When introducing more collectors to the FTDC machinery, there is the possibility to create gaps and a loss in retention. In other words, there are some heavy/intensive collectors (e.g. serverStatus) that will make the FTDC machinery to recollect data each time in a bigger interval, producing a regression in the dimension / richness of the data (more gaps). 

When introducing the router collectors to the MongoD, for instance, the serverStatus will be called twice, one for the shard and one for the router, and this can reproduce the problems explained.

The goal of this ticket is to see if these cases are being reproduced and, if applies, how to solve them.