Details
-
Task
-
Resolution: Won't Fix
-
Minor - P4
-
None
Description
The way to do this in a sharded environment is directly through the mongos. Do not drop the collection from the mongod. While drop does hold the write lock, it doesn't have a large performance overhead, even for large collections. It deletes the extents on disk thus does not have to individually remove each of the individual documents.
Once finished, run flushRouterConfig to make sure that all metadata relating to your sharded cluster is up to date.