Hello,
I have an odd case and hopefully posting here will at least get us to spit ball potential avenues for solutions.
I’m trying to upgrade to version 6.16 from version 5.22 of the Elastic CI stack. We have two identical pipelines, default and test. Both are managed by terraform, via the aws_cloudformation_stack resource. The test pipeline, true to it’s namesake, was upgraded to v6.16 without problems, but the default pipeline had problems when spinning up agents and performing jobs.
The key difference between the two pipelines is within our aws profiles configuration. For default on v5.22, the requirement for the pipeline to work is a buildkite
role, and credential_source = Ec2InstanceMetadata
setting. Such a requirement is not necessary for the test pipeline, (v6.16).
My team was wondering if there’s something under the hood for default pipelines (doubt it, but couldn’t hurt to ask) that would prevent it not running despite configurations being identical.
If you need more info, i’d be happy to provide.