Concurrentcy update in mongodb - node.js

Question

I have used this code on node.js server side to update multiple embedded documents.
DetailerItemGroupModel.find({
"_id": itemGroupId
})
.forEach(function (doc) {
doc.Products.forEach(function (ch) {
//do something before update
});
DetailerItemGroupModel.save(doc);
});
I have the scenario like this :
Client A and B do GET http Method to get the same documents in the same time, then client B doing the update first(call to server which will run the code above) and then client A do the update.
So I want when client A's doing the UPDATE HTTP method to the server they must get the data which was latest (in this case is the document was updated by B) , I mean some how to cancel Client A request and return a bad request to tell the data Client A's going to update was changed by another client. Any way to implement that?
I read about "__v", but not sure when Client A and B send request to update same document at the same time, Does it work with forEach(), I change the code to this
DetailerItemGroupModel.find({
"_id": itemGroupId,
"__v" : {document version}
})
.forEach(function (doc) {
doc.Products.forEach(function (ch) {
//do some thing before update
});
DetailerItemGroupModel.save(doc);
});

Answer 1

The idea is that you want your "Save" to only work on the version that you have (which should be the latest).
Using .save will not help you here.
You can use one of two functions:
update.
findOneAndUpdate
The Idea is you have to do the find and save in one Atomic operation. And in the conditions object, you not only send the _id but you also send the __v which is the version identifier. (Something similar to the below example)
this.update({_id: document._id, __v: document.__v}, {$set: document, $inc: {__v: 1}}, result.cb(cb, function(updateResult) {
...
});
Say you read version 10, now you are ready to update it (save it). It will only work if it finds the document with the particular id and version. If in the meantime the database gets a newer version (version 11) then your update will fail.
BTW, MongoDb now has Transactions, you might want to look into that, because a transaction B would make your code wait until transaction A finishes it's atomic operation, which would be a better model than doing an update and failing and then trying again...