Uncontrolled memory usage growth when covering a large number of service

[dvhh]

covering >20 services ( using the docker container ), the memory usage is increasing steadily until the memory limit is reached (~1GB)

[peterbourgon]

Which memory metric is this?

[peterbourgon]

Can you give https://github.com/peterbourgon/fastly-exporter/releases/tag/v2.2.1 a shot?

[dvhh]

Hello, Sorry for the late reply, will give the version test and will report if same issue occur, unfortunately the Memory metric is an ad-hoc monitoring expressing memory usage in % of the container ( an probably missing a few beats ).

metrics for :

• go_memstats_heap_alloc_bytes (stable)
• go_memstats_heap_idle_bytes (increase during lifetime)
• go_memstats_heap_sys_bytes

These metrics are produced before using the suggested version

[dvhh]

Started running the new version, will update with reported metrics in ~24h

[peterbourgon]

Bear in mind also that the operating system may not necessarily reclaim idle memory and thus steadily increasing values there don't necessarily imply a memory leak.

[dvhh]

current status

[peterbourgon]

Cool, I’ll call this fixed unless you object.

[dvhh]

No issue, the application seems to be more stable in its memory consumption.
Feel free to close the ticket.

[dvhh]

Just to update on the status
based on limits for killing containers (limited to 1GB)

• version 2.2.0 : ~30h uptime for a container
• version 2.2.1 : ~34h uptime for a container
  Thanks for the support, and the clarification about where the issue mainly remain

[dvhh]

version 3.0.1: uptime ~5 days and ongoing, memory usage stable, thanks for your work

[peterbourgon]

Great news, thanks for the reports.