We have a particular kind of error being caused on the server by a client bug in our own system. It’s not relevant to our consideration of application health. But it counts towards our New Relic error rate.
So we’d like to ensure our error rate ignores this kind of error. But we haven’t found a way to do this.
The Go agent allows you to configure which HTTP codes you ignore. But for our case, HTTP code isn’t helpful: errors we want to ignore and errors we want to observe both have the same codes.
The New Relic Web UI allows you to “hide” errors, but that appears not to affect the error rate charts/calculations – only the list of top errors occurring in a given time period.
Some language agents (like Java, I think?) allow you to do Server-Side Agent Configuration in the New Relic Web UI that can get a little more sophisticated with error-ignoring. But Go isn’t one of those agents.
Am I missing anything?