With this change, the flushRouterConfig command is now available on all nodes (shards and config server), so the linked documentation needs to be updated to say that.
In addition, starting in version 3.4, the instructions in points 6 and 7 are not correct anymore. The balancer doesn't need to be restarted, instead, they should be:
6) Clear current cluster meta information. After the jumbo flag has been cleared out from the chunks collection, run flushRouterConfig: <DB.collection> directly against the config server primary.
Currently there is the ability to flush router config on mongos nodes. The Primary nodes in repl sets also hold some cached config information in ChunkManager. Currently the way to 'flush' this information from a Primary is to step down, causing a new Primary to be elected. If the functionality existed that allowed a similar operation to FlushRouterConfig to run on Primaries, then this step down could be avoided in those cases where it currently is used.
- source/tutorial/remove-shards-from-cluster.txt <<- include the flushRouterConfig blurb found in movePrimary here.