Question

I’m new on mongoDB and the point is, that we new realize an bug at our application witch results in multiple mongoDB entries instead of update the (edited) document.
now, the application is online, we realize the bug and trying to manage the trouble which comes with that.

Following situation: @mongoDB there are lots of documents containing this structure:

"_id" : ObjectId("4fd9ede5a6b9579f5b000003"),
"USER" : "my_username",
"matchID" : 18809,
"data1" : 2,
"data2" : 1,
"tippDate" : ISODate("2012-06-14T13:57:57Z"),
"data3" : 0

If the user changes the data at the application, the application inserts an new document instead of updating the existing.

Like that

{
"_id" : ObjectId("4fd9ede5a6b9579f5b000003"),
"USER" : "my_username",
"matchID" : 18809,
"data1" : 2,
"data2" : 1,
"tippDate" : ISODate("2012-06-14T13:57:57Z"),
"data3" : 0
}
{
"_id" : ObjectId("4fd9ede5a6b9579f5b000002"),
"USER" : "my_username",
"matchID" : 18809,
"data1" : 4,
"data2" : 2,
"tippDate" : ISODate("2012-06-14T12:45:33Z"),
"data3" : 0
}

Right now, the bug on application side is solved, but now we have to clean up the database.
The gaol is to keep only the newest record/document from each user.
One way is to handle this on application side: loading all the data from one user, order by date, removing all the data from one user and writing the newest entry back to mongoDB.

But: isn’t it possible to process that an mongoDB like an delete with joints on MySQL?

Thank you for any kind of help or hints!

Answer 1

is’n it possible to process that an mongoDB like an delete with joints on MySQL?

No. MongoDB does not support joins at all.

However, MongoDB does have sorting. So you can run a script to fetch each user, sort them by date and then delete the old ones.

Also, please not that you can override the _id field. It does not have to be an ObjectId(). Based on your description, you have a unique user_name, so why not simply use that as the _id?