0

I have a new server on which i have to copy the old version of the MongoDB already configured. On the old server(3.0.4), there are two shards: one shard has 350GB, and the other has 300GB of data.

I followed the following steps:

  1. Installed MongoDB 3.0.15 on new server
  2. Copied the data from two shards to the new server
  3. then ran the following commands

    mongod    --configsvr --dbpath C:\mongoConfigData/configdb1 --port 27019
    mongod    --configsvr --dbpath C:\mongoConfigData/configdb2 --port 27020
    mongod    --configsvr --dbpath C:\mongoConfigData/configdb3 --port 27021
    
    mongos    --configdb 127.0.0.1:27019,127.0.0.1:27020,127.0.0.1:27021 --port 27017
    
    mongod    --dbpath F:\data --storageEngine wiredTiger --wiredTigerJournalCompressor zlib --wiredTigerCollectionBlockCompressor zlib --port 27010
    
    mongod    --dbpath D:\data --storageEngine wiredTiger --wiredTigerJournalCompressor zlib --wiredTigerCollectionBlockCompressor zlib --port 27011
    
  4. Added the shard and the shard collection details

    sh.addShard("localhost:27010") 
    sh.addShard("localhost:27011") 
    sh.shardCollection("log", {"id" : 1, "eventId": 1})
    

Shard distribution shows both the shards but the total docs and the collection size does not match the one in the initial server. Further the content in the mongodb is not the complete data Why is it so? What am i doing wrong?

  • ..Because sharding information is located at config servers and if you didn't have sharding enabled at old servers (and you didn't transfer config servers from old servers) there was no information what was sharded and where. Right procedure would be "transfer" data from old servers to new cluster using export/import or copyDatabase... – JJussi Feb 6 '18 at 5:58
1

You are not copying the contents of the config servers and effectively creating a new sharded deployment rather than restoring a backup of an existing one. Config servers include essential information about where data lives in a sharded deployment, so you must backup & restore the config server data along with the shards.

Note: there have been changes to sharded cluster configuration in successive releases, so make sure you are using documentation for the correct MongoDB release series. The links below are specific to MongoDB 3.0.

For supported backup approaches see the MongoDB 3.0 Backup & Restore a Sharded Cluster tutorials.

sh.addShard("localhost:27010")

sh.addShard("localhost:27011")

If you've restored a sharded cluster from a backup, the shards will already be configured. If host names have changed from the original deployment, you will need to update the config.shards information via a mongos rather than adding new shards (and then restart the mongod instances in your sharded cluster to ensure the new names are used). See the MongoDB 3.0 Restore a Sharded Cluster tutorial for more specific details.

sh.shardCollection("log", {"id" : 1, "eventId": 1})

If a sharded cluster is already configured, sharding an existing sharded collection will return an error. Since you ran this command starting from fresh config server data, the sh.shardCollection() command will be unaware of how existing data is distributed on the shards and assume that the primary shard for a database with an unsharded collection has the full range of data.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.