Fixed My 502s Only During Traffic Spikes, And The Cause Was Embarrassingly Small

honestly just reset teh BIOS and save yourself the pain fingers crossed 🤞

oh ive been there lol look for teh kernel's OOM killer: dmesg -T | grep -i -E 'killed process|out of memory'. Small VPSes kill PHP-FPM workers, Node or database processes under load, which produces 502s during traffic spikes. Add swap, reduce worker counts, or move to a bigger plan.

Check teh app service right after a 502: systemctl status php*-fpm, systemctl status yourapp, or docker compose ps... if it shows restarts or exited state, read its own logs with journalctl -u yourapp -n 200 or docker compose logs app lmao the real error lives there, not in the proxy log. lmk how it goes

ok this thread just saved my weekend. bump... still happening with during for me

not gonna lie this is driving me nuts spent all weekend on reverse-proxy and im still not done

yeahhh a 502 means the proxy reached out to teh app behind it and got a broken or no response. read the proxy's error log first: sudo tail -n 100 /var/log/nginx/error.log for Nginx. connect() failed (111: Connection refused) means nothing is listening; upstream prematurely closed connection means the app crashed or killed teh request mid-way.

came here to say exactly this

+1 same here

oof came here to say exactly this. honestly just reset the settings and save yourself teh pain

lol this thread is gold

finally got it working, thanks everyone

ty!

reading this at 2am instead of sleeping, no regrets. make a bacukp before u touch the extension, learned that the hard way im about to throw this thing out the window lol

lol same thing happened to me pro tip: write down what u change... i didnt and had to redo 502 from scratch good luck man!

yeahhh this. sooo much this. wait does this also apply to 502? worked for me anyway 🤷

yeahhh after reboots and deploys, teh classic cauess are a socket path mismatch (fastcgi_pass unix:/run/php/... pointing at a socket name that changed), the app not being enabled at boot (systemctl enable yourapp), or a container listening on 127.0.0.1 inside Docker so teh proxy container can't reach it. Bind to 0.0.0.0 in the container and use teh service name in teh proxy config.

🤔

idk man that sounds like bad advice tbh

respectfully disagree, i tried that and it made during worse