Learning by Doing

This issue arises when end user mentioned they couldn't locate certain records and they know the record is indeed in the database and they were using _id to reference the record. Quick investigation of our database, we can see our _id type is split into total count: db.getCollection('upDown').find({}).count() //2403170 count by type 2 which is string db.getCollection('upDown').find({_id:{$type:2}}).count() //1630934 count by not type2 db.getCollection('upDown').find({_id:{$not:{$type:2}}}).count() //772236 check a few sample record, we see the _id is in the default format of bson object, objectId db.getCollection('upDown').find({_id:{$not:{$type:2}}}) This issue was first discovered by end user who uses MongoDB  BI connector to query mongodb data. In the drdl I see we have exposed the field as     - Name: _id       MongoType: string       SqlName: _id       SqlType: varchar from MySQL query, it looks lik...

MongoDB objectid conversion to timestamp Basically we can generate timestamp when this document is inserted by converting the objectid to timestamp: Please look at the following for javascript by  Steve Ridout  on his blog https://steveridout.github.io/mongo-object-time/ Why generate an ObjectId from a timestamp? To query documents by creation date. e.g. to find all comments created after 2013-11-01: db.comments.find({_id: {$gt: ObjectId("5272e0f00000000000000000")}}) Javascript functions var objectIdFromDate = function (date) { return Math.floor(date.getTime() / 1000).toString(16) + "0000000000000000"; }; var dateFromObjectId = function (objectId) { return new Date(parseInt(objectId.substring(0, 8), 16) * 1000); };