EADDRINUSE

[OmgImAlexis]

I'm getting an error like this. The code below is a small example but should show the setup I'm using. I noticed around 5k/s connections I'm seeing the error in 1/5000 connections.

(node:7495) UnhandledPromiseRejectionWarning: Error: listen EADDRINUSE 127.0.0.1:6970
    at Object._errnoException (util.js:1022:11)
    at _exceptionWithHostPort (util.js:1044:20)
    at Server.setupListenHandle [as _listen2] (net.js:1367:14)
    at listenInCluster (net.js:1408:12)
    at doListen (net.js:1517:7)
    at _combinedTickCallback (internal/process/next_tick.js:141:11)
    at process._tickCallback (internal/process/next_tick.js:180:9)

import { LMXBroker, LMXClient } from 'live-mutex';

const wrappedLog = (name: string, level = 'debug') => (...args: any[]) => (log[level] || log.debug)(name, ...args);

let mutexClient;

const getMutex = async () => {
    if (mutexClient) {
        return mutexClient;
    }

    const [broker, client] = await Promise.all([
        new LMXBroker().ensure(),
        new LMXClient().connect()
    ]);

    broker.emitter.on('warning', wrappedLog('broker', 'warn'));
    client.emitter.on('warning', wrappedLog('client', 'warn'));

    mutexClient = client;

    return client;
};

// This function will be called by multiple requests
// I've just simplified to show the issue
const main = async () => {
    // Connect to mutex broker
    const mutex = await getMutex();

    // Lock mutex
    const lock = await mutex.acquireLock('registered').catch(error => {
        log.error(error);
        server.close();
    });

    if (!lock) {
        return;
    }

    // Unlock mutex
    await mutex.releaseLock('registered', { id: lock.id });
};

main().catch(error => {
    console.error(error);
});

[ORESoftware]

you shouldnt (must not) cluster the broker - there needs to be only one broker.

the docs arent that clear on this, i can make that clearer.

if you run only one broker, does the problem occur?

or maybe thats not the problem.
let me take a look at your code tomorrow, i gtg to sleep. are you using the docker version of the broker? how are you running it?

[OmgImAlexis]

Yeah the docs seem hit and miss.

Currently I just need a setup where I can use await and get a lock, then once finished unlock it. I’m using this 100% in node. No docker. I’m also only using a single process but may need to scale out later on.

[ORESoftware]

do you have a full repro? something where I can get it to 5000k/s?

[ORESoftware]

so this is likely what's happening:

const lock = await mutex.acquireLock('registered').catch(error => {
    log.error(error);
    server.close();   // when a lcck cannot be acquired, it will reach here, so the server is restarted
});

I put your code here:
https://github.com/ORESoftware/live-mutex-examples/blob/master/src/dual/repro.ts

the code style in the above is probably better than what you have, which is fine, I can improve the docs

[ORESoftware]

i think the main problem was calling:

   server.close();

you should only call that if you are shutting down the process.

I made a patch, it's here:

• live-mutex@0.2.23

i fixed a bug with regard to client connection

[ORESoftware]

so try your code here:

https://github.com/ORESoftware/live-mutex-examples/blob/master/src/dual/repro.ts

with version live-mutex@0.2.23

and lmk know if it works for you

[OmgImAlexis]

@ORESoftware server.close() is the websocket connection I had open. It needs to happen if the lock isn't acquired.

[OmgImAlexis]

So this seems to work apart from this. I've added catch handlers to all promises I'm using. I'm thinking this may be an error that's thrown in live-mutex itself?

(node:2662) UnhandledPromiseRejectionWarning: lmx client err: client connection timeout after 3000ms.
(node:2662) UnhandledPromiseRejectionWarning: Unhandled promise rejection. This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). (rejection id: 548)

[OmgImAlexis]

Fixed the issue above, I missed the Promise.all.

I still get this though on mass connections. I'm trying to handle 5000 ws connections a second with each needing to use a mutex.

client warning: Connection was closed (and/or a client connection error occurred.)
Couldn't acquire lock for registered:nqkgk LMXClientLockException {
  id: '626bdccc-d587-47ea-9027-96e102d9cf7c',
  key: 'registered:nqkgk',
  code: 'connection_closed',
  message: 'Connection was closed (and/or a client connection error occurred.)',
  stack: 'Connection was closed (and/or a client connection error occurred.)'
}

[ORESoftware]

sure can you share a full repro?

[OmgImAlexis]

I can try but I'm not 100% sure I'll be able to. I'm going to try what you've suggested in the other issue.