Random and semi-persistent ETIMEOUT

[myocytebd]

I ran into a weird problem of keep getting the FetchError below.
(node v12.16.3 (latest LTS) on Windows 7 x64;fetch 2.6.0)
Apparently the stacktrace is async error delivery and useless.

FetchError: request to URL failed, failed, reason: connect ETIMEDOUT 184.27.17.100:80
at ClientRequest.<anonymous> (node_modules\node-fetch\lib\index.js:1455:11)
at ClientRequest.emit (events.js:310:20)
at Socket.socketErrorListener (_http_client.js:426:9)
at Socket.emit (events.js:310:20)
at emitErrorNT (internal/streams/destroy.js:92:8)
at emitErrorAndCloseNT (internal/streams/destroy.js:60:3)
at processTicksAndRejections (internal/process/task_queues.js:84:21)

What I do is like send 20k GET to a server (for static resources) over ten minutes or so.
For each run there are 0-3 fetches went wrong as described below.
0. For every errored fetch, ETIMEOUT comes ~150ms after fetch -- looks weird to me.

1. The errored URLs are persistent during each node run. (Same 20k request for each run)
   fetch same URLs after 5s, and the same FetchError comes during one node run.
2. The errored URLs are random between each node run.
   Say, 1/20k went wrong in first run, and another 2/20k went wrong in second run.
3. I didn't touch any network related configurations of node or fetch.

My questions:

1. For ETIMEOUT that comes so fast, it is really timeout due to client configuration or it could be caused by server response?
2. Any suggestions of where should I start to trouble shoot it? Is there some debug switch for node and fetch?

[JefferyHus]

If I understand you well, you are doing a stress test on your server. In that case it makes total sense to have this kind of errors and it has nothing to do with the package itself. You have to trace your packets going to your server and also keep an eye on your server to see received requests and logs related to each request.

What web server are you using?