You are viewing a single comment's thread.

view the rest of the comments →

0
1

[–] stuff123 0 points 1 point (+1|-0) ago  (edited ago)

That's not exactly a solution, that means the devs are profiling their website to determine the specific error, but what exactly the problem is has not been determined. I think the problem here is the they're using 2 sql servers and some DDOS prevention middleware and expecting that to handle the massive traffic due to the pizzagate controversy on reddit. They need to seriously consider scaling their infrastructure, better sharding of databases, better caching schemes, etc.

This really needs to be discussed somewhere, because right now, I can't find any fleshed out discussion on their infrastructure, and I'm willing to bet there are some community members who have a ton more experience with this sort of problem that would help, but they have to create a sustained discussion...

0
4

[–] Disappointed 0 points 4 points (+4|-0) ago 

Thing is the site performance was just as bad before pizza got flung over here. I don't see it being any worse now than it was before

0
1

[–] stuff123 0 points 1 point (+1|-0) ago 

That's exactly my point, I remember back when there was an influx of redditors for the ellen pao controversy and how the bastards at reddit started outsourcing DDOS attacks against voat just to ensure there would be no migration. However, DESPITE that happening, there STILL was no sustained discussion created to ensure that voat would have a rock solid infrastructure for the future (infrastructure != DDOS protection).

Now, just to be clear, I'm not saying they have to buy $10k worth of AWS server instances, because that's expensive af, and the devs may not have that kind of sustained cash, but there are methods to improve your infrastructure that don't require anything but code. However, if there is no discussion created by the devs on what their infrastructure code looks like, and what they plan on doing, then there can be no improvement offered by the community. And, based on the performance from the ellen pao incident till now, it looks like (perhaps im wrong, but this is just my impression) the devs are really slow at learning how important infrastructure is.