Description of problem:

Right now, a user really doesn't have an intuitive way to infer replication (snapshot synchronization) performance. Such information is essential to determine if SLAs are being met. And, Of course, the lack of metrics available to the user, basically keeps him/her in the dark.

The proposed approach would be to have each CephFS mirror daemon provide a hook (say prometheus export) for the user to poke and fetch the replication metrics at that point in time.