[rabbitmq-discuss] Clustering issue: Error: unable to connect to node rabbit at CNTD3BWEB87O: nodedown

Emile Joubert emile at rabbitmq.com
Wed Jul 11 10:46:52 BST 2012


On 10/07/12 17:33, Wood, Eric (LNG-RDU) wrote:
> Help.

Your earlier emails had
- cookie hash: VWZk1leRufUBRYr1dwMuJg==
but more recently you report
- cookie hash: 0xT5raNnyX73mPb3lJZWuQ==

If the cookie hash reported by rabbitmqctl differs from the hash that
appears in the startup banner then you will be unable to manage that
node. The startup banner contains an ASCII-art rabbit, platform version
numbers and some diagnostic information, such as the location of the
active config file, logfiles, etc. Amongst this diagnostic information
is the cookie hash. Compare this with the information reported by
rabbitmqctl - you will probably find they are different, which means the
cookie file was not copied over correctly. Make sure you follow the
instructions about Erlang cookies on this page:

> I have tried this on several servers and get the same results.  What
> steps are required to execute when I copy the cookie file from one
> server to the next?  I copy it after both installations have been
> completed and both nodes are up and running as "detached" nodes.

You will need to restart the nodes after copying the cookie files. Their
contents is only read at startup, so modifications will not affect a
running broker.

You might be copying the cookie file to the wrong location. If you are
running the broker as a non-system user then the file should be copied
to %HOMEDRIVE%\%HOMEPATH%, or what the Windows API call
GetWindowsDirectory() would return. If running the broker as a service,
as a system user then use %windir% (typically C:\WINDOWS).


More information about the rabbitmq-discuss mailing list