Large number of parallel requests always result in an error about undici HOT 2 CLOSED

I think you are hitting some kind of DDoS protection or kernel limit. Limiting the maximum number of sockets per hosts work:

import { fetch, Agent, setGlobalDispatcher } from 'undici';

setGlobalDispatcher(new Agent({
  connections: 100
}));
const p = []
for (let i = 0; i < 1000; i++) {
  p.push(fetch('http://example.org/').then(re => console.log(re.status)));
}
await Promise.all(p);

• keeping an internal record of pending requests, if an identical request is already pending re-use the response across the multiple pending requests
• keeping a cache of resolved requests and taking advantage of the expiry headers. Note that http://example.org/ has a 1 week expiry

These are not implemented (yet), I would recommend open a different issue to discuss them. They could be a great contribution from a volunteer.

from undici.

keeping an internal record of pending requests, if an identical request is already pending re-use the response across the multiple pending requests
keeping a cache of resolved requests and taking advantage of the expiry headers. Note that http://example.org/ has a 1 week expiry

This sounds like a good interceptor

from undici.