
Here's what log files we have with information what gets into those files:

* couchdb - log message with category couchdb with severity info and above

* debug - all categories except couchdb, xdcr, stats and ns_doctor
with messages of severity debug and above (i.e. all messages). That's
my team's main log file.

* info - all categories except couchdb, xdcr, stats and ns_doctor with
messages of severity info and above

* error - all categories except xdcr with severity error

* xdcr_error - category xdcr and severity error

* xdcr - category xdcr and any severity

* mapreduce_errors - category mapreduce (our views will send js errors from
map/reduce functions here)

* views - category views. We log ns_server's interaction with indexes
here. Initially, to enable Filipe to see what we do without parsing
lots of 'irrelevant' stuff in our logs. We've considered killing it,
but decided to keep for now.

* stats - categories stats and ns_doctor. Stats & ns_doctor states are
  logged here periodically.

All files are set to rotate after 10M. And there's limit of 20 files
per log. So in total thats 200M per log.
