When debugging client issues involving deadlock / slowness in replication failover, we often need to ask for a core dump in order to know what was happening (see SERVER-56756 and SERVER-71520). Ideally, we want currentOp to be able to give us an answer of "what are all of the current running operations, what are they currently doing at the moment, and for those waiting on something, how long have they been waiting, and how long did they spend on each step of their operation until now".
currentOp currently has a progress field for the progress of mapReduce or indexing operations. It would be nice if we can have the progress breakdown for crud/read operations as well. (We could also just start with something simple to provide info about which subsystem, e.g. Query, WT, each operation is currently in.)