Nginx 502 Bad Gateway

[DannyJoris]

I ran into a persistent nginx "502 Bad Gateway" error, seemingly out of the blue, which doesn't go away after restarting ddev.

ddev logs shows the following line:

2023/03/24 19:37:58 [error] 2123#2123: *349 upstream sent too big header while reading response header from upstream, client: 172.18.0.7, server: , request: "GET / HTTP/1.1", upstream: "fastcgi://unix:/run/php-fpm.sock:", host: "spine-health9.vh.ddev.site", referrer: "https://..."

It seems to suggest I should increase the large_client_header_buffers size as done here? https://github.com/ddev/ddev/pull/3034/files#diff-6480821958a08d0a2cb64353db2f99ccb410a30a8d27a7e4d28008b886fe69d0

I've added the following to .ddev/nginx_full/nginx-site.conf, but without success:

client_header_buffer_size 32k;
large_client_header_buffers 4 32k;

I commented out the code I was working on, cleared Drupal caches, but nothing seems to work. Any ideas?

[rfay]

Hi @DannyJoris - you did the right thing checking ddev logs.

It appears that your PHP app is sending an absolutely huge header back to nginx.

I suspect you did the right thing trying those options, but you may need to go far higher than that. And remember you have to ddev restart after changing nginx-site.conf, and that you also have to remove the #ddev-generated` at the top of that file or it will get overwritten. Do check that of course. Did your change survive restart?

But you may have to step-debug through your project and find out what it's doing, because the actual problem will likely affect production.

[DannyJoris]

Thanks for your response, @rfay ! I was able to isolate the issue to a single Drupal Paragraph. I'll put my focus toward debugging that. Thanks again!

[rfay]

Glad you got it sorted out!