A simple stack (example below of a single server) with a single SoftwareConfig resource and single SoftwareDeployment deploys correctly if the template is handed to heat as part of the stack-create call. This appears to be because the resources being added during an update don't make it into the raw_template database table and aren't available to the engine to validate the waitcondition signal.
Log messages below are from a recent single node devstack:
If the SoftwareConfig and SoftwareDeployment are added after the instance is created, as a stack-update, the SoftwareDeployment functions correctly but in /var/lib/upstart/os-collect-config.log I see:
[2014-06-10 02:09:49,050] (heat-config) [DEBUG] Signalling to http://10.24.0.4:8000/v1/signal/arn%3Aopenstack%3Aheat%3A%3A15a41777be7244daacd242a02b862814%3Astacks%2Ftest1%2Fb2585565-e5e4-418b-a83a-8f75a7a642e5%2Fresources%2Ftest-demo-test_heat_sw_deployment?Timestamp=2014-06-10T02%3A09%3A30Z&SignatureMethod=HmacSHA256&AWSAccessKeyId=63902346e9fb4060b2a4174a5c016566&SignatureVersion=2&Signature=i0OqZrVExlfiKcUIqL7qizzR3zn6boavAcmWsg474RE%3D with {"deploy_stdout": "Testing me some heat software config\n", "deploy_stderr": "", "deploy_status_code": 0}
[2014-06-10 02:09:49,769] (heat-config) [DEBUG] Response <Response [400]>
The 400 Bad Request can be seen in the heat logs as "The Resource (test-demo-test_heat_sw_deployment) could not be found in Stack test1". I haven't confirmed this but it seems likely it's being raised in engine/service.py:resource_signal.
template-show does not show the SoftwareConfig or SoftwareDeployment (raw_template doesn't contain the resources either). This may be related to https://bugs.launchpad.net/heat/+bug/1291905 which also seems to revolve around dependencies, but since I'm not manually creating the WaitCondition in this case, the workaround doesn't work.
Template (the networking components aren't relevant, but make things easier to test):
heat_template_version: '2013-05-23'
outputs:
test-demo-FloatingIPaddress:
description: Floating IP assigned
value:
get_attr: [test-demo-FloatingIP, floating_ip_address]
resources:
test-demo:
properties:
flavor: m1.small
image: Ubuntu 13.10 Heat SW config
key_name: devstack
networks:
- port: {get_resource: test-demo-port}
user_data_format: SOFTWARE_CONFIG
type: OS::Nova::Server
test-demo-FloatingIP:
properties: {floating_network_id: 24f9b76f-f292-404d-b359-ce9b209eda03}
type: OS::Neutron::FloatingIP
test-demo-FloatingIpAssoc:
properties:
floatingip_id: {get_resource: test-demo-FloatingIP}
port_id: {get_resource: test-demo-port}
type: OS::Neutron::FloatingIPAssociation
test-demo-port:
properties:
fixed_ips:
- subnet_id: edcd39a2-dcc0-4823-a736-4ca3dd90fdd2
network_id: 02ab2bb0-b165-442d-9b14-6ed20555f846
security_groups:
- default
type: OS::Neutron::Port
Heat SW config (add this as part of stack-update to see the error):
test-demo-test_heat_sw:
properties:
config: '#!/bin/bash
echo "Testing me some heat software config" | tee /tmp/test-heat.txt
'
group: script
inputs:
- {default: None, name: previous}
type: OS::Heat::SoftwareConfig
test-demo-test_heat_sw_deployment:
properties:
config: {get_resource: test-demo-test_heat_sw}
server: {get_resource: test-demo}
type: OS::Heat::SoftwareDeployment
If possible, can you paste the errors from h-eng and h-api-cfn too?