[Cluster] "AssertionError: Resource leak detected" issue from 12.7 still occurs

[Smbc1]

After moving from node v0.10.38 to v0.12.7 I found an error that occurs sometimes (from 0 to 4-5 in a day) only in production. Process output the same with described in this old topics:
nodejs/node-v0.x-archive#8860
nodejs/node-v0.x-archive#9409

Is it fixed or it's so minor that fix is skipped for a half of year? Is there any workarounds (described in #9404 is not working and just increasing damage by stuck all over the clusterized processes).

[Smbc1]

nodejs/node-v0.x-archive#8792

[Smbc1]

Reproduction of error:

1. Create run.js:

var cluster = require('cluster');
cluster.schedulingPolicy = 2; // 1 - none (no bug), 2 - round robin (bug)

if (cluster.isMaster){
    const WORKERS_COUNT = 4;
    for (var i = 0; i < WORKERS_COUNT; i++) {
        cluster.fork();
    }
} else {
    var http = require('http'),
        port = process.env.PORT || '3000';

    var srv = http.createServer(function(req, res) {
        if (Math.random() < 0.1) throw new Error('Test error!');
        res.writeHead(200);
        res.write('hello');
        res.end();
    });
    srv.listen(port, function() {
        console.log('Worker is ready');
    });
}

1. Run: node run.js
2. Run: ab -n10000 -c100 http://localhost:3000/
   Cannot reproduce bug with NODE_CLUSTER_SCHED_POLICY=none (there is schedulingPolicy = 1), only width NODE_CLUSTER_SCHED_POLICY=rr (schedulingPolicy = 2)

Succesfully reproduced on node 4.1.2 and 0.12.7.

[Smbc1]

In addition, if use policy = none AND --expose_gc node crushes in C module:
nodejs/node-v0.x-archive#25648 (comment)
@indutny: it's real problem, because it denies using --expose-gc in production with clusterization at all.
Looks similar to #3551

[indutny]

Relevant: #3551

[bnoordhuis]

@Unterdrucker Does it happen with master or v5.0.0-rc.2 or v4.2.2-rc.1? A fix was landed recently in da21dba and is about to be released. FWIW, I can't reproduce what you're seeing with master.

[Smbc1]

@bnoordhuis yes, it's doesn't happen in 5.0.0, thank you, I am going to try it in production.

[cjihrig]

Sounds like this has been fixed, so I'm going to close this. Please let me know if it needs to be reopened.