Is it possible after a series of errors or excessive access for OLS to run Restart Graceful?

[ofmarconi]

I'm doing a stress test with my OLS via loader.io

When I exaggerate the number of requests I realize that WordPress, even after the test has ended, remains stuck for about 10m ...

But if I give a Graceful Restart it goes back to normal, flying.

I was wondering if on the OLS Panel you can program for when this overload occurs after a few seconds it restarts.

I use Cyberpanel in a droplet.

 

[Cold-Egg]

During the stuck, what did you see the output from the `top` command?

 

[ofmarconi]

I see this while WordPress is frozen.

 

[ofmarconi]

I think it's the database that freezes, right?

Is there anything that can be done in OLS or Cyberpanel to collaborate with this?

When it is frozen and I try to access an image URL the image loads normally, only things that depend on the database that do not respond.

Should Object Cache help with this?
I activated Redis and it didn't help at all.

The strange thing is that without HTTPs Wordpress does not crash.
Maybe it's the rescripts that are redirecting http to https?


(sorry for the English GT)

 

[Cold-Egg]

Hi,

Do you have any cache on this site? Please consider using the LiteSpeed WordPress cache plugin so the load test won't hit the backend Database and PHP engine every time.

 

[ofmarconi]

Hi!

Yes, I'm using, I configured Cyberpanel> One Click Wordpress> I activated the cache in Litespeed



It seems that only with https that the cache is not working.

 

[ofmarconi]

You are marking it as HIT, this is enough to know that the cache is working correctly, right?

My website: https://teste.motionloop.com.br/

 

[Cold-Egg]

Hi,

It is working. Do you still see any sql/lsphp process running high CPU in the background during the test?
Also, I would recommend you not to include any proxy or CDN, e.g. cloudflare in the test, it may slow the benchmark result.

Best,
Eric