
  items:
    - title: Successful checks
      type: text
      text: ${num_up} out of ${num_all}
      metrics:
        - httpcheck_status as $status
      query:
        - uniq($status{http_status_class="2xx"}) as num_all
        - uniq($status{http_status_class="2xx"}) == 1 as num_up

    - type: table
      title: ""
      metrics:
        - httpcheck_status as $status
        - httpcheck_duration as $duration
      query:
        - group by http_url
        - group by host_name
        - sum($status{http_status_class="2xx"}) / sum($status) as availability
        - avg($duration)
      columns:
        availability: { unit: utilization }
      grid:
        items:
          - type: group
            title: Gauges
            columns: 6
            column_width_mode: narrow
            items:
              - title: Status
                type: stat
                metrics:
                  - httpcheck_status as $status
                query:
                  - sum($status{http_status_class="2xx"})
                value_mappings:
                  - op: any
                    text: UNKNOWN
                    color: gray
                  - op: eq
                    value: 0
                    text: DOWN
                    color: red
                  - op: gte
                    value: 1
                    text: UP
                    color: green

          - type: group
            title: General
            items:
              - title: HTTP check result
                metrics:
                  - httpcheck_status as $status
                query:
                  - $status group by http_status_code

              - title: HTTP check duration
                metrics:
                  - httpcheck_duration as $duration
                query:
                  - avg($duration)

metric_monitors:
  - key: http_check_is_down
    name: HTTP check is down
    metrics:
      - httpcheck_status as $status
    query:
      - sum($status{http_status_class="2xx"}) as status_2xx
      - group by http_url
      - group by host_name
    type: threshold
    min_value: 1
    max_value: 1
    num_eval_points: 1
    absent_points: alert
    trend_agg_func: last