Monitor STJ file format

StackState Self-hosted v5.0.x

This page describes StackState version 5.0.

Go to the documentation for the latest StackState release.


Monitors can be attached to any number of elements in the StackState topology to calculate a health state based on 4T data. Each monitor consists of a monitor definition and a monitor function. Monitors are created and managed by StackPacks, you can also create custom monitors and monitor functions outside of a StackPack without having to modify any configuration.

STJ file format

Monitors in StackState are represented textually using the STJ file format. The following snippet presents an example monitor file:

  "_version": "1.0.39",
  "timestamp": "2022-05-23T13:16:27.369269Z[GMT]",
  "nodes": [
      "_type": "Monitor",
      "name": "CPU Usage",
      "description": "A simple CPU-usage monitor. If the metric is above a given threshold, the state is set to CRITICAL.",
      "identifier": "urn:system:default:monitor:cpu-usage",
      "remediationHint": "Turn it off and on again.",
      "function": {{ get "urn:system:default:monitor-function:metric-above-threshold" }},
      "arguments": [{
        "_type": "ArgumentDoubleVal",
        "parameter": {{ get "urn:system:default:monitor-function:metric-above-threshold" "Type=Parameter;Name=threshold" }},
        "value": 90.0
      }, {
         "_type": "ArgumentStringVal",
        "parameter": {{ get "urn:system:default:monitor-function:metric-above-threshold" "Type=Parameter;Name=topologyIdentifierPattern" }},
        "value": "urn:host:/${}"
      }, {
        "_type": "ArgumentScriptMetricQueryVal",
        "parameter": {{ get "urn:system:default:monitor-function:metric-above-threshold" "Type=Parameter;Name=query" }},
        "script": "Telemetry\n.query('StackState Metrics', '')\n.metricField('system.cpu.system')\n.groupBy('')\n.start('-1m')\n.aggregation('mean', '15s')"
      "intervalSeconds": 60

In addition to the usual elements of an STJ file, the protocol version and timestamp, the snippet defines a single node of type Monitor.

The supported fields are:

  • name - a human-readable name that shortly describes the operating principle of the monitor.

  • description - a longer, more in-depth description of the monitor.

  • identifier - a StackState-URN-formatted value that uniquely identifies this monitor definition. For more details see identifier.

  • remediationHint - a short, markdown-enabled hint displayed whenever the validation rule represented by this monitor triggers and results in an unhealthy state.

  • function - the specific monitor function to use as the basis of computation for this monitor. For more details see function.

  • arguments - lists concrete values that are to be used for parameters in the monitor function invocation. For more details and descriptions of commonly used parameters, see arguments.

  • intervalSeconds - dictates how often to execute this particular monitor; new executions are scheduled after the specified number of seconds, counting from the time that the last execution ended. For more details see run interval.

Field information


An important field of the monitor node is the identifier - it is a unique value of the StackState URN format that can be used together with the monitor-specific StackState CLI commands. The identifier should be formatted as follows:

urn : <prefix> : monitor : <unique-monitor-identification>

  • The <prefix> is described in more detail in topology identifiers.

  • The <unique-monitor-identification> is user-definable and free-form.


Each monitor configured in StackState uses a monitor function to compute the health state results that are attached to the elements.

Monitor functions are scripts that accept 4T data as input, check the data based on some internal logic and output health state mappings for the affected topology elements. The function is run periodically by the monitor runner (at the configured intervalSeconds). The monitor function is responsible for detecting any changes in the data that can be considered to change an element's health state.

You can list the available monitor functions using the CLI command:

sts settings list --type MonitorFunction

⚠️ PLEASE NOTE - from StackState v5.0, the old sts CLI has been renamed to stac and there is a new sts CLI. The command(s) provided here are for use with the new sts CLI.

➡️ Check which version of the sts CLI you are running

You can create custom monitor function to customize how StackState processes 4T data.


The arguments defined in the monitor STJ definition should match the parameters defined in the monitor function STJ definition.

The parameter binding syntax is common for all parameter types, and utilizes the following format:

  "_type": "<type-of-the-parameter",
  "parameter": {{ get "<identifier-of-the-function>" "Type=Parameter;Name=<name-of-the-parameter>" }},
  "value": "<value-of-the-parameter>"
  • _type - The type of the parameter.

  • parameter - A reference to the concrete instance of a parameter within a function's parameter list. The Name must match the name specified in the monitor function.

  • value - the value of the parameter to pass to the monitor function.

During an invocation of a monitor function, the parameter value is interpreted and instantiated beforehand with all of the requisite validations applied to it. Assuming it passes type and value validations, it will become available in the body of the function as a global value of the same name, with the assigned value.

  • Parameters marked as required in the monitor function STJ definition must be supplied at least once. If a parameter is not required, then it can be optionally omitted.

  • Parameters marked as multiple in the monitor function STJ definition can be supplied more than once, meaning that they represent a set of values.

Common parameters

Descriptions of parameters that are commonly used by monitor functions can be found below:

Numeric values

The most common and simple monitor function parameter types are numeric values.

To supply a value to the value parameter defined in the monitor function, the monitor STJ definition would look something like the following:

  "_type": "ArgumentDoubleVal",
  "parameter": {{ get "<identifier-of-the-function>" "Type=Parameter;Name=value" }},
  "value": 23.5

Topology Query

Monitor functions that utilize Topology often times take a Topology Query as a parameter.

To supply a value to the topologyQuery parameter defined in the monitor function, the monitor STJ definition would look something like the following:

  "_type": "ArgumentStringVal",
  "parameter": {{ get "<identifier-of-the-function>" "Type=Parameter;Name=topologyQuery" }},
  "value": "type = 'database' OR type = 'database-shard'"

Telemetry query

Monitor functions that utilize telemetry tend to be parameterized with the exact telemetry query to use for their computation. The telemetry query should be built using the StackState Telemetry Script API. The following fields are particularly useful in telemetry queries that are passed to monitor functions:

  • groupBy(fields) - when a monitor will produce a health state for multiple components, use the groupBy field to produce multiple time series as a set of unique values for the defined fields.

  • aggregation(type, interval) - aggregates each time series by the defined type. Each aggregated value is constructed out of a data span the size of the defined interval.

➡️ Learn more about the Telemetry script API

To supply a value to the telemetryQuery parameter defined in the monitor function, the monitor STJ definition would look something like the following. Note that the provided value must utilize the StackState Telemetry Script API and evaluate to a telemetry query, otherwise it will not pass the argument validation that is performed before the function execution begins.

  "_type": "ArgumentScriptMetricQueryVal",
  "parameter": {{ get "<identifier-of-the-function>" "Type=Parameter;Name=telemetryQuery" }},
  "value": "Telemetry.query('StackState Metrics', '').metricField('system.cpu.iowait').groupBy('').start('-10m').aggregation('mean', '1m')"

Topology identifier pattern

Monitor functions that don't process any topology directly still have to produce results that attach to topology elements by way of matching the topology identifier that can be found on those elements. In those cases, one can expect a function declaration to include a special parameter that represents the pattern of a topology identifier.

The topologyIdentifierPattern value supplied to the monitor function should result in a valid topology identifier once processed by the function logic. It therefore likely needs to include various escape sequences of values that will be interpolated into the resulting value by the monitor function:

  "_type": "ArgumentStringVal",
  "parameter": {{ get "<identifier-of-the-function>" "Type=Parameter;Name=topologyIdentifierPattern" }},
  "value": "urn:host:/${}"

The exact value to use for this parameter depends on the topology available in StackState (or more precisely on its identifier scheme), and on the values supplied by the monitor function for interpolation (or more precisely the type of data processed by the function). In the most common case, a topology identifier pattern parameter is used in conjunction with a telemetry query parameter - in this case, the fields used for the telemetry query grouping (listed in its .groupBy() step) will also be available for the interpolation of topology identifier values. For example, consider the following query:

  .query('StackState Metrics', '')
  .groupBy('host', 'region')
  .aggregation('mean', '1m')

The telemetry query above groups its results by two fields: host and region. Both of these values will be available for value interpolation of an exact topology identifier to use, and each different host and region pair can be used either individually or together to form a unique topology identifier. If the common topology identifier scheme utilized by the topology looks as follows, then the different parts of the identifier can be replaced by references to host or region:

# Example identifier as found on a topology element:

# Topology identifier pattern that matches the above example identifier:


The monitor run interval determines how often a monitor logic will be executed. This is configured in the monitor STJ file as a number of seconds using the intervalSeconds field. For example, an intervalSeconds: 60 configuration means that StackState will attempt to execute the monitor function associated with the monitor every 60 seconds. If the monitor function execution takes significant time, the next scheduled run will occur 60 seconds after the previous run finishes.

See also

Last updated