A Mercurius
plugin that exports some execution info in a query.
The additional information currently exported by the plugin are:
- profiling of resolvers execution time
- number of call per resolver
The information is added to the extensions.explain
attribute in the GQL response.
{
extensions: {
explain: {
version: '1.1.1', // The version in package.json
profiler: {
...
},
resolverCalls: {
...
}
}
}
}
The profiler contains the execution time of each resolver called.
data
is anarray
with the definition of the profiler entry:path
is astring
that represents the subpath of the resolverbegin
isnumber
that represents the start time in NANOSECONDSend
isnumber
that represents the end time in NANOSECONDStime
isnumber
that represents the time between begin and end in NANOSECONDS
Example:
{
extensions: {
explain: {
profiler: {
data: [
{
path: 'user',
begin: 100, // time in nanoseconds,
end: 300, // time in nanoseconds,
time: 200, // time in nanoseconds,
},
{
path: 'user.address',
begin: 301, // time in nanoseconds,
end: 400, // time in nanoseconds,
time: 99, // time in nanoseconds,
},
{
path: 'user.status',
begin: 301, // time in nanoseconds,
end: 350, // time in nanoseconds,
time: 49, // time in nanoseconds,
},
...
]
}
...
}
}
}
Every time a resolver is invoked, a counter keeps track of the call and returns a report with resolverCalls object:
data
is anarray
that contains an object for each resolver that has been called:key
is a string that define the resolver.count
is a number that define the number of calls for a resolver.
example
{
extensions: {
explain: {
resolverCalls: {
data:data: [
{ key: "Query.users", count: 1 },
{ key: "User.status", count: 2 },
{ key: "User.contacts", count: 2 },
{ key: "Contact.emails", count: 2 }
]
},
...
}
}
}
npm i fastify mercurius mercurius-explain
import Fastify from 'fastify'
import mercurius from 'mercurius'
import explain, { explainGraphiQLPlugin } from 'mercurius-explain'
const app = Fastify({ logger: true })
const schema = `
type Query {
add(x: Int, y: Int): Int
}
`
const resolvers = {
Query: {
async add(_, { x, y }, { reply }) {
return x + y
}
}
}
app.register(mercurius, {
schema,
resolvers,
graphiql: {
enabled: true,
plugins: [explainGraphiQLPlugin()]
}
})
app.register(explain, {})
app.listen({ port: 3000 })
Test:
curl -X POST -H 'content-type: application/json' -d '{ "query": "{ add(x: 2, y: 2) }" }' localhost:3000/graphql
enabled
:boolean
|function (schema, source, context) => boolean
. Enables or disables the data collection and the enrichment of the response. By default the action is enabled.
If
federated: true
, this field will be ignored and be enabled by default.
Examples:
// Data enrichment disabled
app.register(explain, {
enabled: false
}
// Data are collected and returned only if the request has 'x-mercurius-explain' header
app.register(explain, {
enabled: ({ schema, source, context }) =>
context.reply.request.headers['x-mercurius-explain']
})
-
gateway
:boolean
. If the application is a Mercurius Gateway enables or disables the collection ofextensions.explain
from the services that are part of the gateway that usemercurius-explain
. Use the helper functiongetExplainFederatedHeader
inservice.rewriteHead
in the service definition to enabled federation mode like in the example.It is required to enable extensions collector in the gateway definition. If the federated services are in federation mode it is required to use
rewriteHeaders
to forward thegetExplainFederatedHeader
Examples:
const gateway = Fastify()
await gateway.register(mercuriusGateway, {
gateway: {
services: [
{
name: 'foo',
url: `http://localhost:3000/graphql`,
collectors: {
collectExtensions: true // enabled extensions collector required
},
rewriteHeaders: (headers, context) => {
return { ...getExplainFederatedHeader(context) } // this will forward the federation header to service
}
},
{
name: 'bar',
url: `http://localhost:3001/graphql`,
collectors: {
collectExtensions: true // enabled extensions collector required
},
rewriteHeaders: (headers, context) => {
return { ...getExplainFederatedHeader(context) } // this will forward the federation header to service
}
}
]
},
...opts
})
gateway.register(mercuriusExplain, {
enabled: true,
gateway: true // gateway enabled
})
federated
:boolean
. If the application is a federated service enables or disables federation mode. Federation mode allows the service to trust any request from the gateway. The application will check the presence of a specific header in the request from the gateway and will ignore theenabled
field.
Example:
- Federated service
fastify.register(mercuriusExplain, {
enabled: explainEnabled,
federated: true // enabled federation
})
This function returns the headers required for federation mode.
getExplainFederatedHeader
: function(context)
await gateway.register(mercuriusGateway, {
gateway: {
services: [
{
name: 'foo',
url: `http://localhost:3000/graphql`,
rewriteHeaders: (headers, context) => {
return { ...getExplainFederatedHeader(context) } // this will forward the header to federated service
}
}
]
}
})
In mercurius
it is possibile to add to the self hosted GraphiQL app
the plugin mercurius-explain-graphiql-plugin to show the data returned by mercurius explain
.
This function return the required structure to initialize the plugin.
explainGraphiQLPlugin
: function(options)
options
:null
|object
options.version
:string
. The version of the GraphiQL plugin to be loaded. Default: the same major version of the backend plugin
Example:
import { explainGraphiQLPlugin } from 'mercurius-explain'
app.register(mercurius, {
schema,
resolvers,
graphiql: {
enabled: true,
plugins: [explainGraphiQLPlugin()]
}
})
The explainGraphiQLPlugin
function initializes by default the plugin with the same major version in the package.json
(eg. if the package is 3.4.5
it will load the version ^3
of the GraphiQL plugin).
It's possible to override the version by passing a parameter.
...
plugins: [explainGraphiQLPlugin({version: '3.4.5')]
// or
plugins: [explainGraphiQLPlugin({version: '^4')]
...