Using meteorhacks:cluster my 8 core processor is running the job 8 times

[jamesgibson14]

I just started testing the Meteorhack:cluster package and the multi-core flag, but when I started up my project syced-cron started running all the jobs 8 times (8 core machine).

[zol]

Hey @jamesgibson14 - Did you find a workaround to this? If not, can you provide a repo with a reproduction?

[jamesgibson14]

No I didn't, I just turn off multi-core for now.

On Mon, Sep 14, 2015 at 3:41 PM, Zoltan Olah notifications@github.com
wrote:

Hey @jamesgibson14 https://github.com/jamesgibson14 - Did you find a
workaround to this? If not, can you provide a repo with a reproduction?

—
Reply to this email directly or view it on GitHub
#56 (comment)
.

[babnik63]

I have the same problem. ...!

[aesmon]

Any updates on this problem ?

[ko0stik]

I had the same issue I managed to circumvent by checking the worker number and the environment:

if (parseInt(process.env.CLUSTER_WORKER_ID, 10) === 1 || process.env.NODE_ENV === 'development') { your tasks }

It's dirty, I'm ashamed to show such code but it worked so far.

[chneau]

Can we have an update. I'm not really sure but I have 4 clusters on the same computer doing a mongodump. It looks like some of them don't do the task and let only one do the tasks.
I can see this message << SyncedCron: Not running "SaveDatabase" again. >>

After quick log of the process.env.CLUSTER_WORKER_ID on the job, it looks like the last worker do the task.

[sferoze]

@ko0stik I am using meteorhacks:cluster package as well on a 2 core system.

To prevent the jobs from being scheduled twice I used the code you posted.

Just wondering if you've had any issues with it? Seems to do the trick

[ko0stik]

Hey @sferoze. I stopped using meteorhacks:cluster a while ago when we switched to Galaxy. But it perfectly worked during the whole time we've been using it (until June 2016 I'd say).

[luixal]

Same problem and same solution over here.

In fact, We've noticed that process.env.CLUSTER_WORKER_ID is present in all workers but one (the first launched I guess) so this is working for us:

if (!proces.env.CLUSTER_WORKER_ID) {
  // code here runs only in one instance
}