While bringing new nodes into a replica set or compacting existing nodes the cluster is aware that the node is "recovering" and the status reflects it. What's more: it doesn't cause a warning message in the log files every two seconds. It would be nice to have the cluster aware of the ongoing repair process.

I would also like to see it possible for when a node finishes the repair, it immediately joins back into the cluster and carries on working as normal.

A third feature would be to issue a warning or error if there doesn't appear to be enough space for a repair to even take place (though I understand this isn't so cut and dry).

The motivation behind these requests are to make the repair process a bit easier to automate and leave running without someone staring at it: something that you can just fire and forget on a cluster node over a weekend, for example.