mongoMergeEmptyChunks.js

function mergeEmptyChunks(namespace) {
    assert(isString(namespace) && namespace.length, "No namespace specified");
    var databaseName = namespace.split(".")[0];
    assert(databaseName != namespace, "Namespace should be database.collection");
    var database = db.getSiblingDB(databaseName);
    assert(database, "Database " + databaseName + " not found");

    //print("stop balancer");
    //sh.stopBalancer();
    assert(sh.getBalancerState() == false, "Balancer is not off");
    assert(sh.isBalancerRunning() == false, "Balancer is running");
    
    // based on Maurizio Merli's post
    // https://medium.com/@dash1e/mongodb-merge-all-empty-chunks-6e367680f78d
    
    var configdb = db.getSiblingDB("config");
    var admindb = db.getSiblingDB("admin");

    var minchunk = null;
    var count = 0;
    try {
        configdb.chunks.find({ns : namespace}).sort({min: 1}).forEach(function(chunk) {
            var ds = database.runCommand({
                datasize: namespace,
                keyPattern: { "_id": 1 },
                min: chunk.min,
                max: chunk.max });
                
            if (minchunk != null) {
                if (minchunk.shard != chunk.shard) {
                    while(true) {
                        print("move empty chunk: " + tojsononeline(minchunk)
                              + " to: " + chunk.shard);
                        var res = admindb.runCommand({
                            moveChunk: namespace,
                            bounds: [ minchunk.min, minchunk.max ],
                            to: chunk.shard });
                        if(!res.ok) {
                            print("error: " + res.errmsg);
                            if(res.errmsg == "move failed") {
                                sleep(1000);
                                print("RETRYING");
                                continue;
                            }
                            throw res.errmsg;
                        }
                        minchunk.shard = chunk.shard;
                        break;
                    }
                }
                
                while(true) {
                    var mergeBounds = [ minchunk.min, chunk.max ];
                    print("merge chunk into: " + chunk._id 
                          + " shard: " + chunk.shard 
                          + " bounds: " + tojsononeline(mergeBounds)
                          + " size: " + ds.size 
                          + " #" + count);
                    var res = admindb.runCommand({
                        mergeChunks: namespace,
                        bounds: mergeBounds });
                    if(!res.ok) {
                        print("error: " + res.errmsg);
                        if(res.errmsg.startsWith("could not acquire collection lock")) {
                            sleep(1000);
                            print("RETRYING");
                            continue;
                        }
                        throw res.errmsg;
                    }
                    break;
                }
                minchunk.max = chunk.max;
            }
            
            if (ds.size == 0) {
                if (minchunk == null) {
                    ds.shard = chunk.shard;
                    ds.min = chunk.min;
                    ds.max = chunk.max;
                    minchunk = ds;
                    count = 1;
                    print("min chunk: " + tojsononeline(ds));
                } else {
                    count++;
                }
            } else {
                minchunk = null;
            }
        });
    } finally {
        //print("start balancer");
        //sh.startBalancer();
    }
}

// mergeEmptyChunks("db.collection");