Heat is one of the widest-reaching components within the OpenStack infrastructure to troubleshoot. This is because each template that is built is different and has different dependencies among resources created in the stack that is launched. Further, it is able to depend on or create almost any resource within OpenStack. As with the other components, the best starting place is with the Heat logs in /var/log/heat. This will give you a good indication of where things might not be going correctly.

If a stack is launched and does not successfully complete, but you do not see any errors in the Heat logs, then the issue may be in the instances. Heat has callbacks from the running instances that must work for the orchestration of data and the ordering of instance creation within a stack.

If you review the section in this chapter on the metadata server, you will notice that the post-boot configuration that is run on an instance is delivered by the metadata service. When cloud-init fires and receives post-boot configuration, it will log what it is doing in /var/log/cloud-init.log. There are callbacks into Heat that are included in your cloud-init scripts that must alert Heat that it is time for the next instance in a stack to be created if it is dependent on a previous instance in the stack completing its provisioning. The service that is called back into is the heat-cfn service. If you see failures calling back into the heat-cfn service, you will need to get this URL and verify and troubleshoot this connectivity.