Testing mongo replication on OSX

Question

In two different terminals, I start up mongod with:

mongod --dbpath 1 --port 27001 --smallfiles --oplogSize 50 --logpath log.1 --replSet test
mongod --dbpath 2 --port 27002 --smallfiles --oplogSize 50 --logpath log.2 --replSet test

(data subdirectories 1 and 2 already created).

I go into a third window and run the mongo shell against one of the mongod instances

mongo --port 27001

In the shell, I setup a configuration as

cfg = { _id: "test", members: [ {_id:0, host: "localhost:27001"}, {_id:1, host:"localhost:27002"} ] }

then run:

use admin
rs.initiate(cfg)

The response from the shell is:

{
"startupStatus" : 4,
"info" : "hotest",
"errmsg" : "all members and seeds must be reachable to initiate set",
"ok" : 0
}

The logfile for instance on port 27001 reports:

Sun Nov 18 18:32:56 [rsStart] trying to contact macbookpro.local:27001
Sun Nov 18 18:32:56 [rsStart] couldn't connect to macbookpro.local:27001: couldn't connect to server macbookpro.local:27001
Sun Nov 18 18:32:56 [rsStart] replSet can't get local.system.replset config from self or any seed (yet)

It appears that the rs.initiate(cfg) cannot connect to the mongod instance on port 27001. But when I run mongod without --replSet the mongo shell connects just fine.

It isn't clear where I am screwing up but I will bet it is obvious.

Answer 1

Found it!

It turns out that when one installs MongoDB on OSX with brew a configuration file is created at /usr/local/etc/mongo.conf. In this file there was the line:

# Only accept local connections bind_ip = 127.0.0.1

Well, if you look at the log the rsStart error was:

Sun Nov 18 18:32:56 [rsStart] trying to contact macbookpro.local:27001
Sun Nov 18 18:32:56 [rsStart] couldn't connect to macbookpro.local:27001: couldn't connect to server macbookpro.local:27001
Sun Nov 18 18:32:56 [rsStart] replSet can't get local.system.replset config from self or any seed (yet)

and if I ping macbookpro.local the IP address returned was 192.168.41.1. 192.168.41.1 <> 127.0.0.1 so mongod was refusing to respond. Removing this line from the configuration allowed be to setup replication successfully.

sheesh. I am such an idiot. Looked for hours at this and kept thinking that the machine name when connected to the network was 127.0.0.1. Let this be a lesson to you---don't get old!