rabbitmq-server fails to start after hostname has changed for first time

I am using django-celery for my django project. Last day I have changed my computer's hostname (I am using Ubuntu 12.04, edited file '/etc/hostname'), and after next restart django-celery was failing with error

Consumer: Connection Error: [Errno 111] Connection refused. Trying again in 4 seconds...

After some research on this error I could find that, changing my host name caused this error from here. My rabbitmq startup log shows

file: /var/log/rabbitmq/startup_log

Activating RabbitMQ plugins ...

********************************************************************************
********************************************************************************

0 plugins activated:

ERROR: epmd error for host "jinesh": nxdomain (non-existing domain)

My startup_err file is empty.

when I run

root@jinesh:/home/jinesh# rabbitmqctl list_users
Listing users ...
Error: unable to connect to node rabbit@jinesh: nodedown

DIAGNOSTICS
===========

nodes in question: [rabbit@jinesh]

hosts, their running nodes and ports:
- unable to connect to epmd on jinesh: nxdomain

current node details:
- node name: rabbitmqctl4956@jinesh
- home dir: /var/lib/rabbitmq
- cookie hash: RGhmB2JR1LbZ57j7xWWTxg==

I hope changing the nodename may fix this issue. But I couldn't found a way to do this. Anyone have idea about how solve this issue?

update

while changing hostname you have to change both /etc/hostname and /etc/hosts files.

I reinstalled rabbitmq and solved this issue, Will answer this question.

Remove the old installation of RabbitMQ to fix this problem. Here are steps to reinstall RabbitMQ. These commands are run as the root user:

1. Stop RabbitMQ: rabbitmqctl stop

2. Change /etc/hosts

3. Change /etc/hostname

4. Uninstall old RabbitMQ: dpkg -P rabbitmq-server

5. Remove RabbitMQ’s database: rm -rf /var/lib/rabbitmq

6. Find erlang’s process that is running rabbit: ps ax | grep rabbit

7. Kill the listed process

8. Reinstall RabbitMQ: apt-get install rabbitmq-server

I wrote about these steps on my blog.

REVISION

I moved my blog to a new website.

Thanks to Richard H Fung.

> His steps helped me to solve this issue.

> But I did not have to re-install the rabbitmq.

When I opened my /etc/hosts file I found that IP assigned to my hostname is different than the actual ip(192.168.1.200 [static]).

#/etc/hosts  
127.0.0.1       localhost  
192.168.1.115   HOSTNAME

so I just changed IP address to 192.168.1.200 in my /etc/hosts file and it worked fine.

Richard's answer is good, but you might lose some information in the rabbitmq queues. The following is a possible way to preserve the previous setup of rabbitmq with the new host name.

A Short Answer:

If you want to keep the new host name change, then you can create a rabbitmq-env.conf files in /etc/rabbitmq that references the old host name and all should be good. The following is what should be in the rabbitmq-env.conf file:

NODENAME=rabbit@OLDHOSTNAME

After adding the config file, restart the rabbitmq server then you should be good. (e.g. service rabbitmq-server restart (might need a sudo with ubuntu))

For more details you can read from the rabbitmq website: https://www.rabbitmq.com/man/rabbitmq-env.conf.5.man.html

More Details About the Answer:

I had a similar hostname issues using CentOS... The root issue was when rabbit installs, it references the current host name at the time of install. Since this is a rabbitmq thing, the solution should work for other linux flavors. If you want to see the full answer, you can see it at: https://stackoverflow.com/questions/29501234/rabbtimqadmin-could-not-connect-errno-2-name-or-service-not-known/29502570#29502570

You could solve this problem by either deleting the erlang mnesia database associated with rabbitmq or reinstalling rabbitmq.

I got hint for first solution from rabbitmq mailing list. Excerpt from answer, The Erlang Mnesia database is host specific (because it is a distributed DB). The simplest way to get you fixed is to clear out the database dir.

The second method is the easiest way (not recommend though). To uninstall do

dpkg -P rabbitmq-server

You can refer this link if you would like to know more about installing/removing debian packages.

On Windows, the issue I was facing was due to the McAfee firewall. The exact error was:

> epmd error for host "<HOSTNAME>": address (cannot connect to host/port)

It started working as soon as I disabled the firewall

Remove RabbitMQ database: rm -rf /var/lib/rabbitmq/*

This action solves the problem. I believe somewhere in dumps stored a record with correspondence of nodes host names and ips. This fact causes contradiction if some hosts changed unexpectedly.

My solution was to check and correct the $HOSTNAME and /etc/hostname. It turned out that my router was a little confused and gave me a wrong hostname domain. After restarting this one, I called export HOSTNAME=the.correct.hostname and my rabbit runs without restarting the server os.

In my case I did not have a following entry in /etc/hosts:

127.0.0.1 <hostname>

where <hostname> is the hostname of my machine (as given by the hostname command). After adding that line, RabbitMQ started successfully. I did not have any earlier installations of RabbitMQ.

To solve this problem, I changed my /etc/hostname file with the new hostname then restarted the machine. After that, I ran rm -rf /var/lib/rabbitmq/mnesia/*

Then restarted the service : sudo service rabbitmq-server restart

At this point, it worked for me. If it still don't work, modify the file /etc/rabbitmq/rabbitmq-env.conf by adding those informations :

NODENAME=rabbit@YOUR_NEW_HOSTNAME
NODE_IP_ADDRESS=127.0.0.1
NODE_PORT=5672

Then restart the service : sudo service rabbitmq-server restart

Hope it helps.

When on Windows (sorry about that - I understand the topic is for Ubuntu) just reinstall the RabbitMQ service:

rabbitmq-service.bat remove
rabbitmq-service.bat install

The script is located at C:\Program Files\RabbitMQ Server\rabbitmq_server-<version>\sbin folder.

Or may be just changing the value of HKEY_LOCAL_MACHINE\SOFTWARE\Ericsson\Erlang\ErlSrv\1.1\RabbitMQ\@SName is enough.

Clues: https://github.com/rabbitmq/rabbitmq-server/issues/620

I had the same problem and doing the steps Richard H Fung provided didn't helped me. I also had to uninstall these packages: erlang erlang-epmd rabbitmq-server

zypper remove erlang erlang-epmd rabbitmq-server
rm -rf /var/lib/rabbitmq/*
zypper install erlang erlang-epmd rabbitmq-server
systemctl enable rabbitmq-server.service
systemctl start rabbitmq-server.service

If you have another OS than openSUSE just use your package manager to remove these packages (e.g. apt-get)

The only solution which work for me: install erlang & rabbitmq from deb, so:

First remove:

apt-get purge rabbitmq-server
apt-get purge erlang
apt-get autoremove
reboot

After install wget:

sudo apt-get -y install socat logrotate init-system-helpers adduser
sudo apt-get -y install wget

Install erlang:

wget https://packages.erlang-solutions.com/erlang-solutions_1.0_all.deb
sudo dpkg -i erlang-solutions_1.0_all.deb
sudo apt-get update
sudo apt-get install erlang

Install rabbitmq:

sudo apt-get update
wget https://github.com/rabbitmq/rabbitmq-server/releases/download/v3.7.17/rabbitmq-server_3.7.17-1_all.deb
sudo dpkg -i rabbitmq-server_3.7.17-1_all.deb
rm rabbitmq-server_3.7.17-1_all.deb

Check qpidd daemon, it could be also already hogging the port that rabbitmq needs.

sudo netstat -lnp