Given:
- There are no deployments in the production or staging slots for my Azure Cloud Service (web role)
- WinServer 2012 R2, Azure SDK 2.1, Octopus Server 1.6.3.1723, Octopus TeamCity plugin 1.5.3, TeamCity 8.0.4 (also can repro w/ TC 8.0.1).
When:
- When I trigger an Octopus release from TeamCity to deploy my cloud service into the empty prod slot
Then:
- I expect that the teamcity Octo-deploy step to not continue until the production deployment instances are all up and running.
What actually happens is that the teamcity step completes with “successful” status, but the instances are still not yet in “Running” state in the Azure portal. The octo step takes only 33 seconds to complete, and then my TeamCity step to run the tests against the deployed Cloud Service starts, but they fail because the Cloud Service instances are still being spun up.
This actually works fine when I’m upgrading the same cloud service, i.e., when go through the same steps, with the only difference being that there is already a running deployment in the prod slot of the Cloud Service. In that scenario the Octo task takes ~3 minutes. But when I do it with the empty prod slot, the Octo step takes only 33 seconds. Is there a way to make the “new deployment” scenario block until the instances are alive, the same way the upgrade scenario seems to work?
I’m attaching the Octo logs. One for when I’m deploying into empty prod slot, and one for when I’m upgrading an existing prod deployment.
octo-deploying-into-empty-production-slot-cloudservice-log-2013-11-06.txt (3 KB)