use api to call flamegraph?

[zdyj3170101136]

to reduce overhead, i want to dynamic start profile and stop it.

this http request http://:8081/debug/pprof/allocs would start allocation profiling 60 seconds and stop it.

from aiohttp import web
import asyncio

prof = Tracer()

async def handle(request):
    asyncio.create_task(heap_profile())
    return web.Response(text="Request received")

async def heap_profile():
    prof.__enter__()
    print("started")
    await asyncio.sleep(60)
    prof.__exit__()
    print("stoped")

app = web.Application()
app.add_routes([web.get('/debug/pprof/allocs', handle)])

web.run_app(app)

i know how to produce the bin file by using api.

but how could i use the api to convert bin file to flamegraph?

[pablogsal]

but how could i use the api to convert bin file to flamegraph?

Currently there is no such API. You could shell out to memray to run memray flamegraph ... or use subprocess.run. We can discuss if we can provide such an API in the future but at the time this is the only way that's supported.