MongoDB Index Setup Issues

MongoDB Index Setup Issues


1. The master-Slave library index is not set up synchronously:



MONGODB version 2.6.6, indexing is not synchronized, but is then indexed from the library after the main library is established.
I do not know whether the 3.0 version is the same, the previous use of version 3.0 did not monitor the master and subordinate log. The hole is too big.
5:00 set up to complete the main library is OK.

Official Description:
Background Index Build (page 491) allowed on secondaries. If you initiate a background index builds on a
Primary, the secondaries would replicate the index build in the background.

Automatic rebuild of interrupted index builds after a restart.
–if a standalone or a primary instance terminates during a index build without a clean shutdown, mongod
Now restarts the "index build" when the instance restarts. If the instance shuts down cleanly or if a user kills
The index build, the interrupted index builds does not automatically restart upon the restart of the server.
–if a secondary instance terminates during a index build, the Mongod instance would now restart the interrupted index Bui LD when the instance restarts.
To disable this behavior and use the--noindexbuildretry command-line option.


Indexing from a library is too long, and if the interrupt is indexed from the library, you can restart using the--noindexbuildretry parameter.
I didn't know this parameter at the beginning, so I deleted the previously established index directly in the main library.

It feels good to see people on the internet like this:

1. Stop the secondary server first, restart MongoDB after canceling the--replset parameter and changing the MongoDB port, at which point MongoDB will enter the standalone mode;

2. In standalone mode to run the command Ensureindex index, it is recommended to use the foreground mode of operation;

3. After indexing, shut down secondary server and start in normal way;

Here is the log

Primary

2015-12-17t07:46:53.074+0800 [conn5530345] Index Build (background): 27309700/27314911 99%
。。。

Slave

2015-12-17t07:46:54.985+0800 [REPL Index Builder] Building index in background
2015-12-17t07:46:55.529+0800 [conn2154798] killcursors keyupdates:0 numyields:0 locks (micros) r:73 167ms
2015-12-17t07:46:58.226+0800 [REPL Index Builder] Index build (background): 4700/27314911 0%





2.TTL index deletes data, not a one-time deletion of data beyond the date, but a little bit,

is also the same, always thought that after the index was established, the TTL delete data is a one-time operation, where know a few hours also did not delete the completion, resulting in the database
The pressure is always high.

Later had to change the name, and then set up a new table after the data inserted.
Db. Weblog.renamecollection ("Weblog_bak");
Db. Weblog.ensureindex ({mallid:1,mac:1});
Db. Weblog.ensureindex ({createtime:-1},{expireafterseconds:3600*24*3*30});

var cursor=db. Weblog_bak.find ({_id:{$gte: 24600000}});
while (Cursor.hasnext ()) {
Nc=cursor.next ();
Db. Weblog.insert (NC);
};