502 Bad Gateway errors on nginx server
 

I now have Calibre Content Server running with nginx as a virtual host, which has received a SSL certificate from Certbot for security's sake. They all run on Ubuntu 20.04 on a VPS. And I also set up a startup service that runs Calibre:

Most of the time Calibre runs great, but I get 502 Bad Gateway error about twice or thrice a day. I have logs enabled but calibre.log does not show anything related. nginx's error.log do have reports like:

This can be easily fixed by restarting Calibre server, without having to restart nginx. However, that means I will have to ssh into my VPS and run the command, which is not as easy to do when I am out with only a mobile phone.

Any idea why this might happen? Is there anything I can do to fix this?

You sure there is nothing in the calibre server logs? Does it happen after doing a particular type of activity?

Yes, nothing on calibre.log other than the normal:

Weirdly, 502 Bad Gateway happens quite randomly. Mostly it happens when I am doing nothing at all; went to bed and wake up with 502, for example. Happened twice while I was reading and taking notes, as far as I can remember.

Without anything in the logs I cant really help you. Other than to suggest that you make sure you are running the official calibre binaries and not the distro packaed version of calibre.

Understood. And yes, I am using the official binaries. Will try to get more info if I can.

I think I figured out what went wrong.

So my VPS instance has only 512 MBs of RAM. Although all it does is running Calibre Content Server on nginx, it still runs out of memory through time. And when the memory runs out on a Linux machine, something named Out of Memory (OOM) killer is triggered and kills Calibre, which results in nginx not being able to connect to the backend so finally "502 Bad Gateway".

I added 1GB of swap memory to my instance, and Calibre has not been killed since then.