High load on node library when CPU profiling about help HOT 4 OPEN

Thanks for your response. elastic-apm-node is a module I added to help me tracking down my performance issue 😕
I will try without this module.

from help.

I finally obtained a flamegraph, thanks to 0x. As I feared, it seems my high CPU is mostly coming from my HTTP sending, using axios. I will try using a Worker to avoid blocking my app.
FYI, what we were seeing on elastic-apm-node seems to only be a wrapper, hiding the real culprit.

Thanks.

from help.

After that issue, I brushed up on my knowledge of profiling reports, but I'm still a bit rusty, but here's a go:

 [Shared libraries]:
   ticks  total  nonlib   name
  100357   61.4%          /usr/local/nvm/versions/node/v18.9.0/bin/node
  37636   23.0%          /usr/lib/x86_64-linux-gnu/libc-2.31.so
   4053    2.5%          /usr/lib/x86_64-linux-gnu/libpthread-2.31.so
    339    0.2%          /usr/lib/x86_64-linux-gnu/libstdc++.so.6.0.28
    130    0.1%          [vdso]

This section, at the top, refers to what parts are using how much CPU. As you can see, the CPP libraries are using some, as they are the backbone of Node.js.

Node itself is using a large portion of the CPU, typically in response to whatever code you are running.

Looking down at the report, elastic-apm-node (especially the http-apm-client files) appears to be using a decent amount of CPU, so maybe that's the bottleneck?

Additionally, there's also flamegraphs. These are human-visualizable graphs of CPU usage. They could also help you identify what parts of your script are using the most CPU (https://nodejs.org/en/learn/diagnostics/flame-graphs)

from help.

Thanks for your response. elastic-apm-node is a module I added to help me tracking down my performance issue 😕

I will try without this module.

Haha. Well, according to the profile report, it looks like knex is also using a lot of CPU. I highly recommend flamegraphs to determine what functions are using the most CPU.

from help.