I have a long rant about logging, but the TL;DR is: aggregated logging in general is useless don't bother, everything you need to know should be in your monitoring system. Basically, if something is broken and it's a problem, it will happen again, so you just start logging when something breaks. How do you know something breaks? Log your application metrics. Log your customer experience in the form of metrics and monitoring.
But don't aggregate every code exception and warning. If they aren't affecting your customer experience, who cares? And if they are affecting it, then it should be expressed as an application metric in your monitoring system.
Edit: To make my point a little clearer, I believe that central logging and aggregating should be available, just not turned on all the time.
But don't aggregate every code exception and warning. If they aren't affecting your customer experience, who cares? And if they are affecting it, then it should be expressed as an application metric in your monitoring system.
Edit: To make my point a little clearer, I believe that central logging and aggregating should be available, just not turned on all the time.