Intermittent Data Collection Gaps in Node.js with dd-trace.js

[mogulist]

Hello,

I have been experiencing intermittent data collection gaps while using the Datadog dd-trace.js library to monitor my Node.js application, which is serving a Next.js application. The issue manifests as periods where no data is collected, sometimes lasting for hours or even days. During these periods, metrics like heap memory, CPU usage, and event loop delay are not recorded, leading to gaps in my monitoring graphs.

I have been experiencing this for almost a year. Tried to upgrade dd-trace frequently but in vain.

Let me show you the graphs:

runtime.node.heap_used

runtime.node.mem.rss

runtime.node.cpu.{total | user | system}

Regarding event loop

event_loop.delay.95percentile shows gap, whereas event_loop.delay.{sum | count} doesn't show any gap. It is puzzling why some event_loop metrics show data collection gaps while others consistently collect and display data correctly.

Runtime environment details

• Node.js: 18.18.2
• Netxt.js: 14.2.4
• dd-trace: 5.10.0
• Runs on AWS with K8S, Alpine Linux v3.18
• environment variables injetcted for dd-trace
  • DD_APPSEC_ENABLED=true
  • NODE_OPTIONS: "-r dd-trace/init --max-old-space-size=6144"