_node/stats API does not update successful count after reverting to previous successful config after failure

Hi,

I'm using the new stats API and the automatic config reload. I'm trying to create a monitoring plugin for Icinga (2), Nagios, Shinken, etc. ( https://github.com/widhalmt/check_logstash ) and I'm checking whether the current Logstash configuration is valid.

I see the following behaviour:

* When Logstash is running and I change the config `pipeline.reloads.last_success_timestamp` is updated
* When I break the configuration `pipeline.reloads.last_failure_timestamp` is updated
* When I fix the configuration, Logstash reloads it and works like it should, `pipeline.reloads.last_success_timestamp` is not updated anymore.

What I want to check with my plugin is, if `pipeline.reloads.last_success_timestamp` is newer than `pipeline.reloads.last_failure_timestamp` . Another way would be to integrate `pipeline.reloads.current_config_success` or the like to see whether the current config is working or not.

Right now I have to restart Logstash manually to reset the error information from the node API.

Please tell me if I'm just misinterpreting the API or help me with checking if the current config is vaild.

Cheers,
Thomas