I have installed RabbitMQ on a Debian Linux Squeeze machine, and I would like it to only listen to the localhost interface. I have added
RABBITMQ_NODE_IP_ADDRESS=127.0.0.1
to my /etc/rabbitmq/rabbitmq.conf
file, and that makes it bind to only the localhost interface when listening on the amqp
port (5672). However, it still binds to all interfaces when listening on ports epmd (4369) and 43380:
# lsof -n -a -i -urabbitmq
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
epmd 7353 rabbitmq 3u IPv4 1177662 0t0 TCP *:epmd (LISTEN)
epmd 7353 rabbitmq 5u IPv4 1177714 0t0 TCP 127.0.0.1:epmd->127.0.0.1:50877 (ESTABLISHED)
beam.smp 7365 rabbitmq 10u IPv4 1177711 0t0 TCP *:43380 (LISTEN)
beam.smp 7365 rabbitmq 11u IPv4 1177713 0t0 TCP 127.0.0.1:50877->127.0.0.1:epmd (ESTABLISHED)
beam.smp 7365 rabbitmq 19u IPv4 1177728 0t0 TCP 127.0.0.1:amqp (LISTEN)
How do I prevent this? Do I have to set up iptables, or are there additional RabbitMQ configuration options that will make it do what I want?
Putting the following in
/etc/rabbitmq/rabbitmq-env.conf
will make RabbitMQ and epmd listen on only localhost:It takes a bit more work to configure Erlang to only use localhost for the higher numbered port (which is used for clustering nodes as far as I can tell). If you don't care about clustering and just want Rabbit to be run fully locally then you can pass Erlang a kernel option for it to only use the loopback interface.
To do so, create a new file in
/etc/rabbitmq/
- I'll call itrabbit.config
. In this file we'll put the Erlang option that we need to load on run time.If you're using the management plugin and also want to limit that to localhost, you'll need to configure its ports separately, making the rabbit.config include this:
[ {rabbitmq_management, [ {listener, [{port, 15672}, {ip, "127.0.0.1"}]} ]}, {kernel, [ {inet_dist_use_interface,{127,0,0,1}} ]} ].
(Note RabbitMQ leaves epmd running when it shuts down, so if you want to block off Erlang's clustering port, you will need to restart epmd separately from Rabbit.)
Next we need to have RabbitMQ load this at startup. Open up
/etc/rabbitmq/rabbitmq.conf
again and put the following at the top:This loads that config file when the rabbit server is started and will pass the options to Erlang.
You should now have all Erlang/RabbitMQ processes listening only on localhost! This can be checked with
netstat -ntlap
EDIT : In older versions of RabbitMQ, the configuration file is :
/etc/rabbitmq/rabbitmq.conf
. However, this file has been replaced by therabbit-env.conf
file.To make RabbitMQ listen on localhost / bind only to localhost:
3 Different ways (all equivalent):
Put NODE_IP_ADDRESS=127.0.0.1 in the environment variables file (See http://www.rabbitmq.com/configure.html#define-environment-variables)
Put tcp_listeners and ssl_listeners properties in config file: The configuration entries tcp_listeners and ssl_listeners govern the interfaces that RabbitMQ listens on. An entry for just listening on localhost would be e.g., {tcp_listeners, [{'127.0.0.1', 5672}]} (Syntax might not be correct, check it) http://www.rabbitmq.com/configure.html#config-file
export the env. variable in the startup script (/etc/init.d/rabbitmq-server) export RABBITMQ_NODE_IP_ADDRESS=127.0.0.1
The latter worked for me.
EPMD:
The Epmd program makes distributed parts of Erlang runtime work. If you are building a multi-machine cluster you need to leave them accessible to other nodes and certainly localhost. But it has built-in protection via cookie file.
It hardly ever requires any attention. Just keep in mind that erlang programs (including rabbitmqctl, for example) need to access that port to contact other erlang programs.
But, if you are dealing with financial data or health records, protecting epmd may be a good idea. Default port epmd uses is 4369, other programs connect to it via tcp.
See also: http://www.erlang.org/doc/man/epmd.html#environment_variables
If you need to secure RabbitMQ any further,
Disable the built-in guest account http://www.rabbitmq.com/admin-guide.html#default-state
Consider using SSL and authenticating using the certificate chain
I got these answers from the RabbitMQ community IRC channel.
Would like to thank them.
http://dev.rabbitmq.com/irclog/index.php?date=2011-06-14
Hope the above saves some time for you (it took me 6 hours to find an answer).
If you specify environment variables in the rabbitmq.conf file you have to drop the RABBITMQ_ prefix, so try:
NODE_IP_ADDRESS=127.0.0.1
AFAIK you can't really configure epmd interfaces. You can only set up the epmd port: http://www.erlang.org/faq/how_do_i.html#id55132
With the recent versions of rabbitmq - this is sufficient to bind all ports to localhost (ipv4) only:
/etc/rabbitmq/rabbitmq-env.conf:
/etc/rabbitmq/rabbitmq.conf
And restart rabbitmq.
After this, hostname part of
RABBITMQ_NODENAME
(defaults torabbit@node-hostname
) must resolve to127.0.0.1
forrabbitmqctl
to work. If so -rabbitmqctl status
andrabbitmqctl list_queues
should work and show previously existed data, and you don't need further reading.If this is not so, there are two ways:
node-name entry might be added to
/etc/hosts
:127.0.0.1 node-hostname
, but this might reflect on the other software (e.g. node-hostname is required to resolve to public/other ip address).
hostname part might be set to
localhost
in/etc/rabbitmq/rabbitmq-env.conf
:export RABBITMQ_NODENAME=rabbit@localhost
, but before using this way and restarting rabbitmq - decide if you need existing rabbitmq data. If no data is needed - simple restart of rabbitmq should fix
rabbitmqctl
. Otherwise, here is the script that moves existing rabbitmq data from the old name to the new, it implies restart of rabbitmq (make backup of the data directory before running the script: stop rabbitmq and copy out/var/lib/rabbitmq
directory):More info:
https://www.rabbitmq.com/networking.html
https://github.com/rabbitmq/rabbitmq-server/blob/main/deps/rabbit/priv/schema/rabbit.schema