New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Yep, everything is cached, even html for anonymous sessions. Tiered cache is enabled. Problem is when there is a cache miss Cloudflare will sometimes take very long to fetch from origin. I am talking 10+ seconds to get a 50KB image. This never happened with my previous host (Namecheap shared hosting), and this also doesn't happen all the time with RN, so I have to assume it's an intermittent network issue. CPU and RAM usage are very low.
I have used WordPress since 2009 on a variety of shared hosts. Also used Cloudflare since 2012. Never faced such issues. I have always used servers in the US or EU even though I live in India, and yet the sites rarely ever take over 5 seconds to load. I don't understand why it would sometimes take 2 seconds and sometimes take 20 seconds to load the same 300KB page.
If it was consistently slow I would almost feel better than the intermittent issues
Looks good. I will check it, thanks!
Purging is a bit annoying though.
I'd recommend getting object (db) caching set up, as well as page caching.
The page caching should be set to cache in RAM.
And you should set up good purging rules that clears just what you need to clear. Maybe even pre-loading if your sites aren't very big.
The end goal: Every page that don't change dynamically gets served in pre-compiled HTML straight from RAM, and pages that can't, well, at least there's object caching, speeding it up significantly as well.
This is how I've set up a client's WordPress website that had one issue: It's very low traffic overall, but maybe once a week or so, it spikes, usually a large amount of visitors per second even. This makes sure that it stays cool.
Object caching with redis is already setup (I also tried the sqlite object cache plugin which works well). Page caching with nginx fastcgi is also configured, along with a small snippet to clear the cache on certain events and also after 7 days. Cloudflare caches html on top of it. And overall this setup works great most of the time...until it inexplicably doesn't. Sometimes the server struggles to serve static assets like images which baffles my mind. Even if I lived in Australia and the server is in the US, I shouldn't have to wait 10+ seconds for the server to send a 50KB image when CPU and memory usage are 5% and 30% of the threshold values.
This isn't getting cached? Or even struggles to serve it once per page and cache life?
If it's cached by Cloudflare, there's no problem. If not cached by Cloudflare, it's still usually fine but sometimes it's not and takes very long to load. Basically the issue is intermittent at origin server.
How do you even measure things like that to make sure this happens for other visitors as well?
I've been using BetterStack, StatusCake, etc. to compare providers in general but I never looked at the detail level of a single image!?
And what resource is maxing out at those times?
Yeah, it's not an optimal solution for sure. The site will probably updated once a year if that so it's a small price to pay in this case. Probably should be a static site but it needs updating by a non technical user so...
If it’s once a year, dosen’t really matter
I just check on my PC in Chrome's guest mode (disable cache in dev tools and watch the waterfalls for all assets) after clearing all cache from Cloudflare and fastcgi cache from server, once through Cloudflare proxy and another with Cloudflare development mode enabled. 80-90% of the time the site is fine. Everything loads as it should. But sometimes it will keep spinning forever. I check server's resource usage through htop at the same time and it's all good, no major spikes anywhere. It has become impossible to troubleshoot.
For use cases like this I have begun to experiment with the "Simply Static" WordPress plugin last year. My trial setup is like this: I have a multisite WordPress installation where non technical users can edit their sites as usual and then after edits have been made you have the following choices:
Deploy to Amazon AWS S3
Deploy to BunnyCDN
Deploy to Cloudflare Pages
Deploy to Digital Ocean Spaces
Deploy to GitHub Pages
Deploy to Kinsta Static Site Hosting
Deploy to Netlify
What intrigued me about this is that you basically just need one server which you can secure even better against hacking attempts and then the hosting for the actual sites becomes free because most of these providers charge nothing or just a few cents.
There are integrations for the mentioned providers but if you have a site that is only updated once a year you can choose any static hosting provider because it's not a lot of work to publish it or you can work some automation out through GitHub.
I'm very curious what you as advantaged users think of this setup in terms of performance, etc.
This looks really interesting and could be perfect for my use case. Static sites served directly from a cdn is always going to be quicker than hitting WordPress itself, it'll be interesting to see how the forms plugin works.
For 5/M there are many options, Hetzner, Netcup, Contabo
For Wordpress on LAMP I suggest the Wordops stack. Really easy to install and manage.
https://wordops.net/
@layer7 can you please yabs thisone: Cloud Server - Mid CPU 2Cores-8GB-80GB FRA1
if yabs doesn't tell you anything yet, take a look here:
https://github.com/masonr/yet-another-bench-script