[Rdo-list] RabbitMQ issue when starting Nova service on compute node

[Dave Neary]

Hi,

For complete closure: the Swift issue was because there was an
unfinished install already, when I wiped everything clean and started
again, I got to a different error.

That error was mongod not starting; that was a known issue, solved in a
newer version of Packstack than what was in the RDO repo. The workaround
I used was to disable Ceilometer for the installation, as that was the
only think pulling in MongoDB.

Thanks,
Dave.

On 12/04/2014 02:25 PM, Dave Neary wrote:
> Hi John,
> 
> On 12/04/2014 09:34 AM, John Eckersberg wrote:
>> Dave Neary <dneary at redhat.com> writes:
>>> The fixes would be straightforward: use a non-guest AMQP user and
>>> password, or enable remote connection for the RabbitMQ guest user. But I
>>> can't figure out how to do either of those - I don't think that
>>>
>>>     CONFIG_AMQP_AUTH_USER=amqp_user
>>>     CONFIG_AMQP_AUTH_PASSWORD=PW_PLACEHOLDER
>>>
>>> in the answer file are what I'm looking for, I don't see any way to
>>> update the RabbitMQ config file in amqp.pp
> 
> <snip>
> 
>> If you want to just turn the guest account back on, you could update
>> wherever the top-level rabbitmq puppet class gets called in packstack
>> and set something like...
>>
>> config_variables => {'loopback_users' => '[]'}
> 
> Thanks John! Turned out the main issue was that I was installing this in
> OpenStack instances, and forgot about the security group rules. Since
> those get enforced in the host, not in the guest, it was invisible to me
> - iptables looked fine.
> 
> I did add that line to rabbitmq.conf, with Dan Radez's help, in
> /usr/lib/python2.7/site-packages/packstack/puppet/templates/amqp.pp in
> the rabbitmp class. I still have not had a successful run, but I've been
> hitting a different issue each time.
> 
> My latest issue was due to using floating IP addresses for the hosts -
> mongodb would not bind to that address - so I had to switch to the
> internal IP addresses (unfortunately, as I understand those will not
> stay the same over time).
> 
> After resolving that, I have now hit an issue with Swift ring failing to
> rebalance. I have no idea what that means or how to fix it, the
> information I have suggests that nuking from orbit and restarting is the
> best approach.
> 
> Thanks,
> Dave.
> 
> 

-- 
Dave Neary - NFV/SDN Community Strategy
Open Source and Standards, Red Hat - http://community.redhat.com
Ph: +1-978-399-2182 / Cell: +1-978-799-3338