terraform-provider-helm: First helm release always succeeds and doesn't wait for all pods running
Terraform, Provider, Kubernetes and Helm Versions
Terraform version: 0.14.4 Provider version: 2.0.2 Kubernetes version: AWS EKS 1.18 Helm version: 3
Affected Resource(s)
- helm_release
Debug Output
https://gist.github.com/FischlerA/7930aff18d68a7b133ff22aadc021517
Steps to Reproduce
terraform apply
Expected Behavior
The helm deployment should fail since the pod that is being deployed is running an image that will always fail. (private image which i can’t share)
Actual Behavior
The first time the helm release is deployed it always succeeds after reaching the timeout (5 min), any further deployments fail as they are supposed to after reaching the timeout (5min).
Community Note
- Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
- If you are interested in working on this issue or have submitted a pull request, please leave a comment
About this issue
- Original URL
- State: open
- Created 3 years ago
- Reactions: 15
- Comments: 23 (5 by maintainers)
@jrhouston
We deployed the chart again by using helm directly with
helm install --waitand the behaviour was as expected: After waiting for five minutes, we’ve got an error-message Error: timed out waiting for the condition.I’m running into this too. I pretty regularly have a successful
terraform apply(everything shows successful and complete) and end up withhelm_releaseresources that show~ status = "failed" -> "deployed"on a second run.I had same experiences when I use helm_release in terraform, if something goes wrong, pod status is stay at “pending” or “Error”, “CreateContainer” or some other unusual status for a little longer time, Helm terraform provider will not wait until pods are running, it will exit and reported completed, However terraform state was update as failed.
I can’t speak for everyone, but we haven’t seen this issue in a while.
Is anyone still encountering this issue on the latest version of the provider? I think we fixed this in https://github.com/hashicorp/terraform-provider-helm/pull/727.
Just tried to reproduce this and see the error in provider version v2.0.2 but now I see the appropriate failure diagnostic in v2.6.0.