Quick knowledge tip
We ran into an issue on production where our k8s ingress controller was returning
400 Bad Request on multiple responses a day. After debugging the cause was a header was being sent that was larger than 8Kb, which was rejected (dropped) by NodeJS. Nginx then returned a
400 since the request did not return a valid response.
It was caused by two things:
- An Akamai rule duplicating header values, so we got multiple headers multiple times
- A frontender setting a ludicrous amount of cookies
Both issues are pending testing, so we did a quick fix by expanding Node's HTTP size limit to 16Kbw ith the following param: