Popular posts

@vinaym thanks for reply , You are correct but I missed the metadata from config servers .. ps I wrote a command as the say in mongo community reload config data "flushRouterConfigon" but this command i clear the database.

so I have a shards with data and Mongos can't read my database. hope you understand me clearly

@ahmedash95 are shards connected to mongos...and not able to see the database OR just no shards connected to cluster and no database is available.?

@vinaym the database are not available .. and I can't add shards to mongos because It says can't add shard to because database "xx" exists.

We might need to backup the database on each shard. empty(drop) the database and restore one by one from mongos to the cluster after adding the 3 shards.

What is the database size on each shard?. how may databases? verison? engine?.

@vinaym the first shard has 50 GB , the second one about 40 GB , the third one about 30 GB .. I tried to import shards to culster but he just add 1 and refuse others because the same database name .. i don't know how to till him they are use the same database .. it's just a 1 database .. wierdtrigger engine .. mongo v 2.4

WiredTiger is not with 2.4 version, it works from 3.0 verison. Make sure all mongos, mongo configservers and shards are of same version. Steps to recover is:

1. Backup mongodump all the data on all 3 shards.
2. Drop the databases on all 3 shards.(Instead can start mongo with new datadir, So you dont have to drop database, you still have data safe )
3. Add the 3 shards to the cluster now(there should be no error as its empty).
4. Then Restore the mongodump's taken from 3 shards one by one.
5. This will recover all your data.

NOTE: You can test this in some staging setup with less data and verify steps.

@vinaym Thanks , actually that's what i did and I restored the data from 3 shards ,, but unfortunately I missed about 100 Million document from one collection only. don't know why dumping didn't bring them .. any idea about this ?