My goal is to run a data import against an REST endpoint.
What I don't want is to wait for a request to be resolved before I fire a new one. I want to 'simulate' parallel connections.
And I am not sure I have a basic knowledge problem here.
This code creates child processes:
const numchild = require('os').cpus().length;
const SIZE = 1000;
const SIZE_PER_CHILD = SIZE / numchild;
for (let i = 0; i < numchild; i++) {
const child = child_process.fork('./child.js');
child.send({ start: i * SIZE_PER_CHILD, end: (i + 1) * SIZE_PER_CHILD });
// Some more code...
}
Then per child process I want to generate a random payload for importing and fire it against the REST endpoint:
process.on('message', async function({ start, end }) {
const count = start;
while (count < end) {
const generatedData = 'I was generated! Yay!' + count;
await axios.put('/api/v1/import', generatedData);
count++;
}
});
The above approach would wait for every import request to finish and then fire the next one, until all child imports are done. Not what I want.
Now the endpoint I am firing against should be capable of handling more than the requests I am able to generate like this.
I can rewrite it like this:
process.on('message', async function({ start, end }) {
const count = start;
while (count < end) {
const generatedData = 'I was generated! Yay!' + count;
axios.put('/api/v1/import', generatedData).then(() => console.log('I am done with this one'));
count++;
}
});
The problem with this approach of course would be that all requests are generated within a few seconds and are fired against the endpoint. This is more like DOS-style, I guess. Also not what I want.
What I am hoping to achieve is something like: Have 15 open connections per child process. If a request finished, queue the next request until you have 15 requests pending again.
So I tried this:
process.on('message', async function({ start, end }) {
const count = start;
let queue = [];
while (count < end) {
if (queue.length === 15) {
queue = queue.filter(async (promise) => {
const state = await promiseState(promise);
return state !== 'fulfilled';
});
} else {
const generatedData = 'I was generated! Yay!' + count;
queue.push(axios.put('/api/v1/import', generatedData).then(() => console.log('I am done with this one')));
count++;
}
}
});
function promiseState(p) {
const t = {};
return Promise.race([p, t])
.then(v => (v === t) ? "pending" : "fulfilled", () => "rejected");
}
Also doesn't work and doesn't make sense, right? The filter function returns promises and therefore what I am trying to do doesn't work.
Any way I can achieve this?
Try a p-queue. Below the concurrency is set to 3, meaning max 3 calls are executed at once within this queue:
import PQueue from 'p-queue';
const queue = new PQueue({
concurrency: 3,
});
process.on('message', async function ({ start, end }) {
var calls = [];
var count = start;
while (count < end) {
const generatedData = 'I was generated! Yay!' + count;
calls.push(
queue.add(() => {
return axios
.put('/api/v1/import', generatedData)
.then(() => console.log('I am done with this one'));
})
);
count++;
}
var results = await Promise.all(calls);
});
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With