I assume you compare Nginx + Apache2 setup without optimisations regarding Wordpress and client side vs Lightspeed what is highly optimised regarding Wordpress.
Litespeed working straight out of the box without optimization too. But I did try out a simple HTML site before using WordPress and the results are not far from what I have posted.
Though that’s not for comparison but to show readers that the server is capable. As some may suggest that I get a better server.
Anyways, I need help with this. How to go about it on HestiaCP?
If you go for speed don’t install apache2. It is a know fact that Apache2 is slower than Lightspeed.
Then enable fast cgi cache.
If you would like to use an other web server you are free to modify Hestia for your self and even publish it on a different name. How ever we will not be able to support it…
No comparisons. I just need to see if I can get HestiaCP to work for me. Always looking for alternatives that work and I’m willing to give Hestia a try.
Even when it’s working with Nginx? In this case, Nginx does the heavy lifting while Apache’s advantage is .htaccess access by applications that need it.
Right now I’m interested with what HestiaCP provides and I’m willing to see it work.
I see the HestiaCP NGINX Wordpress template has fast-cgi parameters. Do I still need to turn on fast-cgi in the panel or does the template activate and use fast-cgi for me when using the HestiaCP NGINX Wordpress template?
Before i tried Hestia i did like the look of Cyberpanel and OLS, but had very similar results on the TTFB metric although not even under load, it was just all over the place and frustrating as no obvious reason for it.
However Hestia CP - Apache with Nginx for Static and Redis with some minor ammendments as well as tweaks to the my.cnf and it really does fly. But then im serving via Joomla not Wordpress and want the .htaccess benefits.
I remember the Dev of the Joomla tools i use Akeba wrote a great post on why Nginx only is not always the best route and actually if you learn to tune Apache with Nginx as static you have the best of both worlds so to speak with vastly improved performance.
Unfortunately, I’m not able to make timely responses as there are restrictions on my profile because I’m a new member. Been waiting to respond for like 9 hours now.
But here we go…
I’m not sure whether that’s true or not. But OLS and LiteSpeed are build to kick ass straight out of the box. The reason why I’m saying so is because the turbocharged performance the two give is without any caching.
If you try to install OLS using Direct Admin or aaPanel, it will give you sh*tty performance because the developers have alightly modded it only to mess up its handling of php and static requests. And you have to apply optimizations via .htaccess rewrite rules for it to work.
I will definitely give this a try. Thanks.
Can you post the direct links for the Loader I/O tests? And what was the CPU performance for Cyberpanel+OLS?
Now this comes to optimization. OLS works magic even without Redis/Memcached.
As for optimization, I get 2X better performance when using W3 Total Cache than LSCache (Using Redis or Memcached). And I showed this to the developers and it all has to do with the poor handling of fonts and CSS by LSCache which causes some resources to be render blocking and increase TTFB as well.
So the hype about OLS and Litespeed is true. But the hype about LSCache is false. And without it, my TTFB is excellent.
Thanks, will have to pay attention to this too.
Can you share how you did it? I want to benefit from the use of .htaccess rules. And some of my favorite plugins as well, will need direct access to the nginx.conf file, which requires the user to do it manually. That’s why I love to use .htaccess as it makes plugins configure processes without the need to do it manually.
Done. Initial CPU usage hits 100% for around 5 seconds, then it drops to 50% for another 5 seconds, then finally it drops to an average for 7%. That’s why the first 15 seconds of the Loader I/O results look weird before it stabilizes.
This is a problem with Nginx. It almost looks like it’s always caught sleeping by requests before trying to wake up and get hold of things. But OLS/Litespeed act like they prepared for battle 24/7. And I extensively covered it on aaPanel’s forum.
What I have done with HestiaCP right now is just enable FastCGI. But on aaPanel forum, I did over 10 optimizations on Nginx just to try make it act normal and reduce those weird startup response times.
But still, I couldn’t manage to get on par with OLS on aaPanel where I only enabled php caching (As said earlier, aaPanel have slightly modified OLS and that messes up php caching). Either way, aaPanel+OLS with that one small tweak managed incredible results.
We are working with a very small team and we don’t get paid for our work and we can’t keep the development…
So I am stopping with this discussion we will not change the web server at the moment. Additional support for an extra server type is out of the question. We need to test 5 / 6 server configurations if we want to release a new version. Our time limited.
If you would like to use Hestia use Nginx or Nginx + Apache2 if you want Litespeed use a different one.
I don’t see the reason why keep going on with this discussion…
I do appreciate and respect your hard work at HestiaCP.
As I said earlier, I’m willing to see HestiaCP work for me.
I’ve not requested that you add OLS/Litespeed at all.
I have set these tests just for reference and if there’s any optimizations on HestiaCP that can make the server standout, I’ll be more than happy to give it a try.
Please don’t close the thread as a member has already mentioned ways they’ve managed to optimize Nginx+Apache2 and I’m so interested to get the tips.
