you might be running into an issue with nestJS apps is creating large garbage collections (thats if your service is handling JSON payloads or uses redis). you can add:

NODE_OPTIONS="--trace_gc" or use clinic doctor -- node dist/main.js to be able to debug the high VCPU

you can also redeploy and enabled more detailed logs, so something like:

import { Logger } from '@nestjs/common';

async function bootstrap() {

const app = await NestFactory.create(AppModule, { logger: ['error', 'warn', 'log', 'debug'] });

await app.listen(process.env.PORT || 3000, '0.0.0.0');

Logger.debug('App started with full logging');

}

or CPU level diagnostics:

import { performance, PerformanceObserver } from 'node:perf_hooks';

const obs = new PerformanceObserver((items) => {

console.log(items.getEntries());

});

obs.observe({ entryTypes: ['function'] });

in short, it can also be a thread pool issue when you are performing async work

Thanks, i will run those diagnosis tools.
I use microservices with a separated instance of rabbitmq as a message broker https://docs.nestjs.com/microservices/rabbitmq.

it does explain why - rabbitMQ typically wouldnt have logs but can have high vCPU. look for duplicate consumer tags in rabbitmq, or deploy timestamps vs vCPU spikes. adding the logging above will help pinpoint the exact issue though