You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is this is a general node.js question? Not specific to job-collection?
In general, because node.js is single threaded, expensive calculations should be avoided specifically because the block the event-loop. There is no automatic "suspending" of computation in node.js.
Where job-collection can help is by helping to move expensive computations out of your server's event loop and onto separate vanilla node.js worker processes (either on the same physical server, or on other machines reachable from it). This is the point of the meteor-job npm package. You must manage these worker processes yourself, and must establish and maintain their DDP connections to the server(s) hosting the job-collection. But after that, job-collection will distribute the work, and collect progress, status, results, etc.
fine. I think ill just try scaling horizontally first and try and break up this computation. I'm hoping that there's a way to look at server load before accepting work? Mongo server uses significant amount of ram for each connected meteor.js (DDP, I suppose), looking to double hosting servers as workers is a valid form of laziness.
So if I were to run very expensive computation, what would happen is that it would suspend as soon as something should be done in event-loop?
The text was updated successfully, but these errors were encountered: