3000+ Online on a SSL Site with Redis Cache - FrontEnd is Blazing Fast but Backend Nginx Timeout

v3
#1

Thanks a lot for this Awesome EasyEngine =)

This one saved me a lot of bucks.

Yesterday, We got a peak of more than 3k online on our SSL Wordpress Site with wpredis.

We had to update a post since it was live updates post. But when the number of online users crossed 1k the Admin Freezed

Posts failed to update. After crossing 1.5k Nginx started timing out on the backend.

During all these drama, the frontend was blazing fast.

It was just a 4GB RAM 4 Core VPS Server with Xeon E3 CPU

I informed the Datacenter to make it 8GB RAM and 8 Cores due to Peak Traffic and they made it so.

But still Admin was timing out.

Can you please help me with this?

Any Reason why this issue is caused?

#2

Did you increase ngx_http_limit_req_module in nginx.conf ?

#    limit_req_zone  $binary_remote_addr  zone=app:10m   rate=1r/s;

Read more at:

http://nginx.org/en/docs/http/ngx_http_limit_req_module.html

#3

Thanks for helping me out,Will check it out =)

#4

Sorry, I’m wrong

#5

I’m having a similar experience. The front-end is pretty fast but the back-end is slow. It seems the redis cache is not being well used in the back-end. That seems my impression for now.

When a user is logged in, the server memory spikes up to about 60% to 80%, mainly because of PHP-FPM. I find it very strange considering the caching used and the small number of users online.

#6