`Could not log "process_action.action_controller" event. ArgumentError: wrong number of arguments (0 for 1)` when validation error occured

[a-know]

`Could not log "process_action.action_controller" event. ArgumentError: wrong number of arguments (0 for 1)` when validation error occured

[sakuro]

Should use ActionDispatch::ExceptionWrapper.status_code_for_exception instead of instantiating a new exception from the class name and use its name?

[spscream]

Any workarounds? I have the same issue with logger.

[pxlpnk]

I will look into this the coming weekend. Thanks for the input @sakuro

[pxlpnk]

I looked into it,
using ActionDispatch::ExceptionWrapper.status_code_for_exception would work for a number of errors:
https://github.com/rails/rails/blob/1ca13b47da46247345a0d4547edc495f0cc83bf6/actionpack/lib/action_dispatch/middleware/exception_wrapper.rb#L10-L20

But it would completely ignore the ActiveRecord::NotFound and return a 500 for this.
Maybe there could be a combination of both approaches to catch more of the status codes.

[sakuro]

AR seems that it registers AR specific exceptions to the mappings, right?

https://github.com/rails/rails/blob/1ca13b47da46247345a0d4547edc495f0cc83bf6/activerecord/lib/active_record/railtie.rb#L25

[pxlpnk]

I expected that too after reading the code, but when executing the code with tests there is no AR specific mapping in the exception wrapper. Maybe this just gets loaded when you run rails as an "actual" instance?

[andywenk]

We have the same issue in our production system. Can I help with any info? Any progress here? Thanks a lot!!!

Example stacktrace can be found here: https://gist.githubusercontent.com/andywenk/2aa6e974e637ad93bde9/raw/ab0fdf9c718f5626666252c90df3db9fbcb9e680/gistfile1.txt

[pxlpnk]

Sorry I am currently a bit down under. We are happy about a pull request that would fix this.

[benlovell]

I'm looking into this now. Hold tight.

[benlovell]

@pxlpnk Ideally we'd have at least a few integration specs that load rails in its entirety to cover this end to end, but for now at least I think #134 should resolve this issue.