Clear kResourceStore in timers and intervals

[mcollina]

As noted in this article, the timeout/interval objects keep a reference to the store even after they have been cleared.

I think it would be better to clear up that reference inside the clearTimeout() and clearInterval() functions. We might also want to consider exposing an API for end-users and explicit lifetime tracking.

[mcollina]

cc @nodejs/diagnostics @anonrig

[Qard]

setTimeout should probably actually just clear immediately after running given that it only occurs once. But yes, also clearing in clearInterval makes sense to me too.

[Flarna]

I guess this is not only relevant for timers and intervals, all resources are effected.

const { AsyncLocalStorage, AsyncResource } = require('node:async_hooks');
const als = new AsyncLocalStorage();
let r, p;
als.run("I consume a lot memory", () => {
  r = new AsyncResource("aResource", { requireManualDestroy: true })
  r.emitDestroy()
  p = Promise.resolve()
})
console.log(r, p);

 AsyncResource {
  [Symbol(async_id_symbol)]: 2,
  [Symbol(trigger_async_id_symbol)]: 1,
  [Symbol(kResourceStore)]: 'I consume a lot memory'
} Promise {
  undefined,
  [Symbol(async_id_symbol)]: 3,
  [Symbol(trigger_async_id_symbol)]: 1,
  [Symbol(kResourceStore)]: 'I consume a lot memory'
}

For timers/intervals it's likely easier to find the place to cleanup.
Maybe we can extend the cleanup to some more places?

[mcollina]

Absolutely!

[Flarna]

timers allow similar "leaks" even without AsyncLocalStore if one uses arguments for the handler:

const t = setTimeout(() => {}, 100, "I consume a lot memory");
clearTimeout(t);
console.log(t);
console.log(t._timerArgs);

Timeout {
  _idleTimeout: -1,
  _idlePrev: null,
  _idleNext: null,
  _idleStart: 22,
  _onTimeout: null,
  _timerArgs: [Array],
  _repeat: null,
  _destroyed: true,
  [Symbol(refed)]: true,
  [Symbol(kHasPrimitive)]: false,
  [Symbol(asyncId)]: 2,
  [Symbol(triggerId)]: 1
}
[ 'I consume a lot memory' ]

If handler uses some variable from the outer scope it's even more.