What impact would "HTTP Runtime Metrics" have on my setup?

[janeklb]

I noticed that #2387 was recently merged, and I'm wondering what impact this'll have on the dd-trace setup in my service.

Is the purpose of this change to simplify the dd-trace-js codebase a bit, or does it actually introduce / change some functionality?

Here's how i've got dd-trace configured

DD_PROFILING_ENABLED=true
DD_LOGS_INJECTION=true
DD_RUNTIME_METRICS_ENABLED=true
DD_ENV=<something>
DD_SERVICE=<something>
DD_VERSION=<something>
DD_ENTITY_ID=<something>
DD_AGENT_HOST=<something>

import tracer from 'dd-trace';

tracer.init({ tags: { owner: '<something>' } });
tracer.use('express', {
  blocklist: ['/liveness', '/api/liveness'],
});

export default tracer;

[rochdev]

This change allows us to support additional features like reporting metrics over UDS. It does fallback to DogStatsD over UDP when the endpoint is not available on the agent because the version installed is too old so it shouldn't make any difference for an existing service.

[janeklb]

Thanks @rochdev !

[janeklb]

Do new versions of the agent automatically support UDS, or is there some explicit flag you have to enable (or, on the flip side, is it possible to disable it if it's enabled by default).

[rochdev]

I'm not 100% sure of the conditions when it's enabled or not, but you can find the information here: https://docs.datadoghq.com/developers/dogstatsd/unix_socket/