Memory leak investigation

[hananbs]

Hi
I am using Kong Api Gateway 3.1.1.
I start noticing a memory increase trend the memory just keep getting bigger and bigger.(it's run for 5. days now and keep increasing)

I am using an auto worker count calculation. so we are using 16 workers.

Are there any easy known way / best practice to analyze Kong's memory usage? any way to dump and profile usage?

Using Prometheus metrics of Kong I see that each Kong GW deployment has around :

• 1.4Gb Total Worker Virtual memory - (all 16 worker sum using the metric "kong_memory_workers_lua_vms_bytes")
• 92mb of shared dictionaries (sum(kong_memory_lua_shared_dict_total_bytes))
• besides that, we have another proxy cache which is having max size defined (not suspected to be part of the memory leak cause)

my container of my Kong Api GW is having container_memory_working_set_bytes of around 4.9Gb
the container only runs kong and i do see each Kong worker process using around 300-400mb (having 16)

Any advice on how can I investigate it further?
Any tools to dump & analyze/profile kong/nginx memory usage?

Thanks in advance