Investigate flaky tests

[mcollina]

Tests are really flaky in this project.
I think most of them are because of I/O, timers and reconnects.

They should be investigated and fixed

[mcollina]

cc @marco-ippolito

[marco-ippolito]

Ill look into it, this one seems to be the culprit:

mercurius-gateway/test/retry-failed-services.js

Lines 385 to 448 in eddd8be

	test('gateway - should log error if retry throws', async t => {
	t.plan(1)
	const service1 = await createTestService(
	0,
	userService.schema,
	userService.resolvers
	)
	let service2 = null
	t.context.clock.setTimeout(async () => {
	service2 = await createTestService(
	5114,
	postService.schema,
	postService.resolvers
	)
	}, 2000)
	const gateway = Fastify()
	let errCount = 0
	gateway.log.error = error => {
	if (error.message.includes('kaboom') && errCount === 0) {
	errCount++
	t.pass()
	}
	}
	t.teardown(async () => {
	await gateway.close()
	await service1.close()
	await service2.close()
	})
	await gateway.register(plugin, {
	gateway: {
	services: [
	{
	name: 'user',
	url: `http://localhost:${service1.server.address().port}/graphql`,
	mandatory: false
	},
	{
	name: 'post',
	url: 'http://localhost:5114/graphql',
	mandatory: true
	}
	],
	retryServicesCount: 1,
	retryServicesInterval: 3000
	},
	jit: 1
	})
	gateway.graphqlGateway.addHook('onGatewayReplaceSchema', async () => {
	throw new Error('kaboom')
	})
	await gateway.ready()
	for (let i = 0; i < 200; i++) {
	await t.context.clock.tickAsync(50)
	}
	})

I'll come up with a fix
@mcollina I dont have rights on this repo

[mcollina]

still flaky: https://github.com/mercurius-js/mercurius-gateway/actions/runs/4023965438/jobs/6915447409#step:5:1367

[marco-ippolito]

I'll try another approach on this test probably it's better to rewrite it completely.

[tomi-bigpi]

Possibly related/FYI:

Tests in our application warn with A worker process has failed to exit gracefully and has been force exited. This is likely caused by tests leaking due to improper teardown. Try running with --detectOpenHandles to find leaks. Active timers can also cause this, ensure that .unref() was called on them.

This happens when we have reconnect enabled:

            wsConnectionParams: {
              reconnect: true,

We're disabling reconnects during tests as a workaround for the time being.