Piranha produces 502 bad gateway

[jensbrak]

(This is not an actual question anymore, I solved it but thought the issue and the working solution could come in handy for someone else)

Problem

Symptom: Logging in to Piranha Manager results in a 502 bad gateway error.

Details:

• Upgraded to Piranha 8.3 and corresponding .NET Core version
• Running on a linux machine (debian based)
• Using NginX as a reverse proxy

Solution

Reason: Too large forwarded headers caused the buffers of NginX to overflow. This caused NginX to fail to loading the request that was made with a successful login to the manager.

Fix: In the NginX configuration file for the Piranha site, add the following in the apropriate location target

 proxy_intercept_errors on;
 proxy_buffer_size  128k;
 proxy_buffers      4 256k;
 proxy_busy_buffers_size 256k;

Notes

• Some change following the upgrade to 8.3 resulted in larger forwarded headers but what really caused it I did not follow up
• Indications were that the Antiforgery token of .NET Core was significantly larger but I am not sure it was the actual reason
• There is a possibility that Apache hosted Piranha sites could suffer from this as well, however it seems that Apache has larger buffers by default (if I remember correctly)
• This is not an error specific to Piranha but after I encountered it when upgrading Piranha version, several other people have encountered it as well and asked about it in Gitter. I add the solution here in hope that common hosting related issues with Piranha may be of interst too

[jensbrak]

In the NginX configuration file for the Piranha site, add the following in the apropriate location target

 proxy_intercept_errors on;
 proxy_buffer_size  128k;
 proxy_buffers      4 256k;
 proxy_busy_buffers_size 256k;