More robust handling of Ingress configuration issues

[hostalp]

It sometimes happens that someone makes a mistake in the Ingress resource configuration which results in a critical error on the Nginx server (typical example is duplicate path which leads to duplicate location \"/somepath\" in /etc/nginx/conf.d/namespace-ingress.conf:261\n") - e.g. the Nginx would refuse to reload the configuration at runtime or fail to start.

When this happens in larger environments used by multiple teams, it can be quite annoying as one team affects another one - the Nginx would effectively stop reloading all the configuration and one has to find out why and fix that (which can be a little complicated with multiple teams involved).

Is it possible to make things more robust - for example to either improve the resource validation (and reject these faulty Ingress resources), or by making the Nginx ignore (skip) incorrect conigurations (whole *.conf files generated for each Ingress resource) and just produce error messages for them (thich would break things only for the team that introduced the issue)?

[vepatel]

hey @hostalp, we've an issue around this #7379, can you please add your comments there if you like