[openstack-dev] HEAT - AutoScaling stack waits for resource endlessly and then fails to start.
shardy at redhat.com
Wed Apr 3 08:32:36 UTC 2013
On Wed, Apr 03, 2013 at 01:27:26PM +0530, Skanda Nag wrote:
> I'm able to ping www.google.com or run yum -y makecahe once i'm inside the
> instance, which means my instance has net connectivity.
> My instance has private ip only. It doesn't have a floating ip.
> But still, its timing out waiting for resource completion. Any pointers?
Can you post your /var/log/heat-provision.log again? If you have
connectivity then it can't be getting stuck at makecache as in your previous
post, so we need to see the subsequent log output.
One thing to check is that you have connectivity to the heat-api-cfn service
(which runs on port 8000).
The instance notifies heat of WaitCondition events via the heat-api-cfn
service, so in a setup with nova-network, you have to add an additional
iptables rule allowing access from the instance to the heat-api-cfn via the
bridge device. I'm not sure how this works when using Quantum?
You also need to allow access to the heat-api-cloudwatch service (port 8003)
so the LB instance can report stats to heat.
One test you can do is to install nmap inside the instance and port-scan the
bridge device (but again, I'm not sure you have a bridge device with
Quantum, so scan whatever IP the services will be accessible via).
Also ensure /etc/heat/heat-engine.conf has the IP which the services are
listening on (see GettingStarted linked above, the "Update heat engine
configuration file" section), or the instance won't be able to notify
WaitCondition completion and the stack create will fail with a timeout.
More information about the OpenStack-dev