Swap NGINX with Cloudflare Pingora

trellis
, ,
1

Hey guys, I’m new here. I just discovered Trellis a week ago (too late) and it’s exactly what I was searching for, so thank you so much for that.

I have a question as a newcomer: is it possible to replace NGINX with Pingora? I believe that for those of us using Cloudflare as a proxy for all their domains, it could be very beneficial to also use a reverse proxy developed by Cloudflare itself, which happens to be written in Rust.

Here’s the Pingora open source repo, just launched yesterday :sweat_smile::

Cloudflare Pingora Repo →

Has anyone tried this yet? Please let me know if you have; I’m very interested in learning more about any possible way I can implement it in place of NGINX to see if there are any benefits.

2

Pingora isn’t an off the shelf reverse proxy like Nginx. You would have to build a project in Rust that uses Pingora as a framework, then deploy that compiled binary to your server.

Pingora has a getting started guide and this is just the most basic reverse proxy: pingora/docs/quick_start.md at 8797329225018c4d0ab990166dd020338ae292dc · cloudflare/pingora · GitHub

In my opinion there is zero reason to use Pingora over Nginx unless you have some really specific and custom needs or are a large company.

1 Like
3

Thanks for the clarification @swalkinshaw, I thought that replacing NGINX with Pingora would essentially have the same effect.

I’ve gone down a rabbit hole because of some recent difficulties I’ve been experiencing with several WordPress builds; When I hit ‘update’ on several live WordPress sites, the GUI takes an unusually long time to respond. It’s as if the site freezes momentarily and can’t refresh, sometimes taking up to a minute before the changes are submitted, I thought about the reason might be due to:

  1. A memory issue with NGINX, which is the main reason why Cloudflare created Pingora.

  2. The Litespeed cache server might be at fault

  3. Or it could be that using Litespeed Server + NGINX + Cloudflare is simply not the ideal setup for my wordpress builds.

The challenge is that I rely on Cloudflare, so it’s not an option for me to stop proxying domains through them. So I’m exploring the Trellis approach and I will see whether this issue is specific to the Litespeed server :crossed_fingers:.

4

Yeah we have the exact same issue on our server at AWS. We’re not talking a minute of loading, but more like 10 seconds. We did not have this issue when we used Docker instead of Trellis.

If you ever figure out what is causing this, feel free pinging me because I really want to know.

5

What exactly is the “update”? Is that updating WordPress itself or plugins from a staging or production site (which shouldn’t be possible with the default Bedrock settings)? Is that running a network update with the “Update Now” button?

Consider creating a new topic made for whatever this is in regards to if this isn’t related to Pingora anymore

6

In my case, the site is slow when I press “Update” inside of an article to update/publish the contents.

I’ll create a new thread if the issue persists after we transfer to a new server :slight_smile:

1 Like
7

I don’t think Nginx is the bottleneck here. Wordpress isn’t built for performance and neither are any plugins. Wordpress requires a lot more cpu/memory capacity than any barebones framework but also doesn’t solve the same problems as they do. Wordpress is great for simpler websites that do not require high performance or optimizations. If you’re serving huge amount of data then you probably should look at a different solution than Wordpress.

If your Wordpress builds are slow I would first look debloating them. Writing custom plugins that are more streamlined to what functionality you need is a better option than switching to another language and having to not only rebuild the site but also build the whole backoffice site.