Closed jessereynolds closed 2 years ago
@jessereynolds This is an issue which came out of cloud provider differences between GCP and AWS. Immediate fix is to set lb_ip_mode
to private when deploying to GCP. I've created PR https://github.com/puppetlabs/puppetlabs-pecdm/pull/72 which switches the default to private and will fail the plan if public is set on GCP.
There is a feasible solution for the issue on GCP but it is convoluted, requiring the user to manage their own DNS entries
Describe the Bug
When attempting to spin up a new large architecture on gcp it seems that the plan
peadm::subplans::configure
is either receiving or resolvingcompiler_pool_address
to an empty string.I suspect this is a bug with peadm in the
peadm::certname()
function but not sure. Maybe I just haven't set up my pecdm right?Expected Behavior
A large architecture be spun up in gcp without errors
Steps to Reproduce
Using the following params.json :
Run with
bolt plan run pecdm::provision --params @params.json
Environment
Additional Context
Full output of the run.