An updateMany is performed to override the onCurrentShardSince field on chunk documents.

When the config server steps down while serving the request the client could get a write concern error as the following:

WriteConcernError({
    "code" : 6,
    "codeName" : "HostUnreachable",
    "errmsg" : "operation was interrupted; Error details: { writeConcern: { w: \"majority\", wtimeout: 0, provenance: \"implicitDefault\" } } at config",
    "errInfo" : {}
})

Changing the updateMany to happen as a part of a session with retryWrites:true would solve the issue. (Solution similar the one implemented for SERVER-66077).