How do you connect to a replicaset from a MongoDB shell?

To the best of my knowledge, the mongo command line client will not accept seeds to forward you to the master node, because you may often want to actually operate on the secondary node rather than being forwarded.

However, once connected to any node in the RS, you can discover the RS topology via rs.config() or db.isMaster(). You could then use this information to reconnect to the primary node. Depending on your shell, you might be able to use mongo --eval "db.isMaster()['primary']" to automatically connect to the master.

You can use the "name/seed1,seed2,..." format:

> conn = new Mongo("myReplicaSet/A:27017,B:27017,C:27017")
> db = conn.getDB("test")

This should give you a connection to whichever node is currently primary and handle failover okay. You can specify one or more seeds and it'll find the rest.

Note that (AFAIK) the shell does not allow you to route reads to secondaries with a replica set connection.

Building on the answer by Chris Heald these two bash aliases let me connect to the master with one command (where db1.test.test is one member of the replica set, acme is the database name, mreppy is my account etc) It will fail of course if db1 is down, but it's still handy.

alias whichprimary='mongo db1.test.test/acme --username mreppy --password testtest --quiet --eval "db.isMaster()['"'primary'"']"'
alias connectprimary='mongo -u mreppy -p testtest `whichprimary`/acme'

The quoting in the eval alias is hard, I used How to escape single-quotes within single-quoted strings? for help :-)

In the shell, you can first use:

mongo --nodb

to open a mongo session without connecting to mongo replicaset

Then, like kristina said, then you should be able to use

conn = new Mongo("myReplicaSet/A:27017,B:27017,C:27017")

to connect to a replicaset.

Or eventually put

conn = new Mongo("myReplicaSet/A:27017,B:27017,C:27017")

in your js file and

mongo --nodb yourcode.js

To connect to a replica set Primary use the mongo shell --host option:

mongo --host replicaSetName/host1[:porthost1],host2[:porthost1],host3[:porthost3],etc

For example:

$ mongo --host rs1/john.local:27019,john.local:27018
MongoDB shell version: v3.4.9
connecting to: mongodb://john.local:27019,john.local:27018/?replicaSet=rs1
2017-10-12T14:13:03.094+0000 I NETWORK  [thread1] Starting new replica set monitor for rs1/john.local:27019,john.local:27018
2017-10-12T14:13:03.096+0000 I NETWORK  [thread1] Successfully connected to john.local:27019 (1 connections now open to john.local:27019 with a 5 second timeout)
2017-10-12T14:13:03.096+0000 I NETWORK  [thread1] Successfully connected to john.local:27018 (1 connections now open to john.local:27018 with a 5 second timeout)
rs1:PRIMARY> db
test
rs1:PRIMARY>

Note: From versions 3.4.2 to 3.4.10 there was a bug (SERVER-28072) which prevented specifying the db after when using --host or --port.

You can use the --host param to specify the replSet name and seed list, then mongo will automatically connect to the current primary host.

example:
mongo --host rs0/1.example.com:27017,2.example.com:27017,3.example.com:27017 [dbname]

All you have to do is to use --host and give it one of your hosts in the replicaset but with the name of the replicaset as a prefix.

For example:

mongo --host my_mongo_server1

will connect to my_mongo_server1, it may just be yet another SECONDARY node.

But:

mongo --host my_repl_set_name/my_mongo_server1

will always connect to the PRIMARY node in the replica set, even if it's not my_mongo_server1.

Why? The answer is "replica set monitor". In the example above, mongo shell would connect to the specified node, start a new replica set monitor for the replica set and will use the specified node just to seed it. From there, the monitor will figure out all nodes in the replica set and will switch the connection to the PRIMARY node.

Hope that helped.

I am using v3.4. Also new to mongodb stuff... Although the help info from "man mongo" suggests to use "--host replicaSet/host:port,host:port" url, it does not work for me. However, I can connect to my replicaSet according to official document, as below:

$ mongo "mongodb://c1m,c2m,c3m/?replicaSet=rs0"
MongoDB shell version v3.4.1
connecting to: mongodb://c1m,c2m,c3m/?replicaSet=rs0
2017-02-08T14:46:43.818+0800 I NETWORK  [main] Starting new replica set monitor for rs0/c1m:27017,c2m:27017,c3m:27017
MongoDB server version: 3.4.1
Server has startup warnings:
2017-02-08T13:31:14.672+0800 I CONTROL  [initandlisten]
2017-02-08T13:31:14.672+0800 I CONTROL  [initandlisten] ** WARNING: Access control is not enabled for the database.
2017-02-08T13:31:14.672+0800 I CONTROL  [initandlisten] **          Read and write access to data and configuration is unrestricted.
2017-02-08T13:31:14.672+0800 I CONTROL  [initandlisten]
rs0:PRIMARY>

So I guess the man page of my mongo is outdated (I am using CentOS 7.3).

The answers above are for Mongo 3.2.

According to Mongo 3.4 documentation, the shell was changed a bit:

In 3.2:
mongo --host host1,host2,host3/myRS myDB
or,
mongo --host host1:27017,host2:27017,host3:27017/myRS myDB

In 3.4:
mongo "mongodb://host1,host2,host3/myDB?replicaSet=myRS"
or,
mongo "mongodb://host1:27017,host2:27017,host3:27017/myDB?replicaSet=myRS"

mongodb://< dbuser >:< dbpassword >@example.com:< port >,example2.com:< port >/< dbname >?replicaSet=setname