Getting 413 from Datadog when sending metrics

[aaleksandrov]

Expected Behavior

We send metrics from lambda to Datadog using sendDistributionMetricWithDate (it could be 10k metrics within one invocation). We expect to see metrics on Datadog

Actual Behavior

We receive 413
message | datadog:failed attempt to send metrics to Datadog. HTTP error code: 413

I see a similar issue was raised for the python library and it was fixed by implementing compression. is it possible to do the same?

Steps to Reproduce the Problem

Send too many metrics

Specifications

• We are not using layer, we use this library directly, version 7.104.0
• Node version: latest node20 supported by lambda

Stacktrace

......:/aws/lambda/anomaly-detection-prod
2024/02/27/[$LATEST]3bcabca478e6484eba931ad8693a0dfb
2024-02-27T14:13:49.022Z	5bfee822-85f3-4b9c-a480-ca79831b8afe	ERROR	[dd.trace_id=763581508228917774 dd.span_id=4987919669838945685] {"status":"error","message":"Failed to send metrics to Datadog","name":"Error","stack":"Error: Failed to send metrics to Datadog\n    at Processor.<anonymous> (/var/task/node_modules/datadog-lambda-js/src/metrics/processor.ts:96:13)\n    at step (/var/task/node_modules/datadog-lambda-js/dist/metrics/processor.js:33:23)\n    at Object.throw (/var/task/node_modules/datadog-lambda-js/dist/metrics/processor.js:14:53)\n    at rejected (/var/task/node_modules/datadog-lambda-js/dist/metrics/processor.js:6:65)\n    at processTicksAndRejections (node:internal/process/task_queues:95:5)"}
.js:6:65)    at processTicksAndRejections (node:internal/process/task_queues:95:5)
error

[astuyve]

Hey @aaleksandrov, thanks for this note. We can definitely enable compression, metrics should compress well.

For anyone who may find this ticket and want to work on it, I believe we'd simply need to gzip the payload and set the Content-Encoding: deflate header, as per this pr.

Thanks!

[aaleksandrov]

@astuyve Could you check the related PR please? #492

[astuyve]

This is out in v8.107.0.

Thanks again for the contribution!