How to Speed Up My Site On NodeBB?

srinik

Hello All!

Though the site runs on VPS, it still loads slowly when compared to other sites on the same server. As you might be knowing a slow loading site is a dead one.

In WordPress, there are caching plugins and customization options to enable faster loading of a website.

In a similar way, I'm looking for tips to speed up my nodeBB forum.

PitaJ

Have you looked at the scaling docs?

Scaling - NodeBB Documentation

(docs.nodebb.org)

gotwf

@PitaJ said in How to Speed Up My Site On NodeBB?:

Have you looked at the scaling docs?

Scaling - NodeBB Documentation

(docs.nodebb.org)

I can attest that this config fairly rocks. Proxy for static content access coupled with Redis for session management. Yippie yi-yo kiyah!

OTOH: Targeting KISS, I'd just rock a single mongod and scale vertically, i.e. toss more $$ at it...

Note: Were I to run a sharded/replicated mongodb backend then I'd seriously consider having mongod handle session management and cut redis. My $0.02. You get what you pay for. If'n yer' lucky & I'm having a good day....

srinik

@PitaJ

Thank you ...will look into it.

cryptoethic

Hey guys!

What about cronJobs on multiple instances? Are they executing on every instance of nodebb or only on first for example?

cryptoethic

Ok guys I have found the answer:

In prestart.js there is:

nconf.set('runJobs', nconf.get('isPrimary') && !nconf.get('jobsDisabled'));

And cronJobs are only run:

if (nconf.get('runJobs')) {
	require('./notifications').startJobs();
	require('./user').startJobs();
	require('./plugins').startJobs();
}

So it is well thought-out.

<baris>

@cryptoethic if you have multiple servers, set jobsDisabled: true on all servers except the first one.

PitaJ

@baris is it possible to have a separate instance just for running jobs?

<baris>

It can be done by using a load balancer, for example create 3 servers A, B, C. Set jobsDisabled: true on B & C and add them to the load balancer. A will run the jobs but won't get any traffic.

julian

@PitaJ A separate process just for running jobs would be ideal and future-ready, though we don't do that currently.