v8 heap_size_limit

[sakari]

Would it be possible to add heap_size_limit gauge to the default nodejs metrics that get collected?

It would be useful for displaying an upper limit for heap usage of node processes

[riosje]

take a look to N|solid it is a nodejs alternative 100% compatible which provides telemetry out of the box, no instrumentation needed.
Even the metrics you're looking for and much more.

I wrote a blog post some time ago about how to integrate it with prometheus+grafana, but currently you can connect anything that supports OTLP

[jdmarshall]

{
total_heap_size: 6897664,
total_heap_size_executable: 524288,
total_physical_size: 6356992,
total_available_size: 4340358888,
used_heap_size: 5763352,
heap_size_limit: 4345298944,
malloced_memory: 278632,
peak_malloced_memory: 1114624,
does_zap_garbage: 0,
number_of_native_contexts: 1,
number_of_detached_contexts: 0,
total_global_handles_size: 8192,
used_global_handles_size: 5344,
external_memory: 2302264
}

I support this. Being able to chart response time curves versus max heap size would be a useful comparison for capacity planning and server tuning purposes.

"Should I run 32 instances per machine at N memory limit, or 30 copies and increase the memory limit per instance by 5%?" is just the sort of question I've been asked in the past.