Hi all, I’m back.
Domain is at www.StarCommanderOnline.com
I’ve moved everything over to Google Cloud Compute, and the same problems as Meteor Galaxy.
I have 50 items being watched in subscription, and looks like that’s the bottle neck. Any feedback on why the game is so laggy is much obliged.
Data Intensity.
My MongoDB is maxed at 100% CPU because of the re-subscriptions I’m using.
Luckily, I found this package: https://github.com/kadirahq/subs-manager
Sadly it doesn’t look like the authors around, but the package indeed works. BEAUTIFULLY.
My localhost performs like butter, moving to production now.
Data issue is solved, she runs like butter now.
And it’s all my fault. Although Meteor Galaxy was not AS fast as I thought, which erked me, it’s probably plenty fast at base tier.
The issue was me trying to subscribe users to a post every 15 seconds. But why would one do this if there was a subscription model already? Because I’m dumb and am used to MySQL. The real reason is because I was trying to load in ‘what you can see’ every 15 seconds.
I said screw it, load ALL THE DATA, then I turned on ‘fog of war’ in the helper function instead. So while all objects are there, which is seemingly bad idea - it’s as smooth as butter.
I still have a lot to learn about data subscription.
Cheers Meteor, stay metorishy.
Galaxy has been faster for us compared to other options. Galaxy is hosted on AWS, which is pretty much the standard on the internet.
Also, from how you describe it, it sounds very much like the issue is NOT your application, but your stress on the database. This is not your hosting provider causing this - whether Galaxy or Google or AWS directly. It’s the database that’s being stressed.
Regarding your solution, I can guarantee there are better ways to handle it. But it is very hard as you have not gave much information on how your application works. It’s hard for us to understand what you mean without some examples of what your subscription/publications/methods are, and/or your data structures.
So, stabilized on a Google Compute setup as such:
Meteor Server: 4 CPU cores, 8 gigs ram, running great.
MongoDB Server: 4 CPU cores, 8 gigs of ram as well.
So, I believe the Galaxy servers probably were running OKAY, the DB server hosted at mLab needed more power.
The price on these front ends almost doubles your monthly bill though. As a startup, that’s just insane. Maybe in the future I’ll migrate back for peace of mind. Hope this article is helpful in the future.