Issue Plesk Eating Vultr Server CPU after SSL Installation

[guedzzz]

Hi, I have a Vultr high-frequency server that was working fine for two weeks, until I hire someone to install and secure an SSL for the Plesk installation itself. I have 2 domains only, secured by Let's Encrypt.

Now, I'm getting 102% CPU overloads.

DNS is being handled by Cloudflare with 2 A records and 1 webmail record, only. I was told to turn off the proxy and leave DNS only. Not sure yet if that's best practice and why I have to go that route. I'm trying to become more self-managed, but still got a long way to go.

I noticed the memory and disk usage are also to the max. I'm sure that's what's causing Plesk to crash a lot. Therefore, the server needs to be restarted all the time. I attached the details.

It's important to point out I increase the PHP settings to probably way too high memory limits and post size limits, etc. I lowered it here are the current limits. I did that for both 7.3 and 7.4:

memory_limit = 128M
post_max_size = 128M
max_execution_time = 180
upload_max_filesize = 128M
max_input_time = 180
max_input_vars = 1000

Would anyone be able to shine some light on this matter? Do you need any other information? Please, let me know.

Thanks!

 

[guedzzz]

Also, Plesk is using all the memory limit and all 30GB of disk space in the server. But, there are only two very small sites installed.

 

[mow]

1GB is insufficient nowadays, and the 30GB disk are probably from the logs filling with errors.

Also, if you turn the cloudflare proxy off, you could as well get rid of it completely as there's nothing useful left for it to do.

 

[guedzzz]

1GB is insufficient nowadays, and the 30GB disk are probably from the logs filling with errors.

Also, if you turn the cloudflare proxy off, you could as well get rid of it completely as there's nothing useful left for it to do.

It was the logs and backups. I upgraded to 2048 MB. Is that sufficient?

What would be a good SSL setup that works with CloudFlare proxy? Bc, since we deployed this security update in Plesk, all sites drop.

Thank you!

 

[guedzzz]

1GB is insufficient nowadays, and the 30GB disk are probably from the logs filling with errors.

Also, if you turn the cloudflare proxy off, you could as well get rid of it completely as there's nothing useful left for it to do.

Hi Mow, I updated the memory of the server, but Plesk already is using 1.7GB of it. Is this still normal?

Thanks.

 

[Kaspar]

Navigate to Tools & Settings and bellow the Assistance and Troubleshooting header click on the Process List to view the process list of your server. It shows which running processes use CPU and memory.

If you don't see the Process List link bellow the Assistance and Troubleshooting header, install the Plesk Repair Kit extension.

 

[guedzzz]

Navigate to Tools & Settings and bellow the Assistance and Troubleshooting header click on the Process List to view the process list of your server. It shows which running processes use CPU and memory.

If you don't see the Process List link bellow the Assistance and Troubleshooting header, install the Plesk Repair Kit extension.

Hey, thanks! This is what I got. I'm just starting with Plesk, so I can't tell if there's any red flag just by looking at it.

 

[Kaspar]

Nothing out of the ordinary IMO. How does the process list look when you sort the list on Memory Usage?

 

[guedzzz]

Nothing out of the ordinary IMO. How does the process list look when you sort the list on Memory Usage?

Now I just installed a new plugin in a domain, and the numbers changed a bit. I sorted and then the memory when down right at that second. Lmk what you think.

Also, I added the error I'm getting whenever doing a simple task in WP or I try to install a plugin, for example.

I feel like I can't have more than 2 domains in this plesk, or things could turn out really bad. Is that normal?

Thank you!

 

[mow]

Hi Mow, I updated the memory of the server, but Plesk already is using 1.7GB of it. Is this still normal?

Yes, it is. Unfortunately plesk does not show MemAvailable and the definition of "used" differs from what "free" uses, so it is not directly obvious that there is sufficient RAM (but usage will go up when the server actually does something other than just idle around).

Also, I added the error I'm getting whenever doing a simple task in WP or I try to install a plugin, for example.

That's the point where you go through the logfiles and look for the actual error message(s).

 

[guedzzz]

Ok, so we got to the bottom of it.

1. It needed a better proxy setup to work with Cloudflare. On this one, I can't give much feedback, because I hired someone to do it.
2. The server crashing was because the site was running PHP in FPM Apache and changing to FPM nginx solved the issues, along with increasing the memory limits (according to your own needs).
3. I also upgraded my VPS to use 2 cores, although I've seen people do it with just one. So now I have 2 cores, 2GB RAM, 80BG disk space.


I'm only running 2 WordPress sites, so it should not be crashing like that, even with a lower VPS.

Before you upgrade, try tips #1 #2 first and only after upgrade your VPS. I'm using Vultr, and now I can't go back, so I'm stuck with a more expensive plan, which I will have to make the most out of it.

Thanks for all your guy's support!

 

[mow]

Traffic usually increases over time. You will definitely be using that RAM.
Also, how many workers did you configure at the bottom of that page? Maximum php uses is memory limit x worker count, because that limit is per worker process.

 

[weltonw]

600M...700M...700M...300s...You're setting up yourself for a DOS attack. Make them something sane, (256M, 8/16M, 30-45s)

 

[guedzzz]

600M...700M...700M...300s...You're setting up yourself for a DOS attack. Make them something sane, (256M, 8/16M, 30-45s)

Thank you!

pm.max_children = 30 ok you think?

 

[weltonw]

On a 2 GB machine? No

 

[guedzzz]

600M...700M...700M...300s...You're setting up yourself for a DOS attack. Make them something sane, (256M, 8/16M, 30-45s)

I don't see workers. Perhaps you are talking about pm.max_children? If so, we set it to 30.

 

[guedzzz]

On a 2 GB machine? No

What would you have it set at? Thanks!

 

[weltonw]

Worker = max_children - they're interchangeable.

Probably 2-4.

 

[guedzzz]

Worker = max_children - they're interchangeable.

Probably 2-4.

Got it. I left it at 10 for each site. The guy who worked for me placed it at 30, bc we were not being able to make the server work. I think changing it to FPM nginx fixed a lot of things, and all the limits can work with lower, safer configurations. I still need to get more familiar with how PHP settings affect the DOS security.

I do appreciate all the feedback.

Thanks!

 

[mow]

Rule of thumb is two workers per CPU core ... if you have enough RAM, that is. Adjust one or the other until max_children x memory limit < total RAM.
As those workers are spawned on demand, you won't notice problems until you get enough traffic that the master process starts new workers that immediately abort because kernel says out of memory (you have no safety margin because there's no swap).