Huge Spike in CPU Usage

(Eric) #1

My DO Droplet went into a 504 Error out of no where due to something using a huge amount of CPU.

I tried Power Cycling my droplet but now the website doesn’t even come up and is stuck in a 404 Error. This all happened out of the blue which is strange… any ideas?

(Rafael dos Santos Silva) #2

Disk Space, Discourse Version, Docker Version, Swap Enabled?

(Eric) #3

My droplet has 2GB RAM, 40GB of Disk. Discourse was just recently upgraded to the latest. I originally installed with DO’s One-Click install.

(Eric) #4

I did just try to rebuild the app and this was at the end

“docker: Error response from daemon: failed to create endpoint app on network bridge: Error starting userland proxy: listen tcp bind: address already in use.”

(Rafael dos Santos Silva) #5

Do you have another app running on port 80? Nginx and Apache are good candidates.

(Eric) #6

Seems like it, I just did a sudo netstat for port 80:

tcp 0 0* LISTEN 1022/nginx
tcp6 0 0 :::80 :::* LISTEN 1022/nginx

It was working before as is , did the power cycle affect it?

(Rafael dos Santos Silva) #7

You probably installed nginx somehow. Do you use only the standard install? stop nginx and try the rebuild. If everything works, remove nginx so it doesn’t happen anymore.

(Eric) #8

Thank you for your help. I did try installing nginx separately on this droplet before, but thought I deleted everything. Apparently after the power cycle it brought the service back. I stopped it and now the forum is up again.

I still don’t know what caused the crazy CPU 100% spike that caused the 504 Error earlier that started this chain of events…

(Rafael dos Santos Silva) #9

Try grepping the logs of the time of the spike on the shared folders