This is a load generator for GraphQL subscriptions.
- Run
npm installto install deps - Run
npm run compileto compile tsc which will output the files indistdirectory.
# query to be subscribed
query: |
subscription ($id: Int!) {
test(where: {id: {_eq: $id}}) {
id
name
}
}
# no of connections per seccond
timePeriod: 2
# variables that needs to set while subscribing
variables:
id: '1'
# headers that needs to be set while creating the connection
headers: []
template:
range:
# start vaule of the variables or headers
start: 1
# end value of the variables or headers
end: 10
# headers that needs to be templated using the range start and end
headers: []
# variables that needs to be templated using the range start and end
variables:
- id-
Create the
eventstable using knex migrate. Runknex migrate:latest -
Run the benchmark tool
PG_CONNECTION_STRING=<posgtres_connection_url> ENDPOINT=<graphql_endpoint> LABEL=<unique_label> CONFIG_FILE_PATH=<path_to_config_file> node index.js
-
Create events by making changes in the subscribed table
-
To stop the subscription, execute
NOTIFY benchmarkin benchmark postgres instance which will trigger a notification to stop the benchmark script
exports.up = function(knex, Promise) {
return knex.schema.createTable('events', (table) => {
table.string('label').notNullable(); // unqiue label to identify benchmark
table.integer('connection_id').notNullable(); // connection_id represents the nth connection
table.integer('operation_id').notNullable();
table.integer('event_number').notNullable(); // event_number represents the nth event that was receieved by the client
table.jsonb('event_data').notNullable(); // event_data stores the data that was received this can be used to validate
table.timestamp('event_time', { useTz: true }).notNullable(); // event_time stores the time at which the event was receieved by the client.
table.boolean('is_error').notNullable(); // is_error represents whether the event was error or not.
table.integer('latency'); // latency is not populated by the benchmark tool, but this can be populated by calculating `event_time-event_triggerd_time`
table.unique(['label','connection_id', 'operation_id', 'event_number'])
});
}
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent