Monitoring and Tracing

If something is not working as it should within your GraphQL gateway, you would not want it to go unnoticed.

Monitoring and tracing are essential for debugging and understanding the performance of your gateway.

You can use Gateway plugins to trace and monitor your gateway’s execution flow together with all outgoing HTTP calls and internal query planning.

Healthcheck

Hive Gateway is aware of the usefulness of a health check and gives the user maximum possibilities to use the built-in check.

There are two types of health checks: liveliness and readiness, they both are a health check but convey a different meaning:

Liveliness checks whether the service is alive and running
Readiness checks whether the upstream services are ready to perform work and execute GraphQL operations

The difference is that a service can be live but not ready - for example, server has started and is accepting requests (alive), but the read replica it uses is still unavailable (not ready).

Both endpoints are enabled by default.

Liveliness

By default, you can check whether the gateway is alive by issuing a request to the /healthcheck endpoint and expecting the response 200 OK. A successful response is just 200 OK without a body.

You can change this endpoint through the healthCheckEndpoint option:

gateway.config.ts

import { defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  healthCheckEndpoint: '/healthcheck'
})

Readiness

For readiness check, Hive Gateway offers another endpoint (/readiness) which checks whether the services powering your gateway are ready to perform work. It returns 200 OK if all the services are ready to execute GraphQL operations.

It returns 200 OK if all the services are ready to perform work.

You can customize the readiness check endpoint through the readinessCheckEndpoint option:

gateway.config.ts

import { defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  readinessCheckEndpoint: '/readiness'
})

OpenTelemetry Traces

Hive Gateway supports OpenTelemetry for tracing and monitoring your gateway.

OpenTelemetry is a set of APIs, libraries, agents, and instrumentation to provide observability to your applications.

The following are available to use with this plugin:

HTTP request: tracks the incoming HTTP request and the outgoing HTTP response
GraphQL Lifecycle tracing: tracks the GraphQL execution lifecycle (parse, validate and execution).
Upstream HTTP calls: tracks the outgoing HTTP requests made by the GraphQL execution.
Context propagation: propagates the trace context between the incoming HTTP request and the outgoing HTTP requests.

Usage Example

gateway.config.ts

import { createStdoutExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      // A simple output to the console.
      // You can add more exporters here, please see documentation below for more examples.
      createStdoutExporter()
    ],
    serviceName: 'my-custom-service-name', // Optional, the name of your service
    tracer: myCustomTracer, // Optional, a custom tracer to use
    inheritContext: true, // Optional, whether to inherit the context from the incoming request
    propagateContext: true, // Optional, whether to propagate the context to the outgoing requests
    // Optional config to customize the spans. By default all spans are enabled.
    spans: {
      http: true, // Whether to track the HTTP request/response
      graphqlParse: true, // Whether to track the GraphQL parse phase
      graphqlValidate: true, // Whether to track the GraphQL validate phase
      graphqlExecute: true, // Whether to track the GraphQL execute phase
      subgraphExecute: true, // Whether to track the subgraph execution phase
      upstreamFetch: true // Whether to track the upstream HTTP requests
    }
  }
})

Exporters

You may use one of the following exporters to send the traces to a backend, or create an configure custom exporters and processors.

To use a custom exporter that is not listen below, please refer to Customer Exporters in OpenTelemetry documentation.

In addition, you can fully customize the plugin’s Tracer with any kind of OpenTelemetry tracer, and integrate it to any tracing/metric platform that supports this standard.

A simple exporter that writes the spans to the stdout of the process.

gateway.config.ts

import { createStdoutExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [createStdoutExporter()]
  }
})

An exporter that writes the spans to an OTLP-supported backend using HTTP.

gateway.config.ts

import { createOtlpHttpExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      createOtlpHttpExporter({
        url: 'http://<otlp-endpoint>:4318'
        // ...
        // additional options to pass to @opentelemetry/exporter-trace-otlp-http
        // https://www.npmjs.com/package/@opentelemetry/exporter-trace-otlp-http
      })
    ]
  }
})

An exporter that writes the spans to an OTLP-supported backend using gRPC.

gateway.config.ts

import { createOtlpGrpcExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      createOtlpGrpcExporter({
        url: 'http://<otlp-endpoint>:4317'
        // ...
        // additional options to pass to @opentelemetry/exporter-trace-otlp-grpc
        // https://www.npmjs.com/package/@opentelemetry/exporter-trace-otlp-grpc
      })
    ]
  }
})

Jaeger supports OTLP over HTTP/gRPC, so you can use it by pointing the createOtlpHttpExporter/createOtlpGrpcExporter to the Jaeger endpoint:

gateway.config.ts

import { createOtlpHttpExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      createOtlpHttpExporter({
        url: 'http://<jaeger-endpoint>:4318'
      })
    ]
  }
})

💡

Your Jaeger instance needs to have OTLP ingestion enabeld, so verify that you have the COLLECTOR_OTLP_ENABLED=true environment variable set, and that ports 4317 and 4318 are acessible.

To test this integration, you can run a local Jaeger instance using Docker:

docker run -d --name jaeger \
  -e COLLECTOR_OTLP_ENABLED=true \
  -p 5778:5778 \
  -p 16686:16686 \
  -p 4317:4317 \
  -p 4318:4318 \
  jaegertracing/all-in-one:latest

NewRelic supports OTLP over HTTP/gRPC, so you can use it by configuring the createOtlpHttpExporter/createOtlpGrpcExporter to the NewRelic endpoint:

gateway.config.ts

import { createOtlpHttpExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      createOtlpHttpExporter({
        url: 'http://<newrelic-endpoint>:4318'
      })
    ]
  }
})

💡

For additional information and NewRelic ingestion endpoints, see New Relic OTLP endpoint.

DataDog Agent supports OTLP over HTTP/gRPC, so you can use it by pointing the createOtlpHttpExporter to the DataDog Agent endpoint:

gateway.config.ts

import { createOtlpHttpExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
     createOtlpHttpExporter({
        url: 'http://<datadog-agent-host>:4318'
      })
    ]
  }
})

💡

For additional information, see OpenTelemetry in Datadog.

Zipkin is using a custom protocol to send the spans, so you can use the Zipkin exporter to send the spans to a Zipkin backend:

gateway.config.ts

import { createZipkinExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      createZipkinExporter({
        url: 'http://<zipkin-host>:9411/api/v2/spans'
        // ...
        // additional options to pass to @opentelemetry/exporter-zipkin
        // https://www.npmjs.com/package/@opentelemetry/exporter-zipkin
      })
    ]
  }
})

Batching

All built-in processors allow you to configure batching options by an additional argument to the factory function.

The following configuration are allowed:

true (default): enables batching and use BatchSpanProcessor default config.
object: enables batching and use BatchSpanProcessor with the provided configuration.
false - disables batching and use SimpleSpanProcessor

By default, the batch processor will send the spans every 5 seconds or when the buffer is full.

{
  "scheduledDelayMillis": 5000,
  "maxQueueSize": 2048,
  "exportTimeoutMillis": 30000,
  "maxExportBatchSize": 512
}

You can learn more about the batching options in the Picking the right span processor page.

Reported Spans

The plugin exports OpenTelemetry spans for the following operations:

HTTP Server

💡

This span is created for each incoming HTTP request, and acts as a root span for the entire request. Disabling this span will also disable the other hooks and spans.

By default, the plugin will a root span for the HTTP layer as a span (METHOD /path) with the following attributes for the HTTP request:

http.method: The HTTP method
http.url: The HTTP URL
http.route: The HTTP status code
http.scheme: The HTTP scheme
http.host: The HTTP host
net.host.name: The hostname
http.user_agent: The HTTP user agent (based on the User-Agent header)
http.client_ip: The HTTP connecting IP (based on the X-Forwarded-For header)

And the following attributes for the HTTP response:

http.status_code: The HTTP status code

💡

An error in the this phase will be reported as an error span with the HTTP status text and as an OpenTelemetry Exception.

You may disable this by setting spans.http to false:

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    http: false
  }
}

Or, you may filter the spans by setting the spans configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    http: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onRequest hook.

GraphQL Parse

By default, the plugin will report the validation phase as a span (graphql.validate) with the following attributes:

graphql.document: The GraphQL query string
graphql.operation.name: The operation name

💡

An error in the parse phase will be reported as an error span, including the error message and as an OpenTelemetry Exception.

You may disable this by setting spans.graphqlParse to false:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlParse: false
  }
}

Or, you may filter the spans by setting the spans configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlParse: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onParse hook.

GraphQL Validate

By default, the plugin will report the validation phase as a span (graphql.validate) with the following attributes:

graphql.document: The GraphQL query string
graphql.operation.name: The operation name

💡

An error in the validate phase will be reported as an error span, including the error message and as an OpenTelemetry Exception.

You may disable this by setting spans.graphqlValidate to false:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlValidate: false
  }
}

Or, you may filter the spans by setting the spans configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlValidate: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onValidate hook.

GraphQL Execute

By default, the plugin will report the execution phase as a span (graphql.execute) with the following attributes:

graphql.document: The GraphQL query string
graphql.operation.name: The operation name
graphql.operation.type: The operation type (query/mutation/subscription)

💡

An error in the execute phase will be reported as an error span, including the error message and as an OpenTelemetry Exception.

You may disable this by setting spans.graphqlExecute to false:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlExecute: false
  }
}

Or, you may filter the spans by setting the spans configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    graphqlExecute: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onExecute hook.

Subgraph Execute

By default, the plugin will report the subgraph execution phase as a span (subgraph.execute) with the following attributes:

graphql.document: The GraphQL query string executed to the upstream
graphql.operation.name: The operation name
graphql.operation.type: The operation type (query/mutation/subscription)
gateway.upstream.subgraph.name: The name of the upstream subgraph

In addition, the span will include the following attributes for the HTTP requests;

http.method: The HTTP method
http.url: The HTTP URL
http.route: The HTTP status code
http.scheme: The HTTP scheme
net.host.name: The hostname
http.host: The HTTP host

And the following attributes for the HTTP response:

http.status_code: The HTTP status code

You may disable this by setting spans.subgraphExecute to false:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    subgraphExecute: false
  }
}

Or, you may filter the spans by setting the spans.subgraphExecute configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    subgraphExecute: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onSubgraphHook hook.

Upstream Fetch

By default, the plugin will report the upstream fetch phase as a span (http.fetch) with the information about outgoing HTTP calls.

The following attributes are included in the span:

http.method: The HTTP method
http.url: The HTTP URL
http.route: The HTTP status code
http.scheme: The HTTP scheme
net.host.name: The hostname
http.host: The HTTP host

And the following attributes for the HTTP response:

http.status_code: The HTTP status code

You may disable this by setting spans.upstreamFetch to false:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    upstreamFetch: false
  }
}

Or, you may filter the spans by setting the spans.upstreamFetch configuration to a function:

gateway.config.ts

const openTelemetryConfig = {
  exporters: [
    /* ... */
  ],
  spans: {
    /* ... */
    upstreamFetch: payload => {
      // Filter the spans based on the payload
      return true
    }
  }
}

💡

The payload object is the same as the one passed to the onFetch hook.

Context Propagation

By default, the plugin will propagate the trace context between the incoming HTTP request and the outgoing HTTP requests.

You may disable this by setting inheritContext or propagateContext to false:

gateway.config.ts

import { defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [
      /* ... */
    ],
    // Controls the propagation of the trace context between the incoming HTTP request and Hive Gateway
    inheritContext: false,
    // Controls the propagation of the trace context between Hive Gateway and the upstream HTTP requests
    propagateContext: false
  }
})

Troubleshooting

The default behavor of the plugin is to log errors and warnings to the console.

You can customize this behavior by changing the value of the OTEL_LOG_LEVEL environment variable on your gateway process/runtime.

In addition, you can use the stdout exporter to log the traces to the console:

gateway.config.ts

import { createStdoutExporter, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  openTelemetry: {
    exporters: [createStdoutExporter()]
  }
})

This will log the traces to the console, which can be useful for debugging and troubleshooting.

Prometheus Metrics

Prometheus is a utility for producing, scraping and storage of metrics from services and utilities.

You can use this feature of the gateway to expose and collect metrics from all phases of your GraphQL execution including internal query planning and outgoing HTTP requests.

The metrics gathered are then exposed in a format that Prometheus can scrape on a regular basis on an HTTP endpoint (/metrics by default).

Usage Example

gateway.config.ts

import { defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  prometheus: {
    // Enable the metrics you want to expose
    // The following represent the default config of the plugin.
    metrics: {
      graphql_gateway_fetch_duration: true,
      graphql_gateway_subgraph_execute_duration: true,
      graphql_gateway_subgraph_execute_errors: true,
      graphql_envelop_deprecated_field: true,
      graphql_envelop_request: true,
      graphql_envelop_request_duration: true,
      graphql_envelop_request_time_summary: true,
      graphql_envelop_phase_parse: true,
      graphql_envelop_phase_validate: true,
      graphql_envelop_phase_context: true,
      graphql_envelop_error_result: true,
      graphql_envelop_phase_execute: true,
      graphql_envelop_phase_subscribe: true,
      graphql_envelop_schema_change: true,
      graphql_yoga_http_duration: true
    }
  }
})

You can now start your Hive Gateway and make some requests to it. The plugin will start collecting metrics, and you can access them by visiting the /metrics endpoint.

In most cases, you’ll need to setup a Prometheus server to scrape the metrics from your gateway, we recommend using the official Prometheus Server or tools like Vector.

Grafana Dashboard

If you are using Grafana to visualize your metrics, you can import the published Grafana dashboard for Grafana’s marketplace, or you can use/import this dashboard JSON file directly to easily visualuze the metrics for your gateway.

For additional instructions, please refer to Import dashboards insturction in Grafana documentation.

Reported Metrics

You will find the timing of each phase of the GraphQL execution. If you are not familiar with the lifecycle of a GraphQL operation in the gateway, please refer to the Plugin Lifecycle page. Each plugin hook has a corresponding metric which tracks timings as histograms or summary. You will also find some counters to track the number of requests, errors, and other useful information.

To enable a metric, set the corresponding option to true in the metrics option’s object. You can also provide a string to customize the metric name, or an object to provide more options (see siimon/prom-client documentation). Histogram metrics can be passed an array of numbers to configure buckets.

graphql_yoga_http_duration (default: enabled, type: Histogram)

This metric tracks the duration of incoming (downstream) HTTP requests. It reports the time spent to process each incoming request as a histogram.

It is useful to track the responsiveness of your gateway. A spike in this metric could indicate a performance issue and that further investigation is needed.

Please note that this metric is not specific to GraphQL, it tracks all incoming HTTP requests.

You can use labels to have a better understanding of the requests and group them together. A common filter is to include only statusCode with 200 value and method with POST (the default method for GraphQL requests, but it can also be GET depending on your client setup) value to get execution time of successful GraphQL requests only.

This metric includes some useful labels to help you identify requests and group them together.

Label	Description
`method`	The HTTP method used to request the gateway endpoint. Since GraphQL usually only uses `POST` requests, this can be used to filter out GraphiQL-related requests. It can be any HTTP verb, including disallowed ones. Which means this metric can also be used to track malformed or malicious requests.
`statusCode`	The HTTP status code returned by the gateway. You probably want to filter out non-`200` responses to have a view of the successful requests. This can help you identify which requests are failing and why. Since GraphQL errors are returned as `200 OK` responses, this can be useful to track errors that are not related to the GraphQL, like malformed requests.
`operationName`	If available, the name of the GraphQL operation requested, otherwise `Anonymous`. This can help you identify which operations are slow or failing. We recommend you always provide an operation name to your queries and mutations to help performance analysis and bug tracking.
`operationType`	The type of the GraphQL operation requested. It can be one of `query`, `mutation`, or `subscription`. This can help you differentiate read and write performance of the system. It can for example help understand cache impact.
`url`	The URL of the request. Useful to filter graphql endpoint metrics (`/graphql` by default).

graphql_gateway_fetch_duration (default: enabled, type: Histogram)

This metric tracks the duration of outgoing HTTP requests. It reports the time spent on each request made using the fetch function provided by the gateway. It is reported as a histogram.

This metric can provide insights into the network usage of your gateway. It does not only include requests made to resolve GraphQL operation responses, but also include any other outgoing HTTP requests made by the gateway or one of its plugins. It will for example include requests made to fetch the supergraph schema from the configured Schema Registry.

These metrics include some useful labels to help you identify requests and group them together.

Since they can be heavy, requestHeaders and responseHeaders are disabled by default. You can either set those options to true in the label configuration object to include all headers in the label, but you can also provide a list of header names to include.

Label	Description
`url`	The URL of the upstream request.
`method`	The HTTP method of the upstream request.
`statusCode`	The status code of the upstream response.
`statusText`	The status text of the upstream response.
`requestHeaders`	Disabled by default. A JSON encoded object containing the headers of the upstream request.
`responseHeaders`	Disabled by default. A JSON encoded object containing the headers of the upstream response.

graphql_gateway_subgraph_execute_duration (default: enabled, type: Histogram)

This metric tracks the duration of subgraph execution. It reports the time spent on each subgraph queries made to resolve incoming operations as a histogram.

This metric can provide insights into how the time is spent to resolve queries. It can help you identify bottlenecks in your subgraphs.

Label	Description
`subgraphName`	The name of the targeted subgraph.
`operationType`	The type of the GraphQL operation executed by the subgraph. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation executed by the subgraph. It will be `Anonymous` if no `operationName` is found.

graphql_gateway_subgraph_execute_errors (default: enabled, type: Counter)

This metric tracks the number of errors that occurred during the subgraph execution. It counts all errors found in the response returned by the subgraph execution. It is exposed as a counter.

This metric can help you identify subgraphs that are failing to execute operations. It can help identify issues with the subgraph itself or the communication between the gateway and the subgraph.

Label	Description
`subgraphName`	The name of the targeted subgraph.
`operationType`	The type of the GraphQL operation executed by the subgraph. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation executed by the subgraph. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_phase_parse (default: enabled, type: Histogram)

This metric tracks the duration of the parse phase of the GraphQL execution. It reports the time spent parsing the incoming GraphQL operation. It is reported as a histogram.

Since you don’t have control over the parsing phase, this metric is mostly useful to track potential attacks. A spike in this metric could indicate someone is trying to send malicious operations to your gateway.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_phase_validate (default: enabled, type: Histogram)

This metric tracks the duration of the validate phase of the GraphQL execution. It reports the time spent validating the incoming GraphQL operation. It is reported as a histogram.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_phase_context (default: enabled, type: Histogram)

This metric tracks the duration of the context phase of the GraphQL execution. It reports the time spent building the context object that will be passed to the executors. It is reported as a histogram.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_phase_execute (default: enabled, type: Histogram)

This metric tracks the duration of the execute phase of the GraphQL execution. It reports the time spent actually resolving the response of the incoming operation. This includes the gathering of all the data from all sources required to construct the final response. It is reported as a histogram.

It is the metric that will give you the most insights into the performance of your gateway, since this is where most of the work is done.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_phase_subscribe (default: enabled, type: Histogram)

This metric tracks the duration of the subscribe phase of the GraphQL execution. It reports the time spent initiating a subscription (which doesn’t include actually sending the first response). It is reported as a histogram.

It will notably include the time spent to setup upstream subscriptions with appropriate transport for each source.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_request_duration (default: enabled, type: Histogram)

This metric tracks the duration of the complete GraphQL operation execution. It reports the time spent in the GraphQL specific processing, excluding the HTTP-level processing. It is reported as a histogram.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_request_time_summary (default: enabled, type: Summary)

This metric provides a summary of the time spent on the GraphQL operation execution. It reports the same timing than graphql_envelop_request_duration but as a summary.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_error_result (default: enabled, type: Counter)

This metric tracks the number of errors that was returned by the GraphQL execution.

Similarly to graphql_gateway_subgraph_execute_errors, it counts all errors found in the final response constructed by the gateway after it gathered all subgraph responses, but it also includes errors from other GraphQL processing phases (parsing, validation and context building). It is exposed as a counter.

Depending on the phase when the error occurred, some labels may be missing. For example, if the error occurred during the context phase, only the phase label will be present.

Label	Description
`path`	The path of the field that caused the error. It can be `undefined` if the error is not related to a given field.
`phase`	The phase of the GraphQL execution where the error occurred. It can be `parse`, `validate`, `context`, `execute` (for every operation types including subscriptions).
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_request (default: enabled, type: Counter)

This metric tracks the number of GraphQL operations executed. It counts all operations, either failed or successful, including subscriptions. It is exposed as a counter.

It can differ from the number reported by graphql_yoga_http_duration_sum because a single HTTP request can contain multiple GraphQL operations if batching has been enabled.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_deprecated_field (default: enabled, type: Counter)

This metric tracks the number of deprecated fields used in the GraphQL operation.

Label	Description
`fieldName`	The name of the deprecated field that has been used.
`typeName`	The name of the parent type of the deprecated field that has been used.
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.

graphql_envelop_schema_change (default: enabled, type: Counter)

This metric tracks the number of schema changes that have occurred since the gateway started. When polling is enabled, this will include the schema reloads.

If you are using a plugin that modifies the schema on the fly, be aware that this metric will also include updates made by those plugins. Which means that one schema update can actually trigger multiple schema changes.

graphql_envelop_execute_resolver (default: disabled, type: Histogram)

⚠️

Enabling resolvers level metrics will introduce significant overhead. It is recommended to enable this metric only for debugging purposes.

This metric tracks the duration of each resolver execution. It reports the time spent only on additional resolvers, not on fields that are resolved by a subgraph. It is up to the subgraph server to implement resolver level metrics, the gateway can’t remotely track their execution time.

Label	Description
`operationType`	The type of the GraphQL operation requested. This can be one of `query`, `mutation`, or `subscription`.
`operationName`	The name of the GraphQL operation requested. It will be `Anonymous` if no `operationName` is found.
`fieldName`	The name of the field being resolved.
`typeName`	The name of the parent type of the field being resolved.
`returnType`	The name of the return type of the field being resolved.

Filter resolvers to instrument

To mitigate the cost of instrumenting all resolvers, you can explicitly list the fields that should be instrumented by providing a list of field names to the instrumentResolvers option.

It is a list of strings in the form of TypeName.fieldName. For example, to instrument the hello root query, you would use Query.hello.

You can also use wildcards to instrument all the fields for a type. For example, to instrument all root queries, you would use Query.*.

Troubleshooting

You can observe and troubleshoot the metrics by visiting the /metrics endpoint of your gateway. Run your gateway and execute a few GraphQL operations to produce some metrics.

Then, use the following curl command will fetch the metrics from your gateway:

curl -v http://localhost:4000/metrics

💡

Change http://localhost:4000 to the actual URL of your running gateway.

Customizations

By default, all operations are instrumented, including introspection queries. It is possible to ignore introspection queries for all metrics prefixed by graphql_envelop_ by setting the skipIntrospection option to true.

By default, all labels are enabled, but each one can be disabled to reduce cardinality:

gateway.confing.ts

const prometheusConfig = {
  labels: {
    url: false // remove `url` labels from all relevant metrics
  }
}

By providing a string, you can change the name of the metric. For example, to change the name of the name of the graphql_yoga_http_duration metric to http_request_duration, you would use:

gateway.config.ts

const prometheusConfig = {
  metrics: {
    graphql_yoga_http_duration: 'http_request_duration'
  }
}

By providing an object, you can customize the metric configuration. These configuration objects should be created using the provided factories for each metric type (createCounter, createHistogram, createSummary).

By providing a custom configuration, the default configuration is completely overridden. This means you need to provide all options, including the name and the labels.

You can look at the source code of the plugin to see the default configuration for each metric to use it as a base.

Available options depend on the metric type, and full details about them can be found in the siimon/prom-client documentation.

For example, you can customize the buckets of the graphql_yoga_http_duration histogram metric:

gateway.config.ts

import { register as registry } from 'prom-client'
import { createHistogram, defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  prometheus: {
    metrics: {
      graphql_yoga_http_duration: createHistogram({
        registry,
        histogram: {
          name: 'graphql_yoga_http_duration',
          help: 'Time spent on HTTP connection',
          labels: ['method', 'statusCode', 'operationName', 'operationType'],
          buckets: [0.1, 5, 15, 50, 100, 500]
        },
        fillLabelsFn(params, { request, response }) {
          return {
            method: request.method,
            statusCode: response.status,
            operationType: params.operationType,
            operationName: params.operationName || 'Anonymous'
          }
        }
      })
      // ... rest of metrics ...
    }
  }
})

You can customize the client’s registry by passing a custom registry to the registry option.

gateway.config.ts

import { Registry } from 'prom-client'
 
const myRegistry = new Registry()
 
const prometheusConfig = {
  registry: myRegistry
}

In some cases, the large variety of label values can lead to a huge amount of metrics being exported. To save bandwidth or storage, you can reduce the amount of reported metrics by multiple ways.

Monitor only some phases

Some metrics observe events in multiple phases of the graphql pipeline. The metric with the highest chance causing large amount of metrics is graphql_envelop_error_result, because it can contain information specific to the error reported.

You can lower the amount of reported errors by changing the phases monitored by this metric.

import { execute, parse, specifiedRules, subscribe, validate } from 'graphql'
import { Registry } from 'prom-client'
import { envelop, useEngine } from '@envelop/core'
 
const myRegistry = new Registry()
 
const getEnveloped = envelop({
  plugins: [
    useEngine({ parse, validate, specifiedRules, execute, subscribe }),
    usePrometheus({
      metrics: {
        // To ignore parsing and validation error, and only monitor errors happening during
        // resolvers executions, you can enable only the `execute` and `subscribe` phases
        graphql_envelop_error_result: ['execute', 'subscribe']
      }
    })
  ]
})

Skip observation based on request context

To save bandwidth or storage, you can reduce the amount of reported values by filtering which events are observed based on the request context.

For example, you can only monitor a subset of operations, because they are critical or that you want to debug it’s performance:

import { execute, parse, specifiedRules, subscribe, validate } from 'graphql'
import { envelop, useEngine } from '@envelop/core'
import { usePrometheus } from '@envelop/prometheus'
 
const TRACKED_OPERATION_NAMES = [
  // make a list of operation that you want to monitor
]
 
const getEnveloped = envelop({
  plugins: [
    useEngine({ parse, validate, specifiedRules, execute, subscribe }),
    usePrometheus({
      metrics: {
        graphql_yoga_http_duration: createHistogram({
          registry,
          histogram: {
            name: 'graphql_yoga_http_duration',
            help: 'Time spent on HTTP connection',
            labelNames: ['operation_name']
          },
          fillLabelsFn: ({ operationName }, _rawContext) => ({
            operation_name: operationName
          }),
          shouldObserve: context => TRACKED_OPERATIONS.includes(context?.params?.operationName)
        })
      }
    })
  ]
})

StatsD

You can use @graphql-mesh/plugin-statsd plugin to collect and send metrics to Datadog’s DogStatsD and InfluxDB’s Telegraf StatsD services.

npm i @graphql-mesh/plugin-statsd hot-shots

pnpm add @graphql-mesh/plugin-statsd hot-shots

yarn add @graphql-mesh/plugin-statsd hot-shots

bun add @graphql-mesh/plugin-statsd hot-shots

Compatible with:

Datadog’s DogStatsD server
InfluxDB’s Telegraf StatsD server
Etsy’s StatsD serve

Available metrics:

graphql.operations.count - the number of performed operations (including failures)
graphql.operations.error.count - the number of failed operations
graphql.operations.latency - a histogram of response times (in milliseconds)
graphql.delegations.count - the number of delegated operations to the sources
graphql.delegations.error.count - the number of failed delegated operations
graphql.delegations.latency - a histogram of delegated response times (in milliseconds)
graphql.fetch.count - the number of outgoing HTTP requests
graphql.fetch.error.count - the number of failed outgoing HTTP requests
graphql.fetch.latency - a histogram of outgoing HTTP response times (in milliseconds)

💡

You can also customize the graphql prefix and add custom tags to the metrics.

Usage Example

gateway.config.ts

import { StatsD } from 'hot-shots'
import { defineConfig } from '@graphql-hive/gateway'
import useStatsD from '@graphql-mesh/plugin-statsd'
 
export const gatewayConfig = defineConfig({
  plugins: pluginCtx => [
    useStatsD({
      ...pluginCtx,
      // Configure `hot-shots` if only you need. You don't need to pass this if you don't need to configure it.
      client: new StatsD({
        port: 8020
      }),
      // results in `my-graphql-gateway.operations.count` instead of `graphql.operations.count`
      prefix: 'my-graphql-gateway',
      // If you wish to disable introspection logging
      skipIntrospection: true
    })
  ]
})

Sentry

This plugin collects errors and performance tracing for your execution flow, and reports it to Sentry.

This is how it looks like in Sentry for error tracking:

Example Example

💡

The operation name, document, variables are collected on errors, and the breadcrumbs that led to the error. You can also add any custom values that you need.

To get started with Sentry, you need to create a new project in Sentry and get the DSN:

Start by creating an account and a project in https://sentry.io
Follow the instructions to setup your Sentry instance in your application.
Setup Sentry global instance configuration.
Setup the Envelop plugin.

Then, install the following plugin in your project:

yarn add @sentry/node @sentry/tracing @envelop/sentry

Usage Example

gateway.config.ts

import '@sentry/tracing' // import only once in you entry file!
import { useSentry } from '@envelop/sentry'
import { defineConfig } from '@graphql-hive/gateway'
 
export const gatewayConfig = defineConfig({
  plugins: () => [
    useSentry({
      includeRawResult: false, // set to `true` in order to include the execution result in the metadata collected
      includeResolverArgs: false, // set to `true` in order to include the args passed to resolvers
      includeExecuteVariables: false, // set to `true` in order to include the operation variables values
      appendTags: args => {}, // if you wish to add custom "tags" to the Sentry transaction created per operation
      configureScope: (args, scope) => {}, // if you wish to modify the Sentry scope
      skip: executionArgs => {} // if you wish to modify the skip specific operations
    })
  ]
})

Configuration

startTransaction (default: true) - Starts a new transaction for every GraphQL Operation. When disabled, an already existing Transaction will be used.
renameTransaction (default: false) - Renames Transaction.
includeRawResult (default: false) - Adds result of each resolver and operation to Span’s data (available under “result”)
includeExecuteVariables (default: false) - Adds operation’s variables to a Scope (only in case of errors)
appendTags - See example above. Allow you to manipulate the tags reports on the Sentry transaction.
configureScope - See example above. Allow you to manipulate the tags reports on the Sentry transaction.
transactionName (default: operation name) - Produces a name of Transaction (only when “renameTransaction” or “startTransaction” are enabled) and description of created Span.
traceparentData (default: {}) - Adds tracing data to be sent to Sentry - this includes traceId, parentId and more.
operationName - Produces a “op” (operation) of created Span.
skip (default: none) - Produces a “op” (operation) of created Span.
skipError (default: ignored GraphQLError) - Indicates whether or not to skip Sentry exception reporting for a given error. By default, this plugin skips all GraphQLError errors and does not report it to Sentry.
eventIdKey (default: 'sentryEventId') - The key in the error’s extensions field used to expose the generated Sentry event id. Set to null to disable.

Authorization / Authentication Incremental Delivery (Defer & Stream)