Data recovery using the Oplog log of the MongoDB cluster

(The following methods can only recover part of the data because the Oplog table does not have all of the synchronization logs saved, size limited)

Because the Oplog table (called a table after collection for custom) has no index, I have to select the data for one of the tables I need to recover.
So the Oplog table in each slice is exported separately to another server for processing:

1. Back up:
./mongodump--port 28011-d local-c oplog.rs-o/opt/backup/0706local/

2. Restore to another single node MongoDB server

Mongorestore--port 28011-d temp_local-c shard1_oplog--dir/opt/backup/0706local/local/oplog.rs.bson

......
Mongorestore--port 28012-d temp_local-c shard4_oplog--dir/opt/backup/0706local/local/oplog.rs.bson

How many slices there are, there are a number of newly created tables.

3. Indexing to facilitate query data:
> Db.shard4_oplog.createIndex ({ns:1,op:1})
{
"Createdcollectionautomatically": false,
"Numindexesbefore": 0,
"Numindexesafter": 1,
"OK": 1
}

4. Originally wanted to delete the useless data, but because Oplog originally is a capped table, cannot delete the data, had to do, so deal with it.

> Db.shard4_oplog.remove ({ns: "Ds.tb_monitor", op:{$ne: ' I '}})
Writeresult ({
"Nremoved": 0,
"Writeerror": {
"Code": 20,
"ErrMsg": "Cannot remove from a capped Collection:oplog.shard4_oplog"
}
})


5. Write code, after each row of data query to add to a new table:

About: ForEach:

Step 2:create Backup of 2.6 admin.system.users collection. Copy all documents in the
Admin.system.users (page 286) collection to Theadmin.system.new_userscollection:
Db.getsiblingdb ("admin"). System.users.find (). ForEach (function (Userdoc) {
Status = Db.getsiblingdb ("admin"). System.new_users.save (Userdoc);
if (Status.haswriteerror ()) {
Print (Status.writeerror);
}
}
);

Using a similar method, you write a feature that saves log data to another table:
(The table name is the name of the table to be recovered, the action type is ' I '---the data inserted by the new data is taken out and saved to another new table)

Db.shard3_oplog.find ({ns: "Ds.tb_monitor", op: ' I '}). ForEach (function (res_data) {
Obj_doc = res_data["O"]; #取出oplog中插入的JSON对象
Status = Db.tb_monitor.save (Obj_doc); #保存到新表中.
if (Status.haswriteerror ()) {
Print (Status.writeerror);
}
}
)


6. Then put the newly created table, export, and then restore to the cluster for a new table, and then use the 5 method to save to the source table.

./mongodump-d local-c Shard1_oplog-o/opt/backup/0706local/

Mongorestore--port 28000-d temp_local-c new_collection--dir/opt/backup/0706local/local/shard1_oplog.bson


Db.new_collection.find (). ForEach (function (res_data) {
Status = Db.tb_monitor.save (Res_data);
if (Status.haswriteerror ()) {
Print (Status.writeerror);
}
}
)