A user, Bathinda, is experiencing issues with their website going down after rebooting their server, displaying a blank screen in Firefox and an Error 500 in Chrome. Despite running in safe mode, the issue persists. Bathinda has Nginx outside the container and suspects it may not be the container’s fault. Stopping and starting the container resolves the issue.
Pfaffman suggests a possible race condition between the external Nginx and the Discourse container, but is unsure of the cause. Bathinda confirms having a reverse proxy setup with two containers.
Pfaffman asks if a Docker network is created to connect the containers, but has no suggestions for a solution. itsbhanusharma recommends checking the outer Nginx logs for errors, as well as exploring the inner Nginx logs. Bathinda thanks the users for their input and will investigate the Nginx logs further.
Since I’ve shifted my server from Gcp/1 Vcpu/8gb/50gb to 1 Vcpu/6gb/52gb on Google Console, whenever I reboot my server (for whatever reason. Generally after some imp updates), then I find that my website goes down. Just blank screen in FF, and Error 500 in chrome:
Also I’ve checked that when my site is not working/giving error, the nginx service is running ok in backend/server. But yet outer Nginx’s error msg isn’t displayed in browser.
I’ve Nginx outside the container. And since Nginx’s own error msg also isn’t displayed, I presume perhaps its not the container fault.
Anyhow, if I stop and then start the container, then first outer Nginx msg appears and within minute the website appears. This episode has happened 2-3 times. My website is updated every 5-7 days from front end.
Is this a standard install or do you have a reverse proxy in front of Discourse?
Given that your have some other reverse proxy, it sounds like some kind of race condition where it needs to be running first, or perhaps the external nginx is creating a network that Discourse cannot connect to if discourse starts first
I’ve reverse proxy (i.e. nginx is running outside the disco container). Also I’ve 2 container setup, just in case the problem is related to this somehow.
Though problem isn’t urgent, (only when server is rebooted), is there some simple check/test which could further pin-point the problem.
